#bitfighter IRC Log

Index Search ←Prev date Next date→

IRC Log for 2011-07-24

Timestamps are in GMT/BST.

00:00:00zoombersoommeonee go onlinee
00:04:00zoombermac is so scary to boot up
00:12:00watusimoto Quit (Ping timeout: 252 seconds)
00:16:00raptorhi
00:30:00zoomberit may look like on the outside, the mac booting screen is all nice and colufull
00:30:00zoomberbut its a scary zombie invasion watching your hackintosh try to boot up successfully
00:30:00raptorjust like a linux box!
00:30:00zoomberyeah
00:31:00zoomberfirst the kexts list loads
00:31:00zoomberall the plugins mac loads
00:31:00zoomberthats around the time the macbook shows the apple with the lines going in a circle
00:31:00zoomberthen, when normally you see, "Mac OS X is starting up", you see a bunch of logged items loading
00:32:00zoomberand if something gets wrong, a looooooongg list of a kernal panic and traceback shows
00:32:00watusimoto has joined
00:32:00zoomberhi watusimoto o o o o
00:34:00zoomber is now known as zoomber_laptop
00:34:00Zoomber has joined
00:38:00Flynnn Quit (Quit: This computer has gone to sleep)
00:54:00Flynnn has joined
00:55:00watusimoto Quit (*.net *.split)
00:57:00sam686 has joined
00:57:00ChanServ sets mode +v sam686
00:58:00BitfighterLogBot - Commit 8787da28512e | Author: watusim...@bitfighter.org | Log: Eliminate a whole slew of global vars. Get all UIs from UIManager. Probably has some subtle bugs.
00:58:00BitfighterLogBot - Commit 548f172ab81e | Author: watusim...@bitfighter.org | Log: Merge
00:58:00BitfighterLogBot - Commit 667f6eec7a93 | Author: watusim...@bitfighter.org | Log: Need these too...
00:59:00[1]watusimoto has joined
01:12:00raptoryay for merge
01:16:00raptor[1]watusimoto: can you please explain what you have in mind for the UIManager?
01:16:00raptoralso, was there a reason to chose auto_ptr instead of shared_ptr?
01:45:00raptorcleaning up compile errors...
01:47:00raptori have ~130 linking errors...
01:47:00raptoroh duh
01:47:00raptornew class
02:03:00raptori get one error now when linking, not sure the cause:
02:03:00raptorUIManager.o: In function `MessageUserInterface':
02:03:00raptorUIMessage.h:51: undefined reference to `vtable for Zap::MessageUserInterface'
02:12:00zoomber_laptop Quit (Quit: zoomber_laptop)
02:31:00sam686the editor 016 might be missing heal rate option for turret and forcefield.
02:32:00raptorsam686: thank you for fixing the joystick typo of mine...
02:33:00raptor(axesMask instead of axesDirection)
02:33:00raptorthat is why I shouldn't write code at 2 in the morning...
02:53:00BitfighterLogBot - Commit 5865b6af2f05 | Author: buckyballreaction | Log: Fix compiler errors/warnings on linux
03:53:00[1]watusimotouiManager
03:54:00[1]watusimotoit's a place to get alll your UIs
03:54:00[1]watusimotorather than having a series of global vars
03:54:00raptorah ok
03:54:00[1]watusimotoUIs are now created on an as-needed basis
03:54:00[1]watusimotowhich should make start up quicker
03:54:00[1]watusimotoand, perhaps, reduce memory
03:54:00[1]watusimotoif you don;t visit all the uis in a given session
03:54:00raptoris auto_ptr self-cleaning?
03:54:00[1]watusimotoalso, uis aren't created for server games anymore
03:55:00[1]watusimotoyes
03:55:00raptoroh yay
03:55:00[1]watusimotoit is
03:55:00sam686before all the 016 changes, i think all that ui stuff get created even if running only a dedicated server..
03:55:00raptorit did
03:55:00[1]watusimotoyes
03:55:00[1]watusimotoI haven't tested it, but that shouldn't happen anymore
03:55:00[1]watusimotoI want to simplify getting the uis, but this works for now
03:55:00raptorso i was thinking
03:56:00raptorwe should have some sort of minimum code style we should follow
03:56:00[1]watusimotothere is a separate set for each game: gClientGame, gEditorGame, and gServerGame
03:56:00[1]watusimotothough none should ever be created for gServerGame
03:56:00raptorlike should we:
03:56:00[1]watusimotoand there shouldn;t be much overlap btwn those that are created for client and editor games
03:56:00raptor1. use forward declarations where possible
03:56:00[1]watusimotoof course
03:56:00sam686gClientGame / gServerGame is a pointer, it only creates at certain point of code..
03:57:00raptor2. not use 'using' in headers (or anywhere)
03:57:00[1]watusimotohow do we avoid that one?
03:57:00raptor3. try to not have #include chains
03:57:00[1]watusimotowe should try
03:57:00sam686possibly avoind #include in .h files..
