[taylor]

First new build from me for a while but there will hopefully be a few good things in here.  All of this is in CVS so all future builds should have these changes as well.  The main things that have changed or been fixed (since Phreak's build):

 - Timer changes to fix broken loading bar on some systems that would have one spot lit and then jump to the end.  This change could affect various parts of the game so if anything is either really slow or really fast please report it.  This could fix several speed issues too that seemed to affect only certain systems so if any old bugs are fixed by this please report it.

 - Various memory related fixes that mainly helps developers but should prevent some crashes as well.

 - Don't try to page out ship textures when that ship explodes.  Just like Phreak's build and is reported to fix the EFF explosion stuttering.

 - Various pilot file related fixes to prevent data loss due to mod use.  Now it won't let you in the barracks or the techroom if your current campaign is missing.  Also in the barracks it won't let you access a pilot who is using a campaign that's currently missing.

 - Some ANI related fixes to help guard against crashing when the ANI is corrupted.  If the ANI is corrupted it will stop playing at the first sign of a problem so if something appears to cut off early that would likely be the reason.

 - Make sure that lasers aren't rendered with fog in nebula missions.

 - A fix to not have rendering errors when a laser is animated but it's glow is not.

 - Various sound related crash fixes.


Included in this build are two versions of the release/debug files with the different audio APIs, DirectSound and OpenAL.  The DirectSound filenames are my typical: fs2_open_T-<date>.exe for release builds and fs2_open_T-<date>-dbg.exe for debug builds.  The OpenAL builds have "OAL" between the "_T" and the date.  The plan is to eventually dump DirectSound and go just with OpenAL on all platforms but it needs a lot of testing first.  If you have sound problems or sound related crashes then please specify which build you are using when you report it.

*UPDATED*

 - fixed a bmpman error that would cause anis to not get unloaded properly.  This turned into something like a memory leak as you played from mission to mission.

 - several OpenAL fixes, possible crashes are fixed too but not tested to that fact.

 - prevented display of missile indicators, and offscreen target brackets with disable-hud-except-messages

 - fix issue where a mission with no briefing text/stages might show the text from a previous mission.

 - docking abort on warp out crash fixed by Goober

 - fix anis in briefing screens that didn't show in previous build

 - experimental change to OpenGL where it should reduce redundant image processing when non-DDS textures are already power-of-2, saves some memory too.

 - some attempts to address the various errors reported by redmenace.  Not all fixed but will hopefully give better debug info now so the real problems can be found.


And now for the build: http://icculus.org/~taylor/fso/testing/20050530-win32.rar

[Admiral MS]

Direct Sound is working fine for me, when playing with OpenAL I can hear strange (scratching) sounds that let me think my sound card can't handle it... (Different Problem than reported in the Testers Forum).

And I get many crashes and errors with some campaigns and missions that worked fine with some older builds.
While loading a mission it stopped and nothing else happened.

[redmenace]

I am getting a crash in the campaign mission SM2-04.fs2. fs2_open crashes right after the Uhera explodes. Attached in the relevant data.

[Grimmeehh]

Im still getting crashes with this build when loading missions  from the techroom, a la my other post..

Using the debug mode nothing shows up when it crashes, though apparantly im missing some files (dev only?) such as

Warning: Couldn't open texture 'cyan_glow_small'
referenced by model 'fighter01.pof'

File:U:\src\cvs\fs2_open.testing\code\Model\ModelRead.cpp
Line: 2227

Call stack:
------------------------------------------------------------------
    model_load()    techroom_select_new_entry()    techroom_change_tab()    techroom_init()    game_enter_state()    gameseq_set_state()    game_process_event()    gameseq_process_events()    WinMainSub()    WinMain()    WinMainCRTStartup()    kernel32.dll 7c816d4f()
------------------------------------------------------------------

and other "glows" along with .tbl files, and..

For "fighter01.pof" I couldn't find Fighter01-01a.ani but I did find Fighter01-01a.dds.
tga: Couldn't open 'red_glow_small.tga'

couldn't find particle pcx for Shivan Heavy Laser (and others)

afaik I got all the required files, have i missed something or does none of this matter? right now im assuming its dev only (the U drive reference, and some other references in the debug)

As for the crashing when loading problem, you mentioned that was only on phreaks build so i guessed this might fix it- is anyone else having the problem? (Try loading "As Lightning Falls" and "Exodus" from  the techroom)

[redmenace]

1. Movies are only playing their audio in the game while the loading screen is finished but not disapearing. In other words no video.

