#pentagram@irc.freenode.net logs for 3 Jan 2013 (GMT)

Archive Today Yesterday Tomorrow
Pentagram homepage


[02:58:16] --> Kirben_ has joined #pentagram
[02:58:16] --- ChanServ gives channel operator status to Kirben_
[02:59:42] <-- Kirben has left IRC (Ping timeout: 276 seconds)
[04:25:07] <-- SugarCube has left IRC (Ping timeout: 244 seconds)
[04:38:10] --> SugarCube has joined #pentagram
[08:28:17] <wjp> Kirben_: thanks for updating the snapshot so quickly
[08:28:25] <wjp> oh, and happy new year :-)
[08:39:08] <wjp> (8 year old bug fixed. Whee :-) )
[09:08:30] <-- Kirben_ has left IRC (Ping timeout: 264 seconds)
[09:09:22] --> Kirben has joined #pentagram
[09:09:23] --- ChanServ gives channel operator status to Kirben
[12:51:21] <-- Kirben has left IRC ()
[17:48:06] * wjp ponders
[20:42:22] --> Malignant_Manor has joined #pentagram
[20:44:43] <Malignant_Manor> wjp, Colourless, Would you add ogg support? There is someone on the Exult forums who will provide the music. http://exult.sourceforge.net/forum/read.php?f=1&i=472162&t=339357
[20:50:37] <wjp> I'm going to leave this for Colourless to answer
[20:50:43] <wjp> (but he may be asleep now)
[20:51:06] <wjp> one complication may be the transitions between midi tracks
[21:01:51] <wjp> and, wow, that thread is getting long
[21:02:24] <Malignant_Manor> Yeah
[21:02:47] <Malignant_Manor> Maybe you can have an official beta release even though it is buggy.
[21:08:18] <wjp> there, that one was yours :-)
[21:08:36] * wjp closes one more bug report
[21:09:24] <Malignant_Manor> Now, fix the falling through walls bug.
[21:09:40] <Malignant_Manor> I mean floors, not walls.
[21:10:25] <wjp> I should to add a ton of debugging code to get a handle on those
[21:10:29] <wjp> s/to//
[21:11:39] <Malignant_Manor> Then, there's the Avatar freeze bug that's probably harder to reproduce. https://sourceforge.net/tracker/?func=detail&aid=2104649&group_id=53819&atid=471706
[21:13:25] <wjp> looks like I still have the exact same reaction to that log as in 2008 :-)
[21:14:32] <Malignant_Manor> At least, you can save and reload from that bug.
[21:16:07] <wjp> although actually...
[21:16:31] <wjp> that pid 1 weirdness could be this joystick process I disabled yesterday if that already existed back then
[21:17:08] <wjp> 2006
[21:17:19] <wjp> hm
[21:19:31] <wjp> so I think I'm going to blame that
[21:19:59] <Malignant_Manor> I can check but it could take a long time to produce the bug in the first place.
[21:22:12] <wjp> it at least explains these stack/kill messages for pid 1
[21:24:20] <wjp> I'm not entirely sure which mechanism would then stop the avatar from moving, but it doesn't seem implausible that it could happen
[21:26:49] <wjp> an animation process could maybe get dropped from the active process list and then never wake up the AvatarMoverProcess waiting for that animation to finish
[21:28:12] <wjp> loading would automatically re-add to the active processes list, so that's consistent too
[21:28:44] <wjp> hm, although actually it wouldn't even get saved in that case. So maybe not exactly that
[21:29:34] <Malignant_Manor> It seems like something that you need to debug (with good info) while it happens.
[21:32:53] <wjp> ah, I think I see a way for this to happen
[21:33:26] <wjp> animation process gets PID 1, avatarmoverprocess waits for PID 1 but ends up waiting for JoystickCursorProcess with PID 1
[21:33:47] <wjp> the latter never terminates, so avatarmoverprocess never gets woken up
[21:34:16] <wjp> and judging by the frequency with which PID 1 gets re-used, it seems plausible that at some point an animation process for the avatar would get it
[21:34:45] <wjp> so that makes it very likely that it's fixed since yesterday :-)
[21:34:57] <wjp> well, "fixed" by killing the joystick code
[21:35:27] <Malignant_Manor> Once I compile. I can test to see if it still happens. Of course, this bug can take awhile to happen and has no known way to trigger it.
