Watch, Follow, &
Connect with Us

Please visit our new home
community.embarcadero.com.


Welcome, Guest
Guest Settings
Help

Thread: i m totally disappointed about emb :(



Permlink Replies: 385 - Last Post: Jun 7, 2017 1:14 PM Last Post By: Rudy Velthuis (... Threads: [ Previous | Next ]
loki loki

Posts: 718
Registered: 7/1/02
i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 20, 2017 4:48 AM
I just discover an already know bug in firemonkey android.

this bug is very important (and blocking) because it's create memory
leak every time you use simple java type like string, integer or even
boolean. to gave you an exemple it's like everytime you use a String, an
integer or even a boolean their are never (never) release from the memory

https://quality.embarcadero.com/browse/RSP-14187

it's even worse on android because java permit to have maximun 65000
reference to such type, so your app will crash much faster than when
it's will be out of memory

why i m disappointed it's because this bug was discover in seattle, and
it's still not corrected in berlin (update 2). it's seam this bug is not
the priority of emb because this bug not appear in any demo. in fact you
need to use your app a quite long time dependly of what the app is doing
but if you do http request it's can be just several minutes.

So emb is selling "publisher", look good on the paper but in the
reality, after 6 months of development you see bug like this that
suddenly stop all your developments (you can not see such bug at the
first view)

i don't know who is responsible of the priority at emb, but i think they
work only on new features (like linux) because new feature make sell
(and good publisher presentation) but they make bug correction at lowest
priority and the result is what we have : tones of features but a very
poor and buggy code, often unusable ! just look a question made few days
ago here Has anyone managed to complete a cross-platform app ? resume
quite well the atmosphere

i don't think it's the fault of the emb dev team, but mostly the fault
of the emb management team. these guy just detroy the reputation of
delphi :(

and to finish this bug show how stupid is ARC, without arc this bug will
never be here

steven chesser

Posts: 399
Registered: 4/13/09
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 20, 2017 7:50 AM   in response to: loki loki in response to: loki loki
loki loki wrote:
I just discover an already know bug in firemonkey android.

this bug is very important (and blocking) because it's create memory
leak every time you use simple java type like string, integer or even
boolean. to gave you an exemple it's like everytime you use a String, an
integer or even a boolean their are never (never) release from the memory

https://quality.embarcadero.com/browse/RSP-14187

it's even worse on android because java permit to have maximun 65000
reference to such type, so your app will crash much faster than when
it's will be out of memory

why i m disappointed it's because this bug was discover in seattle, and
it's still not corrected in berlin (update 2). it's seam this bug is not
the priority of emb because this bug not appear in any demo. in fact you
need to use your app a quite long time dependly of what the app is doing
but if you do http request it's can be just several minutes.

So emb is selling "publisher", look good on the paper but in the
reality, after 6 months of development you see bug like this that
suddenly stop all your developments (you can not see such bug at the
first view)

i don't know who is responsible of the priority at emb, but i think they
work only on new features (like linux) because new feature make sell
(and good publisher presentation) but they make bug correction at lowest
priority and the result is what we have : tones of features but a very
poor and buggy code, often unusable ! just look a question made few days
ago here Has anyone managed to complete a cross-platform app ? resume
quite well the atmosphere

i don't think it's the fault of the emb dev team, but mostly the fault
of the emb management team. these guy just detroy the reputation of
delphi :(

and to finish this bug show how stupid is ARC, without arc this bug will
never be here


Another old and nasty bug @ https://quality.embarcadero.com/browse/RSP-15775

Prevents companies like TMS from supporting Android ...
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 20, 2017 1:14 PM   in response to: loki loki in response to: loki loki
Am 20.03.2017 um 12:48 schrieb loki loki:
I just discover an already know bug in firemonkey android.

this bug is very important (and blocking) because it's create memory
leak every time you use simple java type like string, integer or even
boolean. to gave you an exemple it's like everytime you use a String, an
integer or even a boolean their are never (never) release from the memory

https://quality.embarcadero.com/browse/RSP-14187

it's even worse on android because java permit to have maximun 65000
reference to such type, so your app will crash much faster than when
it's will be out of memory

why i m disappointed it's because this bug was discover in seattle, and
it's still not corrected in berlin (update 2). it's seam this bug is not
the priority of emb because this bug not appear in any demo. in fact you
need to use your app a quite long time dependly of what the app is doing
but if you do http request it's can be just several minutes.

So emb is selling "publisher", look good on the paper but in the
reality, after 6 months of development you see bug like this that
suddenly stop all your developments (you can not see such bug at the
first view)

i don't know who is responsible of the priority at emb


I'm quite sure you know the names of the product managers. Don't you?
They should at least in parts be responsible for sheduling work.

I suggest you write them about this issue to raise awarenes.

Greetings

Markus
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 20, 2017 3:07 PM   in response to: Markus Humm in response to: Markus Humm
I'm quite sure you know the names of the product managers. Don't you?

no seriously i don't know them except maybe marco cantu who say exactly
that correction of bug don't make new sell (because customer hope to
receive theses corrections for free) but new features make new sell so
they focus on new feature ... i already say to all of them (in beta
testing program forum) what i think about it, seam they decide to not
re-invite me to the beta tester program ;)
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 21, 2017 12:54 AM   in response to: loki loki in response to: loki loki
loki loki wrote:
I'm quite sure you know the names of the product managers. Don't you?

no seriously i don't know them except maybe marco cantu who say exactly
that correction of bug don't make new sell (because customer hope to
receive theses corrections for free) but new features make new sell so
they focus on new feature ... i already say to all of them (in beta
testing program forum) what i think about it, seam they decide to not
re-invite me to the beta tester program ;)

Your problem is that you don't focus on your issues which are serious but
focus on ranting instead. So people are not inclined to listen.

Again, there is a bug in Delphi, why on Earth you need to rant about ARC
again. It is not ARC that fails, but darn bug in compiler (I think, here I didn't
looked up closely)...

You can have showstopper bugs in compiler regardless of they being ARC
or non-ARC.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 21, 2017 1:50 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Your problem is that you don't focus on your issues which are serious but
focus on ranting instead. So people are not inclined to listen.

yes it's sure that verity always hard to listen ! but remember even if
lies don't hurt at first view, it's will destroy you at the end, and
verity hurt but make you strong at the end ! after making strong childs
of monkey with donkey don't know if it's really possible ...
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 21, 2017 2:03 AM   in response to: loki loki in response to: loki loki
loki loki wrote:
Your problem is that you don't focus on your issues which are serious but
focus on ranting instead. So people are not inclined to listen.

yes it's sure that verity always hard to listen ! but remember even if
lies don't hurt at first view, it's will destroy you at the end, and
verity hurt but make you strong at the end ! after making strong childs
of monkey with donkey don't know if it's really possible ...

????

I have no idea what you mean... it does not matter... you are stubborn enough
and you don't want to listen to people trying to help you.

I am really sorry because of that. You seem to find new issue every few lines
of code you write and it would benefit all if you could keep focus when
reporting and/or asking about those. That includes clear description of
your problem with proper MCVE regardless of where you post about it.

And no useless ranting about unrelated stuff.

ARC is here, that part will not change. You don't have to repeat how ARC
sucks every time you have some issue.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 21, 2017 7:19 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
I am really sorry because of that. You seem to find new issue every few lines
of code you write and it would benefit all if you could keep focus when
reporting and/or asking about those. That includes clear description of
your problem with proper MCVE regardless of where you post about it.

yeaaa i have nothing else to do than reporting thousands or error, and
making clear sample project ... maybe it's emb that will pay me at the
end of the month (and not the opposite by the way) ?

you really think i imagine all of theses bug every day? just today
(really few hours ago) i discussed with an MVP delphi developer who
already work intensively on mobile with delphi, post some article
regarding mobile development with delphi (not like you as far as i know
who never do any serious mobile project, i repeat as far as i know), so
know very well delphi + mobile and he say me he is planning to
investigate XAMARIN ! ... Or like close to all other developpers in
french forum who say me they prefer to learn Android Studio and Xcode (2
languages, 2 codes) to make their mobile app than doing it with delphi
they already know (i don't even speak about the delphi guy who say me he
prefer windev for mobile developement than delphi) ...

you really think i invent all of this ?? i think it's a pity from you
and from delphi who don't want to see the reality

the bug of today (this memoryleak with jstring, jinteger, etc.) you
think it's cost 2 min to discover it ? First when i saw the error log
(and thanks god java did some error log in such case before to crash the
app) i was thinking it's was in my code that i forget to free something,
so i check everything (with this slow compiler) and it's cost me 1 day
to discover the bug. but what make me angry is that someone else already
discover this bug before (1 year ago) AND it's still not yet corrected,
so i really loose 1 day for nothing ! maybe i also invent it :(


ARC is here, that part will not change. You don't have to repeat how ARC
sucks every time you have some issue.

it's suck. i speak about arc because here you understand that the memory
leak is because we leave the compiler release the memory himself when he
think he don't need it anymore and you see that the compiler is not
really clever.
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 21, 2017 8:25 AM   in response to: loki loki in response to: loki loki
loki loki wrote:
I am really sorry because of that. You seem to find new issue every few lines
of code you write and it would benefit all if you could keep focus when
reporting and/or asking about those. That includes clear description of
your problem with proper MCVE regardless of where you post about it.

yeaaa i have nothing else to do than reporting thousands or error, and
making clear sample project ... maybe it's emb that will pay me at the
end of the month (and not the opposite by the way) ?

Well, if you don't report them do you think those issues will magically fix themselves?


you really think i imagine all of theses bug every day? just today
(really few hours ago) i discussed with an MVP delphi developer who
already work intensively on mobile with delphi, post some article
regarding mobile development with delphi

I know you are not imagining those issues. That is why I am trying to make
you see that you are not achieving anything with unfocused rants and
could do so much better if people would take you seriously.

I do take you seriously, but I don't work for Embarcadero so I cannot help
you much.

you really think i invent all of this ?? i think it's a pity from you
and from delphi who don't want to see the reality

I see reality quite well. I can feel your pain. In my reality Delphi does
not support my needs on mobile platforms well.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 21, 2017 9:09 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar

Well, if you don't report them do you think those issues will magically fix themselves?

i already reported 102 issues to the quality.embarcadero.com and this
only for the last year, so with 262 work days by years mean around 1 bug
report every 3 days !!

I know you are not imagining those issues. That is why I am trying to make
you see that you are not achieving anything with unfocused rants and
could do so much better if people would take you seriously.

don't know what you mean by rants, but what sure it's limit insulting
client to hear that on berlin they focus on stuff like getit package
manager or correct bug usefull to everyone (ie access private field from
class helper) and on godzilla they focus on Linux when for example me i
report 1 bug every 3 days for mobile development and close to none of
them have even been investigated.
Van Swofford

Posts: 389
Registered: 6/28/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 21, 2017 8:34 AM   in response to: loki loki in response to: loki loki
loki loki wrote:

I am really sorry because of that. You seem to find new issue every
few lines of code you write and it would benefit all if you could
keep focus when reporting and/or asking about those. That includes
clear description of your problem with proper MCVE regardless of
where you post about it.

yeaaa i have nothing else to do than reporting thousands or error,
and making clear sample project ... maybe it's emb that will pay me
at the end of the month (and not the opposite by the way) ?

you really think i imagine all of theses bug every day? just today
(really few hours ago) i discussed with an MVP delphi developer who
already work intensively on mobile with delphi, post some article
regarding mobile development with delphi (not like you as far as i
know who never do any serious mobile project, i repeat as far as i
know), so know very well delphi + mobile and he say me he is planning
to investigate XAMARIN ! ... Or like close to all other developpers
in french forum who say me they prefer to learn Android Studio and
Xcode (2 languages, 2 codes) to make their mobile app than doing it
with delphi they already know (i don't even speak about the delphi
guy who say me he prefer windev for mobile developement than delphi)
...

you really think i invent all of this ?? i think it's a pity from you
and from delphi who don't want to see the reality

the bug of today (this memoryleak with jstring, jinteger, etc.) you
think it's cost 2 min to discover it ? First when i saw the error log
(and thanks god java did some error log in such case before to crash
the app) i was thinking it's was in my code that i forget to free
something, so i check everything (with this slow compiler) and it's
cost me 1 day to discover the bug. but what make me angry is that
someone else already discover this bug before (1 year ago) AND it's
still not yet corrected, so i really loose 1 day for nothing ! maybe
i also invent it :(


ARC is here, that part will not change. You don't have to repeat
how ARC sucks every time you have some issue.

it's suck. i speak about arc because here you understand that the
memory leak is because we leave the compiler release the memory
himself when he think he don't need it anymore and you see that the
compiler is not really clever.

I believe you just proved Dalija's point quite well.

--
Cheers,
Van

"Good judgment comes from experience, and a lot of that comes from bad
judgment." - Will Rogers
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 21, 2017 9:09 AM   in response to: Van Swofford in response to: Van Swofford

it's suck. i speak about arc because here you understand that the
memory leak is because we leave the compiler release the memory
himself when he think he don't need it anymore and you see that the
compiler is not really clever.

I believe you just proved Dalija's point quite well.

what??
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 21, 2017 11:24 AM   in response to: loki loki in response to: loki loki
loki loki wrote:


it's suck. i speak about arc because here you understand that the
memory leak is because we leave the compiler release the memory
himself when he think he don't need it anymore and you see that the
compiler is not really clever.

I believe you just proved Dalija's point quite well.

what??

Dalija's point was that you (needlessly) rant about how ARC sucks
everytime you write a post. And you proved her point by, well, ranting
how ARC sucks in your reply to her. <g>

Compilers are not really clever, indeed, but they are extremely good at
applying the same principle (e.g. refcounting) without ever forgetting
anything. Humans can easily forget to free an item or to refcount it
(e.g. what you had do for COM interfaces before Delphi got ARC for
interfaces). The compiler won't.

ARC doesn't suck. On the contrary, it works very well, and incessantly.
It enver forgets an AddRef or a Release. But some of the code that was
originally designed to work without ARC sometimes does things that work
against ARC. That sucks. And not all of these situations have been
handled yet.

--
Rudy Velthuis http://www.rvelthuis.de

Lynch's Law: When the going gets tough, everyone leaves.
Mike Margerum

Posts: 573
Registered: 12/1/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 22, 2017 7:05 AM   in response to: loki loki in response to: loki loki
Why don't you just switch to swift or java instead of beating your head
against Delphi?

On 3/21/17 10:19 AM, loki loki wrote:


ARC is here, that part will not change. You don't have to repeat how ARC
sucks every time you have some issue.

it's suck. i speak about arc because here you understand that the memory
leak is because we leave the compiler release the memory himself when he
think he don't need it anymore and you see that the compiler is not
really clever.
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 22, 2017 7:11 AM   in response to: Mike Margerum in response to: Mike Margerum
Mike Margerum wrote:
Why don't you just switch to swift or java instead of beating your head
against Delphi?

I guess guy really likes Delphi ;-)

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 22, 2017 7:27 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
On 3/22/2017 5:11 PM, Dalija Prasnikar wrote:
Mike Margerum wrote:
Why don't you just switch to swift or java instead of beating your head
against Delphi?

I guess guy really likes Delphi ;-)

No delphi is a very good choice to do multi platform development! i m
just sad that because of stupid management team of emb my 6 months
development cost me 1 year :(
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 24, 2017 5:42 AM   in response to: loki loki in response to: loki loki
loki loki wrote:

On 3/22/2017 5:11 PM, Dalija Prasnikar wrote:
Mike Margerum wrote:
Why don't you just switch to swift or java instead of beating your
head >> against Delphi?

I guess guy really likes Delphi ;-)

No delphi is a very good choice to do multi platform development! i m
just sad that because of stupid management team of emb my 6 months
development cost me 1 year :(

I doubt it is management that causes your problems. Others manage to
write mobile apps quite well.

ISTM that your working against ARC instead of going with the flow
(i.e. working with ARC) causes your headaches. Try to understand ARC
before you fight it. <shrug>

And stop ranting. It is hard to take your problems for full if you
constantly rant about how bad the world is to you.
--
Rudy Velthuis http://www.rvelthuis.de

"The Pope!?! How many divisions has he got?"
-- Joseph Stalin
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 24, 2017 6:08 AM   in response to: Rudy Velthuis (... in response to: Rudy Velthuis (...
Rudy Velthuis (TeamB) wrote:
loki loki wrote:

On 3/22/2017 5:11 PM, Dalija Prasnikar wrote:
Mike Margerum wrote:
Why don't you just switch to swift or java instead of beating your
head >> against Delphi?

I guess guy really likes Delphi ;-)

No delphi is a very good choice to do multi platform development! i m
just sad that because of stupid management team of emb my 6 months
development cost me 1 year :(

I doubt it is management that causes your problems. Others manage to
write mobile apps quite well.

To be fair, there is huge number of issues in Delphi for mobile platforms.

Of course, most of them is not ARC related, though ARC does bring
its share of the problems. Main issue with ARC in Delphi is that it
is hard to polish rough ends because there is no Windows based
ARC compiler and compiling/running/testing on mobile takes more time.

Another issue is cross-platform code that caters for non-ARC and ARC
compilers - two don't mix well and and result is sub-optimal when ARC is
concerned.

ISTM that your working against ARC instead of going with the flow
(i.e. working with ARC) causes your headaches. Try to understand ARC
before you fight it. <shrug>

That is one of the problems.

And stop ranting. It is hard to take your problems for full if you
constantly rant about how bad the world is to you.

That too.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 24, 2017 7:08 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
To be fair, there is huge number of issues in Delphi for mobile platforms.

Of course, most of them is not ARC related, though ARC does bring
its share of the problems. Main issue with ARC in Delphi is that it
is hard to polish rough ends because there is no Windows based
ARC compiler and compiling/running/testing on mobile takes more time.

Another issue is cross-platform code that caters for non-ARC and ARC
compilers - two don't mix well and and result is sub-optimal when ARC is
concerned.

sincerely now that i remove arc code myself it's much more easy. so i
don't have any more problem with it but can't imagine how other can code
descent software under delphi with ARC! But when i see that the memory
leak with java type like jstring is still here i guess most of the other
developper don't know (or don't care) about the memory leak they may
have on mobile

take another CONCRETE example, look at the new linux compiler with
tokyo. we have an ISAPI DLL that is a templating system for our website.
it's sad we need IIS and need to buy microsoft on each of our server and
the ability to compile this dll under linux will be very great.

BUT because linux compiler is maded under ARC it's not possible at all
to do such and the linux compiler is completely useless for us. because
we will need to redebug/redo most of our existing code for ARC, test
everything again. also as on win32 their is no ARC the futur development
and testing will be a nightmare (because we test when we developp on our
developpement desktop that is windows). and to finish ARC slow down
everything on multithread software and must be avoided

again you must understand that it's not really arc the problem (except
for the speed) but the problem is that ARC CODE <> NON ARC CODE they are
not compatible and this is the problem !

Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 24, 2017 12:05 PM   in response to: loki loki in response to: loki loki
loki loki wrote:

sincerely now that i remove arc code myself it's much more easy.

What on earth have you done? How did you "remove ARC"?

--
Rudy Velthuis http://www.rvelthuis.de

"Everybody wants to go to heaven, but nobody wants to die."
-- Joe Louis.
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 24, 2017 1:37 PM   in response to: Rudy Velthuis (... in response to: Rudy Velthuis (...
On 3/24/2017 10:05 PM, Rudy Velthuis (TeamB) wrote:
loki loki wrote:

sincerely now that i remove arc code myself it's much more easy.

