Current CVS errors

[0ctane]

Okay, I got everything built. Just so that everyone knows, my attempts at using the cal3d SVN failed with segfaults on login (similar to a problem Pardu had a while ago). I was able to grab the official 0.11.0 codebase and build it.

 

Now, to figure out the audio stuff for OS X. =)

[Grum]

Is no one else seeing this error:

/Users/jeff/programs/el_osx/elc/font.c:1036: error: 'zoom_level' undeclared (first use in this function)

Everyone's building with SKY_FPV_CURSOR, apparently

Shouldn't draw_scene.h be included in font.c since it has the extern call?

 

Also, I had to add #include "global.h" to reflection.c in order for it to find cur_time.

Both fixed.

[Grum]

I got

#0  0x000000000047a90f in build_path_map () at map.c:189
#1  0x000000000047a829 in el_load_map (
file_name=0x7fff4c8bb170 "./maps/cont2map21_maze5.elm")
at map.c:214
#2  0x000000000047a9c3 in change_map (
mapname=0x7fff4c8bb170 "./maps/cont2map21_maze5.elm")
at map.c:243
#3  0x000000000047f891 in process_message_from_server (
in_data=0x2aaab00009a0 "\a\035", data_length=31)
at multiplayer.c:793
#4  0x0000000000478631 in start_rendering () at main.c:155
#5  0x0000000000478a60 in main (argc=1, argv=0x7fff4c8bd728)
at main.c:290 

upon entering (or rather trying to enter) said map, my CVS client is up to date (just checked), I'm building with

default options, plus new_sound, new_actor_scale, minimap, notepad, new_file_io.

Okay, so I didn't wait until tonight.

 

What threw me off was the fact that it was trying to initialize the path finder map. I assumed that it wouldn't do so if loading the map had failed. That'll teach me not to make assumptions So, in the end, I think this issue is exactly the same as this one, i.e. loading the map failed due to a broken check on particle system size. The original issue should already be fixed, and I've now also added a check on the return value of load_map().

 

Can you confirm that it's fixed?

[Florian]

Is no one else seeing this error:

/Users/jeff/programs/el_osx/elc/font.c:1036: error: 'zoom_level' undeclared (first use in this function)

Everyone's building with SKY_FPV_CURSOR, apparently

Ah, that explains why your reflections are broken. Try w/o SKY_FPV_CURSOR 0ctane and you'll hopefully get all the new shader eye candy.

[0ctane]

Ah, that explains why your reflections are broken. Try w/o SKY_FPV_CURSOR 0ctane and you'll hopefully get all the new shader eye candy.

I have not been using SKY_FPV_CURSOR, but I have shaders working. The problem on my Mac (x386) is a bit complicated, as the following screenshots will hopefully illustrate.

 

With reflections turned on, but water shader quality at 0:

Hopefully you see that the terrain not affecting the reflection calculation is gone, poof.

 

Now, if I turn water shader quality to 1 or 2, reflections and the terrain appear just fine. Here is at 2:

Woohoo! So, I just need to make sure that water shader quality != 0 (or someone needs to fix the underlying problem).

 

Now with shadows turned on (w/o shadow mapping), I get random textures appearing in the water:

 

However, if I turn on shadow mapping:

Unfortunately, even though the shadows seem to work, they do not blend properly with the water environment. They are very harsh, and look very peculiar when it is raining.

 

Still, it is progress...

[Torg]

Now with shadows turned on (w/o shadow mapping), I get random textures appearing in the water:

This is similar to someone else running under Windows. They had face textures appearing with shadows.

 

I'm wondering if someone has done something odd in the shadow code.

[Grum]

Thread here.

 

This sort of thing makes me very nervous. Weird textures appearing is one step removed from outright crashing.

[0ctane]

Xaphier was telling me a while ago (right after the 1.4 release) that regular shadows were going to be no longer supported, and that everyone would need shadow mapping in order to do shadows.

 

That thread is similar, although the weird textures are only showing up for water reflections in my case. It seems to rotate through various texture files. Sometimes I see faces, sometimes flag emblems, sometimes books, ....