[21:36:26] <Malignant_Manor> I think I mainly ran and jumped to test it.
[21:37:45] <Malignant_Manor> Now, work on falling through floors. :p
[21:38:21] <wjp> if my hypothesis is right, it would happen from specific savegames
[21:38:47] <Malignant_Manor> The falling through floors or freezing?
[21:38:50] <wjp> the freezing
[21:39:22] <Malignant_Manor> So, I should start a new game.
[21:39:31] <wjp> or find the savegame you could reproduce it from :-)
[21:39:43] <wjp> but it should now be fixed for all savegames
[21:40:18] <wjp> (if I'm right) it could only happen in savegames in which there is a process with pid 1
[21:44:08] <wjp> pity watt isn't here
[21:49:44] --> Kirben has joined #pentagram
[21:49:45] --- ChanServ gives channel operator status to Kirben
[21:50:21] <wjp> by the way, which falling through floor specifically?
[21:59:51] <wjp> #3092602 is funny
[22:00:02] <wjp> that one seems to be a map bug
[22:00:12] <wjp> some of the ground tiles are at different heights
[22:01:13] <Malignant_Manor> I'm not sure. I've had it happen to myself a few times in Pentagram.
[22:02:25] <Malignant_Manor> I have one save that says walk south to fall through floor
[22:02:35] <wjp> yeah, that's the one
[22:02:52] <wjp> I'll have to manually move some map tiles around there
[22:03:03] <wjp> it's possible the original tries a little bit harder to move around the holes
[22:03:53] <Malignant_Manor> I have a save where you can walk on the air.
[22:04:23] <wjp> that sounds interesting
[22:06:13] <Malignant_Manor> I just added it to the tracker. https://sourceforge.net/tracker/?func=detail&aid=3599371&group_id=53819&atid=471706
[22:07:26] <Malignant_Manor> It is probably all ledges.
[22:08:34] <wjp> visually that does look like a step too far
[22:09:04] <Malignant_Manor> I somehow have one save that is way out there.
[22:09:19] <Malignant_Manor> But I cannot take a step in any direction.
[22:10:12] <wjp> the bounding box of the avatar seems to be just on the ledge
[22:10:38] <wjp> so you're positioned pixel-perfect to make that step possible
[22:14:19] <wjp> oh, that jumping one is weird too
[22:14:36] <wjp> interesting rendering issues with the hands when jumping up
[22:14:58] <wjp> ah, you ended up in the floor there
[22:15:43] <wjp> how peculiar
[22:16:16] <wjp> looks like more misplaced floor tiles
[22:16:41] <wjp> so you're possibly not actually standing where it visually appears you are
[22:16:52] <wjp> (due to Z differences in the floor)
[22:18:15] <wjp> same thing for that standing in water
[22:18:25] <wjp> one of the water tiles is at waist-height
[22:18:57] <wjp> looks like I should plan a map fix session :-)
[22:20:31] <wjp> hm, that "shape border" one is strange. I thought we rounded object locations
[22:21:22] <wjp> makes me wonder if that happens in the original
[22:21:33] <wjp> let's see if I remember how to make this teleport script work
[22:25:33] <Malignant_Manor> I don't know why I never reported these in October 2010.
[22:27:54] <wjp> huh, that shape border doesn't happen in the original
[22:28:15] <wjp> I thought I fixed all those object positioning issues :/
[22:28:45] <Malignant_Manor> I just clipped through a wall.
[22:34:31] <wjp> ah, the position is correct but it seems accidentally transparent
[22:35:22] <wjp> hm, what's the English word for the thing of which the border is wrong?
[22:35:47] <Malignant_Manor> I don't know for sure.
[22:36:23] <Malignant_Manor> I know Open_GL clamping can screw up borders
[22:36:29] <wjp> bier? recess?
[22:37:47] <wjp> thanks for all the reports, in any case
[22:37:53] <Malignant_Manor> I found the reproducible clipping error. I will upload it now. You will also notice that the Avatar appears to walk through the roof top tiles that surround the top square.
[22:38:48] <wjp> good night for now
[22:39:34] <Malignant_Manor> Good night.
[22:48:13] <-- Malignant_Manor has left IRC (Quit: ChatZilla 0.9.89 [Firefox 17.0.1/20121128204232])