#exult@irc.freenode.net logs for 15 Jul 2001 (GMT)

Archive Today Yesterday Tomorrow
Exult homepage

[00:00:12] --> matt0 has joined #exult
[02:10:55] --> Kirben has joined #exult
[10:58:43] <-- Kirben has left IRC (Ping timeout for Kirben[co3007967-a.brasd1.vic.optushome.com.au])
[11:01:29] --> Kirben has joined #exult
[13:03:14] <-- Kirben has left IRC (System Meltdown)
[13:26:18] --> wjp has joined #exult
[14:17:00] --> Colourless has joined #Exult
[14:19:21] <Colourless> hi
[14:21:36] <wjp> hi
[14:22:10] * Colourless wonders if there will be more conversation today than yesterday
[14:22:41] <wjp> well, I think we'll only need 2 more sentences for that ;-)
[14:22:51] <Colourless> yeah, we will
[14:23:26] * wjp wasn't feeling very awake yesterday
[14:23:37] <wjp> went to bed at 5am the night before...
[14:23:51] <wjp> which isn't that late for you, I guess :-)
[14:24:01] <Colourless> so what.... i do that every night. ;-)
[14:26:02] <wjp> hmm, yesterday's logs don't even have a scrollbar... that's pretty bad :-)
[14:26:13] <Colourless> that is bad
[14:26:22] * Colourless checks the logs
[14:26:59] <Colourless> i've got a scrollbar here
[14:27:12] * wjp shrinks window
[14:27:25] <wjp> me too ;-)
[14:27:39] <Colourless> if it weren't for all the times that Kirben joined and Left, i wouldn't have one either
[14:28:46] <wjp> hmm, what should I go do today? :-)
[14:29:33] <Colourless> don't ask me, i can't even decide that for myself
[14:30:18] <Colourless> oh look, we've already over doubled the amount of conversation from yesterday
[14:30:28] <wjp> good :-)
[14:30:55] <wjp> if we try hard enough I'm sure we can do even better, though ;-)
[14:31:18] <Colourless> i'm sure we could
[14:31:37] <Colourless> perhaps I could try to summon Jeff.... USECODE!!!!
[14:31:42] <Colourless> damn, it didn't work
[14:32:06] <wjp> :/
[14:32:52] <wjp> Jeff joined friday when we talked about _PALETTE CYCLING_...
[14:32:53] * wjp waits
[14:33:12] <wjp> hmm, didn't work either
[14:33:19] <Colourless> perhaps talking about MAP EDITING would make him come
[14:33:32] * Colourless waits
[14:33:46] <Colourless> doesn't seem to be working either
[14:34:13] <Colourless> it would just be so much easier if he told us the keyword of the day ;-)
[14:34:35] <wjp> hehe :-)
[14:37:41] <Colourless> so did anyone actually fix the palette cycling?
[14:37:52] <wjp> don't think so
[14:38:15] * Colourless checks logs and doesn't seen an entry
[14:38:48] <wjp> there. done
[14:39:09] <Colourless> good
[14:41:36] <wjp> uh oh...
[14:41:45] <Colourless> what!
[14:41:48] <wjp> I can't commit exult.cc...
[14:42:04] <Colourless> oh no. The horror
[14:42:14] <wjp> too much junk in my copy
[14:42:50] * wjp gets a fresh copy
[14:42:51] <Colourless> ooooh
[14:44:37] <wjp> committed
[14:44:49] <Colourless> finally ;-)
[14:45:01] <wjp> :-)
[14:45:37] <wjp> I guess someone should do the scaler patch too
[14:45:51] <Colourless> you can do that too.
[14:45:56] * Colourless doesn't know how
[14:46:13] <wjp> 'man patch' ;-)
[14:46:58] <Colourless> well then, I couldn't have done it even if I wanted to
[14:48:25] <wjp> ok, patch is in... now let's see if everything still works
[14:48:38] <Colourless> :)
[15:16:02] <wjp> it's 16x as slow as the point scaler
[15:16:29] <Colourless> that wouldn't actually surprise me
[15:17:25] <Colourless> how does it compare to the 2xSAI and SuperEagle scalers?
