Jump to content
Eternal Lands Official Forums
Entropy

The Release Candidate is here!

Recommended Posts

Im in Willowvine forest inside [106 313] and cant get out. Door seems not to be working.

Share this post


Link to post
Share on other sites

When I open the hazardous storage at Viorel, there is junk shown int he storage window. The picture says it all, I think: http://img502.imageshack.us/img502/235/clipboard01kt9.png

 

When this corruption happens, I also get corruption in the Stats Window after I put the dung in storage.

 

To reproduce: talk to Viorel, open storage. If you don't see garbage lines, close the storage window, then open it again - you should now see garbage lines in the storage window. Now pick up dung from inventory and drop it in the storage, then press ctrl-a - you should now see corruption in the stats window.

 

Tried to reproduce (not on purpose, just found the guy, talked to him, got corrupted storage window, remembered the report, hit ctrl-a) and am now able to present you the second backtrace of today :confused:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 48003691246432 (LWP 7042)]
0x00000080004664b0 in ?? ()
(gdb) backtrace
#0  0x00000080004664b0 in ?? ()
#1  0x0000000000466f78 in draw_statf ()
#2  0x00000000004672bb in display_stats_handler ()
#3  0x000000000042d241 in draw_window ()
#4  0x000000000042d445 in display_windows ()
#5  0x0000000000425f2a in draw_scene ()
#6  0x0000000000447f8a in start_rendering ()
#7  0x0000000000447fdd in main ()

Share this post


Link to post
Share on other sites

Viorel solution:

 

Index: elc/storage.c
===================================================================
RCS file: /cvsroot/elc/elc/storage.c,v
retrieving revision 1.29
diff -r1.29 storage.c
66c66
<	   if (storage_win > 0) vscrollbar_set_bar_len(storage_win, STORAGE_SCROLLBAR_CATEGORIES, no_storage_categories - STORAGE_CATEGORIES_DISPLAY);
---
>	   if (storage_win > 0) vscrollbar_set_bar_len(storage_win, STORAGE_SCROLLBAR_CATEGORIES, (no_storage_categories - STORAGE_CATEGORIES_DISPLAY) > 0 ? (no_storage_categories - STORAGE_CATEGORIES_DISPLAY) : 1);

Edited by Alia

Share this post


Link to post
Share on other sites

The whole South Redmoon flower patch is bugged up. [158, 235]

 

1. All flowers are visual as Mullein and Daffodils.

2. The wormwood is missing by the house corner.

3. Even though they look as they do, they say harvesting Henbane for example.

4. Most of the flowers on the right/east side closest to storage say "Too far away".

 

All this cannot be intentional? It's messed up and SRM is/was known for it's rich flower patches.

Share this post


Link to post
Share on other sites

Windows XP client occasionally crashes when trying to 'get all' from a bag.

 

1st time:

 

killed a hawk in PV, clicked 'get all', feather appeared in inv. but bag didn't close, and feather was still in it (apparently) Clicked 'get all' second time and client crashed. After reconnecting inventory showed only 1 feather.

I saw this once myself while quickly click-bag/get-all at the bread/gc bag spawn on IP. In my case the bag also didn't close and the client locked-up. I didn't notice if I got the contents or not. I tried several time to reproduce it but I wasn't able to (so I didn't post). But since someone else saw it too...

Share this post


Link to post
Share on other sites

Another bag problem:

I click on one bag, but go to another and open it.

I've seen it twice already.

 

Edit: more information.

After some playing with bags I can get this: two different bags give the same obj_3d_id.

When I click bag in position A i go to bag in position B. When I click bag in position B i open bag B (there were more complicated situations)

If I take bag B, then I can open bag A. When I put another bag in position B, it happens again.

 

Maybe something is wrong with object_under_mouse?

Edited by Alia

Share this post


Link to post
Share on other sites

I crashed last night while walking in Lothalith Fortress (AA). I had just turned eye candy off because the room I was in had several fires and lots of colors etc. lowering my FPS. It was the room with the red carpet runner all the way through it. I stopped walking, turned off eye candy, closed the options window and clicked to walk. My char took about 2 steps and the client crashed. Could not reproduce after restarting the client.

Share this post


Link to post
Share on other sites
Viorel solution:
yay, thanks, one less thing on my TODO list :confused:

the fix is in CVS now

Share this post


Link to post
Share on other sites

Maybe something is wrong with object_under_mouse?

 

If you suspect that, avoid the mouse. As in, use home/delete/insert and drop something to open bag. Is your scenario still valid then?

Share this post


Link to post
Share on other sites

Just an update on crashes:

 

We've found two bugs that may stem from one source (hard to say yet). Through some miraculous bit of chance, I suddenly started crashing last night, consistently, so I was able to debug.

