[Mr_Maniac]

Thursday the 21st April 2005 - Current CVS (but it already was there since some time...):

I have some little Problems (again):
1. When I look into a sun (or when a sun comes into my FOV), the framerate drops dramatically... From 60-120 FPS to 12-18 FPS...
It seems always to be the same...

2. Sometimes, when someone jumps out in missions I get a CTD...

3. This "Texture-Bug" still exists... It's not often, but it exists... Ships get a "strange" texture and/or get translucent...

But the rest of the game runs great!

[WMCoolmon]

What video card you have, and are you using the NVIDIA AGPGART (IIRC) driver?

I'm using a GF6600GT w/ that installed and have been having the problem with almost everything. taylor mentioned he had a lead on a bug but I'm not sure if it's the same one I've been experiencing.

[Mr_Maniac]

I have a GeForce 3 and I am using the kernel AGPGART...
Newest nVidia-Drivers

[taylor]

Originally posted by Mr_Maniac
1. When I look into a sun (or when a sun comes into my FOV), the framerate drops dramatically... From 60-120 FPS to 12-18 FPS...
It seems always to be the same...

I've heard this from many people but have never been able to reproduce it myself.  I assume that it happens with all missions though right?

2. Sometimes, when someone jumps out in missions I get a CTD...

Run in gdb and let us know what the backtrace looks like.  I've had plenty off crashes with someone jumping in (all fixed as far as I know) but never jumping out.

3. This "Texture-Bug" still exists... It's not often, but it exists... Ships get a "strange" texture and/or get translucent...

Do another update and see if it's still a problem.  I noticed last night that every texture (for me at least) that looked wrong was a compressed DDS.  These textures have to sent to the video card as is but if your texture detail level in the options screen wasn't at max then it tried to resize the texture.  This didn't work of course and it only used a portion of the texture data rather than the full texture.  It now ignores this detail setting when loading compressed textures.  If it doesn't fix your problem completely then let me know.

[Mr_Maniac]

Originally posted by taylor

I've heard this from many people but have never been able to reproduce it myself.  I assume that it happens with all missions though right?

Right!

Run in gdb and let us know what the backtrace looks like.  I've had plenty off crashes with someone jumping in (all fixed as far as I know) but never jumping out.

Okay! I'll try that today

Do another update and see if it's still a problem.  I noticed last night that every texture (for me at least) that looked wrong was a compressed DDS.  These textures have to sent to the video card as is but if your texture detail level in the options screen wasn't at max then it tried to resize the texture.  This didn't work of course and it only used a portion of the texture data rather than the full texture.  It now ignores this detail setting when loading compressed textures.  If it doesn't fix your problem completely then let me know.

Hmm... Sounds good...
Oh... The Sathanas looked really frightening with that bug!
The Glowmaps worked, but the rest of the textures/of the ship was totally black...


And now to something completely different:
What's up with beams?
I've noticed that too often, that ships won't fire their beems!
The Colossus didn't fire its beams at the Sathanas (where it should destroy it) and in the mission, where the Sathanas should destroy the Colossus, the Sathanas wouldn't fire!
So I had to cheat to complete those missions!

[Mr_Maniac]

Okay... I've restarted the campaign (I'd played it through)...
And after the first trainig-mission, when I jumped out, I got

Code: [Select]

Program received signal SIGSEGV, Segmentation fault.
model_get_radius(int) (modelnum=0) at model/modelread.cpp:3033
3033            return pm->rad;

Oh... And that FPS-Drop... When I am alone (when I get the full 120 FPS), then the FPS drops about ~20 FPS, when a sun comes into my FOV...

EDIT:
And another one:
In nebula-missions, some effects have "boxes" around them...
So there's something with transparency...

Oh.. And I've played 5-6 missions in a row and it worked fine...
Well.. Okay.. The Colossus destroyed the Hawkwood without beams (with other weapons) and in the same mission, the Arcadia moved(!) and tried to jump out(!)...
(Explanation: A really BIG Jump-Hole opens behind the Arcadia and nothing else happens... The hole AND the Arcadia are staying where they are...
The Arcadia are doing that since a few months or so...)

[taylor]

Originally posted by Mr_Maniac
Okay... I've restarted the campaign (I'd played it through)...
And after the first trainig-mission, when I jumped out, I got

Got a full backtrace?  I'm not sure where that's originating from.

