Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Discuss the source code and development of Spring Engine in general from a technical point of view. Patches go here too.

Moderator: Moderators

abma
Spring Developer
Posts: 3798
Joined: 01 Jun 2009, 00:08

Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by abma »

for details, see the Testing Release Wiki page

changes:
  • fixed win32 installer (it installed always in non-portable mode)
  • fix memory corruption in stacked transports
  • Fix SetUnitWeaponState "range" fails for Cannon
  • fix bumpmapped water on ATI-cards
  • fixed possible lua-errors (for example healthbars)
  • allow resources_map to specify groundFX textures for CEGs using CSimpleGroundFlash
  • added a validation test
the buildbot does a simple validation test, this way current development version hopefully won't break totally. ideas to improve that checks are welcome, maybe some lua-widgets can be added for that.

Bugs and issues
Report in this thread or on Mantis.
Remember to attach infolog.txt as file, if you crash!

Most major bugs are fixed, but there are still known bugs.
Last edited by abma on 07 Sep 2011, 20:11, edited 1 time in total.
User avatar
Deadnight Warrior
Posts: 183
Joined: 08 Jun 2009, 17:59

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by Deadnight Warrior »

I've already told zerver and Kloot about some errors I found in the last testbuild, but since they're present in this build also...

Wake particles aren't rendered any more when you swich from dynamic water to any other. They work when you start the game in any other mode than dynamic. After the bug triggers wake particles are available only in /water 2 or after a engine restart in another water mode. zerver made a ticket on Mantis

FBOs aren't working in MT build, I have no problems with 0.82.7.1

Code: Select all

[f=0000000] [] Warning: FBO-ADVSKY: no/unsupported textures/buffers attached!
[f=0000000] [] Error: [GetSky] error: ADVSKY: FBO is unavailable (falling back to BasicSky)
[f=0000000] [] Warning: FBO-SHADOW: no/unsupported textures/buffers attached!
[f=0000000] [] Error: [CShadowHandler] failed to initialize depth-texture FBO
[f=0000000] [] Warning: FBO-ADVTREE: no/unsupported textures/buffers attached!
[f=0000000] [] Error: Could not create FBO!
[f=0000000] [] Warning: FBO-FARTEXTURE: no/unsupported textures/buffers attached!
[f=0000000] [] Warning: FBO-PERLIN: no/unsupported textures/buffers attached!
No other errors printed, no crashes or stacktrace available.
I use NVidia Verde driver v275.33 on my QuadroFX 3600M

I heard from devs that a new engine feature is causing amphibious units to get slower the deeper in water they are. Ok I got no problems with that, but at some point they get so slow that pathing breaks and unit stops moving ~3 seconds after you give it a move order. And what's the unitdef tag for adjusting this behaviour?

When using /shadows 2 unit shadows get rendered through hills onto terrain behind the hill (as Kloot mentioned, it could be related to dynamic sun shadow problems)

For some reason unknown to me, HealthBars widget doesn't display any stats above units, no errors or warnings either.

If I encounter anything else, I'll report.
User avatar
hoijui
Former Engine Dev
Posts: 4344
Joined: 22 Sep 2007, 09:51

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by hoijui »

good that you report here now.
next time, do it like this right away, as we state everywhere. don't write to individual devs in PM or chat, but in this thread or on mantis.

zerver kind of put two bug report into one, and the one about wake particles is explained in much less detail then you did here.
which in effect means... you did lots of useful tests, but instead of reporting all the detailed results, you told someone else who had to do the report, and most of your findings got lost in the process.
these additional info sound very crucial to me, to find the bug, and maybe even to be able to replicate it.
User avatar
jK
Spring Developer
Posts: 2299
Joined: 28 Jun 2007, 07:30

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by jK »

abma wrote:
  • fix bumpmapped water on ATI-cards
bad formulated you mean "fixed bumpwater on ATIs with atihacks=0"


@Deadnight Warrior
the MT build is not the official one. please test with default one, and report MT errors separately.
abma
Spring Developer
Posts: 3798
Joined: 01 Jun 2009, 00:08

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by abma »

@jk:

for me it didn't work with atihacks=1, too...
User avatar
Forboding Angel
Evolution RTS Developer
Posts: 14673
Joined: 17 Nov 2005, 02:43

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by Forboding Angel »

Will test soon, thx!
User avatar
Deadnight Warrior
Posts: 183
Joined: 08 Jun 2009, 17:59

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by Deadnight Warrior »

Deadnight Warrior wrote:For some reason unknown to me, HealthBars widget doesn't display any stats above units, no errors or warnings either.
Ignore, seems to work on this testbuild, it didn't work on last one.