2. In mystery of the trinity, while using Open GL, if you target the trinity you get a crash.

3. Can't use debug with open GL when you get error messages and select no. You get a black screen.

4. Nebulea Missions are incredibly slow.

5. Crashes and error messages, see attachment

[redmenace]

6. Error Message, while looking at the Orion in the mission "Lion At The Door"

[Trivial Psychic]

Well in TBP 3.2, the standard build won't play the oggs as usual, while the OAL build does as WMC's May 2nd build did, and CTDs as soon as you try to get beyond the pilot select screen.  The only difference is that WMC's build is that with his, after the CTD, the ambient rumble would continue after the program shut down and persist until you reboot... this doesn't occur with Taylor's OAL build.  It seems that OAL is both a problem and a solution for TBP 3.2.  I have no idea how we're gonna solve this one.

[taylor]

Originally posted by redmenace
I am getting a crash in the campaign mission SM2-04.fs2. fs2_open crashes right after the Uhera explodes. Attached in the relevant data.

There is a Mantis bug on this one.  I have yet to have it happen to me so it's a little difficult to back through and find the problem.  Seems to happen only with that one ship though, only in that one mission, and it's not a model issue.

1. Movies are only playing their audio in the game while the loading screen is finished but not disapearing. In other words no video.

Is this OGL or D3D?  I only tested movies on startup (intro) and in the techroom.  Do movies work properly there for you?

3. Can't use debug with open GL when you get error messages and select no. You get a black screen.

Is that during the loading screen?  If so then just wait on it.  It is loading and should go back to a normal screen when it hits the briefing.  If that's not the case, and it's happening in a mission too without ever giving you a screen back then let me know.  I'm assuming that you are running fullscreen?

4. Nebulea Missions are incredibly slow.

In both OGL and D3D?  I only really noticed a big speed drop in D3D.  I'm trying to locate that problem now but if it does happen in OGL then that would make it easier for me to find.

6. Error Message, while looking at the Orion in the mission "Lion At The Door"

Is that with the new HTL model or the original?  If the new one does the old one still give this error?

I'm going through your other messages and logs now and have a few things to fix.  I'll update this weekend.


@Trivial Psychic:  I'll go through the DirectSound code again just in case I'm missing something.  Do you have any more info in the pilot select CTD?  Is it happening with only one pilot or all of them?  New ones too?  Does it seem data related (ie. does the current TBP release or standard FS2 data crash)?  If you can send me the errorlog.txt file then I'll try and track that down and get it fixed.  Also try with -nomusic and see if it still happens if so then try -nosound.  I've had a couple of issues with corrupting audio streams in the OAL build so it could just be that.  If -nomusic fixes it then get me the mainhall music and I'll work out a fix.

[Trivial Psychic]

Originally posted by taylor
@Trivial Psychic:  I'll go through the DirectSound code again just in case I'm missing something.  Do you have any more info in the pilot select CTD?  Is it happening with only one pilot or all of them?  New ones too?  Does it seem data related (ie. does the current TBP release or standard FS2 data crash)?  If you can send me the errorlog.txt file then I'll try and track that down and get it fixed.  Also try with -nomusic and see if it still happens if so then try -nosound.  I've had a couple of issues with corrupting audio streams in the OAL build so it could just be that.  If -nomusic fixes it then get me the mainhall music and I'll work out a fix.

Well, I gave it a try with -nosound and it made it past the pilot select screen, so whatever it is, it has to do with how music oggs interact with the OpenAL implimentation for other sounds.  I've included the errorlog.  Just be aware, that I rename all builds that I install to my own proprietary system.  It is fs2_open_XXXXX_2005-XX-XX_r.exe.  the main XXXXX stream is where the creator's name or short for it will be displayed.  Any special build notes (such as OAL) will be separated by a "_" before the "r".
Since you asked for the mainhall music, HERE it is.

[starfox]

Both Directsound and OpenAL seems to work fine for me. Nebula missions also work without considerable slowdown.

Herc however had little problem, those two purple lights just outside the cockpit were actually seen while on first person mode inside the cockpit. Kinda hard to explain, could somebody just testride the herc in firstperson and see if there's anything...strange going on. Or then the incident above is not a bug but a new feature, or something. Check back later...