And another one:
In nebula-missions, some effects have "boxes" around them...
So there's something with transparency...

I just noticed that too.  I probably broke something with my latest commit but maybe not.  I don't usually play neb missions during testing so the last time I tried it worked but now it doesn't.  That's been a couple of weeks ago though.  I'll try and get it fixed today.

Oh.. And I've played 5-6 missions in a row and it worked fine...
Well.. Okay.. The Colossus destroyed the Hawkwood without beams (with other weapons) and in the same mission, the Arcadia moved(!) and tried to jump out(!)...
(Explanation: A really BIG Jump-Hole opens behind the Arcadia and nothing else happens... The hole AND the Arcadia are staying where they are...
The Arcadia are doing that since a few months or so...)

I've noticed the turret thing too and I'm not sure when that broke.  It seems to be an issue for just some ships.  I think WMCoolmon is on top of it though.

The Arcadia jumping has been doing that for a while as far as I know.  I don't know why exactly but it happens.  I think that there is a transport docked with it which is what actually jumps.  The Arcadia just gets dragged along (after a good while).  I've actually been killed several times in that mission being hit by the Arcadia as it spins around.

[Mr_Maniac]

Originally posted by taylor

Got a full backtrace?  I'm not sure where that's originating from.

No, that was everything gdb gave me...

I just noticed that too.  I probably broke something with my latest commit but maybe not.  I don't usually play neb missions during testing so the last time I tried it worked but now it doesn't.  That's been a couple of weeks ago though.  I'll try and get it fixed today.

Well... It's like this since a few weeks...
I don't know exactly since when...

I've noticed the turret thing too and I'm not sure when that broke.  It seems to be an issue for just some ships.  I think WMCoolmon is on top of it though.

Well... We'll see

The Arcadia jumping has been doing that for a while as far as I know.  I don't know why exactly but it happens.  I think that there is a transport docked with it which is what actually jumps.  The Arcadia just gets dragged along (after a good while).  I've actually been killed several times in that mission being hit by the Arcadia as it spins around.

Yes, it is doing that since a few weeks/months...

Oh... And I have to report another thing:
In the first nebula mission (where I have to follow Zeta 1), after destroying the first shivan group of fighters, Zeta-wing seems to have unlimited afterburner (and they're using it), so I have a hard time following Zeta 1... Even with all energy given to the engines...
But that only happened one time... And I don't know if it will happen again...

EDIT:
Oh, I forgot:
How hard would it be, to use the Changelog-File again?
I like to read the changes, that were made, but I have to look in the cpp-files...

[Mr_Maniac]

And a new problem (thanks WMCoolmon *only a joke*):
When I start fs2_open the transparency of the code-part in the title-screen doesn't work (the background's yellow)...
Then, when I come to the player select screen, the screen flickers madly...
The only things I can see is the playername and the buttons...
But it's unplayable...
Updates that (possibly) caused that:

Code: [Select]

* $Log: 2d.cpp,v $
 * Revision 2.44  2005/04/23 01:17:09  wmcoolmon
 * Removed GL_SECTIONS


EDIT:
I finally was able to make some screenshots!
I also was able to fly a mission... In-mission everything looks okay...
But menus are just... Well... broken...
Screenshots:
Pilot-Selection
Techroom
After a mission (mainhall)

[taylor]

Originally posted by Mr_Maniac
Well... It's like this since a few weeks...
I don't know exactly since when...

Hmm, ok.  I did update to a pure gropengltexture.cpp too which is pretty much when I noticed it.  I'll go back over my old one and see if I can find the change that makes the difference.

How hard would it be, to use the Changelog-File again?
I like to read the changes, that were made, but I have to look in the cpp-files...

The cvs package contains 'rcs2log' which generates a Changelog.  That could be done automatically on the server side (don't know if that will happen though) or just done periodically by me of someone else and then update the file in CVS.  For this case though you can just run rcs2log yourself from the fs2_open directory.  You can either pipe it through 'more' (or 'less' which is my favorite) or just direct the output to a file.

I finally was able to make some screenshots!
I also was able to fly a mission... In-mission everything looks okay...
But menus are just... Well... broken...

Strange.  The menus are the only thing that really uses the section code but it hasn't been compiled in anyway for a while now.  Maybe something got removed that shouldn't have been.  I haven't had a chance to play with it since that change but if anything is missing that shouldn't be then I'll fix it tonight with my other commits.