What on earth have you done? How did you "remove ARC"?

i updated the original delphi source code ! and now i see much (much)
less bug in my app :)

i don't understand why emb stay in this stupid situation with ARC on
mobile and non ARC on desktop ! just to avoid 2 set of DCU they destroy
the reputation of delphi and make mobile development (and linux now) a
nightmare.

the only solution is arc on mobile + desktop or non arc on mobile +
desktop. no need to speak years about it, because this is the only way
to go ! make 2 set of DCU and a compiler option to choose if we want ARC
or not arc
Dave Nottage

Posts: 1,721
Registered: 1/7/00
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 24, 2017 3:07 PM   in response to: loki loki in response to: loki loki
loki loki wrote:

i don't understand why emb stay in this stupid situation with ARC on
mobile and non ARC on desktop ! just to avoid 2 set of DCU they destroy
the reputation of delphi and make mobile development (and linux now) a
nightmare.

the only solution is arc on mobile + desktop or non arc on mobile +
desktop. no need to speak years about it, because this is the only way
to go ! make 2 set of DCU and a compiler option to choose if we want ARC
or not arc

That's just plain incorrect. Just because you're having problems with ARC does not mean that it is not possible to
develop mobile apps with Delphi, with ARC. Many already do, including myself.

--
Dave Nottage [MVP, TeamB]
Hints, tips and tricks at: http://www.delphiworlds.com/blog
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 25, 2017 2:24 AM   in response to: Dave Nottage in response to: Dave Nottage
That's just plain incorrect. Just because you're having problems with ARC does not mean that it is not possible to
develop mobile apps with Delphi, with ARC. Many already do, including myself.

Hmm, what is exactly "many" for you ? Their is some people who can play
chess and who can see thousands of thousands of combination in instant,
maybe be you are from these people, me i m not ... i m not a super
genius, i can't see fastly potential problem (like circular reference) !

the big question: how do you know that their is not memory leak or other
potential problems in your app? their is 0 descent debugging tool to say
you so !

how do you know that the code :

procedure TForm1.btnClick(Sender: TObject);
begin
TTask.Create(procedure
begin
sleep(150);
TThread.Synchronize(nil,
procedure
begin
form1.tag := 1;
end);

end).start;
end;

will create a memory leak !? how do you know all other scenario that
will create a memory leak? and i repeat the question, how do you simply
know that you have a memory leak or not ?

right now under delphi it is not descent programing, this is far west
programing ! maybe you will be lucky and everything will be fine, maybe
not and you app will crash time to time (or will be more slow) but you
will not know why! and maybe it's will be not because of arc, but again
you will not know, because again i repeat 0 descent debugging tool !

But don't say i say ARC is bad, to be honest i don't like verbosity in
the code and the possibility to write my code without any try ...
finally .free end it's so good ! i can't imagine javascript code with
all these try ... finally ! *but now be honest too, did you really
remove the try finally .free end from your code* ?? if not what the hell
is the purpose of the ARC under delphi (that the big question) ??

So can you say me the benefit of ARC ? the real benefit? say me why
your app is better with arc than without arc ! and please be honest
about arc ! i don't speak about theory, where arc is juicy, i speak
about the reality of now with delphi ! arc was introduced from ios, so
many year ago so it's not something new that you can say will improve
with time (time already show everything)

Of course if debugging under android/ios was descent (but that not the
fault of delphi, it's like this) yes maybe, if you developp only you app
for ios/android and not also for desktop/macos, yes maybe, if arc was
also in desktop/macos who have very descent debugging tool (also memory
leak report thanks to fastmm), yes maybe ... but it's a lot of if and
maybe :(

when apple introduce arc (and apple have much much more human resource
than emb to introduce such new feature that is really not easy to
introduce), they let the possibility to developer to stay on non arc,
why emb can't do so also ?

With arc, you can't import your old code from desktop to mobile, where
is the benefit here? with arc you can't remove the try finally from you
code because you still need to debug everything under desktop, where is
the benefit here? with arc you code is more slow under multithread
because of the need to threadlock to increase/decrease the refcount,
where is the benefit here?

concrete perfect example is my win32 isapi dll that i can't migrate to
linux because of arc not compatible with non arc, and that is the
reality! that is just a pity ...

all of this are fact ! you can close your eyes, fact will not change ..
and to finish many is not the number of finger i have ...
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 25, 2017 2:37 AM   in response to: loki loki in response to: loki loki
Am 25.03.2017 um 10:24 schrieb loki loki:


when apple introduce arc (and apple have much much more human resource
than emb to introduce such new feature that is really not easy to
introduce), they let the possibility to developer to stay on non arc,
why emb can't do so also ?

That one you practically answered yourself ;-)
It's due to the bigger ressources Apple has! ;-)
Maintaining 2 sets of libraries for the same platform is not free of
cost and labour!

Greetings

Markus
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 25, 2017 3:53 AM   in response to: Markus Humm in response to: Markus Humm

That one you practically answered yourself ;-)
It's due to the bigger ressources Apple has! ;-)
Maintaining 2 sets of libraries for the same platform is not free of
cost and labour!

but they already do !(!) they add ifdef everywhere to make the same code
compile on desktop and on mobile !
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 25, 2017 5:03 AM   in response to: loki loki in response to: loki loki
Am 25.03.2017 um 11:53 schrieb loki loki:

That one you practically answered yourself ;-)
It's due to the bigger ressources Apple has! ;-)
Maintaining 2 sets of libraries for the same platform is not free of
cost and labour!

but they already do !(!) they add ifdef everywhere to make the same code
compile on desktop and on mobile !

Hello,

issue is: if they enable ARC for FMX on Win32 then VCL needs to be fully
ARC compatible as well, otherwise a 2nd Win32 compiler is necessary.

At least this is my understanding.

Greetings

Markus
Arthur Hoornweg

Posts: 380
Registered: 6/2/98
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 12:23 AM   in response to: Markus Humm in response to: Markus Humm
Markus Humm wrote:

Hello,

issue is: if they enable ARC for FMX on Win32 then VCL needs to be fully
ARC compatible as well, otherwise a 2nd Win32 compiler is necessary.

Absolutely right. The VCL is full of parent<->child relationships.
Jeff Overcash (...

Posts: 1,457
Registered: 9/23/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 10:10 AM   in response to: Arthur Hoornweg in response to: Arthur Hoornweg
Arthur Hoornweg wrote:
Markus Humm wrote:

Hello,

issue is: if they enable ARC for FMX on Win32 then VCL needs to be fully
ARC compatible as well, otherwise a 2nd Win32 compiler is necessary.

Absolutely right. The VCL is full of parent<->child relationships.

Which actually isn't that hard to get around. IBX has a similar type of
relationship between DB accessing components (admin/services, queries etc) and
TIBDatabase and TIBTransaction. There is no need to ever call DisposeOf with
IBX. IBX code requires no changes between ARC and non ARC compilers.

--
Jeff Overcash (TeamB)
(Please do not email me directly unless asked. Thank You)
Learning is finding out what you already know. Doing is demonstrating that you
know it. Teaching is reminding others that they know it as well as you. We are
all leaners, doers, teachers. (R Bach)
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 25, 2017 5:52 AM   in response to: Markus Humm in response to: Markus Humm
Markus Humm wrote:

Am 25.03.2017 um 10:24 schrieb loki loki:


when apple introduce arc (and apple have much much more human
resource than emb to introduce such new feature that is really not
easy to introduce), they let the possibility to developer to stay
on non arc, why emb can't do so also ?

That one you practically answered yourself ;-)
It's due to the bigger ressources Apple has! ;-)
Maintaining 2 sets of libraries for the same platform is not free of
cost and labour!

Apple does not have to. Their objects were already refcounted, so the
switch is between manual and automatic refcounting. AFAIK, on the Mac,
you can still do it manually, if you really really want that, although
I don't know many who would do that. ARC made many things much easier
on the Mac. It also makes many things easier in Delphi, but only if you
don't work against it, like loki seems to do.

--
Rudy Velthuis http://www.rvelthuis.de

"Comparing to another activity is useful if it helps you formulate
questions, it's dangerous when you use it to justify answers."
-- Martin Fowler
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 25, 2017 6:59 AM   in response to: Rudy Velthuis (... in response to: Rudy Velthuis (...
Apple does not have to. Their objects were already refcounted, so the
switch is between manual and automatic refcounting. AFAIK, on the Mac,
you can still do it manually, if you really really want that, although
I don't know many who would do that. ARC made many things much easier
on the Mac. It also makes many things easier in Delphi, but only if you
don't work against it, like loki seems to do.

ok, same question than the one i ask to dave : Can you just gave me only
one advantage of using ARC in delphi right now ?
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 25, 2017 10:50 AM   in response to: loki loki in response to: loki loki
Am 25.03.2017 um 14:59 schrieb loki loki:
Apple does not have to. Their objects were already refcounted, so the
switch is between manual and automatic refcounting. AFAIK, on the Mac,
you can still do it manually, if you really really want that, although
I don't know many who would do that. ARC made many things much easier
on the Mac. It also makes many things easier in Delphi, but only if you
don't work against it, like loki seems to do.

ok, same question than the one i ask to dave : Can you just gave me only
one advantage of using ARC in delphi right now ?

Depends on whether you need to be cross platform compatible or not.
if not you can spare .free calls and try finaly blocks.

Greetings

Markus
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 25, 2017 10:53 AM   in response to: Markus Humm in response to: Markus Humm

Depends on whether you need to be cross platform compatible or not.
if not you can spare .free calls and try finaly blocks.

and if you spare, you will really debug your app only under android ??

really i can't stop smile with all of these guy defending ARC and making
their life complicated and see that none of them are enable just to gave
one simple advantage ;) maybe they are all masochist ;)

Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 25, 2017 3:21 PM   in response to: loki loki in response to: loki loki
Am 25.03.2017 um 18:53 schrieb loki loki:

Depends on whether you need to be cross platform compatible or not.
if not you can spare .free calls and try finaly blocks.

and if you spare, you will really debug your app only under android ??


Yes, sure! The app I'm developing currently can only work under Android
due to APIs only available there or issues on Win32 with the required
technology (on Win32 I'd normally need external hardware and their
firmware often is not compatible with the opposide party they shall
communicate with).

So this app really is tied to Android only.

Greetings

Markus
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 5:04 AM   in response to: Markus Humm in response to: Markus Humm
Markus Humm wrote:

Am 25.03.2017 um 14:59 schrieb loki loki:
Apple does not have to. Their objects were already refcounted, so
the >> switch is between manual and automatic refcounting. AFAIK, on
the Mac, >> you can still do it manually, if you really really want
that, although >> I don't know many who would do that. ARC made many
things much easier >> on the Mac. It also makes many things easier in
Delphi, but only if you >> don't work against it, like loki seems to
do.

ok, same question than the one i ask to dave : Can you just gave me
only one advantage of using ARC in delphi right now ?

Depends on whether you need to be cross platform compatible or not.

No, it doesn't. ARC has this one big advantage in any code, even in
Swift or Objective-C. Of course not in code for a compiler that doesn't
have it.

--
Rudy Velthuis http://www.rvelthuis.de

"I criticize by creation - not by finding fault."
-- Cicero (106-43 B.C.)

loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 5:30 AM   in response to: Rudy Velthuis (... in response to: Rudy Velthuis (...
No, it doesn't. ARC has this one big advantage in any code, even in
Swift or Objective-C. Of course not in code for a compiler that doesn't
have it.

typical Rudy answer without any argument :) you must be politician not
developer !
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 12:22 PM   in response to: loki loki in response to: loki loki
loki loki wrote:

No, it doesn't. ARC has this one big advantage in any code, even in
Swift or Objective-C. Of course not in code for a compiler that
doesn't have it.

typical Rudy answer without any argument :)

I have already given lots of arguments. I am not going to repeat all
that just to argue with you.

--
Rudy Velthuis http://www.rvelthuis.de

"We are what we think. All that we are arises with our
thoughts. With our thoughts, we make the world."
-- Buddha

Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 7:37 AM   in response to: Rudy Velthuis (... in response to: Rudy Velthuis (...
Am 26.03.2017 um 14:04 schrieb Rudy Velthuis (TeamB):
Markus Humm wrote:

Am 25.03.2017 um 14:59 schrieb loki loki:
Apple does not have to. Their objects were already refcounted, so
the >> switch is between manual and automatic refcounting. AFAIK, on
the Mac, >> you can still do it manually, if you really really want
that, although >> I don't know many who would do that. ARC made many
things much easier >> on the Mac. It also makes many things easier in
Delphi, but only if you >> don't work against it, like loki seems to
do.

ok, same question than the one i ask to dave : Can you just gave me
only one advantage of using ARC in delphi right now ?

Depends on whether you need to be cross platform compatible or not.

No, it doesn't. ARC has this one big advantage in any code, even in
Swift or Objective-C. Of course not in code for a compiler that doesn't
have it.

That was what I was refering to!
In Delphi you can only exploit it, if you do not need your code to stay
compatible with any of the non Nexgen compilers and you know this!
No need to go nitpicking here. Better tell us what EMBT has messed up in
their Android support in Tokyo... Because that's really severe/unuseable
in the current state. It doesn't seem to be noticeable in smallish "toy"
applications, but it's quite visible in real world sized apps.

Just came up with the idea that comparing a Berlin build of one of the
EMBT own mobile games against a Tokyo one could help, as for these EMBT
has the source code and thus a real world example and in a jump'n run or
arcade style game that effect should be visible easily... :-(

Greetings

Markus
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 7:49 AM   in response to: Markus Humm in response to: Markus Humm
That was what I was refering to!
In Delphi you can only exploit it, if you do not need your code to stay
compatible with any of the non Nexgen compilers and you know this!
No need to go nitpicking here. Better tell us what EMBT has messed up in
their Android support in Tokyo... Because that's really severe/unuseable
in the current state. It doesn't seem to be noticeable in smallish "toy"
applications, but it's quite visible in real world sized apps.

Just came up with the idea that comparing a Berlin build of one of the
EMBT own mobile games against a Tokyo one could help, as for these EMBT
has the source code and thus a real world example and in a jump'n run or
arcade style game that effect should be visible easily... :-(

about tokyo, the real problem is *how theses bugs did pass the quality
check* before to make tokyo in production ? i mean it's look like no one
in the dev team check any android application before to release tokyo ?
and worse as they completely refactor the plateform.android.pas (the
heart of the android support) they must be aware to even double check
all the android application and not just their hello world app !

it's scary :(

Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 8:22 AM   in response to: loki loki in response to: loki loki
Am 26.03.2017 um 16:49 schrieb loki loki:
That was what I was refering to!
In Delphi you can only exploit it, if you do not need your code to stay
compatible with any of the non Nexgen compilers and you know this!
No need to go nitpicking here. Better tell us what EMBT has messed up in
their Android support in Tokyo... Because that's really severe/unuseable
in the current state. It doesn't seem to be noticeable in smallish "toy"
applications, but it's quite visible in real world sized apps.

Just came up with the idea that comparing a Berlin build of one of the
EMBT own mobile games against a Tokyo one could help, as for these EMBT
has the source code and thus a real world example and in a jump'n run or
arcade style game that effect should be visible easily... :-(

about tokyo, the real problem is *how theses bugs did pass the quality
check* before to make tokyo in production ? i mean it's look like no one
in the dev team check any android application before to release tokyo ?
and worse as they completely refactor the plateform.android.pas (the
heart of the android support) they must be aware to even double check
all the android application and not just their hello world app !

it's scary :(


I fully agree!
Maybe that's the reason they need Nick as new boss of development...

Greetings

Markus
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 8:32 AM   in response to: Markus Humm in response to: Markus Humm
Maybe that's the reason they need Nick as new boss of development...

dependently if he is pro ARC or not :)
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 12:20 PM   in response to: loki loki in response to: loki loki
loki loki wrote:
Maybe that's the reason they need Nick as new boss of development...

dependently if he is pro ARC or not :)

I guess that his personal opinions will not matter. It is too late
to turn the ship around.

The only thing that can be "discussed" is how and when will ARC
be brought to currently non-ARC platforms.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 12:33 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
On 3/26/2017 10:20 PM, Dalija Prasnikar wrote:
loki loki wrote:
Maybe that's the reason they need Nick as new boss of development...

dependently if he is pro ARC or not :)

I guess that his personal opinions will not matter. It is too late
to turn the ship around.

The only thing that can be "discussed" is how and when will ARC
be brought to currently non-ARC platforms.

yes exactly :) this why i say we don't need to think an eternity, we
need to make 2 set of dcu to have the possibility to have arc on desktop
(or not for old program and people like me who want speed)

that all ! 5 seconds on thinking and basta
Lars Fosdal


Posts: 156
Registered: 10/26/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 27, 2017 7:13 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Dalija Prasnikar wrote:
The only thing that can be "discussed" is how and when will ARC
be brought to currently non-ARC platforms.

Yesterday isn't soon enough. Two different memory life cycle models makes cross platform code a bloody mess.

--
http://plus.lars.fosdal.com
Delphi Developers Google+ Community: https://plus.google.com/communities/103113685381486591754 (8500+ members)
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 27, 2017 7:20 AM   in response to: Lars Fosdal in response to: Lars Fosdal
Lars Fosdal wrote:
Dalija Prasnikar wrote:
The only thing that can be "discussed" is how and when will ARC
be brought to currently non-ARC platforms.

Yesterday isn't soon enough. Two different memory life cycle models makes cross platform code a bloody mess.

Agreed.

Problem is that devs working only on Windows might disagree.

They are comfy with status quo and see no reason to change.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 27, 2017 7:27 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar

Problem is that devs working only on Windows might disagree.

They are comfy with status quo and see no reason to change.

off course, who want to redo all their code ! they will never agree with
arc ... maybe they will agree for new project (maybe), this why their is
no other way that make arc as an option (and it's really easy to do)

the stupidity of arc under linux that block the migration of existing
win32 server app code to linux show how pity is the actual situation ...
Paul TOTH

Posts: 196
Registered: 1/2/17
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 27, 2017 7:41 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Le 27/03/2017 à 16:20, Dalija Prasnikar a écrit :
Lars Fosdal wrote:
Dalija Prasnikar wrote:
The only thing that can be "discussed" is how and when will ARC
be brought to currently non-ARC platforms.

Yesterday isn't soon enough. Two different memory life cycle models makes cross platform code a bloody mess.

Agreed.

Problem is that devs working only on Windows might disagree.

They are comfy with status quo and see no reason to change.

well even when they plan to use mobile dev, why should they change their
habits (and code) ?

BTW: Marco seems to like [weak] and [unsafe] for Interfaces
http://blog.marcocantu.com/blog/2016-april-weak-unsafe-interface-references.html

but not for ARC objects
<<It should be only used outside the System unit in very rare
situations. It is considered dangerous and its use is not recommended as
no code associated with reference counting is generated.>>

http://docwiki.embarcadero.com/RADStudio/Tokyo/en/Automatic_Reference_Counting_in_Delphi_Mobile_Compilers#The_Unsafe_Attribute

Delphi is getting very complex...
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 27, 2017 7:52 AM   in response to: Paul TOTH in response to: Paul TOTH

Delphi is getting very complex...

yes and all of this for close to 0 gain ! no speed gain (total
opposite), no code simplification (the total of the total opposite right
now), no less memory leak (without kidding, no word to say how opposite
it is)

all of this is just for theory and maybe because swift did like this and
it's good to do like other :(
Mike Margerum

Posts: 573
Registered: 12/1/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 27, 2017 9:57 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Problem is that devs working only on Windows might disagree.