[Ermabwed]

I got

#0  0x000000000047a90f in build_path_map () at map.c:189
#1  0x000000000047a829 in el_load_map (
file_name=0x7fff4c8bb170 "./maps/cont2map21_maze5.elm")
at map.c:214
#2  0x000000000047a9c3 in change_map (
mapname=0x7fff4c8bb170 "./maps/cont2map21_maze5.elm")
at map.c:243
#3  0x000000000047f891 in process_message_from_server (
in_data=0x2aaab00009a0 "\a\035", data_length=31)
at multiplayer.c:793
#4  0x0000000000478631 in start_rendering () at main.c:155
#5  0x0000000000478a60 in main (argc=1, argv=0x7fff4c8bd728)
at main.c:290 

upon entering (or rather trying to enter) said map, my CVS client is up to date (just checked), I'm building with

default options, plus new_sound, new_actor_scale, minimap, notepad, new_file_io.

Okay, so I didn't wait until tonight.

 

What threw me off was the fact that it was trying to initialize the path finder map. I assumed that it wouldn't do so if loading the map had failed. That'll teach me not to make assumptions So, in the end, I think this issue is exactly the same as this one, i.e. loading the map failed due to a broken check on particle system size. The original issue should already be fixed, and I've now also added a check on the return value of load_map().

 

Can you confirm that it's fixed?

Yes, it's fixed

[Florian]

I switched to 0ctane's cal3d 0.11 Framework yesterday.

 

Now I get lots of those during "initializing actor definitions":

el_osx(931,0xa000d000) malloc: *** Deallocation of a pointer not malloced: 0xad28aa94; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug

 

So I set MallocBadFreeAbort=1 so that gdb aborts and prints a bt:

 

(gdb) bt full

#0 0x9003d66c in kill ()

No symbol table info available.

#1 0x9010e8cf in raise ()

No symbol table info available.

#2 0x9010d422 in abort ()

No symbol table info available.

#3 0x9000558f in free ()

No symbol table info available.

#4 0x04fc8c31 in operator delete (ptr=0xad28aa94) at ../../../../gcc-4.2.1/libstdc++-v3/libsupc++/del_op.cc:49

ptr = (void *) 0x0

#5 0x04f9ed9b in std::string::_Rep::_M_destroy (this=0xad28aa94, __a=@0xbfffdf5b) at /usr/src/gcc-4.2.1-build/i386-apple-darwin8.10.1/libstdc++-v3/include/ext/new_allocator.h:97

this = (basic_string<char,std::char_traits<char>,std::allocator<char> >::_Rep * const) 0x0

#6 0x04fa030e in std::string::assign (this=0xbfffe080, __str=@0xbfffdf90) at /usr/src/gcc-4.2.1-build/i386-apple-darwin8.10.1/libstdc++-v3/include/bits/basic_string.h:238

this = (basic_string<char,std::char_traits<char>,std::allocator<char> > * const) 0xbfffe080

__str = (const basic_string<char,std::char_traits<char>,std::allocator<char> > &) @0x0: Cannot access memory at address 0x0

 

Anyways, without MallocBadFreeAbort=1 EL loads just fine but spams STDERR ...

[Grum]

This is not the full backtrace, is it? The deepest frame (#6) is still in the c++ library. Can you post a few more (or all, if it isn't 13 pages of backtrace) levels so that we can see where in EL this is triggered?

[Florian]

This is the full backtrace ...

 

I found the reason for this behaviour (and a crash when exiting the client):

- I use gcc 4.2.1 in /usr/local/*

- 0ctane uses Apple's gcc 4.0.x in /usr/*

- I link to /usr/local/lib/* but 0ctane's Frameworks are linked to /usr/lib/*

 

somewhere the stdc++ lib gets messed up.

 

Now I recompiled with Apple's standard settings and the malloc issues are gone, also the client-exit-crash.

Edited October 3, 2007 by Florian

[0ctane]

I found the reason for this behaviour (and a crash when exiting the client):

- I use gcc 4.2.1 in /usr/local/*

- 0ctane uses Apple's gcc 4.0.x in /usr/*

- I link to /usr/local/lib/* but 0ctane's Frameworks are linked to /usr/lib/*

Yeah, I have to use all the standard Apple libraries since I cannot depend on people having installed additional software. This is one reason that I no longer use fink or opendarwin, since I cannot afford to have my libraries all messed up when deploying a product.

