#pentagram@irc.freenode.net logs for 17 Jun 2005 (GMT)

Archive Today Yesterday Tomorrow
Pentagram homepage


[00:06:12] --> Kirben has joined #pentagram
[00:06:12] --- ChanServ gives channel operator status to Kirben
[01:00:37] <-- Lightkey has left IRC (Read error: 145 (Connection timed out))
[01:08:46] --> Lightkey has joined #Pentagram
[01:28:44] <servus> Mr. William Jones Palenstein.
[01:29:30] <megawatt> woohoo. got Some level of pool testing... although it seems that pooled vs. unpooled is dead even in speed until all the needed pools are allocated
[01:29:32] <megawatt> ???
[01:30:18] <megawatt> servus is intentially messing up names?
[01:31:18] <megawatt> anyway... after the pools are allocated, there is a definite improvement.... I think I'm gonna tweak my test numbers a bit to keep it more realistic though
[01:32:15] <megawatt> (I never see more than 3 192 byte pools during gameplay, the test created 19
[01:34:47] --> fufufu has joined #pentagram
[01:35:10] <-- fufufu has left IRC (Client Quit)
[01:46:52] <megawatt> ok... having the test repeat 10000 times was a bad idea.... taking forever
[01:48:05] <megawatt> yeah.. an 8 minute test of allocation isn't called for
[01:52:21] <megawatt> Unpooled Allocation: 228558 ms
[01:52:21] <megawatt> Pooled Allocation: 141005 ms
[01:55:26] <megawatt> considering that the test is forming a linked list while allocating to be sure to clean up after itself.. I'd say pooling is roughly twice the speed of normal allocation
[01:56:31] <megawatt> only repeat 100 times now
[01:56:33] <megawatt> Unpooled Allocation: 2221 ms
[01:56:33] <megawatt> Pooled Allocation: 1236 ms
[02:00:08] <megawatt> well.. need to see about a few speed enhancement tweaks and a process to remove empty pools after a delay, and it will be close to ready to commit.
[02:03:35] <megawatt> in the meantime.... http://fallofrome.dyndns.org:8080/mem.diff
[02:05:55] <megawatt> Anything I'm doing poorly or needs to change stylistically? (I haven't coded anything real for a little while now and fear I'm getting rusty on the skills that weren't that impressive in the first place)
[03:27:05] <servus> Poor megawatt, talking to his lonesome.
[04:00:19] <megawatt> bah...
[04:06:50] <servus> : o)
[04:18:38] <sbx> megawatt's whois says his name is Matthew. Matt is short for Matthew. If you take "matt" and turn the m upside down it looks like "watt". How peculiar.
[04:24:18] <megawatt> yes, indeed... peculiar.
[04:25:18] <megawatt> "They're on to me... quickly! Alternate nick!"
[04:25:24] <megawatt> heh
[04:33:01] <sbx> I'm just in suspicious connect-the-dots mode since playing Deus Ex. (properly subtitled "Conspiracy")
[04:45:37] <megawatt> I noticed you attempted to stay with Unatco. It's a shame you can't be evil in the game... although I tried my best.. I went on a killing spree of every non-essential character.
[04:46:00] <megawatt> And everyone still treated me like a good guy.
[05:09:35] <-- oaQ^ has left IRC (adams.freenode.net irc.freenode.net)
[05:11:11] --> oaQ^ has joined #pentagram
[05:13:32] <sbx> I thought about doing that but was afraid it would get me less skill points. Something to try next game. >:) (I'm still on HellsKitchen pt3 though)
[05:26:43] <Kirben> Skill points are mainly exploration based, will lost some credits if you slaughter the NPCs in early missions though.
[05:28:59] <Kirben> Sometimes worth not killing some enemies, instead just knocking them unconcious. Some enemies have useful items, I seem to recall at least one mib/wib with good stuff.
[05:51:01] <sbx> That's why I didn't blow up the terrorists holding hostages. They had shotguns. :)
[05:51:25] <sbx> didn't know you could knock out mib though
[05:57:13] <servus> Good game, sbx :)
[06:24:46] <sbx> yep
[10:39:08] --> Colourless has joined #Pentagram
[10:39:08] --- ChanServ gives channel operator status to Colourless
[10:43:26] <Colourless> willem are you around?
[11:30:02] <wjp> I am now
[11:31:13] <sbx> hi wjp
[11:48:48] <Colourless> would it be possible to split module.mk into 2 different files. one that purely contains the module reqs that could be used by all the make files
[11:49:56] <Colourless> hmm. but there isn't much in module.mk..
[11:51:22] <Colourless> actually it would probably end up being too difficult
[12:03:18] <wjp> we could put some extra variables in objects.mk
[12:03:43] <wjp> so the pentagram_OBJ assignment in module.mk would just use a singe variable from objects.mk
[12:10:46] <Colourless> yeah something like that might clean things up a bit for other systems
[12:11:32] <wjp> the list of directories/sets-of-objects has become somewhat larger than it originally was :-)
[12:12:03] <Colourless> the msvc makefile i've been making is rather complex.
[12:12:15] <Colourless> supports different build options
[12:12:28] <Colourless> can build in DEBUG or RELEASE
[12:12:50] <wjp> ah, nice
[12:12:52] <Colourless> with optional VERBOSE output and ability to specify DONT_USE_PRECOMPILED_HEADER
[12:13:09] <Colourless> also uses different objects depending on the build mode
[12:13:23] <Colourless> debug will create .obj_dbg files and release will make .obj files
[12:13:35] <Colourless> can do alot of stuff
[12:13:49] <Colourless> just doing something like this is all i need to do that
[12:13:50] <Colourless> PENTAGRAM_OBJS=$(PENTAGRAM_OBJS:.o=.obj_dbg)
[12:16:11] <Colourless> sucks though that nmake and gnu make have differences in syntax
[12:49:59] <wjp> hm, yeah, so need to separate the various parts of the makefiles properly
[12:56:32] <wjp> one other Makefile-related issue is 'make dist'
[12:57:40] <wjp> it'll mainly just be 'get every .cpp/.h file, all .txt files in docs/, all .ini files in data/ and some misc files, but doing it somewhat cleanly would be nice
[12:58:09] <wjp> maybe have common.mk (included in each module.mk) automatically take all .cpp/.h/.ini/.txt files, and specify the others in module.mk per directory
[13:00:52] <wjp> oh, and all .mk files, and Makefile in all subdirs
[13:01:29] <wjp> that should take care of everything except the docs/xml dir, the root dir and some data files
[13:02:07] <wjp> ...and .c/.cc files
[13:02:25] <wjp> ...but not data.h :-)
[14:36:34] <Colourless> time for me to go to bed. discuss it tomorrow
[14:36:42] <-- Colourless has left IRC ("casts improved invisibility")
[14:46:14] <-- Kirben has left IRC ("System Meltdown")
[18:11:46] --- sbx is now known as sbx|afk
[18:24:22] --> trin has joined #pentagram
[18:25:14] <trin> howdy :)
[19:11:09] --> Dark-Star has joined #pentagram
[19:11:38] <-- trin has left IRC (Read error: 104 (Connection reset by peer))
[23:18:08] <-- Dark-Star has left IRC ()