03:57:00[1]watusimotoI did 1 and 3 pretty well with uimanager
03:58:00[1]watusimotoall the includes aer in the .cpp file; the .h has only forward declarations
03:58:00raptorI have a question about forward declarations...
03:58:00raptorare they really needed anymore?
03:58:00[1]watusimotoaren;t they?
03:58:00raptori mean - they were used when RAM was at a premium
03:58:00[1]watusimotoI thought they were used to avoid includes
03:58:00raptorwell yes
03:58:00[1]watusimotoand make compilation faster
03:58:00[1]watusimotonot to save memory
03:59:00raptori read that they were used to save memory
03:59:00raptornot necessarily to make compilation faster...
03:59:00[1]watusimotoalso, ui manager should handle having different UIs if we have different clientGames
03:59:00sam686take a look at point.h, there is no includes (except string), and it has a BitStream..
03:59:00raptoralso, they tend to goof up IDE 'open declaration' ability sometimes
03:59:00[1]watusimotoopen declaration ???
04:00:00raptorctrl-click a class and you go right to it/...??
04:00:00raptoris that not available in vc++?
04:00:00[1]watusimotoI've mapped that to alt-enter
04:01:00sam686for me, i can right click class, then "Go to definition"
04:01:00raptoryes
04:01:00raptorthat
04:01:00[1]watusimotothat
04:01:00[1]watusimotoyes, they do goof that up
04:01:00raptorforward decs can goof that up sometimes...
04:01:00[1]watusimotobut they;re better than including stuff
04:02:00[1]watusimotohelps avoid .h chains
04:02:00raptorthis is informative:
04:02:00raptorhttp://stackoverflow.com/questions/2632601/why-are-forward-declarations-necessary
04:02:00raptorthe first answer, at least
04:04:00raptorI'm still not convinced that forward decs make things all that faster anymore
04:04:00raptorbut RAM reduction I can see
04:04:00sam686avoiding includes reduce compiling memory, and can speed up compiler..
04:06:00sam686in most .h files, having #ifdef to only include .h once, can be slow if having to include more then once, has to read whole file line by line until it find #endif
04:07:00raptori guess that's true...
04:08:00sam686some tnl .h files have this that might speed up a little bit: fndef _TNL_TYPES_H_ include "tnlTypes.h" #endif
04:08:00raptorheader header guards?
04:08:00raptorhaha
04:09:00raptorthat's actually a good idea...
04:09:00[1]watusimotobut if I have a header that includes another header, and I modify that second included header, it forces a recompile of everything that includes it
04:09:00raptoryep
04:09:00raptorsuch is the nature of headers, right?
04:09:00raptorohhhh
04:09:00raptorok
04:09:00[1]watusimotoso it might not speed compiling from the ground up, but can cause major recompilation for minor changes
04:09:00[1]watusimotoadvanced declrations avoid that
04:10:00raptorso i full compile in 40 seconds, i forgot that you guys take several minutes..
04:10:00[1]watusimotoif I can just do and adv decl of a class, and I change that class, it doesn;t require a recompile
04:10:00sam686it has to recompile all .cpp files that use changed .h files, even if included indirectly..
04:10:00[1]watusimotoright
04:10:00[1]watusimotoexactly
04:10:00[1]watusimotobesides, raptor, if you are recompiling everyting on your fast computer, you are still hastening the heat death of the universe
04:11:00[1]watusimotoby using adv declr, we can slow that a bit
04:11:00raptorhaha
04:11:00raptorabsolute entropy
04:13:00raptorok well - i will make an effort to use forward decs, then
04:13:00raptoreven if i think they are ugly and slow down my class search time
04:13:00sam686maybe #pragma once could be used insted of define guard in .h files. i see this: http://en.wikipedia.org/wiki/Pragma_once
04:14:00raptorthat is not guaranteed to work on older versions of gcc (and by extension mac...)
04:14:00raptorsadly
04:14:00sam686thats one problem, doesn�t work on all different compilers..
04:15:00sam686plus, who uses outdated linux if most can often upgrade to free?
04:15:00raptormost VPS...
04:15:00raptorbut that will be resolved in a year when they all update their systems to centos 5
04:15:00raptorer centos 6
04:16:00raptor(I think)
04:16:00[1]watusimotoI think I've just eliminated the need for gEditorgame
04:16:00raptoras for style #2 (using namespaces):
04:16:00raptorhttp://stackoverflow.com/questions/1452721/why-is-using-namespace-std-considered-a-bad-practice-in-c
04:16:00raptorit isn't for performance, but for avoidance of method collisions
04:16:00raptorwell - it isn't for performance any more...
04:17:00[1]watusimotoI don;t want to prefix all our stuff with TNL:: or std::
04:18:00raptorit was originally for performance reasons in header files
04:18:00sam686having to prefix TNL:: and all that stuff can make files bigger size..
04:18:00raptorbecause it would load the whole namespace in each header
04:19:00raptori started doing it for a couple headers, but i wanted to talk to you guys about it first before I kept doing it
04:20:00raptorcode style differences is started to hurt my brain a little...