 

* Crashes from blood in many types of creatures, but only in specific situations

* Crashes from fighting specific types of creatures (armed orcs and giants are known, there may be some more), also only in specific situations.

 

The "only in specific situations" aspect hints at a buffer overflow.

 

The first bug can be narrowed down in code to actor_scripts.c, the lines starting at blood_level= and ending at ec_create_impact_blood. The ec_create_impact blood is known not to cause a problem (it happens even if you return right away), so it looks like something with the use of CalSkeleton (cal3d) -- either how we're making use of cal3d, or something wrong with cal3d itself.

 

The second bug hasn't been whittled down yet, but it happens whether eye candy is on or not, and only happens when fighting. So, it's not an eye candy bug, but that doesn't let us know exactly what kind of bug it is :icon13: Since it happens in fighting, actor_scripts seems a good place to start. However, it is interesting to note that it's creature-specific. Perhaps this, too, is a bug either in cal3d or in how we're using cal3d.

 

Either way, I'll work more on this come this evening. Huzzah -- reproducability!

Share this post


Link to post
Share on other sites

I just got the client today and was on the test server, I noticed I couldnt get my stats above my compus and clock. So someone told me to change my screen size to the largest one avalibly, I did but when my screen resized I couldnt take at all. I could still fight, and enter new maps, but I couldnt talk or exit the game no matter what I tried. I was in Glacmor, dont know what happened.

Edited by RunTime

Share this post


Link to post
Share on other sites

Another bag problem:

I click on one bag, but go to another and open it.

I've seen it twice already.

 

Edit: more information.

After some playing with bags I can get this: two different bags give the same obj_3d_id.

When I click bag in position A i go to bag in position B. When I click bag in position B i open bag B (there were more complicated situations)

If I take bag B, then I can open bag A. When I put another bag in position B, it happens again.

 

Maybe something is wrong with object_under_mouse?

 

Very interesting..

Can you reproduce it though?

 

Ok, found and fixed the chinstrap penguin bug, thanks for reporting it.

It will take effect at the next test server restart.

Share this post


Link to post
Share on other sites

Cannot really give coord right now but it's impossible to get out of the new Temple of Suffering, in willowine forest that is. (You enter it at some ruin like thing)

And also in the new? Tirnym catacomb, the exit doesn't work either. (You enter it at the graveyard, it's a door in the ground)

Both are just enters and are very easy to find and not dependant on any coord, but if you want coords for the enter I can get them I guess.

Share this post


Link to post
Share on other sites
I just got the client today and was on the test server, I noticed I couldnt get my stats above my compus and clock. So someone told me to change my screen size to the largest one avalibly, ...
This is not a bug. The Stats in HUD only works with resolutions 1024x768 and greater. This is because the skill list would otherwise overlap with the quickbar. The skill list was recently expanded to include engineering, and therefore the resolution restriction was placed on the stats.
... I did but when my screen resized I couldnt take at all. I could still fight, and enter new maps, but I couldnt talk or exit the game no matter what I tried. I was in Glacmor, dont know what happened.
I am not quite sure what you mean, "couldnt take at all". alt-x did not work for exiting the game? Are you saying the the text entry line was off the screen?

Share this post


Link to post
Share on other sites

Map bug:

Port Anitora 219 189.

There are some thing that seems to be suppose to be a table/showcase with 3 vials standing on it. But the whole top of the table is missing, it's like the vials are floating in the air.

Same goes for some "table" at 229, 200.

 

Update:

Glacmor - Wave Neck 66 299, you dissappear to underground.

Glacmor - Glacmor Skill Academy 345 134, you can walk through the "rock"/railing by the stairs. (Note that this is NOT inside, it's outside)

Iscalrith - 213 195, inside the ice stone/mountain thingy.

Edited by Beaverhunter

Share this post


Link to post
Share on other sites
I just got the client today and was on the test server, I noticed I couldnt get my stats above my compus and clock. So someone told me to change my screen size to the largest one avalibly, ...
This is not a bug. The Stats in HUD only works with resolutions 1024x768 and greater. This is because the skill list would otherwise overlap with the quickbar. The skill list was recently expanded to include engineering, and therefore the resolution restriction was placed on the stats.

 

I suppose using a smaller font, would be new feature, rather then a bug fix? It's not that I (and others) have small screens, but like stuff around the client, that we use 800x600. It just about fits if you start with y=11 in hud.c:960. I agree it ain't pretty then. Changing the vertical bounding box for the font (and reducing fontsize) to 13 rather then 15 should solve the problem entirely.