They are comfy with status quo and see no reason to change.

Spring4D makes non arc Delphi a pleasure to use. I dont think I would
be using Delphi if this did not exist.
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 27, 2017 7:14 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
They are comfy with status quo and see no reason to change.

Which exactly is the reason to change? I really would like to know.

If I have a 2 million LOC codebase working nicely since Delphi 5 times and then EMBT decides to force ARC down your throat. What exactly I'm gaining?

If you have a good answer to that, I suggest EMBT to hire you immediately as Chief Marketing Officer. You really deserve it.
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 1:37 AM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
They are comfy with status quo and see no reason to change.

Which exactly is the reason to change? I really would like to know.

If I have a 2 million LOC codebase working nicely since Delphi 5 times and then EMBT decides to force ARC down your throat. What exactly I'm gaining?

Probably not much... I know what I would gain, and what Delphi as
general purpose language would gain... but not all people would agree with me.

Delphi non-ARC memory management model is also broken beyond description.
We are just used to it so it does not look broken but it is.

I am talking about reference counting for interfaces. Because of that you cannot
mix interface and object references, or you have to really have deep knowledge
of the system. That prevents any newcomers experienced in OOP to use Delphi
without tripping over most basic stuff.

ARC would fix that broken part once and for all.

I know that you don't want to change your code. Nobody does. If what I wrote
above means nothing in your development process then I cannot give you
anything more. For me it does mean something, but I am fine either way.

If you have a good answer to that, I suggest EMBT to hire you immediately as Chief Marketing Officer. You really deserve it.

At this point the there are no good answers... the only answer is evolution
already happened and dinosaurs are on their way to extinction... whether
they like it or not.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 1:53 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar

At this point the there are no good answers... the only answer is evolution
already happened and dinosaurs are on their way to extinction... whether
they like it or not.

non arc code it is not dinosaurs ! it's cheetah !!
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 2:00 AM   in response to: loki loki in response to: loki loki
loki loki wrote:

At this point the there are no good answers... the only answer is evolution
already happened and dinosaurs are on their way to extinction... whether
they like it or not.

non arc code it is not dinosaurs ! it's cheetah !!

You do know that cheetah is endangered species. They are genetically doomed.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 2:06 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar

non arc code it is not dinosaurs ! it's cheetah !!

You do know that cheetah is endangered species. They are genetically doomed.

yes this is sad :) poaching is very very bad !!!
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 2:07 AM   in response to: loki loki in response to: loki loki
On 3/28/2017 12:06 PM, loki loki wrote:

non arc code it is not dinosaurs ! it's cheetah !!

You do know that cheetah is endangered species. They are genetically doomed.

yes this is sad :) poaching is very very bad !!!

i mean yes this is sad :( and not :)
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 2:30 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
At this point the there are no good answers... the only answer is evolution
already happened and dinosaurs are on their way to extinction... whether
they like it or not.

Why don't you just stop using Courier New as your editor font? Doesn't make any sense to keep using this when is scientifically proven that there are at least one thousand fonts better to the eyes than that old crap.

Does the paragraph above make any sense to you? No? Hum. So that's how I see yours.
Arthur Hoornweg

Posts: 380
Registered: 6/2/98
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 12:25 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Dalija Prasnikar wrote:

ARC would fix that broken part once and for all.

So would a [weak] attribute for interfaces probably.
Paul TOTH

Posts: 196
Registered: 1/2/17
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 12:30 AM   in response to: Arthur Hoornweg in response to: Arthur Hoornweg
Le 31/03/2017 à 09:25, Arthur Hoornweg a écrit :
Dalija Prasnikar wrote:

ARC would fix that broken part once and for all.

So would a [weak] attribute for interfaces probably.

http://blog.marcocantu.com/blog/2016-april-weak-unsafe-interface-references.html
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 1:39 AM   in response to: Alexandre Machado in response to: Alexandre Machado
On 3/28/2017 5:15 AM, Alexandre Machado wrote:
They are comfy with status quo and see no reason to change.

Which exactly is the reason to change? I really would like to know.

If I have a 2 million LOC codebase working nicely since Delphi 5 times and then EMBT decides to force ARC down your throat. What exactly I'm gaining?

If you have a good answer to that, I suggest EMBT to hire you immediately as Chief Marketing Officer. You really deserve it.

would be also interesting to hear the answer of this question from the
actual embt chief marketing officer ;)
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 2:33 AM   in response to: loki loki in response to: loki loki
would be also interesting to hear the answer of this question from the
actual embt chief marketing officer ;)

The answer of people that don't have any valid argument is this: "evolution". What does that mean? Change for the sake of change? Really?
I wonder if that's how they justify charging thousands of work hours to their customers, or to their employers share holders...
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 3:14 AM   in response to: Alexandre Machado in response to: Alexandre Machado
On 3/28/2017 12:33 PM, Alexandre Machado wrote:
would be also interesting to hear the answer of this question from the
actual embt chief marketing officer ;)

The answer of people that don't have any valid argument is this: "evolution". What does that mean? Change for the sake of change? Really?
I wonder if that's how they justify charging thousands of work hours to their customers, or to their employers share holders...

Lars Fosdal start a discussion on google+ about arc

https://plus.google.com/u/0/+LarsFosdal/posts/Bh4utBUXSwc

The ARC vs non-ARC situation adds yet another layer of complexity to
creating cross platform code. The Linux compiler uses ARC - while
Windows and OSX do not.

Guess what that does to the Linux support: It falls way down on the
viability list for cross platform code for servers and services. Note
that I don't mind ARC - I just don't get why they did not move to ARC
for all platforms.

They could duplicate the libs, put the old ones into care & maintenance,
and introduce
arc.system. *
arc.vcl. *
arc.data. *
and so forth...

The current situation is pretty much ridiculous.

Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 3:37 AM   in response to: loki loki in response to: loki loki
loki loki wrote:
On 3/28/2017 12:33 PM, Alexandre Machado wrote:
would be also interesting to hear the answer of this question from the
actual embt chief marketing officer ;)

The answer of people that don't have any valid argument is this: "evolution". What does that mean? Change for the sake of change? Really?
I wonder if that's how they justify charging thousands of work hours to their customers, or to their employers share holders...

Lars Fosdal start a discussion on google+ about arc

https://plus.google.com/u/0/+LarsFosdal/posts/Bh4utBUXSwc

The ARC vs non-ARC situation adds yet another layer of complexity to
creating cross platform code. The Linux compiler uses ARC - while
Windows and OSX do not.

Guess what that does to the Linux support: It falls way down on the
viability list for cross platform code for servers and services. Note
that I don't mind ARC - I just don't get why they did not move to ARC
for all platforms.

They could duplicate the libs, put the old ones into care & maintenance,
and introduce
arc.system. *
arc.vcl. *
arc.data. *
and so forth...


Because as long as there are two systems you are pretty much screwed anyway.
And having separated code is harder to maintain. Embarcadero is already
stretched thin, they don't need and cannot afford extra work.

Complete removal of non-ARC compiler is the tipping point where ARC will
start to function.

The current situation is pretty much ridiculous.

Agreed.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 4:43 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Complete removal of non-ARC compiler is the tipping point where ARC will
start to function.

ahahahah this will be the tipping point where delphi developper will
become like the dinausaur :) maybe 1 or 2 will be still here ;)
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 1:43 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Because as long as there are two systems you are pretty much screwed anyway.
And having separated code is harder to maintain. Embarcadero is already
stretched thin, they don't need and cannot afford extra work.

What do you use Delphi for? Win32, Win64? I know that you don't use Delphi for Android and iOS and MacOS, because you were one doing the most fierce opposition against FMX, ARC and all that stuff when it was released.
I remember uncountable posts where you clearly said that you would never use FMX for iOS/MacOS and "Swift is sooooo good" and bla bla bla.
Now what? You want EMBT to pretty much screw every body relying on their Win32/64 compiler to save you from what exactly?
This is a genuine question. I want to understand why someone that doesn't use Delphi for Android/iOS/MacOS is worried about this.
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 5:38 AM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
Because as long as there are two systems you are pretty much screwed anyway.
And having separated code is harder to maintain. Embarcadero is already
stretched thin, they don't need and cannot afford extra work.

What do you use Delphi for? Win32, Win64? I know that you don't use Delphi for Android and iOS and MacOS, because you were one doing the most fierce opposition against FMX, ARC and all that stuff when it was released.
I remember uncountable posts where you clearly said that you would never use FMX for iOS/MacOS and "Swift is sooooo good" and bla bla bla.
Now what? You want EMBT to pretty much screw every body relying on their Win32/64 compiler to save you from what exactly?
This is a genuine question. I want to understand why someone that doesn't use Delphi for Android/iOS/MacOS is worried about this.

I was never against ARC. I was also not against FMX as idea, but
FMX implementation is buggy as hell combined with horrible
kitchen sink design.

I was against removal of 8-bit strings from mobile compilers and
I was against ZBS that served no purpose at all.

8-bit strings are back (except for AnsiStrings, but they don't make too
much sense on other platforms anyway.)

I will repeat again. I never asked for ARC, but I have no problem with it
and I actually like ARC as memory management very much.

I could not use Delphi for mobile because lack of 8-bit strings because
my internal library is built on top of UTF8String.

My first mobile projects were internal ones that absolutely required my library.
No UTF8String - no luck. Instead of using Android for that internal projects
we switched to Windows tablets. Also I knew FMX was buggy and lacked
features since I tried to port some stuff on OSX. It was not worth the trouble
so I never finished porting those. For my internal Android projects I could
live with FMX bugs until they are polished but lack of UTF8Strings was
showstopper.

For my other non-internal mobile projects I had different set of requirements
Delphi could not satisfy - support for Tegra and Intel chips - that was
showstopper for those.

Adding FMX bugs on top, there were just too many things that didn't work for me.

I never had problem with fact mobile compilers introduced ARC.

I only have some issues with DisposeOf that was not implemented properly.

So I am not saying that Delphi ARC compiler is perfect in sense it does not
have some issues that needs to be polished and that it lacks some features.
But those can be polished. And those features could be polished faster
with single ARC compiler.

Why am I interested to see Delphi ARC compiler on all platforms? Simple,
because I know how nice is to work with proper ARC memory management.
I would love to have that on Windows even though I would have to adapt
some code.

Also, since Delphi is cross-platform tool, it only makes sense that people
use it on other platforms too. That can bring more users and keep existing
ones tied to Delphi. More users means more secure future for Delphi.
As Delphi developer I am interested in that. One of core requirements for
more developers using Delphi is resolving issues on other platforms.
One of the issue is debugging ARC code that cannot be properly done
without Windows ARC compiler, and another is fixing FMX.

And as far as Swift is concerned, it is not all peachy, but it has some nice features
that could also be implemented in Delphi. That is why I usually bring Swift
up in relation to Delphi.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
Remy Lebeau (Te...


Posts: 8,950
Registered: 12/23/01
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 9:08 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Dalija wrote:

I could not use Delphi for mobile because lack of 8-bit strings
because my internal library is built on top of UTF8String.

There are other ways to work with UTF-8 than just UTF8String. That shouldn't
have been a showstopper.

--
Remy Lebeau (TeamB)
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 9:51 AM   in response to: Remy Lebeau (Te... in response to: Remy Lebeau (Te...
Remy Lebeau (TeamB) wrote:
Dalija wrote:

I could not use Delphi for mobile because lack of 8-bit strings
because my internal library is built on top of UTF8String.

There are other ways to work with UTF-8 than just UTF8String. That shouldn't
have been a showstopper.

Not really when you use UTF8String as base string type. I have been doing
that since Delphi 7 and Unicode transition was piece of cake.

Due to memory requirements some of my heavier data processing apps have,
switching to conversion on input/output was not an option because UTF-16
would double memory consumption and hit the memory limit.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 12:40 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar

Not really when you use UTF8String as base string type. I have been doing
that since Delphi 7 and Unicode transition was piece of cake.

Due to memory requirements some of my heavier data processing apps have,
switching to conversion on input/output was not an option because UTF-16
would double memory consumption and hit the memory limit.

can't believe i was thinking i was the only one in the delphi world to
notice that Unicode string use 2x more memory than ansistring ! and for
mobile developpement where memory really matter the decision to banish
the ansiString was really "questionable" ... anyway the introduction of
utf8string to berlin was too late for me because i already spend lot of
time to convert all my ansitring library to unicode :(
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 1:04 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Due to memory requirements some of my heavier data processing apps have,

<Irony ON>
Come on! You? Worried about memory? It is just a few bytes here and there, Dalija...
Memory is cheap and computers have evolved so much since you first designed this application. That shouldn't be a problem to you anymore ;-)
<Irony OFF>

See how it feels when someone without a minimal idea about the requirements decides to explain how a "few bytes here and there" won't affect anything and you should just "get over it"?
How does it feel?
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 1:46 PM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
Due to memory requirements some of my heavier data processing apps have,

<Irony ON>
Come on! You? Worried about memory? It is just a few bytes here and there, Dalija...

Processing large textual data files is not just few bytes here and there. You are complaining
about TList<T> used in TComponent that is already quite heavy class not suitable for
high volume processing. For that you can write your own containers like I have done.

Yes, I could do my processing with byte arrays or memory mapped files or something else.
Point is I don't have to.

You will never see me complaining about Unicode in VCL/FMX because just like I don't
use thousands and thousands of TComponent instances I also don't store huge amount
of data in my GUI.

Memory is cheap and computers have evolved so much since you first designed this application. That shouldn't be a problem to you anymore ;-)
<Irony OFF>

See how it feels when someone without a minimal idea about the requirements decides to explain how a "few bytes here and there" won't affect anything and you should just "get over it"?
How does it feel?

It still feels the same like before. You are still trying to prove your point
using idiotic examples comparing apples to oranges.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 1:54 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Processing large textual data files is not just few bytes here and there. You are complaining
about TList<T> used in TComponent that is already quite heavy class not suitable for
high volume processing. For that you can write your own containers like I have done.

You still don't understand. TList is a multi-purpose container of data. It has nothing to do with TComponent.
I used TComponent as en example of how a bloated TList can compromise many other classes, in cascade, relying on it.
You are taking one example I gave in a different thread, and using as it was my main point. You are wrong.

Yes, I could do my processing with byte arrays or memory mapped files or something else.
Point is I don't have to.

You will never see me complaining about Unicode in VCL/FMX because just like I don't
use thousands and thousands of TComponent instances I also don't store huge amount
of data in my GUI.

Neither do I.

Memory is cheap and computers have evolved so much since you first designed this application. That shouldn't be a problem to you anymore ;-)
<Irony OFF>

See how it feels when someone without a minimal idea about the requirements decides to explain how a "few bytes here and there" won't affect anything and you should just "get over it"?
How does it feel?

It still feels the same like before. You are still trying to prove your point
using idiotic examples comparing apples to oranges.

You are wrong again. The problem is that you only know apples (your reality). You actually think that you do know mine and you think that it is an orange. But it is not.
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 12:37 PM   in response to: Remy Lebeau (Te... in response to: Remy Lebeau (Te...
I could not use Delphi for mobile because lack of 8-bit strings
because my internal library is built on top of UTF8String.

There are other ways to work with UTF-8 than just UTF8String. That shouldn't
have been a showstopper.

and some people say i m crazy because i redo myself all the Firemonkey
control :) yes their is always solution, but at with cost :(
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 12:59 PM   in response to: Remy Lebeau (Te... in response to: Remy Lebeau (Te...
There are other ways to work with UTF-8 than just UTF8String. That shouldn't
have been a showstopper.

Dalija thinks that switching from manual memory management to ARC is a breeze, "evolution", and "you must adapt or die".
On the other hand, she also thinks that getting rid of UTF8String in her base library is a show stopper, impossible, unthinkable!
It is interesting how funny things can be.... LOL
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 1:34 PM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
There are other ways to work with UTF-8 than just UTF8String. That shouldn't
have been a showstopper.

Dalija thinks that switching from manual memory management to ARC is a breeze, "evolution", and "you must adapt or die".
On the other hand, she also thinks that getting rid of UTF8String in her base library is a show stopper, impossible, unthinkable!
It is interesting how funny things can be.... LOL

We all have our reasons for doing things one way or another.

Again I didn't ask for ARC, again I don't think transition to ARC does
not take time and money. I am merely saying at some point in the future
this change will inevitably happen.

Point is there can only be one compiler, having two different management
system in not sustainable in the long run. There ARC switch is not an option
just like Unicode switch was not an option.

Sometimes changes are breaking ones because there is no other way to
make that change (or evolve the language). On the other hand some
breaking changes are totally unnecessary because they don't stand in a way
of progress or other must have features.

Unicode and ARC belong to former, removal of 8-bit strings, ZBS and private
access through helpers are the later.

Also, UTF8 is Unicode. It is current highly adopted and widely spread standard
it is not me clinging to some obsolete format refusing to change. The fact
UTF8 was again introduced to mobile compilers support that. I am sure that
decision was not made just to shut me up.

Like I said, I explored my options and changing my core library just to port some
of my projects to Android was not viable option. Cheaper solution was to use
Windows tablets.

I never supported breaking changes just for the sake of change. ARC is not
that kind of change. While we can discuss whether or not move to ARC was
good or bad idea considering existing Delphi code-bases, I can see advantages
of having ARC as memory management system.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 1:42 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
We all have our reasons for doing things one way or another.

Yes, we do.
What about a campaign asking EMBT to remove UTF8Strings from Win32/Win64 compiler?
(You know, that old, obsolete string type. We should really have only one string type, don't you think?)
How does it affect you?
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 2:00 PM   in response to: Alexandre Machado in response to: Alexandre Machado
On 3/29/2017 11:43 PM, Alexandre Machado wrote:
We all have our reasons for doing things one way or another.

Yes, we do.
What about a campaign asking EMBT to remove UTF8Strings from Win32/Win64 compiler?
(You know, that old, obsolete string type. We should really have only one string type, don't you think?)
How does it affect you?

when they remove from delphi all the tiny usefull ansiString function
(like inttostr, strtoint, tstringlist, etc.) they in some way already
removed the support for utf8string/ansiString ... at this time i was
very frustrated! I spend many days to rebuild all these tiny functions
for ansiString (mostly by simply copy/past the old code they simply
remove from delphi after the unicode migration instead of keeping it
somewhere).

i don't know how this affect dalija, but me it's was very bad ...
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 2:14 PM   in response to: loki loki in response to: loki loki
loki loki wrote:
On 3/29/2017 11:43 PM, Alexandre Machado wrote:
We all have our reasons for doing things one way or another.

Yes, we do.
What about a campaign asking EMBT to remove UTF8Strings from Win32/Win64 compiler?
(You know, that old, obsolete string type. We should really have only one string type, don't you think?)
How does it affect you?

when they remove from delphi all the tiny usefull ansiString function
(like inttostr, strtoint, tstringlist, etc.) they in some way already
removed the support for utf8string/ansiString ... at this time i was
very frustrated! I spend many days to rebuild all these tiny functions
for ansiString (mostly by simply copy/past the old code they simply
remove from delphi after the unicode migration instead of keeping it
somewhere).

i don't know how this affect dalija, but me it's was very bad ...

I'm not asking to remove it. I'm just trying to make her see that "people who live in glass houses shouldn't throw stones".