[redmenace]

Originally posted by taylor

There is a Mantis bug on this one.  I have yet to have it happen to me so it's a little difficult to back through and find the problem.  Seems to happen only with that one ship though, only in that one mission, and it's not a model issue.
Yes, basically. I also noticed a texture flickering issue as well.

Is this OGL or D3D?  I only tested movies on startup (intro) and in the techroom.  Do movies work properly there for you?
I believe it is both in open GL and D3D. Unless I post otherwise, assume that the intro and the techroom movies work fine.

Is that during the loading screen?  If so then just wait on it.  It is loading and should go back to a normal screen when it hits the briefing.  If that's not the case, and it's happening in a mission too without ever giving you a screen back then let me know.  I'm assuming that you are running fullscreen?
I will take a look into that. Assume that is goes back unless I post otherwise.

In both OGL and D3D?  I only really noticed a big speed drop in D3D.  I'm trying to locate that problem now but if it does happen in OGL then that would make it easier for me to find.
D3D. And the more violent the nebulea the slower it is. IE more lightning. On recently destroyed ships debris the electrical lightning seems to slow down the game as well.

Is that with the new HTL model or the original?  If the new one does the old one still give this error?
I will take a look into that. Assume that is the HTL model that is the cause.
I'm going through your other messages and logs now and have a few things to fix.  I'll update this weekend.

In regaurds to number 2, I will check the problem with targeting the Trinity.

[redmenace]

In number 3 it is not going back. In window mode it does continue to load. I will double check. But I did notice that it does take a longer than normal to load.

In bug 2 I have the fs.log and errorlog attached to this post.

In bug 6, there is a wing departing the hanger. This, as it turns out, is not caused by the HTL model. It is strictly related to fs2_open IMO. I tried with out the VPs being used.

[redmenace]

I got 2 additional crashes during one during Slaying the Ravana regaurding vec3d like the one Lion at the door.

Second was during Into the Maelstrom and concerned a failed assertion. Unfortunatly I did not get a copy of the fs.log before I started up fs2_open.

[redmenace]

During Fient! Perry! Reposte! I got 1 crash with no error message and one nonfatal error message.

[taylor]

That vecmat thing is a pain.  It only seems to happen for me in HTL mode so perhaps we can just do without the warning there unless there is actually something to be fixed.  Bobboau would know better than me about that though.  It's not a big deal on Linux since the warning won't interrupt your game.  On Windows is a bigger deal since you actually have to deal with it right then.  Enough bug reports have been filed where the non-fatal vecmat warning is the reported crash that I think it's time for fixing or deleting.

The "Fient! Perry! Reposte!" crash is a model_get() problem where it seems to be NULL but there isn't a NULL check there.  The real problem of course is that shouldn't ever happen.  I'll have to try and recreate that because it's going to need some backtracking to find out why it happened.  I'll go ahead and add the extra NULL check but it won't really help the problem.  Does it occur in a reproducable way or just sometimes (or one time)?

[redmenace]

I will play the mission a few times, but it appears to be fairly random.

[DaBrain]

Ok, the stutter bug is fixed, but the OpenAL version won't let me start TBP for some reason.

Well, I'll report more later.

[redmenace]

I had a crash during "A Game of Tag."

I was warping out and then Wokka...

[taylor]

Originally posted by DaBrain
the OpenAL version won't let me start TBP for some reason.

The current release or the beta?  I haven't had any problems with the current release but please send an errorlog.txt or something so I can figure out what's wrong.  That one errorlog that Trivial Psychic posted gave me a clue that has been fixed but I don't know why it would have gotten to that point in the first place.  I'm making a couple of OpenAL changes now so perhaps that will fix it.

Originally posted by redmenace
I had a crash during "A Game of Tag."

A docking bug.  Adding a NULL check should prevent the Assert(), or an outright crash that would occur in release builds.  Was any ship currently docked or had you called for a support ship at the time you jumped out?


A new build will go up today with as many of these fixes as I've been able to handle.

[Goober5000]

Originally posted by taylor
A docking bug.  Adding a NULL check should prevent the Assert(), or an outright crash that would occur in release builds.  Was any ship currently docked or had you called for a support ship at the time you jumped out?

This is probably my department, but the reason there's an Assert and not a NULL check is because this should never ever ever happen.  Therefore I echo taylor's request for information.