 

Fix for today's cvs, when compiling debug build (-Wall -Werror) [edit] found a few more[/edit]:

Index: frustum.c
===================================================================
RCS file: /cvsroot/elc/elc/frustum.c,v
retrieving revision 1.28
diff -u -r1.28 frustum.c
--- frustum.c   17 May 2007 00:09:09 -0000	  1.28
+++ frustum.c   17 May 2007 21:01:21 -0000
@@ -465,7 +465,7 @@
	MATRIX4x4 modl;														 // This will hold our modelview matrix
	MATRIX4x4 clip;														 // This will hold the clipping planes
	unsigned int cur_intersect_type;
-	   VECTOR3 p1, p2, p3;
+	   /* XXX Unused: VECTOR3 p1, p2, p3; */

	if (main_bbox_tree->intersect[INTERSECTION_TYPE_DEFAULT].intersect_update_needed == 0) return;
#else
Index: gl_init.c
===================================================================
RCS file: /cvsroot/elc/elc/gl_init.c,v
retrieving revision 1.113
diff -u -r1.113 gl_init.c
--- gl_init.c   11 May 2007 05:56:09 -0000	  1.113
+++ gl_init.c   17 May 2007 21:01:21 -0000
@@ -1037,8 +1037,8 @@
#endif

	if(have_vertex_buffers){
-			   e3d_object * obj;
#ifndef		NEW_E3D_FORMAT
+			   e3d_object * obj;
			for(i=0;i<cache_e3d->max_item;i++){
					if(!cache_e3d->cached_items[i])continue;
					obj=cache_e3d->cached_items[i]->cache_item;
Index: elwindows.c
===================================================================
RCS file: /cvsroot/elc/elc/elwindows.c,v
retrieving revision 1.90
diff -u -r1.90 elwindows.c
--- elwindows.c 11 May 2007 05:56:09 -0000	  1.90
+++ elwindows.c 17 May 2007 21:01:22 -0000
@@ -698,7 +698,7 @@
void   destroy_window(int win_id)
{
	window_info *win;
-	   widget_list *widget, *next;
+	   /* XXX Unused: widget_list *widget, *next; */

	if(win_id < 0 || win_id >= windows_list.num_windows)	return;
	if(windows_list.window[win_id].window_id != win_id)	 return;
Index: io/e3d_io.c
===================================================================
RCS file: /cvsroot/elc/elc/io/e3d_io.c,v
retrieving revision 1.12
diff -u -r1.12 e3d_io.c
--- io/e3d_io.c 9 May 2007 22:50:04 -0000	   1.12
+++ io/e3d_io.c 17 May 2007 21:01:22 -0000
@@ -82,7 +82,9 @@
{
	e3d_header header;
	e3d_material material;
+#ifdef USE_EXTRA_TEXTURE
	e3d_extra_texture extra_texture;
+#endif
	char cur_dir[1024];
	int i, j, l, idx, mem_size, vertex_size, material_size;
	int file_pos, indicies_size, float_count, index_size, v_size, m_size;

 

If this is mangled, click while I'm online :icon13:

Edited by RallosZek

Share this post


Link to post
Share on other sites

If you suspect that, avoid the mouse. As in, use home/delete/insert and drop something to open bag. Is your scenario still valid then?

 

If i open bag by dropping it opens right bag. I log what obj_id and bag id I open, so i'm pretty sure problem in with wrong

obj_3d_id.

 

Entropy: I have no exact recipe how to do it, but I can reproduce it. I just walk at bag spawn on ip and get all / drop

items. After 1-2 mins I encounter this bug :icon13:

 

Edit: btw, it is even more interestin: if I click on on such confusing bag standing on one point I go to one bag, but if i stand on another point i can go to a different bag!

 

Entropy: AFAIK it doesn't happen with old client (that uses old 3d format). And it seems that it begins when new bag spawns (although i'm not sure in it)

Edited by Alia

Share this post


Link to post
Share on other sites
I just got the client today and was on the test server, I noticed I couldnt get my stats above my compus and clock. So someone told me to change my screen size to the largest one avalibly, ...
This is not a bug. The Stats in HUD only works with resolutions 1024x768 and greater. This is because the skill list would otherwise overlap with the quickbar. The skill list was recently expanded to include engineering, and therefore the resolution restriction was placed on the stats.
one note, as of about an hour ago (in CVS, not the release candidate yet), you can have statbars in 800x600, but you have to turn off the digital clock

Share this post


Link to post
Share on other sites
one note, as of about an hour ago (in CVS, not the release candidate yet), you can have statbars in 800x600, but you have to turn off the digital clock

Thanks for doing this. I thought about it, but I was a little lazy.

 

Shadowing bug on OS X:RC2.png

Previous build without NEW_E3D_FORMAT:RC1.png

 

As seen on the table, there is a face missing and the shadow is a few units below the ground. The sign has its shadow messed up too.

[edit] I am at Port Anitora

Edited by 0ctane

Share this post


Link to post
Share on other sites

Glacmor - Wave Neck 66 299, you dissappear to underground.

