#bitfighter IRC Log

Index Search ←Prev date Next date→

IRC Log for 2011-08-31

Timestamps are in GMT/BST.

00:08:00Zooomba Quit (Quit: Page closed)
00:32:00Flynnn has joined
00:43:00karamazovapy has joined
01:30:00karamazovapy Quit (Ping timeout: 245 seconds)
01:57:00Flynnn Quit (Quit: This computer has gone to sleep)
03:29:00raptor has joined
03:29:00ChanServ sets mode +o raptor
03:42:00raptorsam686: did your server move?
03:43:00sam686yes.. seems to be better.
03:43:00raptorwhat was wrong before?
03:44:00sam686before, my server was running off my computer, but that eats some of my CPU, and can sometimes lag when i am downloading or uploading something heavy..
03:45:00sam686players seems to equally like my server, even when moved..
03:46:00sam686note that bot zone generating seg fault is only on server 96.44.134.183, does not seem to occur on other server..
03:47:00raptoryeah, i think it has to do with the version of gcc used
03:47:00sam686http:// 208.107.52.15 is my computer IP address..
03:48:00sam686could try building with -O1 instead of -O2 and see if the error is still there, because debug build don't seg fault..
03:49:00raptoryeah, i'll try that...
03:50:00raptorthat server is super slow...
03:50:00raptorlatent, at least
03:50:00sam686my computer is about 4 slower then the server...
03:54:00raptorok started a compile...
03:55:00raptorare you planning on merging your random level code into 016?
03:55:00sam686probably yes..
03:56:00raptorit's been almost 500 revisions since last release..
03:56:00raptormore than 500
03:57:00raptor531
03:58:00raptorI think issue 115 is very important: http://code.google.com/p/bitfighter/issues/detail?id=115
03:58:00raptorbut i'm not sure how to start it...
03:58:00sam686just do cd exe
03:58:00sam686then ./bitfighterd
03:59:00sam686then maybe ctrl+z and bg then to keep it running adter quitting ssh, disown
03:59:00sam686or, you could gdb
03:59:00raptorok, started server
03:59:00sam686good, i see it..
04:00:00sam686crash?
04:00:00raptory
04:00:00raptorsame place, too
04:00:00sam686the problem is bot zone generating loading cp_badlands
04:00:00sam686try O0 instead of O2? if there is such a thing as -O0
04:01:00raptoryes, there is Os and O0
04:01:00raptori think O0 is the same as nothing
04:01:00sam686i wonder what bug does gcc have with optimizing?
04:02:00raptori don't think it is a gcc bug necessarily
04:04:00sam686can try searching: http://www.google.com/search?q=gcc+floating+point+inaccurate+with+optimization+bug
04:06:00raptori am going ot compile with -frounding-math
04:09:00sam686since triangle.c only have a floating point problem, maybe extra compile options things only needed for triangle.c
04:09:00raptoryes, but I don't want to splice up the Makefile for that one piece if i can help it..
04:10:00raptorok ready to test?
04:11:00sam686ok
04:11:00raptorstarted
04:11:00sam686ok, it died again..
04:11:00raptorsegfault
04:11:00raptorok, i'm going to try with just triangle.c without -O2
04:12:00raptorok server up
04:13:00raptorno crash?
04:14:00sam686i don't see any crash..
04:14:00sam686seems to be working now
04:14:00raptorok fine - i'll adjust the Makefile...
04:14:00raptori don't know what to do in windows...
04:16:00sam686it doesn't happen in windows, unless compiled with fast floating point model..
04:17:00sam686even if i use fast floating point model then i can simply use #pragma float_control( precise, on ) in triangles.c
04:41:00sam686in triangles.c, can try #define REAL long double instead of what it was defined, and see if the error is gone..
04:41:00raptorwhat line?
04:42:00sam686line 218
04:42:00raptorhow did you come to this conclusion?
04:42:00sam686it might be the floating point being too inaccurate..