[Florian]

I still get those crashes quite often:

 

#0  0x03d1034a in CalAnimation::getCoreAnimation ()
No symbol table info available.
#1  0x03d26fe9 in CalMixer::updateSkeleton ()
No symbol table info available.
#2  0x03d2836e in CalModel::update ()
No symbol table info available.
#3  0x00029df3 in cal_render_actor (act=0x16d9a200) at cal.c:345
	pCalRenderer = (struct CalRenderer *) 0x92b23141
	meshCount = -1073748464
	meshId = -1073748408
	submeshCount = -1833815653
	points = {...}
	meshNormals = {...}
	meshTextureCoordinates = {...}
	meshFaces = {...}
	skel = (struct CalSkeleton *) 0x1a72a1b0
	_mesh = (struct CalMesh *) 0x0
	_coremesh = (struct CalCoreMesh *) 0x0
	_weaponmesh = (struct CalCoreMesh *) 0x3f800000
	_shieldmesh = (struct CalCoreMesh *) 0x0
	reverse_scale = 115.75
#4  0x000adce2 in draw_actor_shadow (actor_id=0x16d9a200) at shadows.c:463
	x_pos = 115.5
	y_pos = 170
	z_pos = 0
	x_rot = 0
	y_rot = 0
	z_rot = 0
#5  0x000adde5 in display_actors_shadow () at shadows.c:486
	act = (actor *) 0x16d9a200
	i = 2
	x = 114
	y = 164
#6  0x000adf09 in display_shadows () at shadows.c:523
No locals.
#7  0x000ae606 in render_light_view () at shadows.c:678
	cur_intersect_type = 0
#8  0x0005919d in display_game_handler (win=0x5194400) at gamewin.c:954
	main_count = 450341
	times_FPS_below_3 = 0
	fps = {39, 39, 39, 38, 40}
	shadows_were_disabled = 0
	eye_candy_was_disabled = 0
	str = "\000\000\000\000\000\000\000\000\000\000\200?\000\000\000\000È\220\233\021¹\204\233\021\001\000\000\000(èÿ¿I\205\233\021 ÀÚ\021\000\000\000\000\000\000\000\000\000\000\200?Dê\003\005Dèÿ¿\031\000\000\000\001\000\000\000 ÀÚ\021\004\000\000\000 ÀÚ\021Õ-²\222\210y\236\021\f ²¢Hèÿ¿jË\235\021 ÀÚ\021\000\000\000\000\001\000\000\000¦\211\236\021ÐéÚ\021Pèÿ¿xèÿ¿\032.²\222 ÀÚ\021á\r\000\000úøø=\000\000\000\000¾¾¾>   >\230èÿ¿±T²\222 ÀÚ\021\000\000\200?\000\000\200?\000\000\200?"
	i = 0
	any_reflection = 2
	mouse_rate = 1
#9  0x00049292 in draw_window (win=0x5194400) at elwindows.c:1135
	ret_val = 0
	W = (widget_list *) 0x597aec0
#10 0x00049b4a in display_window (win_id=0) at elwindows.c:1302
No locals.
#11 0x000461b8 in display_windows (level=1) at elwindows.c:72
	id = -1
	next_id = -9999
	i = 0
#12 0x0003d0d9 in draw_scene () at draw_scene.c:116
No locals.
#13 0x000819cf in start_rendering () at main.c:168
	event = {
 type = 24 '\030', 
 active = {
type = 24 '\030', 
gain = 126 '~', 
state = 18 '\022'
 }, 
 key = {
type = 24 '\030', 
which = 126 '~', 
state = 18 '\022', 
keysym = {
  scancode = 2 '\002', 
  sym = SDLK_UNKNOWN, 
  mod = 1191414493, 
  unicode = 13204
}
 }, 
 motion = {
type = 24 '\030', 
which = 126 '~', 
state = 18 '\022', 
x = 2, 
y = 0, 
xrel = 0, 
yrel = 0
 }, 
 button = {
type = 24 '\030', 
which = 126 '~', 
button = 18 '\022', 
state = 176 '°', 
x = 2, 
y = 0
 }, 
 jaxis = {
type = 24 '\030', 
which = 126 '~', 
axis = 18 '\022', 
value = 2
 }, 
 jball = {
type = 24 '\030', 
which = 126 '~', 
ball = 18 '\022', 
xrel = 2, 
yrel = 0
 }, 
 jhat = {
type = 24 '\030', 
which = 126 '~', 
hat = 18 '\022', 
value = 176 '°'
 }, 
 jbutton = {
type = 24 '\030', 
which = 126 '~', 
button = 18 '\022', 
state = 176 '°'
 }, 
 resize = {
type = 24 '\030', 
w = 2, 
h = 0
 }, 
 expose = {
type = 24 '\030'
 }, 
 quit = {
type = 24 '\030'
 }, 
 user = {
type = 24 '\030', 
code = 2, 
data1 = 0x0, 
data2 = 0x47038add
 }, 
 syswm = {
type = 24 '\030', 
msg = 0x2
 }
}
	done = 0
	network_thread_data = {0x13517290, 0x36c19c}
	network_thread = (SDL_Thread *) 0x169d4e00
	message_queue = (queue_t *) 0x13517290
#14 0x00081d9c in SDL_main (argc=1, argv=0x5907050) at main.c:290
No locals.
#15 0x0018b615 in -[SDLMain applicationDidFinishLaunching:] (self=0x591e230, _cmd=0x19e450, note=0x59115b0) at SDLmain.m:197
	status = -1868196602