Dalija advocates for something that will force many companies to abandon Delphi.
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 2:28 PM   in response to: loki loki in response to: loki loki
Dalija is thinking about making a plastic surgery to make Delphi more beautiful.
However, Delphi is bleeding. And bleeding badly. You just have to stop the bleeding, not make it beautiful.
STOP breaking and removing things! Stop pushing away customers that have been using Delphi for decades.
Fix what is wrong, improve what needs to be improved, but stop breaking stuff!
People won't pay 4k and then spend months updating and retesting their huge code base! Nobody will do that!
Arthur Hoornweg

Posts: 380
Registered: 6/2/98
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 12:27 AM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
There are other ways to work with UTF-8 than just UTF8String. That shouldn't
have been a showstopper.

Dalija thinks that switching from manual memory management to ARC is a breeze, "evolution", and "you must adapt or die".
On the other hand, she also thinks that getting rid of UTF8String in her base library is a show stopper, impossible, unthinkable!
It is interesting how funny things can be.... LOL


Given that UTF8 strings are extremely common in the computing world I'd hate to miss them too.

Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 12:31 AM   in response to: Arthur Hoornweg in response to: Arthur Hoornweg
Given that UTF8 strings are extremely common in the computing world I'd hate to miss them too.

Nothing should be removed. That's my opinion.

But some people think that "whatever I don't use or don't care about can be removed, deprecated, bloated, refactored. But don't touch my stuff"
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 4:20 AM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
Given that UTF8 strings are extremely common in the computing world I'd hate to miss them too.

Nothing should be removed. That's my opinion.

But some people think that "whatever I don't use or don't care about can be removed, deprecated, bloated, refactored. But don't touch my stuff"

I really have a lot of work to do and very little time at the moment
and this kind of discussion where you are constantly twisting my words
and where I have to repeat same things over and over again don't
serve much purpose.

Still, I will say it once more: I am never for removing things that can
stand on their own and don't stand in a way of other language features.

UTF8String type is such feature. It has nothing to do with it being "my stuff".

When it comes to ARC, two compilers don't work. You cannot have it
both, you cannot have a switch. At the end only one must be chosen.
Cross-platform code kind of works but it just follows non-ARC memory
model with ARC glued on top of it. Staying like that would be the worst
scenario we could have.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 3:50 AM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
would be also interesting to hear the answer of this question from the
actual embt chief marketing officer ;)

The answer of people that don't have any valid argument is this: "evolution". What does that mean? Change for the sake of change? Really?
I wonder if that's how they justify charging thousands of work hours to their customers, or to their employers share holders...

Evolution is change. Not all changes are good, but that does not mean they
are bad either.

Change already happened. I am not saying it is good and I am not saying it is bad.
I am just saying you cannot do anything about it.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 4:04 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Change already happened. I am not saying it is good and I am not saying it is bad.
I am just saying you cannot do anything about it.

Yes, we do. Some people just use other tools to build new APP. That's what we can do. That also some companys do.
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 9:01 AM   in response to: wenjie zhou in response to: wenjie zhou
Am 28.03.2017 um 13:05 schrieb wenjie zhou:
Change already happened. I am not saying it is good and I am not saying it is bad.
I am just saying you cannot do anything about it.

Yes, we do. Some people just use other tools to build new APP. That's what we can do. That also some companys do.

And not all of those are cross platform.
And some of those use ARC as well and some even use a QC who might kick
in at the wrong time.

Not everything else is better.

Where is your real problem?
You can in your own code still use try / finally and free mechanism and
thus have cross platform compatible code.

So what exact problem do you have?

Greetings

Markus
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 9:07 AM   in response to: Markus Humm in response to: Markus Humm

Where is your real problem?
You can in your own code still use try / finally and free mechanism and
thus have cross platform compatible code.

huuuu but we speak from years now that non arc code (try / finally) is
not compatible with arc code
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 6:40 PM   in response to: Markus Humm in response to: Markus Humm
So what exact problem do you have?

No problem. Some of old customer will try ARC. Some of them will use new tool. But seldom new users will use Delphi to build new APP. That's OK for everyone.
Just keep away old customers, it's very nice. And this will be reflected in its profits.
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 12:51 PM   in response to: wenjie zhou in response to: wenjie zhou
wenjie zhou wrote:
Change already happened. I am not saying it is good and I am not saying it is bad.
I am just saying you cannot do anything about it.

Yes, we do. Some people just use other tools to build new APP. That's what we can do. That also some companys do.

So use something else if you think it will be better for you.

What I meant to say is that there is nothing you can do to change
the fact Delphi is moving in ARC direction.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 3:26 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
What I meant to say is that there is nothing you can do to change
the fact Delphi is moving in ARC direction.


nooo i can't believe we both agree on the same think, delphi is moving
in the arc direction :)

I fully of the fully (of the fully) agree of this with you! and i even
have an (nonofficial) video of this moving to arc of delphi:
https://www.youtube.com/watch?v=BQ33CFyWFKo

loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 4:44 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Evolution is change. Not all changes are good, but that does not mean they
are bad either.

yes maybe, but arc with delphi is not good and it's really bad :)
Paul TOTH

Posts: 196
Registered: 1/2/17
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 7:41 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Le 28/03/2017 à 12:50, Dalija Prasnikar a écrit :
Alexandre Machado wrote:
would be also interesting to hear the answer of this question from the
actual embt chief marketing officer ;)

The answer of people that don't have any valid argument is this: "evolution". What does that mean? Change for the sake of change? Really?
I wonder if that's how they justify charging thousands of work hours to their customers, or to their employers share holders...

Evolution is change. Not all changes are good, but that does not mean they
are bad either.

Change already happened. I am not saying it is good and I am not saying it is bad.
I am just saying you cannot do anything about it.

that was the position of Borland for Delphi for .Net when win32 was
considered as a dead platform...at least they tried to make a .Net
compiler as close to the win32 as possible.

now the ARC/non-ARC question has nothing to do with the underlaying
platform. it's just an internal behavior it should not create so much
confusion and be a developper choice...and so a compiler option.

unfortunatly choices have been made that lead to bad situation like this one

var
[unsafe] obj: TObject;
begin
obj := TObject.Create; // same code for ARC and non-ARC because of
[unsafe] attribute
obj.Free; // for ARC it just put 0 into obj without freeing the object.
end; // memory leak !
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 9:03 AM   in response to: Paul TOTH in response to: Paul TOTH
Am 28.03.2017 um 16:41 schrieb Paul TOTH:


unfortunatly choices have been made that lead to bad situation like this one

var
[unsafe] obj: TObject;
begin
obj := TObject.Create; // same code for ARC and non-ARC because of
[unsafe] attribute
obj.Free; // for ARC it just put 0 into obj without freeing the object.
end; // memory leak !

Hello,

you fail to tell us why that object has been declared as [unsafe].
If there shouild not be a valid reason for this you should stop complaining.

If you present a valid reason for this attribute we can continue talking.

In your simple case removing that attribute fixes your memory leak.

Greetings

Markus

loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 9:08 AM   in response to: Markus Humm in response to: Markus Humm
On 3/28/2017 7:03 PM, Markus Humm wrote:
var
[unsafe] obj: TObject;
begin
obj := TObject.Create; // same code for ARC and non-ARC because of
[unsafe] attribute
obj.Free; // for ARC it just put 0 into obj without freeing the object.
end; // memory leak !

20 years i do programing underder delphi and i need several minute to
understand clearly all the implication of this 3 lines of code ... yea
delphi become very complex !!!!
Paul TOTH

Posts: 196
Registered: 1/2/17
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 9:40 AM   in response to: Markus Humm in response to: Markus Humm
Le 28/03/2017 à 18:03, Markus Humm a écrit :
Am 28.03.2017 um 16:41 schrieb Paul TOTH:


unfortunatly choices have been made that lead to bad situation like this one

var
[unsafe] obj: TObject;
begin
obj := TObject.Create; // same code for ARC and non-ARC because of
[unsafe] attribute
obj.Free; // for ARC it just put 0 into obj without freeing the object.
end; // memory leak !

Hello,

you fail to tell us why that object has been declared as [unsafe].
If there shouild not be a valid reason for this you should stop complaining.

If you present a valid reason for this attribute we can continue talking.

In your simple case removing that attribute fixes your memory leak.


Oh I see, it's not a bug, it's a functionality...we just don't know
which one for now :)

Free on an unsafe object should AT LEAST raise a warning because on
other situations it do something.

Greetings

Markus
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 11:32 AM   in response to: Paul TOTH in response to: Paul TOTH
Am 28.03.2017 um 18:40 schrieb Paul TOTH:
Le 28/03/2017 à 18:03, Markus Humm a écrit :
Am 28.03.2017 um 16:41 schrieb Paul TOTH:


unfortunatly choices have been made that lead to bad situation like this one

var
[unsafe] obj: TObject;
begin
obj := TObject.Create; // same code for ARC and non-ARC because of
[unsafe] attribute
obj.Free; // for ARC it just put 0 into obj without freeing the object.
end; // memory leak !

Hello,

you fail to tell us why that object has been declared as [unsafe].
If there shouild not be a valid reason for this you should stop complaining.

If you present a valid reason for this attribute we can continue talking.

In your simple case removing that attribute fixes your memory leak.


Oh I see, it's not a bug, it's a functionality...we just don't know
which one for now :)

Free on an unsafe object should AT LEAST raise a warning because on
other situations it do something.

Hello,

this is something else.
But you initially didn't describe the situation you were after well
enough. Sorry!

Greetings

Markus
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 12:49 PM   in response to: Paul TOTH in response to: Paul TOTH
Paul TOTH wrote:
Le 28/03/2017 à 12:50, Dalija Prasnikar a écrit :
Alexandre Machado wrote:
would be also interesting to hear the answer of this question from the
actual embt chief marketing officer ;)

The answer of people that don't have any valid argument is this: "evolution". What does that mean? Change for the sake of change? Really?
I wonder if that's how they justify charging thousands of work hours to their customers, or to their employers share holders...

Evolution is change. Not all changes are good, but that does not mean they
are bad either.

Change already happened. I am not saying it is good and I am not saying it is bad.
I am just saying you cannot do anything about it.

that was the position of Borland for Delphi for .Net when win32 was
considered as a dead platform...at least they tried to make a .Net
compiler as close to the win32 as possible.

That was completely different situation. Embarcadero is not trying to jump
on technology wagon forced from outside. They are evolving the language
as their own choice.

now the ARC/non-ARC question has nothing to do with the underlaying
platform. it's just an internal behavior it should not create so much
confusion and be a developper choice...and so a compiler option.

There is no choice. There cannot be choice. Choice implies two compilers and
two sets of everything. And current situation shows that in reality cross-compiler
code is not sustainable option.

unfortunatly choices have been made that lead to bad situation like this one

var
[unsafe] obj: TObject;
begin
obj := TObject.Create; // same code for ARC and non-ARC because of
[unsafe] attribute
obj.Free; // for ARC it just put 0 into obj without freeing the object.
end; // memory leak !

In order to be compatible with old compiler code Delphi has some rules.
And they are not that complex.

First rule of all is that Free on ARC translates to nil. It is obvious that above
code won't work. Above code leaks not because of "bad" choices but because
of your false expectations.

I am not even saying that you should have known upfront above code will not
work, or that people might not be confused at times, but it is clear as day that
this is not ARC problem or its bad implementation.

Whether or not ARC compiler should, or could issue warning here may not
be as simple as that. If you think warning is in order file QP report.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 1:46 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Evolution is change. Not all changes are good, but that does not mean they
are bad either.

Change already happened. I am not saying it is good and I am not saying it is bad.
I am just saying you cannot do anything about it.

I pay you my next month's salary if you write this exact phrase to your major customer whenever he asks something about "why is this costing me so much?"

Deal?
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 27, 2017 7:23 AM   in response to: Lars Fosdal in response to: Lars Fosdal
On 3/27/2017 5:13 PM, Lars Fosdal wrote:
Dalija Prasnikar wrote:
The only thing that can be "discussed" is how and when will ARC
be brought to currently non-ARC platforms.

Yesterday isn't soon enough. Two different memory life cycle models makes cross platform code a bloody mess.

We already saw how coherency went missing in the past years with the
push to adopt the ARC memory manager despite it being not the right
choice for highly parallel applications (Delphi still has too much focus
on "client" applications, even when it tries to appeal to server
applications developers, like now with the linux support).

I don't understand why they just start to add ARC in mobile without
thinking 5 min that it's not compatible with non arc code and in this
way will never arrive to desktop without a big pain like the disaster of
the unicode migration. The managing team don't have any brain or what :(

wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 4:10 AM   in response to: loki loki in response to: loki loki
loki loki wrote:
On 3/27/2017 5:13 PM, Lars Fosdal wrote:
Dalija Prasnikar wrote:
The only thing that can be "discussed" is how and when will ARC
be brought to currently non-ARC platforms.

Yesterday isn't soon enough. Two different memory life cycle models makes cross platform code a bloody mess.

We already saw how coherency went missing in the past years with the
push to adopt the ARC memory manager despite it being not the right
choice for highly parallel applications (Delphi still has too much focus
on "client" applications, even when it tries to appeal to server
applications developers, like now with the linux support).

I don't understand why they just start to add ARC in mobile without
thinking 5 min that it's not compatible with non arc code and in this
way will never arrive to desktop without a big pain like the disaster of
the unicode migration. The managing team don't have any brain or what :(


I agree with you, but such expression is too straight.
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 5:01 AM   in response to: wenjie zhou in response to: wenjie zhou

I agree with you, but such expression is too straight.

we will discuss about it

when you will start to report a new bug every 3 days,

when you will have in your project around 30 original delphi unit that
you manually corrected yourself and cost you month of debuging,

when instead of seeing you bug corrected (even those super trivial like
https://quality.embarcadero.com/browse/RSP-15426 where just the keyword
raise is missing) you will see usefull bug like accessing private
members from class helper corrected,

when you will lost all your week end to guess what is wrong in the brand
new Tokyo,

when you will lost months of development because of ARC introduced in
ridiculous way,

when you will need to redo all your library because a guy decide to
forbid anistring in mobile compiler to finally add it years after...

and when you will say this to them year after year, realease after
release, you will this that they not listen and nothing change

Contrary as some other, i work and leave with Delphi (from delphi 3) and
when i see people destroying it, it's not make me happy !

wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 9:13 AM   in response to: loki loki in response to: loki loki
loki loki wrote:

I agree with you, but such expression is too straight.

we will discuss about it
So, I try not to introduce new features into formal work. i just stand and see.
They did not have the spirit to respect the work of the customers. And as the result, you will see such things happen again and again.
So, what can you do? Post something here is really meanless. They lack spirit. Even under pressure they make some change(UTF8).
They still can not stand in the customer's point of view . There's two kinds of man in their team. The one is saler, they only care money.
The two is idealist , they only want to make a new language they like. So, old classic Delphi is dead. No one inherited the spirit of classic Delphi in their team.

The only thing you can do is, STOP. Stop follow them. Stay on the version you like.
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 3:50 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
I guess that his personal opinions will not matter. It is too late
to turn the ship around.

Let me see here... what is easier?
a) Remove ARC from mobile which will force half dozen Delphi/mobile developers to rewrite their toy application
b) Force ARC down the throat of all other Win32/Win64 Delphi which have huge codebases, you know, the ones that pay their salaries

?

"Oh, sure, (b) is much easier! and also looks modern! That's evolution!"

PS: From Danny Thorpe, explaining why Borland failed (https://www.quora.com/Why-did-Borland-fail):

"In the height of the enterprise transformation, I asked Del Yocam, one of many interim CEOs after Kahn, "Are you saying you want to trade a million loyal $100 customers for a hundred $1 million customers?" Yocam replied without hesitation "Absolutely."

Now, Embarcadero/Idera wants to do exactly the opposite (although similar): Sacrifice all their paying customers to hunt 20 y.o. script kids that consider manual memory management so 90's... Good luck.
I do know the end result of that. And believe me, you will hear me saying "I told you", and of course you and Rudy and many others will deny it, as people always do when things go incredibly wrong.
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 5:43 PM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
I guess that his personal opinions will not matter. It is too late
to turn the ship around.

Let me see here... what is easier?
a) Remove ARC from mobile which will force half dozen Delphi/mobile developers to rewrite their toy application
b) Force ARC down the throat of all other Win32/Win64 Delphi which have huge codebases, you know, the ones that pay their salaries

?

"Oh, sure, (b) is much easier! and also looks modern! That's evolution!"

They will never choose B. B means they make a mistake decision. As decision maker, This is unacceptable.
The most likely thing is, to do more on this way and cause a lot of facts to force you to change.
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 5:54 PM   in response to: wenjie zhou in response to: wenjie zhou
They will never choose B. B means they make a mistake decision. As decision maker, This is unacceptable.

No... (a) would mean that they took the wrong way back then when they decided by the ARC way.

(b) means: "We don't care if we are going to break your 2 million LOC code. Just deal with it."

Of course nobody will ever choose to rewrite such a big code base to comply with ARC. You just rewrite it using something else.
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 9:06 PM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
They will never choose B. B means they make a mistake decision. As decision maker, This is unacceptable.

No... (a) would mean that they took the wrong way back then when they decided by the ARC way.

(b) means: "We don't care if we are going to break your 2 million LOC code. Just deal with it."

Of course nobody will ever choose to rewrite such a big code base to comply with ARC. You just rewrite it using something else.

You are right. Sorry for my bad english. I say the opposite of what I mean. Yes. they will choose (a). Never (b).
For (b). VCL and Desktop become a obstacle for them. If there are not so many VCL applications . They really do not need "DisposeOf", which is an inexplicable existence even in ARC.
It's really great irony.
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 12:08 AM   in response to: wenjie zhou in response to: wenjie zhou
Today i lost again 10 min to understand


while MyObjectQueue.count > 0 do begin
aObject := MyObjectQueue.extract;
AObject.free; // << will not free now because strangely refcount = 2
aObject := nil; // << will not free now because strangely refcount = 2
end;

and also on each iteration, at the instruction aObject :=
MyObjectQueue.extract; we do around 10 _InstCopy, _InstClear,
atomicincrement, atomicdecrement, again atomicincrement,
atomicdecrement, again atomicincrement, atomicdecrement, etc...

i guess it's because the compiler create temp object

that an huge gain in speed and simplicity for sure !!

and their is still people to defend the ARC :( can't believe it !

wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 3:48 AM   in response to: loki loki in response to: loki loki
loki loki wrote:
Today i lost again 10 min to understand


while MyObjectQueue.count > 0 do begin
aObject := MyObjectQueue.extract;
AObject.free; // << will not free now because strangely refcount = 2
aObject := nil; // << will not free now because strangely refcount = 2
end;

and also on each iteration, at the instruction aObject :=
MyObjectQueue.extract; we do around 10 _InstCopy, _InstClear,
atomicincrement, atomicdecrement, again atomicincrement,
atomicdecrement, again atomicincrement, atomicdecrement, etc...

i guess it's because the compiler create temp object

that an huge gain in speed and simplicity for sure !!

and their is still people to defend the ARC :( can't believe it !


Think about "procedure of object", All of such "event" is really an object + function address. And such object should be add reference count too.
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 5:56 AM   in response to: loki loki in response to: loki loki
loki loki wrote:
Today i lost again 10 min to understand


while MyObjectQueue.count > 0 do begin
aObject := MyObjectQueue.extract;
AObject.free; // << will not free now because strangely refcount = 2
aObject := nil; // << will not free now because strangely refcount = 2
end;

and also on each iteration, at the instruction aObject :=
MyObjectQueue.extract; we do around 10 _InstCopy, _InstClear,
atomicincrement, atomicdecrement, again atomicincrement,
atomicdecrement, again atomicincrement, atomicdecrement, etc...

i guess it's because the compiler create temp object

that an huge gain in speed and simplicity for sure !!

and their is still people to defend the ARC :( can't believe it !