04:43:00sam686using long double means the compile usually will never need to convert and round the x86 80-bit floating point to slammer memory..
04:45:00sam686http://en.wikipedia.org/wiki/Long_double - x86 have always supported 80-bot floating point internally..
04:45:00koda Quit (Quit: koda)
04:46:00raptorok started server again...
04:47:00raptorsegfault
04:47:00sam686guess that didn't fix the problem..
04:48:00raptortry now
04:49:00sam686it segfault?
04:49:00raptoryeah, rats
04:49:00raptori found a gcc option on that system that was different from mine, and i was hoping that turning it off would solve it...
04:51:00raptori have an idea...
04:52:00raptorline 211 kept us from reaching double precision
04:52:00raptorso we are using float
04:52:00sam686what file in line 211?
04:52:00raptortriangle.c
04:54:00sam686could just command out #define SINGLE to give Double a try...
04:54:00raptorthat's what i'm doing
04:54:00sam686but, when i try double, i get bot zone generation failure..
04:55:00raptoryeah, i get a segfault with it...
04:55:00raptorrats
05:03:00raptori think ultimately we should use poly2tri...
05:03:00raptorbut i'm almost done with the Makefile adjustment for triangle
05:06:00sam686i see, if removing #define SINGLE in triangles.c, then also must remove #define SINGLE from triangles.h
05:06:00raptorah... maybe that was my problem...
05:06:00raptorlet me try that, too
05:07:00raptorah good, no segfault
05:08:00sam686why do we even need #define SINGLE and #define REAL double stuff in triangle.c when it is already in triangle.h? i would remove #define REAL double from triangle.c
05:08:00raptorsomehow it's not showing up in server list anymore...
05:09:00raptor /connect doesn't work either
05:09:00sam686maybe it died?
05:10:00raptori just restarted it
05:10:00raptorlike the IP dissappeared
05:11:00sam686did it freeze trying to generate bot zones?
05:11:00raptorok, still segfaults
05:11:00raptoryes
05:11:00raptorthat is what happened :(
05:21:00sam686though, i get compile errors when i change REAL to double in GeomUtils.cpp
05:21:00sam686when changing REAL to double in triamgle.h
05:21:00raptoryeah, that isn't the solution
05:22:00raptori think the proper solution is to just use integer math and upscare the points, like what clipper requires
05:22:00raptorupscale
05:23:00raptorok pushed the Makefile hack
05:28:00BFLogBot - Commit 3e017dfddbd2 | Author: buckyballreaction | Log: Disable gcc optimizations for Triangle. Seems to fix rare crash with generating zones on some levels with some versions of gcc. Seems to be floating point related. Maybe we need to try poly2tri again?
05:41:00raptorwell good night..
05:42:00raptor Quit (Remote host closed the connection)
06:43:00watusimoto has joined
07:48:00BFLogBot - Commit e3498491932a | Author: sam8641 | Log: triangle now use double, seems to fix most triangle errors
08:02:00LordDVG has joined
08:13:00watusimoto Quit (Ping timeout: 264 seconds)
12:57:00LordDVG Quit (Remote host closed the connection)
13:42:00raptor has joined
13:42:00ChanServ sets mode +o raptor
14:06:00raptorhi sam686, your server is down...
14:44:00BFLogBot - Commit 0f7ff2bf7efb | Author: buckyballreaction | Log: Fix crash if no music files are found
15:14:00ZoomberImac has joined
15:14:00ChanServ sets mode +v ZoomberImac
15:15:00raptorsam686: do you think I can revert my Makefiel changes? it looks like triangle is working now on cp_badlands with your latest commit
15:16:00ZoomberImaclol sam left his test server up on 96 server
15:16:00raptorthat was my fault
15:16:00raptori was trying to crash it...
15:16:00raptorok it's shut down
15:18:00ZoomberImaclol its ok
15:18:00ZoomberImaci dont mind
15:22:00ZoomberImaclooks like my server must have just crashed recently