[0ctane]

Going along with my previous post about random textures in the water, I just compiled with SKY_FPV_CURSORS, and I notice the sky is rotating through textures also. It is nighttime in game, so I don't know exactly what is going on. Also, it seems like with SKY_FPV_CURSORS, I no longer get reflections in the water, which might be related.

 

My bad:

[11:44:26] File './textures/clouds.bmp' not found
[11:44:26] File './textures/cloud_detail.bmp' not found
[11:44:26] File './textures/cloud_detail_alpha.bmp' not found
[11:44:26] File './textures/moonmap.bmp' not found
[11:44:26] File './textures/BrightSun.bmp' not found

Where can I get these? Found them.

Edited October 3, 2007 by 0ctane

[Beaverhunter]

Oki, been away abit but no errors except for the link errors are still present, that is:

 

1>Linking...
1>cal3d.lib(cal3d.dll) : error LNK2005: _CalSkeleton_GetBoneBoundingBox already defined in cal3d_wrapper.obj
1>cal3d.lib(cal3d.dll) : error LNK2005: _CalSkeleton_CalculateBoundingBoxes already defined in cal3d_wrapper.obj
1>C:\dev\Eternal Lands\EternalLands.exe : fatal error LNK1169: one or more multiply defined symbols found

 

 

I can get elc to compile without errors if I comment these in cal3d_wrapper.cpp:

extern "C" CAL3D_WRAPPER_API void CalSkeleton_GetBoneBoundingBox(CalSkeleton *self, float *min, float *max)
{
self->getBoneBoundingBox(min, max);
}

extern "C" CAL3D_WRAPPER_API void CalSkeleton_CalculateBoundingBoxes(struct CalSkeleton *self)
{
self->calculateBoundingBoxes();
}

 

So are these functions even needed?

With 0.11 rc2 they seem to be included inside cal3d.lib already.

Do they differ in some way? If so, any suggestions to what I should do?

[0ctane]

Since it seems like many people are using the SKY_FPV_CURSORS define, I gave it a try. It looks awesome (at first) but I also lose reflections.

 

Compiled without FPV, reflections on:

 Scene and actors reflect fine with shaders.

Compiled with FPV, reflections off:

Obviously no reflections, but water is very bright.

Compiled with FPV, reflections on:

Still no reflections (although turned on), and water has become brown (reflecting early morning sky?).

 

I still get the harsh shadowing from the environment, although actor shadows are soft.