Works for me just fine.

I guess issue is in code you didn't posted here. What is exactly
MyObjectQueue type?

Again you fail to provide proper MCVE...

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
Paul TOTH

Posts: 196
Registered: 1/2/17
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 10:59 PM   in response to: Alexandre Machado in response to: Alexandre Machado
Le 29/03/2017 à 03:58, Alexandre Machado a écrit :
I guess that his personal opinions will not matter. It is too late
to turn the ship around.

Let me see here... what is easier?
a) Remove ARC from mobile which will force half dozen Delphi/mobile developers to rewrite their toy application
b) Force ARC down the throat of all other Win32/Win64 Delphi which have huge codebases, you know, the ones that pay their salaries

?

"Oh, sure, (b) is much easier! and also looks modern! That's evolution!"

PS: From Danny Thorpe, explaining why Borland failed (https://www.quora.com/Why-did-Borland-fail):

"In the height of the enterprise transformation, I asked Del Yocam, one of many interim CEOs after Kahn, "Are you saying you want to trade a million loyal $100 customers for a hundred $1 million customers?" Yocam replied without hesitation "Absolutely."

Now, Embarcadero/Idera wants to do exactly the opposite (although similar): Sacrifice all their paying customers to hunt 20 y.o. script kids that consider manual memory management so 90's... Good luck.
I do know the end result of that. And believe me, you will hear me saying "I told you", and of course you and Rudy and many others will deny it, as people always do when things go incredibly wrong.

note that they have the choice of adding non-ARC compiler for mobile
application with FMX that IS non-ARC compatible, or switching win32
compiler to ARC with VLC that is NOT ARC compatible.

Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 5:48 AM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
I guess that his personal opinions will not matter. It is too late
to turn the ship around.

Let me see here... what is easier?
a) Remove ARC from mobile which will force half dozen Delphi/mobile developers to rewrite their toy application
b) Force ARC down the throat of all other Win32/Win64 Delphi which have huge codebases, you know, the ones that pay their salaries

?

"Oh, sure, (b) is much easier! and also looks modern! That's evolution!"

PS: From Danny Thorpe, explaining why Borland failed (https://www.quora.com/Why-did-Borland-fail):

"In the height of the enterprise transformation, I asked Del Yocam, one of many interim CEOs after Kahn, "Are you saying you want to trade a million loyal $100 customers for a hundred $1 million customers?" Yocam replied without hesitation "Absolutely."

Now, Embarcadero/Idera wants to do exactly the opposite (although similar): Sacrifice all their paying customers to hunt 20 y.o. script kids that consider manual memory management so 90's... Good luck.
I do know the end result of that. And believe me, you will hear me saying "I told you", and of course you and Rudy and many others will deny it, as people always do when things go incredibly wrong.

In order to survive Delphi needs to attract new developers. Yes, transition
can be painful and I am sure that some developers would get lost along
the way.

I would also rather have million loyal $100 customers. That is why I am
constantly repeating that all Delphi SKU (including Starter) should support
all platforms out of the box without additional cost so more people could
get involved and use it without the need to buy expensive Enterprise SKU.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 7:07 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
In order to survive Delphi needs to attract new developers.

Of course.

But who on earth told you that the user base will grow if Win32/64 compilers become ARC-enabled?
Do you have any real trustworthy evidence for that or it is just your personal opinion?

You are just gambling. If/when this change happens lots of people will jump off the boat. And you really believe that an equal (or greater) number of developers used to free tools will start using Delphi because, you know, "now with ARC Delphi is cool again"?
Let me tell you one thing that might shock you: Delphi is not cool by any standards, doesn't matter how memory management is done.
By the way, C and C++ combined are still on the top. How on earth can these new developers, unable to handle manual memory management, cope with C???

Oh... but C and C++ allow you to plug-in a garbage collector in a library and voilá! EXACTLY where C/C++ did it right and Delphi did it wrong. Using C/C++ you can decide how memory management is done.
Delphi is (or should be!) a C/C++ contender not C#/Java/Python/Swift/whatever.

Why Delphi is not cool anymore? Well I can tell you why it was cool before:
- In the old days, doesn't matter which tool any other programmer was using, I knew that my compiled code was as fast as or faster than the rest.
Is it true anymore? No, it is not. Why it isn't? Because code generation is not important and compiler evolution stopped completely.
- In the old days, I knew that my application would use less memory than the rest.
I can't even say this nowadays with everything being bloated by a factor of 2 or more!

And finally (finally!), here is one thing that I believe firmly:

The "we want to attract new users by making our language look more modern"-stuff doesn't work either.

Guess who wrote this??? Does the name "Simon Kissel" ring the bell? He wrote that, 4 years ago. Unbelievable that we still have to repeat this.
Remember his alternative roadmap? Interesting to read it again, after 10 years.
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 10:41 PM   in response to: Alexandre Machado in response to: Alexandre Machado
Good talking.
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 29, 2017 11:39 PM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
In order to survive Delphi needs to attract new developers.

Of course.

But who on earth told you that the user base will grow if Win32/64 compilers become ARC-enabled?
Do you have any real trustworthy evidence for that or it is just your personal opinion?

Which part of "two memory management models in same code base don't work"
you didn't understand?

It is not question what will happen with Windows user base it is question about
other platforms user base. Other platforms cannot prosper until base frameworks
have to maintain non-ARC compatibility.

You are just gambling. If/when this change happens lots of people will jump off the boat. And you really believe that an equal (or greater) number of developers used to free tools will start using Delphi because, you know, "now with ARC Delphi is cool again"?
Let me tell you one thing that might shock you: Delphi is not cool by any standards, doesn't matter how memory management is done.
By the way, C and C++ combined are still on the top. How on earth can these new developers, unable to handle manual memory management, cope with C???

Oh... but C and C++ allow you to plug-in a garbage collector in a library and voilá! EXACTLY where C/C++ did it right and Delphi did it wrong. Using C/C++ you can decide how memory management is done.
Delphi is (or should be!) a C/C++ contender not C#/Java/Python/Swift/whatever.

I don't know how C++ is handling that, but I know that different memory management
systems require different coding patterns. You can plug in different memory manager
but that implies all libraries and all code must be able to handle such memory model.

That means separate ecosystems. Such thing may work in C++ with wast libraries
to choose from, but it is not viable solution for Delphi because of its smaller user base
and smaller number of 3rd party vendors.

Why Delphi is not cool anymore? Well I can tell you why it was cool before:
- In the old days, doesn't matter which tool any other programmer was using, I knew that my compiled code was as fast as or faster than the rest.
Is it true anymore? No, it is not. Why it isn't? Because code generation is not important and compiler evolution stopped completely.
- In the old days, I knew that my application would use less memory than the rest.
I can't even say this nowadays with everything being bloated by a factor of 2 or more!

It is not about being cool, it is about developing faster because you can focus
on your code instead of memory micromanagement.

Current bloat has nothing to do with ARC per-se.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 12:10 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
It is not about being cool, it is about developing faster because you can focus
on your code instead of memory micromanagement.

Nonsence. Without the quality, "faster" means produce waste faster. When themself can't guarantee their quality. How dare to say "developing faster because you can focus on your code."
The first thing is, Prove me that "you"(Embarcadero) can use ARC for developing faster with HIGH QUALITY. Let's us put the efficiency and bloated aside. Did ARC lead them produce more stable product ? NO. And the process lasted for several years. What's the wonderful ARC do? Does it really resolved memory leaks ? NO, Just because of ARC. There are new memory leaks. See: https://quality.embarcadero.com/browse/RSP-16985
By the way. Is manual memory management the most important bottle of modern program development? No, never.
When somebody don't know manual memory management, How can you expect him to produce a good product . Especially for server applications. We are not just make toys.

The members in Embarcadero are not the most clever mans in the world. Why they not just supply basic rules, and let the those more clever people do more for delphi. Just like FastMM, just like JCL.
Force the TObject ARC lead only one way to go. Old fashion TObject + IInterface make things with various possibilities .
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 11:13 AM   in response to: wenjie zhou in response to: wenjie zhou
Am 30.03.2017 um 10:50 schrieb wenjie zhou:
It is not about being cool, it is about developing faster because you can focus
on your code instead of memory micromanagement.

Nonsence. Without the quality, "faster" means produce waste faster. When themself can't guarantee their quality. How dare to say "developing faster because you can focus on your code."
The first thing is, Prove me that "you"(Embarcadero) can use ARC for developing faster with HIGH QUALITY. Let's us put the efficiency and bloated aside. Did ARC lead them produce more stable product ? NO. And the process lasted for several years. What's the wonderful ARC do? Does it really resolved memory leaks ? NO, Just because of ARC. There are new memory leaks. See: https://quality.embarcadero.com/browse/RSP-16985
By the way. Is manual memory management the most important bottle of modern program development? No, never.
When somebody don't know manual memory management, How can you expect him to produce a good product . Especially for server applications. We are not just make toys.

The members in Embarcadero are not the most clever mans in the world. Why they not just supply basic rules, and let the those more clever people do more for delphi. Just like FastMM, just like JCL.
Force the TObject ARC lead only one way to go. Old fashion TObject + IInterface make things with various possibilities .

Hello,

I don't really think the quality issues it currently has are because of
ARC. I rather think they're because of management and changing the
internal development model etc.
If they'd use really experienced developers there would be way less
failures in the product I think and that has ZERO to do with ARC. ARC is
only a straw man brought up by you and 1-2 others repeatedly until
nobody wants to read your posts anymore because they don't really
contain any new things besides repeating the same complaints over and over.

Wouldn't it be more wise to go back to your IDE and work on your GUI
controls library?

Greetings

Markus
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 10:45 PM   in response to: Markus Humm in response to: Markus Humm
Wouldn't it be more wise to go back to your IDE and work on your GUI
controls library?

GUI controls library? In your eyes, All development is GUI controls? I understand now why you think ARC is good.

" I think and that has ZERO to do with ARC"
Why ARC, you talk me that do not need care about memory. So, it's significant. And with such significant, IDE still can not with high quality.
How can you prove me the significant?

If they'd use really experienced developers there would be way less
failures in the product I think
If they have really experienced developers, Manual memory management will be a problem?
For other companies, with ARC, can they hire inexperienced development to produce high quality product?
Now, we come back, what does ARC do? Tell me please.
Lajos Juhasz

Posts: 755
Registered: 3/14/14
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 12:04 AM   in response to: wenjie zhou in response to: wenjie zhou
wenjie zhou wrote:

Wouldn't it be more wise to go back to your IDE and work on your GUI
controls library?

GUI controls library? In your eyes, All development is GUI controls?
I understand now why you think ARC is good.

" I think and that has ZERO to do with ARC"
Why ARC, you talk me that do not need care about memory. So, it's
significant. And with such significant, IDE still can not with high
quality. How can you prove me the significant?

If they'd use really experienced developers there would be way less
failures in the product I think
If they have really experienced developers, Manual memory management
will be a problem? For other companies, with ARC, can they hire
inexperienced development to produce high quality product? Now, we
come back, what does ARC do? Tell me please.

Ok you have different requirement than that EMB is targeting. But
please answer one really easy question what does all these message has
to do with a Firemoney a graphics library. This discussion doesn't
belong here. End of the story.

Ok I do understand that real developers are developing only web
servers. Unfortunately that also has nothing to do with FireMonkey.
Please also if ARC is so devil and that hard to do how does all the
developers using other languages manage it and how on earth a library
such as Indy was able to support the new compilers?

Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 8:44 AM   in response to: Lajos Juhasz in response to: Lajos Juhasz
Am 31.03.2017 um 09:04 schrieb Lajos Juhasz:


Ok you have different requirement than that EMB is targeting. But
please answer one really easy question what does all these message has
to do with a Firemoney a graphics library. This discussion doesn't
belong here. End of the story.

Ok I do understand that real developers are developing only web
servers. Unfortunately that also has nothing to do with FireMonkey.
Please also if ARC is so devil and that hard to do how does all the
developers using other languages manage it and how on earth a library
such as Indy was able to support the new compilers?


+1

Greetings

Markus
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 1:21 PM   in response to: Lajos Juhasz in response to: Lajos Juhasz
Lajos Juhasz wrote:

Please also if ARC is so devil and that hard to do how does all the
developers using other languages manage it and how on earth a library
such as Indy was able to support the new compilers?

Exactly. It is not the devil, it is only problematic for those who try
to work against it, out of some baseless belief that it should be
avoided.
--
Rudy Velthuis http://www.rvelthuis.de

"The world is a tragedy to those who feel, but a comedy to those
who think." -- Horace Walpole (1717-1797)
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 5:20 PM   in response to: Lajos Juhasz in response to: Lajos Juhasz

Ok you have different requirement than that EMB is targeting. But
please answer one really easy question what does all these message has
to do with a Firemoney a graphics library. This discussion doesn't
belong here. End of the story.

Firemoney ? Firemoney for Linux? TObject(ARC) only works for Firemoney ?

Please also if ARC is so devil and that hard to do how does all the
developers using other languages manage it and how on earth a library
such as Indy was able to support the new compilers?

I have no doubt about the use of ARC on Firemoney. But not TObject(ARC). Affect all types of programs.
I even suggest all objects(in Firemoney) has a same base class(Maybe name is TARCObject), Leave TObject there.
Lajos Juhasz

Posts: 755
Registered: 3/14/14
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 5:46 PM   in response to: wenjie zhou in response to: wenjie zhou
wenjie zhou wrote:


Ok you have different requirement than that EMB is targeting. But
please answer one really easy question what does all these message
has to do with a Firemoney a graphics library. This discussion
doesn't belong here. End of the story.

Firemoney ? Firemoney for Linux? TObject(ARC) only works for
Firemoney ?

I am sorry to inform you that at the moment there is no Firemonkey on
Linux. Delphi supports only console applications on Linux.

[snip]
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 6:18 PM   in response to: Lajos Juhasz in response to: Lajos Juhasz
Lajos Juhasz wrote:
wenjie zhou wrote:


Ok you have different requirement than that EMB is targeting. But
please answer one really easy question what does all these message
has to do with a Firemoney a graphics library. This discussion
doesn't belong here. End of the story.

Firemoney ? Firemoney for Linux? TObject(ARC) only works for
Firemoney ?

I am sorry to inform you that at the moment there is no Firemonkey on
Linux. Delphi supports only console applications on Linux.

[snip]

I am sorry too. Linux is ARC-enabled. No Firemoney , but ARC-enabled.
So, his talking has no persuasion

Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 11:40 PM   in response to: Lajos Juhasz in response to: Lajos Juhasz
Am 02.04.2017 um 02:46 schrieb Lajos Juhasz:
wenjie zhou wrote:


Ok you have different requirement than that EMB is targeting. But
please answer one really easy question what does all these message
has to do with a Firemoney a graphics library. This discussion
doesn't belong here. End of the story.

Firemoney ? Firemoney for Linux? TObject(ARC) only works for
Firemoney ?

I am sorry to inform you that at the moment there is no Firemonkey on
Linux. Delphi supports only console applications on Linux.

And even if it were different there were no Firemoney for Linux, unless
EMBT starts to implement some money management stuff.... ;-)

There might be Firemonkey some day though ;-)
(somebody recently even demonstrated that one can get some of the FMX
primitives like TRectangle to work on Linux)

Greetings

Markus
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 2, 2017 8:51 AM   in response to: wenjie zhou in response to: wenjie zhou
wenjie zhou wrote:


Ok you have different requirement than that EMB is targeting. But
please answer one really easy question what does all these message
has to do with a Firemoney a graphics library. This discussion
doesn't belong here. End of the story.

Firemoney ? Firemoney for Linux?

Any kind of money is welcome. It should not be burnt, though. <g>
--
Rudy Velthuis http://www.rvelthuis.de

"... one of the main causes of the fall of the Roman Empire was
that, lacking zero, they had no way to indicate successful
termination of their C programs." -- Robert Firth
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 3, 2017 1:27 PM   in response to: Rudy Velthuis (... in response to: Rudy Velthuis (...
Am 02.04.2017 um 17:51 schrieb Rudy Velthuis (TeamB):
wenjie zhou wrote:


Ok you have different requirement than that EMB is targeting. But
please answer one really easy question what does all these message
has to do with a Firemoney a graphics library. This discussion
doesn't belong here. End of the story.

Firemoney ? Firemoney for Linux?

Any kind of money is welcome. It should not be burnt, though. <g>

+1

Do you already know the story of the really washed money?
North Korean diplomants a few years ago brought quite a lot of 1 Dollar
notes back home (as their luggage won't be searched). The notes got
wahsed and reprinted as 100 Dollar notes as the size is the same and the
only real safety mechanism is the paper used.

It finally was revealed by some banker who had suspicions about some
note and when that was looked at in detail it was found out that the
printing quality was too good actually ;-)

Greetings

Markus
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 8:44 AM   in response to: wenjie zhou in response to: wenjie zhou
Am 31.03.2017 um 07:45 schrieb wenjie zhou:
Wouldn't it be more wise to go back to your IDE and work on your GUI
controls library?

GUI controls library? In your eyes, All development is GUI controls? I understand now why you think ARC is good.

No. But all you and loki loki seem to be able is to repeatedly whine
about how bad ARC is.

We got your opinion already by now.
Further complaining like that which doesn't really bring much new
concepts, ideas, solutions, points to the table is just a waste of time.

" I think and that has ZERO to do with ARC"
Why ARC, you talk me that do not need care about memory. So, it's significant. And with such significant, IDE still can not with high quality.
How can you prove me the significant?

Hm?
IDE is not using ARC. And many of the issues the IDE has are not memory
management related.


If they'd use really experienced developers there would be way less
failures in the product I think

If they have really experienced developers, Manual memory management will be a problem?
For other companies, with ARC, can they hire inexperienced development to produce high quality product?
Now, we come back, what does ARC do? Tell me please.

ARC makes memory management easier for some kinds of people. I do not
have problems with manual management, but I see some kinds of people who
have this and who see managing memory a waste of time. Not every trend
is a good one, but you might not be able to attract many young
developers if you don't try to make stuff as easy as possible. That's at
least how some folks see it. So tell me which JavaScript developer cares
about memory management?

The other alternative would be a GC but that would be worse.

Greetings

Markus
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 6:07 PM   in response to: Markus Humm in response to: Markus Humm
We got your opinion already by now.
Further complaining like that which doesn't really bring much new
concepts, ideas, solutions, points to the table is just a waste of time.

Nothing new. ARC exists for more than ten years in Delphi. Nothing new. Just bloated and slow down.

Hm?
IDE is not using ARC. And many of the issues the IDE has are not memory
management related.
OK, It's wonderful. In other word. New ARC exists for 5 years.
And no any product uses ARC for embarcadero themselves. Their compiler, their IDE, their tools.
But, force every customers to use ARC.


ARC makes memory management easier for some kinds of people.
Such people do not suitable for native programming. The best position for them is script development.

I do not have problems with manual management, but I see some kinds of people who have this and who see managing memory a waste of time.
Please calm down and have a think.
How much time you wasted for managing memory. If you spend a lot of time writing code instead of thinking. That's you problem.
Don't pass the buck to managing memory.

Not every trend is a good one, but you might not be able to attract many young
developers if you don't try to make stuff as easy as possible.