[15:17:58] <wjp> hmm, dunno
[15:19:00] <wjp> I'll try the old bilinear now, and the others after that
[15:19:26] * Colourless want to make a new bilinear scaler
[15:19:37] <Colourless> and perhaps an interlaced bilinear scaler as well
[15:20:20] <Colourless> i think that it should be possible to squeze out quite a bit of performance out of the bilinear scaler
[15:20:32] <Colourless> that is of course, theoretical
[15:26:15] <wjp> old bilinear: 22.9 ms/call, new one: 16.8 ms/call
[15:26:49] <Colourless> ok improvement
[15:29:11] <wjp> trying 2xSaI...
[15:35:48] <wjp> 21.2 ms/call
[15:36:00] <Colourless> ouch
[15:36:06] <Colourless> faster than bilinear
[15:36:09] <Colourless> Old one
[15:36:15] <wjp> yeah
[15:38:05] <wjp> those Time_queue:add/remove are really slow...
[15:38:24] <wjp> 0.98/0.84 ms/call, respectively
[15:38:33] * Colourless has a look
[15:39:26] <Colourless> using a vector and an i
[15:39:29] <Colourless> ignore that
[15:40:23] <Colourless> uses a list and an iterator
[15:41:30] <wjp> maybe we should use a more efficient priority queue
[15:41:35] <Colourless> a linked list would be much faster
[15:41:52] <wjp> maybe a heap?
[15:42:34] <Colourless> perhaps.
[15:44:24] <wjp> the STL priority_queue seems to use a heap internally
[15:49:11] <Colourless> IMO we should attempt to avoid using STL
[15:52:08] <wjp> shall I commit the bilinear scaler?
[15:52:18] <Colourless> yeah, it's faster
[15:53:46] <Colourless> Derek writes some really tight code. I wouldn't have though that the 2xSaI would have been faster than our old Bilinear
[16:05:57] <wjp> brb
[16:06:06] <Colourless> k
[16:38:43] <wjp> b
[16:40:38] <Colourless> wb
[17:22:18] --- wjp is now known as wjp|dinner
[18:01:50] --- wjp|dinner is now known as wjp
[18:01:53] <wjp> back
[18:01:59] <Colourless> wb
[18:02:03] <wjp> thanks
[18:24:24] <wjp> hey, the grades for that 5-person team project just came in... seems we got a 10 (!?)
[18:24:36] <wjp> (ie. maximum grade)
[18:25:14] * wjp doesn't get it
[18:25:26] <Colourless> get what?
[18:25:48] <wjp> I wouldn't have given us much more than a 7
[18:26:05] <Colourless> perhaps everyone else was much worse than you
[18:26:13] <wjp> could very well be :-)
[18:26:40] <wjp> I heard a few horror stories about other teams' programs
[18:50:33] <wjp> hmm, I've just profiled a somewhat 'average' exult session
[18:50:56] <Colourless> and?
[18:50:59] <wjp> time_queue::add/remove are only called ~2K times
[18:51:11] <wjp> (in a few minutes of gameplay)
[18:51:50] <wjp> and the times per call aren't even high enough to register
[18:52:09] <Colourless> not really worth touching then
[18:52:13] <wjp> so apparently the queue stays relatively empty during 'normal gameplay'
[18:53:07] <Colourless> it should stay pretty empty. Only things like scheduled usecode really use the queue. There are probably other things as well
[18:53:19] <wjp> animations too
[18:53:27] <wjp> schedules, probably
[18:54:17] <Colourless> i think schedules are handled elsewhere
[18:54:30] <wjp> could be
[18:54:36] <Colourless> i beleive they are directly changed by the clock
[19:18:24] <Colourless> i'll be off now
[19:18:35] <-- Colourless has left IRC (leaving)
[19:24:32] --> sty has joined #exult
[19:24:37] <sty> ave!
[19:24:43] <wjp> hi
[19:29:45] <sty> actors.h have some problems ...
[19:30:05] <wjp> it does?
[19:30:28] <sty> alot
[19:30:58] <sty> it gives to me compile errors
[19:32:00] <sty> (the one just downloaded from CVS)
[19:32:22] <wjp> works for me...
[19:32:43] <sty> i've done also make clean
[19:32:51] <wjp> what errors do you get?
[19:32:55] <sty> it had always compiled before yesterday
[19:33:19] <sty> In file included from Actor_gump.cc:24:
[19:33:19] <sty> ../actors.h:115: `Chunk_object_list' was not declared in this scope
[19:33:19] <sty> ../actors.h:115: `old_chunk' was not declared in this scope
[19:33:19] <sty> ../actors.h:115: `Chunk_object_list' was not declared in this scope
[19:33:21] <sty> ../actors.h:115: `new_chunk' was not declared in this scope
[19:33:22] <sty> ../actors.h:116: parse error before `,'
[19:33:47] <sty> have i to relaunch ./configure ?