[Florian]

Linker warnings:

 

/usr/bin/ld: warning multiple definitions of symbol _CalCoreAnimation_Scale
cal3d_wrapper.o definition of _CalCoreAnimation_Scale in section (__TEXT,__text)
/Library/Frameworks/cal3d.framework/cal3d(single module) definition of _CalCoreAnimation_Scale
/usr/bin/ld: warning multiple definitions of symbol _CalCoreMesh_Scale
cal3d_wrapper.o definition of _CalCoreMesh_Scale in section (__TEXT,__text)
/Library/Frameworks/cal3d.framework/cal3d(single module) definition of _CalCoreMesh_Scale
/usr/bin/ld: warning multiple definitions of symbol _CalCoreSkeleton_Scale
cal3d_wrapper.o definition of _CalCoreSkeleton_Scale in section (__TEXT,__text)
/Library/Frameworks/cal3d.framework/cal3d(single module) definition of _CalCoreSkeleton_Scale
/usr/bin/ld: warning multiple definitions of symbol _CalMixer_RemoveAction
cal3d_wrapper.o definition of _CalMixer_RemoveAction in section (__TEXT,__text)
/Library/Frameworks/cal3d.framework/cal3d(single module) definition of _CalMixer_RemoveAction
/usr/bin/ld: warning multiple definitions of symbol _CalSkeleton_CalculateBoundingBoxes
cal3d_wrapper.o definition of _CalSkeleton_CalculateBoundingBoxes in section (__TEXT,__text)
/Library/Frameworks/cal3d.framework/cal3d(single module) definition of _CalSkeleton_CalculateBoundingBoxes
/usr/bin/ld: warning multiple definitions of symbol _CalSkeleton_GetBoneBoundingBox
cal3d_wrapper.o definition of _CalSkeleton_GetBoneBoundingBox in section (__TEXT,__text)
/Library/Frameworks/cal3d.framework/cal3d(single module) definition of _CalSkeleton_GetBoneBoundingBox

 

cal3d 0.11, latest CVS

patch to remove warnings: http://www.superfloh.dyndns.org/ELwiki/index.php/Cal3d

Edited October 3, 2007 by Florian

[Learner]

Linker warnings:

 

/usr/bin/ld: warning multiple definitions of symbol _CalCoreAnimation_Scale
cal3d_wrapper.o definition of _CalCoreAnimation_Scale in section (__TEXT,__text)
/Library/Frameworks/cal3d.framework/cal3d(single module) definition of _CalCoreAnimation_Scale
/usr/bin/ld: warning multiple definitions of symbol _CalCoreMesh_Scale
cal3d_wrapper.o definition of _CalCoreMesh_Scale in section (__TEXT,__text)
/Library/Frameworks/cal3d.framework/cal3d(single module) definition of _CalCoreMesh_Scale
/usr/bin/ld: warning multiple definitions of symbol _CalCoreSkeleton_Scale
cal3d_wrapper.o definition of _CalCoreSkeleton_Scale in section (__TEXT,__text)
/Library/Frameworks/cal3d.framework/cal3d(single module) definition of _CalCoreSkeleton_Scale
/usr/bin/ld: warning multiple definitions of symbol _CalMixer_RemoveAction
cal3d_wrapper.o definition of _CalMixer_RemoveAction in section (__TEXT,__text)
/Library/Frameworks/cal3d.framework/cal3d(single module) definition of _CalMixer_RemoveAction
/usr/bin/ld: warning multiple definitions of symbol _CalSkeleton_CalculateBoundingBoxes
cal3d_wrapper.o definition of _CalSkeleton_CalculateBoundingBoxes in section (__TEXT,__text)
/Library/Frameworks/cal3d.framework/cal3d(single module) definition of _CalSkeleton_CalculateBoundingBoxes
/usr/bin/ld: warning multiple definitions of symbol _CalSkeleton_GetBoneBoundingBox
cal3d_wrapper.o definition of _CalSkeleton_GetBoneBoundingBox in section (__TEXT,__text)
/Library/Frameworks/cal3d.framework/cal3d(single module) definition of _CalSkeleton_GetBoneBoundingBox

 

cal3d 0.11, latest CVS

patch to remove warnings: http://www.superfloh.dyndns.org/ELwiki/index.php/Cal3d

Since not everyone is running the same version of Cal3D yet, the extra functions should be #ifdef'd out, not deleted.

[0ctane]

Linker warnings:

...

cal3d 0.11, latest CVS

patch to remove warnings: http://www.superfloh.dyndns.org/ELwiki/index.php/Cal3d

Florian,

Are you using my framework? I have it built against the stable 0.11.0 release. I noticed problems with the SVN, but 0.11.0 seems happy.

