Engine Testing - 94.1.1-1121-g7c05c85 (10. Sep 2013)

[jK]

New testing release for preparing 95.0!

bugfixes:
- fixed desync related to ASSIMP
- fixed SSE crashing in ThreadPool on Windows

download

changelog.txt

There were a few breaking changes since last test release, see here

See the Testing Release Wiki page for general info about how to obtain the release, and an archive of all testing releases since the last stable one.

If you find a bugs, please report to Mantis.

Please attach infolog.txt as file, if you crash!

Note: MacOSX buildslave is down atm, so no builds are available for it.

[Beherith]

Is it OK if I test it on an intel hd 4000 with ForceIntelShaderSupport?

[Silentwings]

Epic win, I already have this build. Great job with all the new stuff.

[Funkencool]

Code: Select all

[f=0000000] [Chili] Error: in `BAR's Main Menu`:line0 : [string "LuaUI/Widgets/chili/controls/control.lua"]:139: bad argument #1 to 'DeleteRBO' (RBO expected, got nil)
[f=0000000] [Chili] Error: stacktrace:
	[C]: in DeleteRBO
	[string "LuaUI/Widgets/chili/controls/control.lua"]:139: in Dispose
	[C]: in DeleteRBO
	[string "LuaUI/Widgets/chili/controls/control.lua"]:139: in Dispose

Spring hung for a minute or two. By the time I closed it, my infolog had about 60,000 lines of this repeated (for each chili control).

[Google_Frog]

The testing wiki page is quite out of date. There are three easy steps with absolutely no skill required.

• Get a lobby which has proper multi-engine support.
• Join the BLEEDING EDGE springie, you now have the correct test engine version.
• Host a game with this engine and using whatever game you want.

[Jools]

First time I tried to launch the game it hung and had to be killed with process explorer. Second time it worked but there's a lot of error output, for example these:

Code: Select all

[f=0000000] Warning: [ReturnFalse] deprecated field!

Also, is it possible to remove clutter like this from infolog (almost half the infolog is full of those lines):

Code: Select all

[ArchiveScanner] Warning: Archive U:\bin\Spring\Data\games\bota17beta10_1.sd7: The cost for reading a 2nd-class meta-file is too high: Units/CORE_T2_VEH_Hedgehog.FBI

Also, I noticed that the commander can no longer D-Gun into water. Is this a bug?

[PicassoCT]

Code: Select all

[f=0000342] Warning: [GetClass] name "portalstorm" does not match any class
[f=0000342] Warning: [GetClass] name "portalstorm" does not match any class
[f=0000342] Warning: [GetClass] name "portalstorm" does not match any class
[f=0000342] Warning: [GetClass] name "portalstorm" does not match any class
[f=0000342] Warning: [GetClass] name "portalground" does not match any class
[f=0000343] Warning: [GetClass] name "blackerThenSmoke" does not match any class
[f=0000343] Warning: [GetClass] name "flames" does not match any class
[f=0000343] Warning: [GetClass] name "blackerThenSmoke" does not match any class
[f=0000343] Warning: [GetClass] name "glowSmoke" does not match any class
[f=0000344] Warning: [GetClass] name "blackerThenSmoke" does not match any class
[f=0000344] Warning: [GetClass] name "flames" does not match any class
[f=0000344] Warning: [GetClass] name "blackerThenSmoke" does not match any class

Spamed with every ceg call..

+ Not working Spring Lobby.

[Silentwings]

I also have spam of things like "[f=0000596] Warning: [GetClass] name "reclaimshards1" does not match any class" accompanying CEG calls.

edit: This is fixed with -1228, nice job :)

[Forboding Angel]

Still has this bug: http://springrts.com/mantis/view.php?id=3845

[raaar]

same problem here:

Code: Select all

...
[f=0022875] Warning: [GetClass] name "bplasmaballbloom" does not match any class
[f=0023091] Warning: [GetClass] name "bplasmaballbloom" does not match any class
[f=0025330] Warning: [GetClass] name "bplasmaballbloom" does not match any class
[f=0025438] Warning: [GetClass] name "featureblastwrapper" does not match any class
...