[19:34:44] <wjp> I think your actors.h may be outdated
[19:35:01] <wjp> line 115 doesn't have any Chunk_object_list things here
[19:35:16] <sty> so CVS is not updating regularly
[19:35:54] <wjp> does 'cvs -z3 update -d' give any lines other than 'P <filename>' ?
[19:35:59] <sty> cvs -d:pserver:anonymous@cvs.exult.sourceforge.net:/cvsroot/exult -z3 update -Pd
[19:36:18] <wjp> you can skip the -d.... argument if you're in the main exult dir.
[19:36:41] <sty> is it right ?
[19:37:12] <wjp> yeah
[19:37:27] <sty> no it does not give nothing but P ..... and cvs server:updating....
[19:37:59] <wjp> did it say 'P actors.h' this time?
[19:38:49] <sty> no it didn't
[19:39:01] <wjp> could you do a 'cvs status actors.h' ?
[19:40:23] <wjp> (interesting things are 'status', 'working revision' and any 'sticky' things, if they're not "(none)")
[19:42:44] <sty> let's do rm actors.h :))
[19:55:24] <wjp> does it work now?
[19:56:15] <sty> no
[19:56:31] <wjp> same errors?
[19:57:09] <sty> can't update actors.h
[19:57:18] <sty> even if i deleted it
[19:57:19] <wjp> can't?
[19:57:32] <wjp> why not?
[19:57:54] <sty> cvs seems to not recognize the file "updatable" so it doesn't update it
[19:58:24] <wjp> if you don't have actors.h, it should just download it
[19:58:59] <wjp> what happens if you run "cvs -z3 update actors.h" from the main exult dir?
[20:01:04] <sty> it seems to work no
[20:01:05] <sty> w
[20:01:25] <wjp> from what directory were you running the update commands first?
[20:03:19] <sty> exult
[20:03:40] <wjp> the dir. in which configure.in, exult.cc, etc... are?
[20:04:21] <sty> /..... / exult
[20:04:53] <wjp> that doesn't tell me much... I don't know your directory structure :-)
[20:05:03] <wjp> I have several directories named 'exult', for example...
[20:05:03] <sty> does it matter ?
[20:05:23] <sty> i started with "checkout" command
[20:05:28] <sty> checkout exult
[20:05:31] <wjp> well, 'cvs update' has to be run from the directory that contains configure.in
[20:06:03] <wjp> while you usually run 'cvs checkout' one level higher
[20:06:44] <sty> ]# cvs -z3 update -P
[20:06:52] <sty> cvs update: No CVSROOT specified! Please use the `-d' option
[20:06:52] <sty> cvs [update aborted]: or set the CVSROOT environment variable.
[20:07:04] <sty> this happens in /.../exult dir
[20:07:23] <sty> yea i've done checkout one level higher
[20:07:38] <sty> i must everytime specify -d .....
[20:07:47] <sty> remember that i don't login
[20:08:28] <wjp> it should remember the cvsroot
[20:09:49] * wjp tries an anonymous checkout
[20:11:07] <wjp> ok, I'm now in /data/wpalenst/tmp, running cvs -z3 -d:pserver:anonymous@cvs..... co exult
[20:11:18] <wjp> which creates a /data/wpalenst/tmp/exult
[20:11:44] <wjp> from /data/wpalenst/tmp/exult I can now run a 'cvs -z3 update -d' without having the specify the CVSROOT
[20:13:27] <sty> quote the entire command of the checkout
[20:14:42] <wjp> cvs -z3 -d:pserver:anonymous@cvs.exult.sourceforge.net:/cvsroot/exult co exult
[20:14:51] <wjp> (run from /data/wpalenst/tmp)
[20:16:07] <sty> did you made the exult dir first ?
[20:16:14] <wjp> no
[20:16:25] <wjp> 'cvs co' created it
[20:25:35] <wjp> I have to go
[20:25:39] <wjp> see you later
[20:25:58] <sty> cya
[20:26:00] <sty> thanks
[20:26:05] <-- wjp has left IRC ([x]chat)
[22:12:05] <-- sty has left IRC ()