Who tells you young developers need that? For ARC, in many case, it seemed easy. But some case, you should deeply knows how ARC works, or you will failed,
in this case, ARC is more complex than manual managing memory.
e.g. GetMem and New for TSomeRecord, And in the TSomeRecord, it has a TObject member. (ARC) For GetMem, you should not FreeMem directly, you should set nil for the member.
If you calls New, you can Dispose directly. How do you have to explain for young developers. It's more complex.

Think about another example in book <Design Patterns>, Flyweight for word processor, A large number of small objects.
Just one traversal can cause serious performance problems. How can those young peoples know why?

And, on the other side. The compiler is used by companys, not those young peoples. No jobs, no new comers.

That's at least how some folks see it. So tell me which JavaScript developer cares
about memory management?
What is JavaScript ? JavaScript can works on different platform with no any change. No different binaries. Even Java can do this.
Can Delphi do this? I assume Delphi become the same easy with JavaScript . In this respect, Delphi will always behind JavaScript.
As a new comer, which one do you choose?

So, you don't catch the key point. As Native language. The best advantage is performance and less memory.
Native Delphi can reach the heights JavaScript will never reach. And sacrifice a lot of performance chasing easy goes wrong.
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 11:55 PM   in response to: wenjie zhou in response to: wenjie zhou
Am 02.04.2017 um 03:12 schrieb wenjie zhou:


Hm?
IDE is not using ARC. And many of the issues the IDE has are not memory
management related.
OK, It's wonderful. In other word. New ARC exists for 5 years.
And no any product uses ARC for embarcadero themselves. Their compiler, their IDE, their tools.
But, force every customers to use ARC.

Wrong. Their libraries do implement it so they have to use it. Just
Win32/VCL is not ARC enabled yet so the IDE cannot use it yet. But you
know this. Some of the tools they create use it as well, for instance
the client for mobile devices which can instantly display on a mobile
device what you do in form designer. I'm quite sure that's only one tool
where they use it.



ARC makes memory management easier for some kinds of people.
Such people do not suitable for native programming. The best position for them is script development.

Maybe. But neither of us has the numbers what percentage of the
newcomers would be affected.


I do not have problems with manual management, but I see some kinds of people who have this and who see managing memory a waste of time.
Please calm down and have a think.
How much time you wasted for managing memory. If you spend a lot of time writing code instead of thinking. That's you problem.
Don't pass the buck to managing memory.

Hm? I already did a fair amount of bug fixing in manual memory
management code, in quite some part due to sloppy programming of a way
older than me developer. Somew of that wouldn't have been necessary if
ARC would have been available.


Not every trend is a good one, but you might not be able to attract many young
developers if you don't try to make stuff as easy as possible.

Who tells you young developers need that?

Where are the numbers telling you that those youngsters happily long for
manual memory management? You don't have those and I don't have any
numbers myself, but in some cases I have eyes. Youngsters what to have
things working easy. ARC makes it easy for a lot of cases. That it makes
a few ones harder they will not notice at the start. That is a problem,
but it cannot be fixed easily as it would require that fresh programmers
have a complete mental image of that field right from the start in order
to pick manual memory management even if that requires to write a few
more lines here and there. But: programming is about knowing so many
things that having a complete mental image right from the start is illusory.


And, on the other side. The compiler is used by companys, not those young peoples. No jobs, no new comers.

No programmers for certain languages = company switches technology for
the next project to be able to hire new programmers. Means: development
environment needs to be attractive to youngsters. The compiler is not
used by a company, it is always used by the person sitting in front of
the compuert or having set up the automated build machine to use the
compiler.


That's at least how some folks see it. So tell me which JavaScript developer cares
about memory management?
What is JavaScript ? JavaScript can works on different platform with no any change. No different binaries.
Even Java can do this.

Only to a certain extend and not on all platforms.

Can Delphi do this? I assume Delphi become the same easy with JavaScript . In this respect, Delphi will always behind JavaScript.
As a new comer, which one do you choose?

So, you don't catch the key point. As Native language. The best advantage is performance and less memory.
Native Delphi can reach the heights JavaScript will never reach. And sacrifice a lot of performance chasing easy goes wrong.

I do get this point quite well, even if it doesn't looks like this to
you. But it's the same old issue over again: you don't have the numbers
about these youngster developers and their attitude but you think your
opinion about how the product should work is the only good one.

Greetings

Markus
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 2, 2017 1:19 AM   in response to: Markus Humm in response to: Markus Humm

I do get this point quite well, even if it doesn't looks like this to
you. But it's the same old issue over again: you don't have the numbers
about these youngster developers and their attitude but you think your
opinion about how the product should work is the only good one.

I really don't want to talk anymore. But for "the product should work is the only good one" is just you. What i want just ARC and NON_ARC.
Both of them. And you need only ARC. Who think the only good one? That's really funny.
C++ claims to be a multi paradigm programming language. Do not force user. Give the user full option. Who in the hell thinks only good one for Delphi?
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 2, 2017 4:29 AM   in response to: wenjie zhou in response to: wenjie zhou
Am 02.04.2017 um 10:19 schrieb wenjie zhou:

I do get this point quite well, even if it doesn't looks like this to
you. But it's the same old issue over again: you don't have the numbers
about these youngster developers and their attitude but you think your
opinion about how the product should work is the only good one.

I really don't want to talk anymore. But for "the product should work is the only good one" is just you. What i want just ARC and NON_ARC.
Both of them. And you need only ARC. Who think the only good one? That's really funny.
C++ claims to be a multi paradigm programming language. Do not force user. Give the user full option. Who in the hell thinks only good one for Delphi?

You can go down that route but that has consequences:

1. it needs more personal ressources and more testing because of the
added "modes"

2. more options makes it harder to get for new developers: which one
shall I choose? => needs more education/training

3. bigger code base has more chances to contain bugs

While I don't hink the current situation is good and having both options
available would be nice I don't think getting the latter is feasible.
I'm realist about that.

Greetings

Markus
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 2, 2017 5:17 AM   in response to: wenjie zhou in response to: wenjie zhou
wenjie zhou wrote:

I do get this point quite well, even if it doesn't looks like this to
you. But it's the same old issue over again: you don't have the numbers
about these youngster developers and their attitude but you think your
opinion about how the product should work is the only good one.

I really don't want to talk anymore. But for "the product should work is the only good one" is just you. What i want just ARC and NON_ARC.
Both of them. And you need only ARC. Who think the only good one? That's really funny.
C++ claims to be a multi paradigm programming language. Do not force user. Give the user full option. Who in the hell thinks only good one for Delphi?

You cannot have both of them. Period.

Current, cross-platform RTL/FMX "kind of works" code is broken
by design. It cannot be properly fixed as long as it has to work
on both ARC and non-ARC compilers.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
Paul TOTH

Posts: 196
Registered: 1/2/17
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 2, 2017 6:36 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Le 02/04/2017 à 14:17, Dalija Prasnikar a écrit :
wenjie zhou wrote:

I do get this point quite well, even if it doesn't looks like this to
you. But it's the same old issue over again: you don't have the numbers
about these youngster developers and their attitude but you think your
opinion about how the product should work is the only good one.

I really don't want to talk anymore. But for "the product should work is the only good one" is just you. What i want just ARC and NON_ARC.
Both of them. And you need only ARC. Who think the only good one? That's really funny.
C++ claims to be a multi paradigm programming language. Do not force user. Give the user full option. Who in the hell thinks only good one for Delphi?

You cannot have both of them. Period.

I didn't notice that you are in charge about that question.

Current, cross-platform RTL/FMX "kind of works" code is broken
by design. It cannot be properly fixed as long as it has to work
on both ARC and non-ARC compilers.
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 2, 2017 7:21 AM   in response to: Paul TOTH in response to: Paul TOTH
Paul TOTH wrote:
Le 02/04/2017 à 14:17, Dalija Prasnikar a écrit :
wenjie zhou wrote:

I do get this point quite well, even if it doesn't looks like this to
you. But it's the same old issue over again: you don't have the numbers
about these youngster developers and their attitude but you think your
opinion about how the product should work is the only good one.

I really don't want to talk anymore. But for "the product should work is the only good one" is just you. What i want just ARC and NON_ARC.
Both of them. And you need only ARC. Who think the only good one? That's really funny.
C++ claims to be a multi paradigm programming language. Do not force user. Give the user full option. Who in the hell thinks only good one for Delphi?

You cannot have both of them. Period.

I didn't notice that you are in charge about that question.

It is not a question whether I am in charge or not.

You all think I am saying things because I want them to be one
way or the other. I am merely stating some facts.

When ARC compiler got out I thought (just like devs at Embarcadero
thought) that with some modification in the old code you can make
it work across all compilers and all will live happy ever after.

Since then, I have gathered insights and experience and I can say
that is not possible in reality. Yes, code that must work under ARC
could be patched to the point it can work without major issues,
but catering for non-ARC compiler will constantly trow us curve
balls.

As long as we have two compilers, cross-compiler code will be total
and utmost crap. Yes I am exaggerating a bit here, but I feel it is
important to get that message through.

I don't know whether people in charge are already aware of above
situation or they are still fooling themselves, but eventually someone
will have to make tough choice and retire non-ARC compiler for good.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 2, 2017 9:03 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Dalija Prasnikar wrote:

Paul TOTH wrote:
Le 02/04/2017 à 14:17, Dalija Prasnikar a écrit :
wenjie zhou wrote:

I do get this point quite well, even if it doesn't looks like
this to >>> you. But it's the same old issue over again: you don't
have the numbers >>> about these youngster developers and their
attitude but you think your >>> opinion about how the product
should work is the only good one.

I really don't want to talk anymore. But for "the product should
work is the only good one" is just you. What i want just ARC and
NON_ARC. >> Both of them. And you need only ARC. Who think the
only good one? That's really funny. >> C++ claims to be a multi
paradigm programming language. Do not force user. Give the user
full option. Who in the hell thinks only good one for Delphi?

You cannot have both of them. Period.

I didn't notice that you are in charge about that question.

It is not a question whether I am in charge or not.

You all think I am saying things because I want them to be one
way or the other. I am merely stating some facts.

Exactly.

--
Rudy Velthuis http://www.rvelthuis.de

"In theory, theory and practice are the same. In practice,
they’re not."
-- Yogi Berra
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 2, 2017 9:03 AM   in response to: Paul TOTH in response to: Paul TOTH
Paul TOTH wrote:

Le 02/04/2017 à 14:17, Dalija Prasnikar a écrit :
wenjie zhou wrote:

I do get this point quite well, even if it doesn't looks like
this to >>> you. But it's the same old issue over again: you don't
have the numbers >>> about these youngster developers and their
attitude but you think your >>> opinion about how the product should
work is the only good one.

I really don't want to talk anymore. But for "the product should
work is the only good one" is just you. What i want just ARC and
NON_ARC. >> Both of them. And you need only ARC. Who think the only
good one? That's really funny. >> C++ claims to be a multi paradigm
programming language. Do not force user. Give the user full option.
Who in the hell thinks only good one for Delphi?

You cannot have both of them. Period.

I didn't notice that you are in charge about that question.

She is not "in charge", but she and others, me included, have
explained, many times already, that IT IS NOT VIABLE. It will simply
not happen. So people can keep on nagging and nagging and ignore the
simple facts, but that won't change anything. It will not happen.

Oh, Apple can do it. But their objects have been reference counted
MANUALLY for ages already, and you can still link automatic RC with
manual RC code. But for Delphi, it is not possible, unless they'd
completely rewrite EVERY library (and 3rd parties too) to use reference
counting EVERYWHERE. Then you can really throw away your legacy Delphi
code. ALL of it.

So no, it won't happen. Forget it. Not because I am "in charge", but
simply because I understand why it won't happen.

--
Rudy Velthuis http://www.rvelthuis.de

"if you keep doing what you've always done, you'll keep getting
what you've always gotten."
-- Jim Rohn
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 2, 2017 8:56 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Dalija Prasnikar wrote:

wenjie zhou wrote:

I do get this point quite well, even if it doesn't looks like
this to you. But it's the same old issue over again: you don't
have the numbers about these youngster developers and their
attitude but you think your opinion about how the product should
work is the only good one.

I really don't want to talk anymore. But for "the product should
work is the only good one" is just you. What i want just ARC and
NON_ARC.

You cannot have both of them. Period.

I think some people read that, but simply ignore it, because they don't
like it. It has been explained many times already, but somehow they
don't care about the arguments and keep on nagging and nagging and
nagging ad nauseam.

Before, it was the same with the "switch" for Unicode. People simply
did not want to understand the arguments.

Well, I hope that one day, reality will slowly enter their brains. <g>
--
Rudy Velthuis http://www.rvelthuis.de

"Good code is short, simple, and symmetrical - the challenge
is figuring out how to get there." -- Sean Parent
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 2, 2017 8:53 AM   in response to: wenjie zhou in response to: wenjie zhou
wenjie zhou wrote:

And no any product uses ARC for embarcadero themselves. Their
compiler, their IDE, their tools.

Well, the tools are Windows-based and either non-GUI, or they are VCL.
And there is no Delphi with ARC on Windows yet.

Would you like that?

--
Rudy Velthuis http://www.rvelthuis.de

"I never forget a face, but in your case I'll be glad to make an
exception." -- Groucho Marx
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit] [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 3, 2017 1:30 PM   in response to: Rudy Velthuis (... in response to: Rudy Velthuis (...
Am 02.04.2017 um 17:53 schrieb Rudy Velthuis (TeamB):
wenjie zhou wrote:

And no any product uses ARC for embarcadero themselves. Their
compiler, their IDE, their tools.

Well, the tools are Windows-based and either non-GUI, or they are VCL.
And there is no Delphi with ARC on Windows yet.

Would you like that?

Like I already explained: there are already at least some tools for
other platforms than Windows which use FMX and need to use ARC.

Greetings

Markus
Paul TOTH

Posts: 196
Registered: 1/2/17
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 12:15 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Le 30/03/2017 à 08:39, Dalija Prasnikar a écrit :
Alexandre Machado wrote:
In order to survive Delphi needs to attract new developers.

Of course.

But who on earth told you that the user base will grow if Win32/64 compilers become ARC-enabled?
Do you have any real trustworthy evidence for that or it is just your personal opinion?

Which part of "two memory management models in same code base don't work"
you didn't understand?

fine, if it's not possible, let's remove ARC.

It is not question what will happen with Windows user base it is question about
other platforms user base. Other platforms cannot prosper until base frameworks
have to maintain non-ARC compatibility.

You think that other platforms cannot prosper until they use the SAME
base frameworks, so again, let's remove ARC


You are just gambling. If/when this change happens lots of people will jump off the boat. And you really believe that an equal (or greater) number of developers used to free tools will start using Delphi because, you know, "now with ARC Delphi is cool again"?
Let me tell you one thing that might shock you: Delphi is not cool by any standards, doesn't matter how memory management is done.
By the way, C and C++ combined are still on the top. How on earth can these new developers, unable to handle manual memory management, cope with C???

Oh... but C and C++ allow you to plug-in a garbage collector in a library and voilá! EXACTLY where C/C++ did it right and Delphi did it wrong. Using C/C++ you can decide how memory management is done.
Delphi is (or should be!) a C/C++ contender not C#/Java/Python/Swift/whatever.

I don't know how C++ is handling that, but I know that different memory management
systems require different coding patterns. You can plug in different memory manager
but that implies all libraries and all code must be able to handle such memory model.

you don't give good argument pro ARC, you tell that you didn't claim
ARC, so let's remove it and you'll have a single framework everywhere.

That means separate ecosystems. Such thing may work in C++ with wast libraries
to choose from, but it is not viable solution for Delphi because of its smaller user base
and smaller number of 3rd party vendors.

Why Delphi is not cool anymore? Well I can tell you why it was cool before:
- In the old days, doesn't matter which tool any other programmer was using, I knew that my compiled code was as fast as or faster than the rest.
Is it true anymore? No, it is not. Why it isn't? Because code generation is not important and compiler evolution stopped completely.
- In the old days, I knew that my application would use less memory than the rest.
I can't even say this nowadays with everything being bloated by a factor of 2 or more!

It is not about being cool, it is about developing faster because you can focus
on your code instead of memory micromanagement.

BTW I wonder why old style code could mix TObject and Interface that do
not use the same memory management while new code couldn't mix TObject
and TManagedObject for instance that would use ARC.

Current bloat has nothing to do with ARC per-se.

so what is it about ?
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 12:22 AM   in response to: Paul TOTH in response to: Paul TOTH
Current bloat has nothing to do with ARC per-se.

so what is it about ?

Good question.
Let me guess, "Current bloat" because of "generic", Why "generic"? It because of "ARC".
Lars Fosdal


Posts: 156
Registered: 10/26/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 4:47 AM   in response to: wenjie zhou in response to: wenjie zhou
wenjie zhou wrote:
Current bloat has nothing to do with ARC per-se.

so what is it about ?

Good question.
Let me guess, "Current bloat" because of "generic", Why "generic"? It because of "ARC".

Generics and ARC are not related at all.

--
http://plus.lars.fosdal.com
Delphi Developers Google+ Community: https://plus.google.com/communities/103113685381486591754 (8500+ members)
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 4:59 AM   in response to: Lars Fosdal in response to: Lars Fosdal
Current bloat has nothing to do with ARC per-se.

so what is it about ?

Good question.
Let me guess, "Current bloat" because of "generic", Why "generic"? It because of "ARC".

Generics and ARC are not related at all.
You don't know we discuss in another post. Now, please try to answer a question: Why replace TList with TList<T> in classes.pas ?
TList<T> is slower and bloated than TList. And TList works perfectly in classea.pas more than ten years. Why replace it with TList<T>?
With a variety of T. Code bloated. With different TList<T>, RTTI for those types bloated. What's the benifit?
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 11:09 AM   in response to: wenjie zhou in response to: wenjie zhou
Am 30.03.2017 um 13:59 schrieb wenjie zhou:
Current bloat has nothing to do with ARC per-se.

so what is it about ?

Good question.
Let me guess, "Current bloat" because of "generic", Why "generic"? It because of "ARC".

Generics and ARC are not related at all.
You don't know we discuss in another post. Now, please try to answer a question: Why replace TList with TList<T> in classes.pas ?
TList<T> is slower and bloated than TList. And TList works perfectly in classea.pas more than ten years. Why replace it with TList<T>?
With a variety of T. Code bloated. With different TList<T>, RTTI for those types bloated. What's the benifit?

You have already been told, but since you don't really want to listen to
the arguments of the other side either we need to tell you again:

- TList<T> provides more functionality than TList. Some may like to use
this and some don't have any need for it.
- TList<T> is type safe, TList is not.
- One of the strenghts of Delphi is type safety as it removes certain
kinds of programming errors

But I'm quite sure I didn't tell you anything new and that you'll keep
on with your useless rants. If that RTTI is of any use here I can't
really tell, I just don't know.

Greetings

Markus
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 12:59 PM   in response to: Markus Humm in response to: Markus Humm
- TList<T> provides more functionality than TList. Some may like to use
this and some don't have any need for it.

I don't have anything against generics or TList<T>. What really pisses me off is deprecating one of the most basic core classes of RTL/VCL and force everyone to use a bloated one.

Also, are you writing new code? If yes, then sure, go ahead and use TList<T>. Do you have old code which uses TList and has been working perfectly for the last 10 years, why should you replace it with a TList<T>?

a) Make it slower and bloated
b) Introduce new bugs in a tested, bug free code
c) Spend uncountable man hours which are really scarce
d) All answers above?