15:25:00ZoomberImacraptor: expect both sam and my server up in 10 or less minutes
15:25:00raptorok
15:49:00raptorgibebd is back
15:49:00raptorgibbed
15:49:00raptorbut not Bitfighter sam686
15:58:00ZoomberImachmm
15:58:00ZoomberImaci must have made a mistake in the cron script
15:58:00ZoomberImaci made one for sam but didnt really ever test it
16:26:00karamazovapy has joined
16:47:00ZoomberImachi karamazovapy
16:47:00karamazovapyhi ZoomberImac
16:50:00ZoomberImac is now known as Im
16:51:00Im is now known as ZisTotallyBillyD
16:51:00ZisTotallyBillyD is now known as billyDeewilliams
16:58:00billyDeewilliamshey karamazovapy, do you still have that "welcome to kserv, the level chagne password is newlev" map?
16:58:00billyDeewilliamsi'm stuck on how you get those cursive barrier letters
16:59:00billyDeewilliamsare you actually using forumlas?
17:00:00karamazovapy Quit (Read error: Connection reset by peer)
17:00:00LordDVG has joined
17:00:00LordDVG Quit (Changing host)
17:00:00LordDVG has joined
17:01:00LordDVG Quit (Remote host closed the connection)
17:02:00LordDVG has joined
17:02:00LordDVG Quit (Changing host)
17:02:00LordDVG has joined
17:35:00LordDVG Quit (Remote host closed the connection)
17:38:00LordDVG has joined
17:38:00LordDVG Quit (Changing host)
17:38:00LordDVG has joined
18:12:00billyDeewilliamsraptor: you around?
18:12:00raptory
18:12:00billyDeewilliams***ROBOT ERROR*** in robots/timer.bot ::: Error loading file: robots/timer.bot:21: unfinished string near '" '. Shutting robot down.
18:12:00raptorwhat was your command? and on what server?
18:13:00billyDeewilliamsslash addbot 0 timer
18:13:00billyDeewilliamsi think its related to just "timer.bot"
18:14:00billyDeewilliamsall the other bots work fine
18:14:00raptorwhat server?
18:14:00billyDeewilliamsi made a clone "user levels" server
18:14:00billyDeewilliamsi need to test the time ticks of it
18:14:00raptorit could be that timer isn't on user levels
18:15:00billyDeewilliamsi made a clone of the server, so there are two user levels servers on
18:15:00billyDeewilliamsok, i may have found the problem
18:15:00billyDeewilliamsaaah ok
18:16:00billyDeewilliamsits a bug when using nano editor
18:16:00billyDeewilliamsbreaks lines when they are too long for your comamnd window
18:16:00billyDeewilliamsso, someone used nano on the timer bot, and one of its lines got seperated
18:16:00billyDeewilliamsso now, it works
18:17:00raptoryou can turn that off in nano
18:17:00raptormight be a commandline option
18:19:00billyDeewilliamsah, ok
18:19:00billyDeewilliamsthanks
18:45:00sam686[SavedServerPasswords] doesn't work for server names starting with ;
18:45:00sam686i see lots of ; 3=
18:45:00sam686 ; 3=
18:45:00sam686 ; 3=
18:45:00sam686 ; 3=
18:46:00raptorhi sam686
18:49:00sam686the bot zone triangle.c works much better in double type, doesn't error with optimization, and doesn't error with fast floating point model, like it did with single type..
18:49:00sam686i will go, will be back in 2 hours..
18:49:00raptoryes, i've been testing
18:49:00raptordo you think that i can revert my Makefile change?
18:50:00sam686yes, if you want to..
18:50:00raptorok
18:50:00raptorsee you later
18:54:00BFLogBot - Commit cc2206bdb86a | Author: buckyballreaction | Log: Revert Makefile hack for compiling Triangle without optimizations since sam found a better solution
20:30:00billyDeewilliams Quit (Quit: billyDeewilliams)
22:02:00LordDVG Quit (Remote host closed the connection)
22:04:00raptor Quit (Remote host closed the connection)
22:49:00Flynnn has joined
23:27:00raptor has joined
23:27:00ChanServ sets mode +o raptor
23:52:00Flynnn Quit (Quit: This computer has gone to sleep)

Index Search ←Prev date Next date→

These logs were automatically created by BFLogBot on irc.freenode.net.