It seems it affects only CEG called from gadgets with Spring.SpawnCEG, and prevents them from being displayed.

Also, some of my explosions seem WAY bigger than in build 1087 (check attachment), and for some CEG the some of the components aren't being drawn, but it varies from CEG to CEG. The same "inherited" components called with

Code: Select all

	[_std]{class=CExpGenSpawner;[properties]{pos=0,0,0;delay=0;damage=d1;explosionGenerator=custom:STDNARROWBLASTEFFECTS;}air=1;ground=1;water=1;underwater=1;unit=1;count=1;}

are displayed for a CEG and not for another, and it happens every time:
- cannon shot blasts are missing the "class=CSimpleParticleSystem;" sparks and dust and the [smoke]
- red light laster blasts are missing the "class=CExploSpikeProjectile;" spikes, but show the CSimpleParticleSystem

maybe it has something to do with the damage scaling..

Also, the fog of war refresh rate is still the same, i reopened 3916

grrr...

[raaar]

flowers

[Forboding Angel]

He already explained why.

It actually is updating properly, but the problem is that it is simply too expensive to do it with the straight up engine method (as I understand it).

Therefore, a shader approach *should* be much faster.

Correct me if I'm wrong, JK, but that is the way I understood your explanation.

[smoth]

ah, jK posted a comment, let's see.. (check attachment)

how nice. How about explaining why?

Anyone else care to comment?

Comment: don't try and drum up drama

[raaar]

The reason being performance could be implied, yes, but either way he should be explicit and provide more information.

there's a thread from 2010 where some people talk about performance drop related with los/fog of war enabled, but that's it.

It doesn't make much sense : on the other thread i tested increasing the rate and it seems it has a negligible effect on performance even on my old computer. Fog-of-war in spring has had this behaviour since i remember spring from 2009-2010, maybe earlier.

either way, unless more people show interest in this, it's going to keep being postponed. He whined at the mere fact of reopening the issue on mantis (it isn't fixed..).

[smoth]

it is nice and all but has nothing to do with the thread. Please consider making a NEW thread to champion your issue.

[greammeakel]

So once again. I forget who I was talking to. But as this has to do with testing 94.1.1-1131 blah blah blah. Here is a file to show you all where building Mac OSX engine stops.

also. why no RTF file support. so file is in PDF now :D

[Jools]

One more comment: it seems this testing build is quite slow on loading. It seems to do more scanning of my PC before the single player menu appears. During this time the screen goes black and you cannot do anything, and this can last for up to 1-2 minutes.

Is new spring scanning more stuff than before? Maybe 3 out of 4 times, the game doesn't finish loading before win7 determines that is has hung. Maybe reduce some scanning of nonsense...

[Silentwings]

It's probably that whenever you get a new engine build or (maybe>?) change your .sdd substantially it has to re-hash all the files in your .sdd(s) one by one; and you wouldn't normally be making it do that. sd7/sdz/etc scan quicker because they are a single file.

With .sdd's, I used to have to wait a couple of minutes on the first time I ran a new engine, but now I got a SSD :p

[raaar]

put the CEG problem on mantis

still happens on build 1138. However, the other one related to

Code: Select all

[f=0022875] Warning: [GetClass] name "bplasmaballbloom" does not match any class

seems fixed.

[jK]

One more comment: it seems this testing build is quite slow on loading. It seems to do more scanning of my PC before the single player menu appears. During this time the screen goes black and you cannot do anything, and this can last for up to 1-2 minutes.

Is new spring scanning more stuff than before? Maybe 3 out of 4 times, the game doesn't finish loading before win7 determines that is has hung. Maybe reduce some scanning of nonsense...

No, it just creates the window now (and shows black) before it is starts the ArchiveScanner ...