[WMCoolmon]

It looks like gr_clear and gr_flip are messed up, but I don't remember any of the GL_SECTIONS code being in there.

[taylor]

Yeah, I'm not sure what the deal is.  I looked over the CVS log from your stuff and it all looked fine to me.  My current tree doesn't have this problem so perhaps I fixed it without noticing.  When I get the chance to commit later today we'll know for sure.

About the only thing I see that's wrong is that popups are creating a slighly offset background image.  That's new to me, but may be something that I did with my current changes.

[Mr_Maniac]

PROBLEM:
I've just noticed, that the debug-version doesn't write the fs2_open.log anymore...
I tried to run the debug-build to see if I can get information about the recent errors (which still exist ) and always when I looked into the fs2_open.log, it looked the same...
Today I looked at the file and.... Well... Last write to the file: 13. Apr....
Even if I delete it, it will not write to fs2_open.log...

EDIT:
MY FAULT!
I didn't run "make clean" before making the debug-build...
So it just linked "normal" files...

[fizz]

Originally posted by taylor

The cvs package contains 'rcs2log' which generates a Changelog.  That could be done automatically on the server side (don't know if that will happen though)

I'm with Maniac here! Having an up-to-date change log makes following development so much easier. Who do we need to bug for this to happen?

[Mr_Maniac]

Okay... Now everything looks normal again!
THX!

[Mr_Maniac]

New problem...
SEGFAULT when I try to start a mission...
If campaign or simulator... Always...
(Since two or three days)

gdb is useless as usual...
Only output:

Code: [Select]

(gdb) file fs2_open_r
Reading symbols from /home/maniac/.fs2_open/fs2_open_r...done.
Using host libthread_db library "/lib/libthread_db.so.1".
(gdb) run
Starting program: /home/maniac/.fs2_open/fs2_open_r
continue

Program received signal SIGSEGV, Segmentation fault.
0xb7c5c8eb in strlen () from /lib/libc.so.6
(gdb) continue
Continuing.

Program terminated with signal SIGSEGV, Segmentation fault.
The program no longer exists.
(gdb)

[taylor]

Type "bt" the (gdb) prompt, should give more info.  And this happens in the readyroom (continuing a campaign) and the simroom right?  Are you trying the same mission in the simroom or different ones?

[Mr_Maniac]

It happens when I click on "Start Mission"... The Briefing works..
Ship and Weapon-loadout works, too...
But as soon as I click on "Start", fs2_open crashes...

And fs2_open refuses to close... It remains in front of everything else...
No matter if it gets run in fullscreen or window...
And when I kill it and type "bt", then gdb gives me the same output ("0xb7c5c8eb in strlen () from /lib/libc.so.6") than before...

[Mr_Maniac]

Okay... Now I've recompiled fs2_open (again )...
This time gdb gave me some better output:

Code: [Select]

#0  0xb7c5c8eb in strlen () from /lib/libc.so.6
#1  0x0807d1e0 in NavPoint::GetInteralName (this=0x8413490) at autopilot.cpp:187
#2  0x0807d3a0 in DistanceTo (nav=9) at autopilot.cpp:805
#3  0x0807db94 in NavSystem_Do () at autopilot.cpp:483
#4  0x08051c3a in game_simulation_frame () at freespace.cpp:5137
#5  0x08055967 in game_frame (paused=0) at freespace.cpp:5617
#6  0x08055d73 in game_do_frame () at freespace.cpp:6091
#7  0x0809dd6d in gameseq_process_events () at gamesequence.cpp:657
#8  0x08057afa in WinMainSub (argc=9, argv=0x9) at freespace.cpp:8561
#9  0x08057b2f in main (argc=9, argv=0x9) at freespace.cpp:8623

[taylor]

You do have to "bt" before you kill it.

And this problem just looks like a NULL problem, one that I've fixed before.  I was getting a ton off NULL errors with the autopilot stuff before but something new is slipping through I guess.  I thought that I had that problem fixed already though I had never experienced it myself.  Add a:

Code: [Select]

if ( ((waypoint_list*)target_obj)->name == NULL )
    return NULL;
near the top of NavPoint::GetInteralName() and see if that helps any.  Also, what mission is this with?  I've only had these nav problems in a couple of nebula missions.  Probably needs to be a better fix somewhere else in the code but if this works I'll go ahead and commit until the real problem can be tracked down.