 

Few more spots:

You are in Glacmor - Wave Neck [45,295]

You are in Glacmor - Wave Neck [46,294]

You are in Glacmor - Wave Neck [47,293]

You are in Glacmor - Wave Neck [48,292]

You are in Glacmor - Wave Neck [49,291]

You are in Glacmor - Wave Neck [50,290] < above ground again.

 

You are in Glacmor - Wave Neck [52,291]

You are in Glacmor - Wave Neck [53,292]

You are in Glacmor - Wave Neck [54,293] < above ground again.

 

I think that spot goes a bit further, cause a polar just walked outof that rock :icon13:

 

Update, revisiting some oldies:

You are in Irinveron [60,464]

 

Most of the rocks on that border are not solid.

 

Random crashes when walking in irinveron. Now, I normally do not use shadows and both of these happened with shadows on. Will see if I can reproduce without shadows, cause even when run under gdb, the backtrace isn't very useful:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xba33000 (LWP 100164)]
0x29eb0140 in ?? ()
(gdb) bt
#0  0x29eb0140 in ?? ()
#1  0x29665000 in ?? ()
#2  0x000001fe in ?? ()
#3  0x0000002a in ?? ()
#4  0x082658c8 in meshFaces.3 ()
#5  0x28e7d74a in _nv001254gl () from /usr/X11R6/lib/libGLcore.so.1
#6  0x08265820 in meshTextureCoordinates.2 ()
#7  0x082658c8 in meshFaces.3 ()
#8  0x00000000 in ?? ()
#9  0x00001e00 in ?? ()
#10 0x00001e00 in ?? ()
#11 0x00000000 in ?? ()
#12 0x00000000 in ?? ()
#13 0x00000207 in ?? ()
#14 0x000000ff in ?? ()
#15 0x00000000 in ?? ()
#16 0x00000070 in ?? ()
#17 0x00000004 in ?? ()
Previous frame inner to this frame (corrupt stack?)

Edited by RallosZek

Share this post


Link to post
Share on other sites

keyboard bug(from pre RC) where you aren't able to use keyboard.

 

I've looked into this problem: It happens this way.

When you press a key you receive SDL_KEYDOWN event, but we handle it ONLY if have_keyboard_focus is true

which is set on/off by receiving SDL_ACTIVEEVENT. Usually we receive it when we actiave/deactivate our window,

but when we load map we dont receive this event on window activation (why? yes we dont poll events while

loading map, but they should wait in the queue...)

 

Anyway how can we receive SDL_KEYDOWN if we dont have focus? :)

If this check was done for a reason, tell it, if not here is fix for 2coolfool's problem

 

Index: elc/events.c
===================================================================
RCS file: /cvsroot/elc/elc/events.c,v
retrieving revision 1.133
diff -r1.133 events.c
79,81d78
< 			if(!have_keyboard_focus){
< 				break;  //don't have focus, so we shouldn't be getting keystrokes
< 			}
108d104
< 				have_keyboard_focus = event->active.gain;

Edited by Alia

Share this post


Link to post
Share on other sites

The reason we did that is because SDL sometimes steals the keys from other applications even if it doesn't have the focus, so people were complaining about their bank passwords and stuff appearing in the el chat.. It almost happened to me too, so I understand them.

Share this post


Link to post
Share on other sites

revision 1.133

date: 2007/03/23 21:11:30; author: ttlanhil; state: Exp; lines: +12 -16

don't accept keystrokes if we don't have keyboard focus. remove un-needed #ifdef's

 

I think this fixes a problem when typing in an application like yahoo messenger (with always on top?), it's echo'd on el. Seems to only happen on windows, never seen it myself.

 

As for the crash above, I can't reproduce it at night, with or without shadows. Here's the scenario (reproduced twice with daylight):

0. Enable shadows

1. Enter and exit blacksmith in Irinveron [322,544]

2. mapwalk to the forge in the south [248,129]

 

Somewhere along the way (different points) I crashed. From what I can tell, 3 different causes:

1) bug in nvidia's shadow support (nvidia GeForce4 4200 Go/AGP/SSE2, OpenGL version string: 1.5.8 NVIDIA 96.31, FreeBSD 6.2-STABLE)

2) bug in el's shadow support

3) I encountered something along the way, that crashed me two times and didn't see what it was.

 

When I relog in and reenable shadows, I didn't crash again. So I hope someone can reproduce at daylight, with and/or without shadows.

Relevant settings:

#no_adjust_shadows = 0

#use_shadow_mapping= 1

#max_shadow_map_size = 1

#clouds_shadows = 1

#shadow_map_size = 1

Edited by RallosZek

Share this post


Link to post
Share on other sites
Guest
This topic is now closed to further replies.

  • Recently Browsing   0 members

    No registered users viewing this page.

×