Now when I recall where I've experienced shadows getting rendered through hills, it was on 0.83 testbuild when sun turned by ~120┬░. As the building shadows follow sun direction they eventually turned "around the corner" and got behind a hill.

As for the MT errors and FBOs, it seems to work correctly if I disable MT loading in springsettings.cfg (aka. LoadingMT=0), no driver updates/downgrades needed.
Google_Frog
Moderator
Posts: 2464
Joined: 12 Oct 2007, 09:24

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by Google_Frog »

Healthbars http://springrts.com/mantis/view.php?id=2511

Apparently fixed but still set to feedback. See if the widget is loaded and, if you can, do some debugging to see if the bars are just not drawn.
abma
Spring Developer
Posts: 3798
Joined: 01 Jun 2009, 00:08

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by abma »

@Google_Frog:
http://springrts.com/mantis/view.php?id=2511#c7312 , healtbars are fixed, but chat commands are still reordered...
User avatar
jK
Spring Developer
Posts: 2299
Joined: 28 Jun 2007, 07:30

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by jK »

abma wrote:chat commands are still reordered...
only in spring MT (+ unknown commands)
Deadnight Warrior wrote:As for the MT errors and FBOs, it seems to work correctly if I disable MT loading in springsettings.cfg (aka. LoadingMT=0), no driver updates/downgrades needed.
LoadingMT has nothing to do with spring MT, it is a multithreaded loadingscreen available in all spring builds.
abma
Spring Developer
Posts: 3798
Joined: 01 Jun 2009, 00:08

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by abma »

@jk: thx for clarify

@Deadnight Warrior:

what doesn't work with LoadingMT=1 in spring single threaded? is it the same issue than here: http://springrts.com/phpbb/viewtopic.php?f=11&t=26704 ? can you please attach a infolog.txt with the error inside?
User avatar
Deadnight Warrior
Posts: 183
Joined: 08 Jun 2009, 17:59

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by Deadnight Warrior »

abma wrote:what doesn't work with LoadingMT=1 in spring single threaded?
No it's not for ST it's for MT build. I had black terrain and units and lots of FBO errors/warnings which I posted in my first post. Even with LoadingMT turned off I still have problems with invisible S3O models in MT build. (zerver told me to try playing with those MT settings)

For ST build I found that Link limits are too steep for playing an offline match against AIs, even with up/down speed of >1MB/s and packet limit of 1024. E323AI is notorious for command spam. RAI performed better. zerver told me that those limits can be disabled by setting them to -1, and that they where introduced because some widgets can spam a lot of commands that can crash spring. If those link limits can be disabled what's the purpose of having them in the first place?
User avatar
hoijui
Former Engine Dev
Posts: 4344
Joined: 22 Sep 2007, 09:51

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by hoijui »

there is already a bug report for the problem with AIs that you mention. we generally call it bandwidth limit.
http://springrts.com/mantis/view.php?id=2541

the limit is not just for widgets, but also for AIs. it is kind of meant as an incentive for AI devs to limit themselfs in commands per frame.
the problem with AIs in particular is, that when their team has lots of units, and they give a new move command to each unit, each frame, then they cause a lot of work for the path-finder.

as described in the mantis report, the current implementation is not a solution for that problem.
User avatar
AF
AI Developer
Posts: 20687
Joined: 14 Sep 2004, 11:32

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by AF »