EMBT answer is (d), believe it or not!!
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 10:48 PM   in response to: Markus Humm in response to: Markus Humm
Tell me why replace TList with TList<T> in classes.pas.

In my opinion, you and me both want Delphi to be better. But you just see some benefit. Did not try to think deeply about the problems it brings.
Another question, do you really use IInterface to supply ARC by own in your applications? If you do. You will understand, it's no need to make all TObject to be ARC.

Someone had mentioned COM on Windows is ARC. Let me guess what language develop COM. C++ (non-ARC) but can do ARC realization.
ARC is really really not an new thing. And ARC exits for long time in Delphi. And now decision just remove non-ARC form Delphi.
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 1:29 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
It is not question what will happen with Windows user base it is question about
other platforms user base. Other platforms cannot prosper until base frameworks
have to maintain non-ARC compatibility.

you really think that is removing non-arc support from delphi that it's
will make it prospering ? i don't know why me i think it's correcting as
fast as possible all the bugs that will make it growing. and their is
zero bug reported by client because of non-arc code contrary to all the
bugs reported BECAUSE of arc ;)


That means separate ecosystems. Such thing may work in C++ with wast libraries
to choose from, but it is not viable solution for Delphi because of its smaller user base
and smaller number of 3rd party vendors.

exactly, i m curious how many years (decades?) it's will take to see
library like jvcl fully support arc, especially when now close to only
one or 2 developers continue to maintain it (thank to them by the way)!
don't know if they will find the time to redevelop/re debug everything,
time where they are not payed for that, and especially on code that was
not even made by themself


Why Delphi is not cool anymore? Well I can tell you why it was cool before:
- In the old days, doesn't matter which tool any other programmer was using, I knew that my compiled code was as fast as or faster than the rest.
Is it true anymore? No, it is not. Why it isn't? Because code generation is not important and compiler evolution stopped completely.

ARC will not help this by making the compiled code exactly 10x more slow
than the equivalent non arc code

- In the old days, I knew that my application would use less memory than the rest.

unicode migration didn't help this too :)

It is not about being cool, it is about developing faster because you can focus
on your code instead of memory micromanagement.

no, you don't developpe faster with arc, you just avoid some try ...
finally .free end that all, but you will loose much much more time than
the 2 secondes to write these lines by always thinking if you don't have
a circular reference somewhere (really really easy to have with
anonymous method)
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 11:10 AM   in response to: loki loki in response to: loki loki
Am 30.03.2017 um 10:29 schrieb loki loki:
It is not question what will happen with Windows user base it is question about
other platforms user base. Other platforms cannot prosper until base frameworks
have to maintain non-ARC compatibility.

you really think that is removing non-arc support from delphi that it's
will make it prospering ? i don't know why me i think it's correcting as
fast as possible all the bugs that will make it growing. and their is
zero bug reported by client because of non-arc code contrary to all the
bugs reported BECAUSE of arc ;)

I'm quite sure the IDE bugs and some of the other bugs I reported are
not because of ARC...
...and I did report quite a number of them!

Greetings

Markus
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 11:04 PM   in response to: Markus Humm in response to: Markus Humm
I'm quite sure the IDE bugs and some of the other bugs I reported are
not because of ARC...
...and I did report quite a number of them!

I'm quite sure. ARC can not help them to increase quality.
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 4:05 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Which part of "two memory management models in same code base don't work"
you didn't understand?

It is not question what will happen with Windows user base it is question about
other platforms user base. Other platforms cannot prosper until base frameworks
have to maintain non-ARC compatibility.

So you are confirming everything that we (the non-ARC people) have been saying for years:
"A huge code base designed for manual memory management cannot be easily ported to ARC model"
So in your infinite wisdom you actually think that EMBT should drop manual memory management which, - I'm deeply sure -
is what 98% of all Delphi code out there is written for, to allow the other 2% to flourish?? Really?
And you call this "evolution"?
Lars Fosdal


Posts: 156
Registered: 10/26/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 30, 2017 11:19 PM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
Which part of "two memory management models in same code base don't work"
you didn't understand?

It is not question what will happen with Windows user base it is question about
other platforms user base. Other platforms cannot prosper until base frameworks
have to maintain non-ARC compatibility.

So you are confirming everything that we (the non-ARC people) have been saying for years:
"A huge code base designed for manual memory management cannot be easily ported to ARC model"
So in your infinite wisdom you actually think that EMBT should drop manual memory management which, - I'm deeply sure -
is what 98% of all Delphi code out there is written for, to allow the other 2% to flourish?? Really?
And you call this "evolution"?

Dude, you are taking one statement and turning into something different.

"two memory management models in same code base don't work"
does NOT translate into
"A huge code base designed for manual memory management cannot be easily ported to ARC model"

The porting is not really that complicated. Trying to cater for both models in the same sourcecode, is.

--
http://plus.lars.fosdal.com
Delphi Developers Google+ Community: https://plus.google.com/communities/103113685381486591754 (8500+ members)
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 12:14 AM   in response to: Lars Fosdal in response to: Lars Fosdal
Dude, you are taking one statement and turning into something different.

"two memory management models in same code base don't work"
does NOT translate into
"A huge code base designed for manual memory management cannot be easily ported to ARC model"

The porting is not really that complicated. Trying to cater for both models in the same sourcecode, is.

So explain to me, please: How can be that porting is easy but maintaining (what already works) is hard?
Dude, by your own words: changing and retesting 100% of your code base is easy, while leaving it untouched is hard? LOL
Their code is already compatible with 2 memory models. I'm not asking them do even touch the code. You are asking them to
refactor 100% of RTL and VCL, plus test 100% of the refactored code, but you truly believe that what you ask is easy, and I'm the one asking for something impossible?
Lars Fosdal


Posts: 156
Registered: 10/26/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 12:55 AM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
Dude, you are taking one statement and turning into something different.

"two memory management models in same code base don't work"
does NOT translate into
"A huge code base designed for manual memory management cannot be easily ported to ARC model"

The porting is not really that complicated. Trying to cater for both models in the same sourcecode, is.

So explain to me, please: How can be that porting is easy but maintaining (what already works) is hard?
Dude, by your own words: changing and retesting 100% of your code base is easy, while leaving it untouched is hard? LOL
Their code is already compatible with 2 memory models. I'm not asking them do even touch the code. You are asking them to
refactor 100% of RTL and VCL, plus test 100% of the refactored code, but you truly believe that what you ask is easy, and I'm the one asking for something impossible?

Again, twisting statements is a really lousy way of constructing an argument.

What I am saying is that I don't want to have to care about the life cycle model. I just want to deal with ONE model, across all platforms.
How they do it is up to them. If they want to have the double model in the same source base - I assume we won't see the full ARC benefits, because that means the superflous try/finallys have to remain in many places.
Of the two models, ARC would be my preference, and yes, it would mean refactoring the whole codebase.
We disagree, let's leave it at that.

--
http://plus.lars.fosdal.com
Delphi Developers Google+ Community: https://plus.google.com/communities/103113685381486591754 (8500+ members)
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 4:09 AM   in response to: Lars Fosdal in response to: Lars Fosdal
Lars Fosdal wrote:
Alexandre Machado wrote:
Dude, you are taking one statement and turning into something different.

"two memory management models in same code base don't work"
does NOT translate into
"A huge code base designed for manual memory management cannot be easily ported to ARC model"

The porting is not really that complicated. Trying to cater for both models in the same sourcecode, is.

So explain to me, please: How can be that porting is easy but maintaining (what already works) is hard?
Dude, by your own words: changing and retesting 100% of your code base is easy, while leaving it untouched is hard? LOL
Their code is already compatible with 2 memory models. I'm not asking them do even touch the code. You are asking them to
refactor 100% of RTL and VCL, plus test 100% of the refactored code, but you truly believe that what you ask is easy, and I'm the one asking for something impossible?

Again, twisting statements is a really lousy way of constructing an argument.

What I am saying is that I don't want to have to care about the life cycle model. I just want to deal with ONE model, across all platforms.
How they do it is up to them. If they want to have the double model in the same source base - I assume we won't see the full ARC benefits, because that means the superflous try/finallys have to remain in many places.
Of the two models, ARC would be my preference, and yes, it would mean refactoring the whole codebase.
We disagree, let's leave it at that.

+1

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 3, 2017 5:33 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Again, twisting statements is a really lousy way of constructing an argument.

+1

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar

Dalija,

here in this same thread you mentioned comments I did in another thread (to be precise in a private area), completely out of the context, to make your point. I wonder if this can be considered also "statement twist", or this rule is only valid for me....
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 4, 2017 5:12 AM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
Again, twisting statements is a really lousy way of constructing an argument.

+1

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar

Dalija,

here in this same thread you mentioned comments I did in another thread (to be precise in a private area), completely out of the context, to make your point. I wonder if this can be considered also "statement twist", or this rule is only valid for me....

What your comments exactly did I twist?

It is hard to follow such long conversations that spread along several
threads. It is possible that I misunderstood or interpreted wrongly.

But when you directly quote my words and then make something else
out of them I can hardly treat that as honest mistake.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 7:51 PM   in response to: Lars Fosdal in response to: Lars Fosdal
Again, twisting statements is a really lousy way of constructing an argument.

What I am saying is that I don't want to have to care about the life cycle model. I just want to deal with ONE model, across all platforms.
How they do it is up to them. If they want to have the double model in the same source base - I assume we won't see the full ARC benefits, because that means the superflous try/finallys have to remain in many places.
Of the two models, ARC would be my preference, and yes, it would mean refactoring the whole codebase.
We disagree, let's leave it at that.

Do you want care about how many memory your app used and waste? Do you care efficiency of you app? Not every thing should solved by compiler.
If you think auto life cycle is so important , there's a lot of dynamic languages you can chose, They do not have the advantage of efficiency and memory utilization.
Because these languages are not designed for this purpose at the beginning of the design.

Once again, As a native language. what's the most advantage compared with such dynamic languages.
To sacrifice performance to study those dynamic languages. You will foud Delphi become useless.
Reasons:
As native, Delphi had to compiler & link for different platform, and deploy many binaries. As dynamic languages , they do not need any chang. That's the huge disparity. And that is insurmountable barrier .
So, what's advantage for the native? Without doubt, performance and memory.
"Your preference" would heart Delphi deeply.

If I remember correctly, IInterface was introduced in Delphi3. And we deal with two mode(ARC+non_ARC) more than ten years.
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 11:11 PM   in response to: wenjie zhou in response to: wenjie zhou
Am 01.04.2017 um 04:52 schrieb wenjie zhou:

As native, Delphi had to compiler & link for different platform, and deploy many binaries. As dynamic languages , they do not need any chang. That's the huge disparity. And that is insurmountable barrier .
So, what's advantage for the native? Without doubt, performance and memory.

What has ARC to do with memory usage?
You can still control that.
Performance: maybe yes because the reference counting mechanism needs to
be protected.

Greetings

Markus
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 11:48 PM   in response to: Markus Humm in response to: Markus Humm
What has ARC to do with memory usage?
You can still control that.

DisposeOf not free memory. How can you control that?
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 4:59 AM   in response to: wenjie zhou in response to: wenjie zhou
wenjie zhou wrote:
What has ARC to do with memory usage?
You can still control that.

DisposeOf not free memory. How can you control that?

By fixing your code.

If DisposeOf is used properly and in right place memory will
be released. If not you have bug you have to fix.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 5:32 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
DisposeOf not free memory. How can you control that?

By fixing your code.

If DisposeOf is used properly and in right place memory will
be released. If not you have bug you have to fix.

How can i know that? How can i detect it?
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 6:06 AM   in response to: wenjie zhou in response to: wenjie zhou
wenjie zhou wrote:
DisposeOf not free memory. How can you control that?

By fixing your code.

If DisposeOf is used properly and in right place memory will
be released. If not you have bug you have to fix.

How can i know that? How can i detect it?

Well, that is the harder part on non-Windows platforms. There are
different tools to find memory leaks and it depends on the platform.

If you are suspecting specific instance you can also use another
reference with unsafe attribute to inspect desired instance through
debugger. If memory is cleared then all contained managed references
will also be cleared - strings for instance.

I have only barely touched real memory leak inspection so I cannot
give you full report on what and how.

So far the easiest way to find leaks (for me) was on Linux with Valgrind.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 6:11 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Dalija Prasnikar wrote:
wenjie zhou wrote:
DisposeOf not free memory. How can you control that?

By fixing your code.

If DisposeOf is used properly and in right place memory will
be released. If not you have bug you have to fix.

How can i know that? How can i detect it?

Well, that is the harder part on non-Windows platforms. There are
different tools to find memory leaks and it depends on the platform.

If you are suspecting specific instance you can also use another
reference with unsafe attribute to inspect desired instance through
debugger. If memory is cleared then all contained managed references
will also be cleared - strings for instance.

I have only barely touched real memory leak inspection so I cannot
give you full report on what and how.

So far the easiest way to find leaks (for me) was on Linux with Valgrind.

But for non-ARC, we do not need such tools. We can detect it in Delphi.
ARC lead location error more difficult.
It will be very difficult for you to locate the reference number if it does not match
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 6:19 AM   in response to: wenjie zhou in response to: wenjie zhou
wenjie zhou wrote:

But for non-ARC, we do not need such tools. We can detect it in Delphi.
ARC lead location error more difficult.
It will be very difficult for you to locate the reference number if it does not match

You would not need such tools either if all compilers would be ARC.
It is far easier to find reference cycles than leaks from zombies.

This is part that can be improved. Core problem is not ARC itself but
detection tools. You would have equal issues with manual memory
management if you would not have leak tools on your disposal.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 6:47 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
You would not need such tools either if all compilers would be ARC.

It is far easier to find reference cycles than leaks from zombies.
Show me. Mobile is ARC only.

You would have equal issues with manual memory
management if you would not have leak tools on your disposal.
It's only about leak. But if the reference count is smaller than expected. And the object be freed premature.
You will hard to check it.
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 7:06 AM   in response to: wenjie zhou in response to: wenjie zhou
wenjie zhou wrote:
You would not need such tools either if all compilers would be ARC.

It is far easier to find reference cycles than leaks from zombies.
Show me. Mobile is ARC only.

Again, you will depend on native memory tools that will show your
memory usage is increasing. There are also tools for reference cycle
leak checks. Like https://bitbucket.org/shadow_cs/delphi-leakcheck

You would have equal issues with manual memory
management if you would not have leak tools on your disposal.
It's only about leak. But if the reference count is smaller than expected. And the object be freed premature.
You will hard to check it.

Again, it is not ARC that is a problem. You would have same issues
finding leaks under manual memory management on other platforms.

And if your reasoning goes along the lines but I can debug that on Windows
then having Windows ARC compiler is equally good solution to the problem.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 7:35 AM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Again, you will depend on native memory tools that will show your
memory usage is increasing. There are also tools for reference cycle
leak checks. Like https://bitbucket.org/shadow_cs/delphi-leakcheck

Thank you. I will study it.

Again, it is not ARC that is a problem. You would have same issues
finding leaks under manual memory management on other platforms.

I guess you mean wild pointer.

Finally, it still can not proved ARC is more better. Slow down is without doubt. Break compatibility still there.
If there is a large number of small objects, just a cycle can cause catastrophic performance barriers.
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 3:38 AM   in response to: Markus Humm in response to: Markus Humm
What has ARC to do with memory usage?

ahahah because arc make memory leak :)

More seriously because arc is responsible of freeing the object (when
reference count reach 0), not you anymore (even via .free) and this is a
big change! of course you can still play with disposeOF but this
requiered to update all your previous code

by the way you will be also suprised to know how the [weak] is
implemented. in fact delphi keep in memory the list of all weak pointer,
so using weak to remove the refcount can even be more worse than keeping
the refcount

Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 10:34 AM   in response to: loki loki in response to: loki loki
Am 01.04.2017 um 12:38 schrieb loki loki:
What has ARC to do with memory usage?

ahahah because arc make memory leak :)

More seriously because arc is responsible of freeing the object (when
reference count reach 0), not you anymore (even via .free) and this is a
big change! of course you can still play with disposeOF but this
requiered to update all your previous code

by the way you will be also suprised to know how the [weak] is
implemented. in fact delphi keep in memory the list of all weak pointer,
so using weak to remove the refcount can even be more worse than keeping
the refcount


Hm? You simply need to know who's holding a strong reference and those
need to free it someday.

Greetings

Markus
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 1, 2017 11:51 AM   in response to: Markus Humm in response to: Markus Humm

Hm? You simply need to know who's holding a strong reference and those
need to free it someday.

if it's was so easy ... the problem that on the paper, on pretty simple
scenario ARC look very simple. but in reality it's absolutely not the case !

now look this bug in delphi who make access violation when you use TEdit
under ios, off course all of this because of ARC :

https://quality.embarcadero.com/browse/RSP-14086

and start to have a head hake to understand where the bug is (solution
is even written)

Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 3, 2017 5:39 PM   in response to: Markus Humm in response to: Markus Humm
What has ARC to do with memory usage?

It is not ARC per se. But the whole ARC snowball: ARC forced some core changes in VCL. In the end: Your application performs poorer and the memory requirements are higher.
By how much? Nobody actually knows, but some very basic tests here and there are worrying. But the "Let's move to ARC" people just ignores everything. They are basically gambling.
The very core point is: Who wants to pay for a new IDE and compiler that will make your application slower and more memory hungry? Your customers don't care if you use ARC or have to track pointers in your mind.
So good luck trying to make your customers see that your product version 2.0 is slower and requires more memory, but they should be happy because now your product has ARC.
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 8, 2017 8:03 AM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
What has ARC to do with memory usage?

It is not ARC per se. But the whole ARC snowball: ARC forced some core changes in VCL. In the end: Your application performs poorer and the memory requirements are higher.
By how much? Nobody actually knows, but some very basic tests here and there are worrying. But the "Let's move to ARC" people just ignores everything. They are basically gambling.

Everything is a gamble, otherwise COBOL would still be used at
large.

Memory management models are not one size fits all. I believe
you were the one mentioned having more 100$ customers is better
than having few 10000$ customers.

Platforms like iOS and Android are exactly the platforms where there
is possibility for small customers growth. And those are also platforms
where all native tools have some sort of automatic memory management.

Number of Windows applications out there also would not suffer from having
ARC. So besides "changing code" factor, in reality only smaller fragment of
Delphi target applications would suffer from performance issues with ARC
and for all others automatic memory management would be beneficial.

Not because we don't know how to handle memory management, but because
we could develop applications faster if we don't have to fiddle with it.

My Delphi 7 application upgraded to Delphi XE doubled in exe
size and memory requirements also grew. It was obviously
working fine without Unicode strings and enhanced RTTI that
contributed to bloat. It was also working fine without generics.

Initial cost of porting was pure waste of time. When I was finished
the port and was happy about it, my boss asked am I nuts or what
because I spent time working on something and as end result I
had application that did nothing new and doubled its resource
requirements.

My response was that port will enable me to use some new features
and that I will be able to implement new features faster and more reliable.
And that growth of my application needs is nothing comparing to
growth of computer processing powers during last 10 years since
app was released. And that soon proved to be true.

Sure, I still think there is plenty of room for various optimizations
that would have impact on my app size and speed, but none of these
are true showstoppers.

And for some apps I still use Delphi 7. Not because I could not port my code
(I had) but because for them increase in size (and performance) was relevant.

