Planning for 0.77 Release

[Auswaschbar]

GML re-enabled in BuildServ since r6423 at zerver's request.

http://spring.clan-sy.com/mantis/view.php?id=1045
And some people have one or two fps whith gml enabled while with gml disabled, they have more > 100.
And for some people it crashes right after start.

Please disable it again and wait until its fixed.

edit: a status report for 0.77 for those who don't read development-log:
Apart from the bug with gml mentioned above, we don't have had any crashes, its very stable.
The only thing wich prevents an release now is a (or maybe more, didn't know yet) sync error occuring after some time ingame.

[dizekat]

Well threading is certainly great idea, but the way i see it... a chance to increase performance by some 10% , combined with *dramatic* decrease in stability, which always happens when something non-threaded is fixed for multithreading. Multithreading is not easy, infact, almost impossible to get right unless architecture is written with minimizing thread interaction and locking in mind.

TLS shoudn't even be needed at all in properly written from scratch multi-threaded application (say a Thread class has its 'this' , where it stores thread-local data).

[Tobi]

+1 for not blocking release if only multithreading problems are remaining, just release non-GML version then (and build GML version later).

[clericvash]

+1 for not blocking release if only multithreading problems are remaining, just release non-GML version then (and build GML version later).

I second that, work can always continue after and can release a new build later, better to get the other features out the door for everyone to use/test.

Multithreading would be great, but not something that is totally needed.

[bibim]

Ok I re-disabled it...

btw, shouldn't "GML=false" be the default in current build system since GML is currently bugged (and it seems it won't be used in next Spring release anyway) ? Zerver, I can add you a BuildServ option to build with GML=true if you want...

And btw too, commit r6327 deleted rts/spring.desktop but it is still referenced in SConstruct, so linux install fails.

[imbaczek]

bibim: add an option for gml, the default is of less importance IMHO.

[zerver]

Ok I re-disabled it...
Zerver, I can add you a BuildServ option to build with GML=true if you want...

That would be nice. MT bugs should not stop you from releasing anything. I just wanted some error reports, which I have got now. It may be that CA uses other GL calls than BA and those happen to be incorrectly implemented.

Edit: Does buildserv build from the .77 branch or trunk? The .77 branch contains an old GML version...

[YokoZar]

bibim: add an option for gml, the default is of less importance IMHO.

If we're disabling it for the release, then we should default to having it off for the release (and then back on in SVN for the next release)

[Argh]

Both SpringLobby and TASClient were broken on clean install, 6423. Different errors with each. This is not specific to P.U.R.E.- glad I caught it though.

SpringLobby reports that "The procedure entry point _ZNK8wxWindow20GetThemedBorderStyleEv could not be located in the dynamic link library wxmsw28u_core_gcc_custom.dll".

TASClient gets into a nasty revolving-logic error after reporting it cannot find the /Tips folder.

[Argh]

Went back to 6359. TASClient works, same error with SpringLobby. Guess that's going to have to be the release version for now, until this gets looked into...

[bibim]

Edit: Does buildserv build from the .77 branch or trunk? The .77 branch contains an old GML version...

It can build any branch or tag.

bibim: add an option for gml, the default is of less importance IMHO.

If we're disabling it for the release, then we should default to having it off for the release (and then back on in SVN for the next release)

Yeah, that's what I think too...

[Satirik]

tasclient tips bug is fixed anyway the tips.txt should be added to the installer

and i repeat my question : is the game speed auto adjust fixed in 0.77 ? because it's a critical bug most people with bad comp can't play battles after 30min

[Auswaschbar]

tasclient tips bug is fixed anyway the tips.txt should be added to the installer

and i repeat my question : is the game speed auto adjust fixed in 0.77 ? because it's a critical bug most people with bad comp can't play battles after 30min

Yes.

------------------------------------------------------------------------
r6425 | Auswaschbar | 2008-09-15 12:30:53 +0200 (Mo, 15. Sep 2008) | 3 lines

* fix slowdown when CPU usage is high
* #ifdef commit r6421 because it doesn't compile

[zerver]

Yeah, that's what I think too...

I guess we could make an MT version in conjunction with this release too but it will be very limited in what it can do. My testing so far is based on BA only, with lua widgets disabled.

[Satirik]

tasclient tips bug is fixed anyway the tips.txt should be added to the installer

and i repeat my question : is the game speed auto adjust fixed in 0.77 ? because it's a critical bug most people with bad comp can't play battles after 30min

Yes.

------------------------------------------------------------------------
r6425 | Auswaschbar | 2008-09-15 12:30:53 +0200 (Mo, 15. Sep 2008) | 3 lines

* fix slowdown when CPU usage is high
* #ifdef commit r6421 because it doesn't compile

nice

[bibim]

bibim: add an option for gml, the default is of less importance IMHO.

I've added 2 commands in BuildServ to manage alternate build profiles. So now anyone can create its own set of config flags and build Spring with it. That way it's more generic and could also be usefull for other things that GML.

[tberthel]

6359 seems pretty stable. So, what is keeping it from being released as .77

FYI, the GML fails for me too.

[Crayfish]

Now would be a really good time to release by the looks of it, PURE #1 on ModDB and all.

[Kloot]

So, what is keeping it from being released as .77

The only thing wich prevents an release now is a (or maybe more, didn't know yet) sync error occuring after some time ingame.

[AF]

Argh is not alone in that I too get those errors when using springlobby from the spring installer