We ran into this years ago in the AI community

Did nobody wonder why all the classic AIs had command queues and per frame throttles? AAI even went as far as tracking what a unit should be doing and then checking every other frame of that was what it was actually doing
zerver
Spring Developer
Posts: 1358
Joined: 16 Dec 2006, 20:59

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by zerver »

Deadnight Warrior wrote:zerver told me that those limits can be disabled by setting them to -1, and that they where introduced because some widgets can spam a lot of commands that can crash spring. If those link limits can be disabled what's the purpose of having them in the first place?
When you are running a game server, the clients cannot disable the incoming limit on the host side. That is the purpose; the clients can only change the settings on their side of the connection.
User avatar
Forboding Angel
Evolution RTS Developer
Posts: 14673
Joined: 17 Nov 2005, 02:43

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by Forboding Angel »

Ok, synopsis... not good.

There is a lot of good, but a lot of extremely bad decisions.

First off. 15 minute game, this is my ENTIRE infolog:
[f=0000000] [] Error: Joystick 0 not found
[f=0000000] [] Warning: AI Interface property 'supportsLookup' could not be found.
[f=0000000] [] Warning: AI Interface property 'supportsLookup' could not be found.
[f=0000000] [] Error: [LoadFeaturesFromMap] unknown map feature type ""
[f=0000000] [Sound] Warning: Unable to open audio file: FailedCommand
[f=0000000] [Sound] Warning: CSound::GetSoundId: could not find sound: FailedCommand
I cannot troubleshoot with an infolog like that. Infolog should report EVERYTHING! I have no idea why in the world you would castrate the 1 and almost only troubleshooting tool we have.

---

Loading screen... Yeah sure, it looks all fine and good. Now change it back. I did loadscreens specifically for the previous method. So did many other projects. Why you decided that changing it was a good idea I will never know, but at the very least you should have A: asked us, B: given the option to revert it. Seriously, is a progress bar so damn difficult? Don't display any text, just have a progress bar that progresses a notch further at every loading step. It's not that hard. I should not be the only one capable of suggesting this.

Springsettings was not in my portable that I downloaded. Is that an oversight?

Game speed changes and viewradius changes no longer report in the chat. This is bad on so many levels. This also goes back to infolog reporting.

Ogg soundstream is broken again. Seriously guys, it's not freaking funny. Please fix the playing of ogg files. Any halfway crappy windows vb app can play an mp3 file without sucking. Why is a big production like spring so bad at it?

You have NO BUSINESS RELEASING this in it's current state. These are gigantic breaking bugs. We don't need another round of 7 .x releases because it wasn't done right the first time.

Why has no one else mentioned this stuff? Am I honestly the only one who bothered to look?
Google_Frog
Moderator
Posts: 2464
Joined: 12 Oct 2007, 09:24

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by Google_Frog »

Why has no one else mentioned this stuff? Am I honestly the only one who bothered to look?
Many just use mantis if the bug is not something people may want to discuss.

For example infolog:
http://springrts.com/mantis/view.php?id=2658
User avatar
Forboding Angel
Evolution RTS Developer
Posts: 14673
Joined: 17 Nov 2005, 02:43

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by Forboding Angel »

Ack durr ok, my mistake. So sorry.
User avatar
FLOZi
MC: Legacy & Spring 1944 Developer
Posts: 6240
Joined: 29 Apr 2005, 01:14

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by FLOZi »

RE: Loading Screen, plan is eventually to have it fully customisable via lua; I don't know the current progress on that.
User avatar
knorke
Posts: 7971
Joined: 22 Feb 2006, 01:02

Re: Engine Testing - 6. September 2011 (0.82.3-3386-gefb020b)

Post by knorke »

No matter what I type into the console, I got nothing to show up in chat. No matter if "/shadows 1" or "blabla"
So basically there is no chat at all (BA & XTA)

Is that what http://springrts.com/mantis/view.php?id=2658 is about?
Most chat and commands no longer show up on the console or in infolog.

Also changing camera mode via ctrl+Fx keys activates the function of the Fx key too. ie ctrl+F2 for overhead camera enables path map.
Post Reply

Return to “Engine”