The very core point is: Who wants to pay for a new IDE and compiler that will make your application slower and more memory hungry? Your customers don't care if you use ARC or have to track pointers in your mind.
So good luck trying to make your customers see that your product version 2.0 is slower and requires more memory, but they should be happy because now your product has ARC.

Until you cannot see the benefits of bringing your code to ARC then
don't. Do it sequentially and when you are ready. Rome wasn't built in a day.
Non-ARC compilers are not going anywhere any time soon.

And number of 3rd party vendors already have updated their libraries with
Tokyo Linux support. I guess porting it is not that hard after all.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
wenjie zhou

Posts: 417
Registered: 6/28/02
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 9, 2017 7:36 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
Until you cannot see the benefits of bringing your code to ARC then
don't. Do it sequentially and when you are ready. Rome wasn't built in a day.
Non-ARC compilers are not going anywhere any time soon.

Non-ARC compiler is a complete pseudo proposition. ARC for what? For objects. For records? For Integer?
ARC should not exist in compiler. The smart pointer in C++ has been the perfect interpretation of all this.
Rely on the compiler to realization ARC only prove the ability to express this language is not good.

C has no class/object. So is there a ARC compiler for C?
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :( [Edit]
Click to report abuse...   Click to reply to this thread Reply
  Posted: Apr 10, 2017 5:29 AM   in response to: wenjie zhou in response to: wenjie zhou
wenjie zhou wrote:

Until you cannot see the benefits of bringing your code to ARC then
don't. Do it sequentially and when you are ready. Rome wasn't built
in a day. Non-ARC compilers are not going anywhere any time soon.

Non-ARC compiler is a complete pseudo proposition. ARC for what? For
objects. For records? For Integer? ARC should not exist in compiler.
The smart pointer in C++ has been the perfect interpretation of all
this.

The smart pointer in C++ relies on the compiler too. The compiler must
insert code that calls the destructor when a class (or struct a.k.a
record) leaves scope. That is more or less what ARC does too.

--
Rudy Velthuis http://www.rvelthuis.de

"Giving money and power to government is like giving whiskey
and car keys to teenage boys."
-- P. J. O'Rourke
Arthur Hoornweg

Posts: 380
Registered: 6/2/98
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 12:44 AM   in response to: Alexandre Machado in response to: Alexandre Machado
Alexandre Machado wrote:
In order to survive Delphi needs to attract new developers.

Of course.

But who on earth told you that the user base will grow if Win32/64 compilers become ARC-enabled?
Do you have any real trustworthy evidence for that or it is just your personal opinion?


Well it is true that transitioning from other memory-managed languages to Delphi would be simpler in the ARC scenario. Remobjects seems to be quite successful with their Oxygene language. I personally would choose ARC over GC any time because ARC automatically minimizes RAM usage whereas GC gobbles it up all the time. So an ARC version of Delphi is more desirable than a GC version like Oxygene IMHO.

Alexandre Machado

Posts: 1,683
Registered: 8/10/13
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 31, 2017 12:50 AM   in response to: Arthur Hoornweg in response to: Arthur Hoornweg
Remobjects seems to be quite successful with their Oxygene language.

I don't have any data that can confirm or refute this.

I personally would choose ARC over GC any time because ARC automatically minimizes RAM usage whereas GC gobbles it up all the time. So an ARC version of Delphi is more desirable than a GC version like Oxygene IMHO.

Of course. It doesn't mean that I actually want any of them.
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 25, 2017 11:03 AM   in response to: loki loki in response to: loki loki
loki loki wrote:
Apple does not have to. Their objects were already refcounted, so the
switch is between manual and automatic refcounting. AFAIK, on the Mac,
you can still do it manually, if you really really want that, although
I don't know many who would do that. ARC made many things much easier
on the Mac. It also makes many things easier in Delphi, but only if you
don't work against it, like loki seems to do.

ok, same question than the one i ask to dave : Can you just gave me only
one advantage of using ARC in delphi right now ?

Right now... no... because of cross-platform code. There are some gains
if you don't need to have your code compatible for non-ARC compiler.

But full advantages will appear only when all compilers move to ARC.

I am saying all along, that right now you are paying the price for ARC but
you don't have any gains.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 25, 2017 12:51 PM   in response to: Dalija Prasnikar in response to: Dalija Prasnikar
But full advantages will appear only when all compilers move to ARC.

I am saying all along, that right now you are paying the price for ARC but
you don't have any gains.

tss tss you made my laugh :) arc was introduced 6 years ago (yes 6 YEARS
ago, like an eternity in programing) and we still not see any advantage
nor any hope for the futur ... it's a pity :( pity that you don't
understand/share my point of view and stay in the deny :(
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 25, 2017 2:14 PM   in response to: loki loki in response to: loki loki
loki loki wrote:
But full advantages will appear only when all compilers move to ARC.

I am saying all along, that right now you are paying the price for ARC but
you don't have any gains.

tss tss you made my laugh :) arc was introduced 6 years ago (yes 6 YEARS
ago, like an eternity in programing) and we still not see any advantage
nor any hope for the futur ... it's a pity :( pity that you don't
understand/share my point of view and stay in the deny :(

It does not matter how long ago it was introduced. It cannot be used to its
fullest potential as long as you have cross-platform code for non-ARC compilers.

As long as there are two sets of compilers that will remain so. I have never stated
otherwise.

That also has nothing to do how good/bad ARC is. There is nothing wrong with
ARC as memory management model. Problem is with two compilers.

ARC works fine, and ARC could work fine. It only has to get rid of non-ARC
code that is holding it down.

--
Dalija Prasnikar
https://twitter.com/dalijap
https://plus.google.com/+DalijaPrasnikar
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 5:00 AM   in response to: loki loki in response to: loki loki
loki loki wrote:

Apple does not have to. Their objects were already refcounted, so
the switch is between manual and automatic refcounting. AFAIK, on
the Mac, you can still do it manually, if you really really want
that, although I don't know many who would do that. ARC made many
things much easier on the Mac. It also makes many things easier in
Delphi, but only if you don't work against it, like loki seems to
do.

ok, same question than the one i ask to dave : Can you just gave me
only one advantage of using ARC in delphi right now ?

The fact you don't have to take care of memory management anymore. That
is the one huge advantage. I love it. It is not that I can't manage
memory manually, but it is always nice not having to do that.

--
Rudy Velthuis http://www.rvelthuis.de

"Everyone likes to say Hitler did this and Hitler did that. But
the truth is Hitler did very little. He was a world class
asshole, but the evil actually done, from the death camps to
World War Two, was all done by citizens who were afraid to
question if what they were told by their government was the
truth or not, and who because they did not want to admit to
themselves that they were afraid to question the government,
refused to see the truth behind the Reichstag Fire, refused to
see the invasion by Poland was a staged fake, and followed
Hitler into national disaster."
-- Michael Rivero
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 5:32 AM   in response to: Rudy Velthuis (... in response to: Rudy Velthuis (...
On 3/26/2017 3:00 PM, Rudy Velthuis (TeamB) wrote:
loki loki wrote:

Apple does not have to. Their objects were already refcounted, so
the switch is between manual and automatic refcounting. AFAIK, on
the Mac, you can still do it manually, if you really really want
that, although I don't know many who would do that. ARC made many
things much easier on the Mac. It also makes many things easier in
Delphi, but only if you don't work against it, like loki seems to
do.

ok, same question than the one i ask to dave : Can you just gave me
only one advantage of using ARC in delphi right now ?

The fact you don't have to take care of memory management anymore. That
is the one huge advantage. I love it. It is not that I can't manage
memory manually, but it is always nice not having to do that.

but you read a little what we say before to answer ? right now their is
close to 0 lines of code without the try ... finally .free end because
on desktop (where at least we debug) their is no arc support ..
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 11:15 AM   in response to: loki loki in response to: loki loki
loki loki wrote:

but you read a little what we say before to answer ? right now their
is close to 0 lines of code without the try ... finally .free end
because on desktop (where at least we debug) their is no arc support
..

There is now, for Linux. As Dalija already said: the problem with
cross-platform is that you must code for both manual memory management
and for ARC. That means you keep on using try-finally, Free, etc.
unless you only compile for ARC platforms.

But ARC can generally work with such code as well. That is why Free and
DisposeOf were defined as they are.

--
Rudy Velthuis http://www.rvelthuis.de

"Real Programmers always confuse Christmas and Halloween because
Oct31 == Dec25 !" -- Andrew Rutherford
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 12:30 PM   in response to: Rudy Velthuis (... in response to: Rudy Velthuis (...

There is now, for Linux. As Dalija already said: the problem with
cross-platform is that you must code for both manual memory management
and for ARC. That means you keep on using try-finally, Free, etc.
unless you only compile for ARC platforms.

But ARC can generally work with such code as well. That is why Free and
DisposeOf were defined as they are.

i stop to discuss with you, because you are just a wall and it's simply
wasting time
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 27, 2017 11:44 AM   in response to: loki loki in response to: loki loki
Am 26.03.2017 um 21:30 schrieb loki loki:

There is now, for Linux. As Dalija already said: the problem with
cross-platform is that you must code for both manual memory management
and for ARC. That means you keep on using try-finally, Free, etc.
unless you only compile for ARC platforms.

But ARC can generally work with such code as well. That is why Free and
DisposeOf were defined as they are.

i stop to discuss with you, because you are just a wall and it's simply
wasting time

No, this time he's not.
His point that EMBT added .free and DisposeOf to ARC is exactly to be
able to use the old coding stile until they can ARC enable the
win32/win64 platform as well.

Rome was not built in a day either!
As much as I want improvements to Delphi I want them in a good quality
either and looking at the Android debakle right now I am for not
creating new stuff too quickly!

Greetings

Markus
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 27, 2017 2:34 PM   in response to: Markus Humm in response to: Markus Humm

No, this time he's not.
His point that EMBT added .free and DisposeOf to ARC is exactly to be
able to use the old coding stile until they can ARC enable the
win32/win64 platform as well.

but old code, desktop code, don't have disposeOF, so in any case you
need to rewrite all your old code, so it's useless!

if at least emb did think 5 min and put the content of .disposeOF inside
the .free maybe it's will be a little more accurate (i think 95-98% of
old code will work in arc without any change in this way).

Rome was not built in a day either!

Rome was already build, because delphi was already one of the best
developpent tool in it's time ! now i m ashame to say that i work with
delphi ...

As much as I want improvements to Delphi I want them in a good quality
either and looking at the Android debakle right now I am for not
creating new stuff too quickly!

it's what i say for years now ! i complain last year about this in beta
program forum ... maybe because of this (i speak frankly what i think, i
m not mvp so i m free to say my opinion), this year they decide to not
invite me and seam no-one else did try the android compilation and you
see now the disaster ! disaster that is still not yet admitted
officially by emb ... bug report
(https://quality.embarcadero.com/browse/RSP-17162) is still in reported
state !

we don't need new feature! we don't need more complexity! we need
quality! quality! quality! and arc it's a typical example of a great new
feature that was introducing without any thinking and that is completely
useless and fully painless right now
Lajos Juhasz

Posts: 755
Registered: 3/14/14
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 8:17 AM   in response to: loki loki in response to: loki loki
loki loki wrote:

if at least emb did think 5 min and put the content of .disposeOF
inside the .free maybe it's will be a little more accurate (i think
95-98% of old code will work in arc without any change in this way).

DisposeOf should not be required so often, only when forms and some
other objects are created where a "hidden" reference can remain.

For a normal code how it's now everything should work even on a desktop
compiler. It's a pity that Embarcadero never published the ARC enabled
desktop compiler. They have it and keep it in secret. That way
developers could test their code.
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 9:07 AM   in response to: Lajos Juhasz in response to: Lajos Juhasz
Am 28.03.2017 um 17:17 schrieb Lajos Juhasz:
loki loki wrote:

if at least emb did think 5 min and put the content of .disposeOF
inside the .free maybe it's will be a little more accurate (i think
95-98% of old code will work in arc without any change in this way).

DisposeOf should not be required so often, only when forms and some
other objects are created where a "hidden" reference can remain.

For a normal code how it's now everything should work even on a desktop
compiler. It's a pity that Embarcadero never published the ARC enabled
desktop compiler. They have it and keep it in secret. That way
developers could test their code.

Really? Any proof about they have it?

Greetings

Markus
Lajos Juhasz

Posts: 755
Registered: 3/14/14
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 12:49 PM   in response to: Markus Humm in response to: Markus Humm
Markus Humm wrote:

Am 28.03.2017 um 17:17 schrieb Lajos Juhasz:
loki loki wrote:

if at least emb did think 5 min and put the content of .disposeOF
inside the .free maybe it's will be a little more accurate (i think
95-98% of old code will work in arc without any change in this
way).

DisposeOf should not be required so often, only when forms and some
other objects are created where a "hidden" reference can remain.

For a normal code how it's now everything should work even on a
desktop compiler. It's a pity that Embarcadero never published the
ARC enabled desktop compiler. They have it and keep it in secret.
That way developers could test their code.

Really? Any proof about they have it?

Greetings

Markus

I am right now unable to locate any. Allen Bauer wrote once about his
private build with ARC enabled but cannot find the post right now (if I
remember correctly that was in the time frame when I was sure that he
was only joking about zero based strings).
loki loki

Posts: 718
Registered: 7/1/02
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 9:11 AM   in response to: Lajos Juhasz in response to: Lajos Juhasz
On 3/28/2017 6:17 PM, Lajos Juhasz wrote:
loki loki wrote:

if at least emb did think 5 min and put the content of .disposeOF
inside the .free maybe it's will be a little more accurate (i think
95-98% of old code will work in arc without any change in this way).

DisposeOf should not be required so often, only when forms and some
other objects are created where a "hidden" reference can remain.

hahahah just count the number of disposeOF you have in the original
delphi source code :)
Lajos Juhasz

Posts: 755
Registered: 3/14/14
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 12:45 PM   in response to: loki loki in response to: loki loki
loki loki wrote:

On 3/28/2017 6:17 PM, Lajos Juhasz wrote:
loki loki wrote:

if at least emb did think 5 min and put the content of .disposeOF
inside the .free maybe it's will be a little more accurate (i think
95-98% of old code will work in arc without any change in this
way).

DisposeOf should not be required so often, only when forms and some
other objects are created where a "hidden" reference can remain.

hahahah just count the number of disposeOF you have in the original
delphi source code :)

Of course there are now since the codebase is used by ARC and non ARC
compilers. We will have to wait for only ARC compilers. If there will
be such a release then we can return and talk about the case of
DisposeOf.
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 28, 2017 9:06 AM   in response to: loki loki in response to: loki loki
Am 27.03.2017 um 23:34 schrieb loki loki:

No, this time he's not.
His point that EMBT added .free and DisposeOf to ARC is exactly to be
able to use the old coding stile until they can ARC enable the
win32/win64 platform as well.

but old code, desktop code, don't have disposeOF, so in any case you
need to rewrite all your old code, so it's useless!

if at least emb did think 5 min and put the content of .disposeOF inside
the .free maybe it's will be a little more accurate (i think 95-98% of
old code will work in arc without any change in this way).

Rome was not built in a day either!

Rome was already build, because delphi was already one of the best
developpent tool in it's time ! now i m ashame to say that i work with
delphi ...

As much as I want improvements to Delphi I want them in a good quality
either and looking at the Android debakle right now I am for not
creating new stuff too quickly!

it's what i say for years now ! i complain last year about this in beta
program forum ... maybe because of this (i speak frankly what i think, i
m not mvp so i m free to say my opinion), this year they decide to not
invite me and seam no-one else did try the android compilation and you
see now the disaster ! disaster that is still not yet admitted
officially by emb ... bug report
(https://quality.embarcadero.com/browse/RSP-17162) is still in reported
state !

we don't need new feature! we don't need more complexity! we need
quality! quality! quality! and arc it's a typical example of a great new
feature that was introducing without any thinking and that is completely
useless and fully painless right now

Hello,

if you voiced your opinion a bit differently things might look
differently and I'm also not sure that its > 1 year since your FT
participation so your NDA might still be valid.

Another question: are you on SA? if yes, you can always request access
to "the outer FT" afaik.

Greetings

Markus
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 27, 2017 2:53 PM   in response to: loki loki in response to: loki loki
loki loki wrote:


There is now, for Linux. As Dalija already said: the problem with
cross-platform is that you must code for both manual memory
management and for ARC. That means you keep on using try-finally,
Free, etc. unless you only compile for ARC platforms.

But ARC can generally work with such code as well. That is why Free
and DisposeOf were defined as they are.

i stop to discuss with you, because you are just a wall

I could say the same about you. <g>

--
Rudy Velthuis http://www.rvelthuis.de

"Sometimes it pays to stay in bed on Monday, rather than
spending the rest of the week debugging Monday's code."
-- Christopher Thompson
Markus Humm

Posts: 4,982
Registered: 11/9/03
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 7:41 AM   in response to: Rudy Velthuis (... in response to: Rudy Velthuis (...
Am 26.03.2017 um 14:00 schrieb Rudy Velthuis (TeamB):
loki loki wrote:

Apple does not have to. Their objects were already refcounted, so
the switch is between manual and automatic refcounting. AFAIK, on
the Mac, you can still do it manually, if you really really want
that, although I don't know many who would do that. ARC made many
things much easier on the Mac. It also makes many things easier in
Delphi, but only if you don't work against it, like loki seems to
do.

ok, same question than the one i ask to dave : Can you just gave me
only one advantage of using ARC in delphi right now ?

The fact you don't have to take care of memory management anymore. That
is the one huge advantage. I love it. It is not that I can't manage
memory manually, but it is always nice not having to do that.

Hm?
Don't take care of? And what about those situations where you need to
declare something as weak to avoid a cycle?

Coding like this led to issues like in the BDS2006 IDE which (because of
not having understood .NET string handling good enough) sometimes ate
all available memory (up to the 2 GB a normal 32 bit process gets) and
got to an halt until the GC finally kicked in and slowly released stuff.

Yes, ARC is different to GC, but that cylces thing with weak should be
known to write proper ARC programs.

Greetings

Markus
Rudy Velthuis (...


Posts: 7,591
Registered: 9/22/99
Re: i m totally disappointed about emb :(
Click to report abuse...   Click to reply to this thread Reply
  Posted: Mar 26, 2017 11:38 AM   in response to: Markus Humm in response to: Markus Humm
Markus Humm wrote:

Am 26.03.2017 um 14:00 schrieb Rudy Velthuis (TeamB):
loki loki wrote:

Apple does not have to. Their objects were already refcounted, so
the switch is between manual and automatic refcounting. AFAIK, on
the Mac, you can still do it manually, if you really really want
that, although I don't know many who would do that. ARC made many
things much easier on the Mac. It also makes many things easier in
Delphi, but only if you don't work against it, like loki seems to
do.

ok, same question than the one i ask to dave : Can you just gave me
only one advantage of using ARC in delphi right now ?

The fact you don't have to take care of memory management anymore.
That is the one huge advantage. I love it. It is not that I can't
manage memory manually, but it is always nice not having to do that.

Hm?
Don't take care of? And what about those situations where you need to
declare something as weak to avoid a cycle?

Well, cycles are indeed the only thing you must care about, with
reference counting. The same actually applies to interfaces, but there,
it is not often a problem. With objects, it can be. <shrug>

--
Rudy Velthuis http://www.rvelthuis.de

"Everything has been figured out, except how to live."
-- Jean Paul Sartre
Dalija Prasnikar

Posts: 2,281
Registered: 11/9/99
Re: i m totally disappointed about emb :(