 

I agree with Learner. Maybe we need (eek) a CAL3D_SVN define.

[Grum]

I agree with Learner. Maybe we need (eek) a CAL3D_SVN define.

And of course there's no easy way to detect the installed library version in C code, since it's only exported in a C++ header file. Which means we can't do this automatically, so we'd have to add a yet another define in the makefiles. I was just trying to get rid of a few of them...

 

EDIT: and anyway, the version nr is not #defined, so we can't use it for conditional compilation anyway

Edited October 4, 2007 by Grum

[Sayre]

Not entirely certain where to put this, but it is a current CVS issue so...

 

I am compiling with the new sound and effects etc. I can't uncheck the enable sound effects checkbox, hence the sound effects are always active.

 

bug or feature?

 

S.

[Torg]

Not entirely certain where to put this, but it is a current CVS issue so...

Heh. That's 2. Now I am kinda worried.

Aisy chose the NEW_SOUND thread. Probably the best place because it isn't default yet.

 

I am compiling with the new sound and effects etc. I can't uncheck the enable sound effects checkbox, hence the sound effects are always active.

When did you update CVS?? I fixed the init code last night (or at least I thought I did).

 

bug or feature?

Hehehe, bug. Sometimes you do need to turn off the sounds.

[Aislinn]

Not entirely certain where to put this, but it is a current CVS issue so...

Heh. That's 2. Now I am kinda worried.

Aisy chose the NEW_SOUND thread. Probably the best place because it isn't default yet.

 

I am compiling with the new sound and effects etc. I can't uncheck the enable sound effects checkbox, hence the sound effects are always active.

When did you update CVS?? I fixed the init code last night (or at least I thought I did).

I did a fresh cvs a few minutes ago and my dims sound button was fixed. (I edited my post in the other thread, not sure if you saw that).

[Florian]

 

Are you using my framework? I have it built against the stable 0.11.0 release. I noticed problems with the SVN, but 0.11.0 seems happy.

 

Yes, definitely your Debug-Build of cal3d 0.11:

#0  0x04809e53 in CalAnimation::getCoreAnimation (this=0x0) at /Users/jeff/Desktop/cal3d-0.11.0/macosx/../src/cal3d/animation.cpp:51
No locals.
#1  0x0482c4c0 in CalMixer::updateSkeleton (this=0x1a6ae190) at /Users/jeff/Desktop/cal3d-0.11.0/macosx/../src/cal3d/mixer.cpp:463
	pCoreAnimation = (CalCoreAnimation *) 0x17930fa0
	listCoreTrack = (list<CalCoreTrack*,std::allocator<CalCoreTrack*> > &) @0x17930fb8: {
 <_List_base<CalCoreTrack*,std::allocator<CalCoreTrack*> >> = {
_M_impl = {
  <allocator<std::_List_node<CalCoreTrack*> >> = {
	<new_allocator<std::_List_node<CalCoreTrack*> >> = {<No data fields>}, <No data fields>}, 
  members of _List_base<CalCoreTrack*,std::allocator<CalCoreTrack*> >::_List_impl: 
  _M_node = {
	_M_next = 0x17930ee0, 
	_M_prev = 0x179345f0
  }
}
 }, <No data fields>}
	iteratorCoreTrack = {
 _M_node = 0x17930fb8
}
	pSkeleton = (CalSkeleton *) 0x19769e70
	vectorBone = (vector<CalBone*,std::allocator<CalBone*> > &) @0x19769e74: {
 <_Vector_base<CalBone*,std::allocator<CalBone*> >> = {
_M_impl = {
  <allocator<CalBone*>> = {
	<new_allocator<CalBone*>> = {<No data fields>}, <No data fields>}, 
  members of _Vector_base<CalBone*,std::allocator<CalBone*> >::_Vector_impl: 
  _M_start = 0x1a6ae3d0, 
  _M_finish = 0x1a6ae45c, 
  _M_end_of_storage = 0x1a6ae45c
}
 }, <No data fields>}
	iteratorAnimationAction = {
 _M_node = 0x1a6ae1ac
}
	iteratorAnimationCycle = {
 _M_node = 0x348
}
...