Sign in Join

Live chat for Free Pascal, Lazarus, Delphi and every Pascal programmer.

Mailing lists / fpc-devel / 2006-11

AW: Bug 0007281 / MemLeak with interfaces

8 messages · 5 people · –

Reply by email Subscribe to this list Your subscription

Read-only archive. Use j and k to move between messages.

Helmut Hartl
It is not usefull to do. There won't be a new release anymore from the fixes branch.

Your statement concludes that 2.04 is not usable for production environment.
-> Except you do not use interfaces - or write programs which are only commandlinetools, not expected to run for some time.

This also means we cannot release Zeos Database Objects for Freepascal, which is quite sad. We cannot recommend an average ZEOS user what to do to
get a memory stable and usable compiler environment.

FPC 2.0.2 is not usable -> Dataset has bugs. (but is memstable) FPC 2.0.4 is not usable -> Dataset is OK (but it's not memstable) (A real world deamon app with about 80.000 lines of code runs about 2 hours,
before crashing the machine.)
FPC 2.1.1 is not usable, as it is the development branch.

For our company it's not a problem we use a patched 2.0.2. But in our forum we have lots of request for a stable environment.

So what we have done is what the FPC site recommends:

Download Daily Update of Release Tree (version 2.0.x) These snapshots contain the latest bug fixes, without any major new feature. This should be more stable than the development snapshots (and even the last official release), but there is still no guarantee that

these >snapshots are bugfree.

We have 1446 registered users in the zeosforum and many of them are using/ starting to use freepascal apart from delphi & kylix.

Maybe you can give me a recommendation what to say to them? I personally don't want to say them "It's not usefull to use fpc"

It's maybe even usefull for FPC people as we offered to contribute the whole ZEOS Library with the Lazarus/FPC modified LGPL license.

Reply by email

Florian Klaempfl
to Helmut Hartl
> It is not usefull to do. There won't be a new release anymore from the > fixes branch.

What Peter wanted to say is: there won't be any 2.0.x releases but we want the next release to be 2.2.0 instead.

Your statement concludes that 2.04 is not usable for production environment.

Well, depends on what you want to do ;) The 2.0.x series has certain problems with memory handling depending on your application. Depending on the version either variants, widestrings, vararrays or interfaces leak.

Show quoted text 10 lines
> Download Daily Update of Release Tree (version 2.0.x) > These snapshots contain the latest bug fixes, without any major new > feature. This should be more stable than the development snapshots (and > even the last official release), but there is still no guarantee that these >snapshots are bugfree. We have 1446 registered users in the zeosforum and many of them are using/ starting to use freepascal apart from delphi & kylix. Maybe you can give me a recommendation what to say to them?

I merged the patch but help us to test 2.1.1 as much as possible based on the revisions known to be good
(http://www.freepascal.org/wiki/index.php/Tested_Unstable_Revisions) so we can get 2.2.0 out asap (head is currently broken due to heavy rewrites).

The perfect solution for your users would be if you run daily regression tests with trunk and tell users which revisions work good with Zeos (currently 5191 is a good bet) and tell us if you get worse test results.

Reply by email

Bart Hotmail
to Florian Klaempfl: What Peter wanted to say is: there won't be any 2.0.x releases but we want the next…

Does that mean we can soon expect single static assignment?

--Bart

Earlier message 45 lines
----- Original Message ----- From: "Florian Klaempfl" To: "FPC developers' list" Sent: Tuesday, November 07, 2006 9:21 PM Subject: Re: AW: [fpc-devel] Bug 0007281 / MemLeak with interfaces > Helmut Hartl schrieb: >>> It is not usefull to do. There won't be a new release anymore from the >>> fixes branch. > > What Peter wanted to say is: there won't be any 2.0.x releases but we want > the next release to be 2.2.0 instead. > >> >> Your statement concludes that 2.04 is not usable for production >> environment. > > Well, depends on what you want to do ;) The 2.0.x series has certain > problems with memory handling depending on your application. Depending on > the version either variants, widestrings, vararrays or interfaces leak. > >>> Download Daily Update of Release Tree (version 2.0.x) >>> These snapshots contain the latest bug fixes, without any major new >>> feature. This should be more stable than the development snapshots (and >>> even the last official release), but there is still no guarantee that >> these >snapshots are bugfree. >> >> We have 1446 registered users in the zeosforum and many of them are >> using/ starting to use freepascal apart from delphi & kylix. >> >> Maybe you can give me a recommendation what to say to them? > > I merged the patch but help us to test 2.1.1 as much as possible based on > the revisions known to be good > (http://www.freepascal.org/wiki/index.php/Tested_Unstable_Revisions) so we > can get 2.2.0 out asap (head is currently broken due to heavy rewrites). > > The perfect solution for your users would be if you run daily regression > tests with trunk and tell users which revisions work good with Zeos > (currently 5191 is a good bet) and tell us if you get worse test results. > _______________________________________________ > fpc-devel maillist - > http://lists.freepascal.org/mailman/listinfo/fpc-devel >

Reply by email

Peter Vreman
to Bart Hotmail
Show quoted text 8 lines
>>> It is not usefull to do. There won't be a new release anymore from the >>> fixes branch. > > What Peter wanted to say is: there won't be any 2.0.x releases but we > want > the next release to be 2.2.0 instead. > Does that mean we can soon expect single static assignment?

SSA is only partly implemented. Unless you are running a scientific benchmark you won't notice it.

Reply by email

Peter Vreman
to Helmut Hartl: Your statement concludes that 2.04 is not usable for production environment. -> Except…
Show quoted text 30 lines
> It is not usefull to do. There won't be a new release anymore from the > fixes branch. Your statement concludes that 2.04 is not usable for production environment. -> Except you do not use interfaces - or write programs which are only commandlinetools, not expected to run for some time. This also means we cannot release Zeos Database Objects for Freepascal, which is quite sad. We cannot recommend an average ZEOS user what to do to get a memory stable and usable compiler environment. FPC 2.0.2 is not usable -> Dataset has bugs. (but is memstable) FPC 2.0.4 is not usable -> Dataset is OK (but it's not memstable) (A real world deamon app with about 80.000 lines of code runs about 2 hours, before crashing the machine.) FPC 2.1.1 is not usable, as it is the development branch. For our company it's not a problem we use a patched 2.0.2. But in our forum we have lots of request for a stable environment. So what we have done is what the FPC site recommends: >Download Daily Update of Release Tree (version 2.0.x) >These snapshots contain the latest bug fixes, without any major new >feature. This should be more stable than the development snapshots (and >even the last official release), but there is still no guarantee that these >snapshots are bugfree.

We simply don't have the time to maintain 2 branches. Now the focus is put on getting a 2.2.0 release. Also creating a FPC release for all the supported platforms takes a huge time. It is not like a simple unix source .tar.gz that we can release.

Reply by email

Jonas Maebe
to Peter Vreman: SSA is only partly implemented. Unless you are running a scientific benchmark you won't…
Peter Vreman wrote
> Does that mean we can soon expect single static assignment? SSA is only partly implemented. Unless you are running a scientific benchmark you won't notice it.

It's actually (partial) static single use which we have. But other optimizations to the register variable handling are much more noticeable (at least on cpus which have a reasonable set of registers, although even on x86 there are some improvements).

Jonas

Reply by email

Helmut Hartl
If anybody is willing to do so, he'll get svn write access so feel

free

to volounteer ;)

I am maybe interested in doing so - if I am bright enough to do it. :-)

helmut

Reply by email