04:20:00raptorI suggest we write a wiki page on code style and put these things in there; as well as things like 3-space indentation
04:21:00sam686having to prefix TNL:: 100 times can make file size 500 bytes bigger..
04:21:00raptorOK how about this: i'll do a crazy for loop that'll compile a class 1000 times and we'll do some testing to see what slows down the compiler
04:30:00raptorhere is the script i will use and pass it to 'time': http://pastie.org/2262099
04:34:00raptorhere is the first test with using 'using namespace' in the header: http://pastie.org/2262110
04:34:00raptorsorry, that is the class file i am using
04:38:00raptorresults with putting using namespace in header: http://pastie.org/2262115
04:46:00raptorok, empirical evidence that not using 'using namespace' has no effect: http://pastie.org/2262115
04:46:00raptornow for forward decs...
04:50:00sam686i will be testing speed of compiler if using namespace TNL to remove TNL:: in point.h and see if there any speed differences..
04:52:00raptoryeah - you can do it, i found no difference with or without it, so why not make our code more legible
04:52:00raptorwe just have to not conflict with method names in the namespace
04:53:00raptorok, testing 250 compile iterations with including 'Point.h'
05:03:00sam686using namespace TNL and removing TNL:: in point.h, gives me 6 min 16 sec, keeping it the way it is gaves me 6 min 20 sec, very small speed difference..
05:07:00raptorOK
05:08:00raptorwhen i see the TNL:: in a header, i'll remove then and go back to 'using namespace'
05:12:00sam686one possible way to speed up compiling is to combine multiple .cpp files into one .cpp..
05:12:00raptoryes, that would work - would use lots more RAM, too
05:12:00sam686but then, it will need more memory to compile a big .cpp file..
05:30:00raptorok results: http://pastie.org/2262198
05:30:00raptori find no speed advantage with forward declaring
05:31:00raptorhowever, the advantage I could see would be with IDE auto-detection of what classes need to be recompiled; but does that offset the muddling with 'open definition/declaration' ?
05:32:00sam686maybe the one .cpp file you was testing included point.h, making forward declaration no speed gain except on file that don not include point.h
05:32:00raptorplease reword that
05:32:00sam686did you test speed of full rebuild or only one cpp file?
05:33:00raptorjust one cpp file
05:33:00sam686testing only one .cpp may not be enough and might be inaccurate.
05:34:00raptori did 10 cpp in one instance
05:35:00sam686maybe your CPU is too fast, the hard disk is randomly slowing down compiling throwing timing off..
05:35:00raptoryes - i think i'm just too fast
05:36:00raptorbut I also think that the difference found on your end might only be a couple seconds out of 5 minutes, as well
05:36:00raptorbut i did three tests with each battery
05:36:00raptorto throw off any randomness
05:37:00raptorif I had seen even 1 second differenc in some place, then that would have been good
05:37:00raptorbut the average times are only off from each other roughly 200ms
05:38:00sam686i think it is because of your computer being 64-bit, your linux might be 64-bit if i am right
05:39:00raptoryes, i run everything 64bit
05:39:00raptorhow do you suppose that affects it compared to 32bit?
05:40:00sam686not sure..
05:41:00sam686i guess linux compiler have some tricks that speed up compiling, and windows compile some things differently..
05:41:00raptoryeah
05:41:00raptorcomparing mine to yours is futile
05:41:00raptori think
05:41:00raptorbut now you know why I don't think forward decs do much other than clutter code...
05:42:00raptorbut on your end, they might do a lot
05:43:00sam686since i have 1.5 GB ram, i can try having one .cpp file include multiple .cpp files, i really do see speed increase of compiling while i am in the middle of doing it..
05:43:00raptori wonder if there is some utility that will combine all the .h/.cpp into one file for us - like sqlite does
05:44:00[1]watusimoto Quit (Ping timeout: 258 seconds)
05:45:00sam686i frequently running into Redifinition static Vector<DatabaseObject *> fillVector; zap/gameobject.cpp line 215 and others..
05:46:00raptoryuk
05:54:00raptorok, well good night
05:54:00raptor Quit (Remote host closed the connection)
06:26:00zoomber_laptop has joined
06:42:00Zoomber Quit (Read error: Connection reset by peer)
06:43:00Zoomber has joined
06:45:00Zoomber Quit (Client Quit)
06:45:00Zoomber has joined
06:46:00Zoomber has left
07:08:00Flynnn Quit (Quit: This computer has gone to sleep)
07:08:00zoomber_laptophey karamazovapy you still on?
07:49:00BitfighterLogBot - Commit 6a456a4a0eb3 | Author: sam8641 | Log: Remove multiple static fillVector, now only one global fillVector
07:51:00zoomber_laptop Quit (Quit: zoomber_laptop)
10:04:00BitfighterLogBot - Commit add15d67aedc | Author: sam8641 | Log: Fix idle problem from shared fillVector, moved MAX_FILE_NAME_LEN, fix project file
12:49:00BitfighterLogBot - Commit 03734a261818 | Author: sam8641 | Log: Fix turret hostile team error, fix burst won't damage
14:10:00watusimoto has joined
14:25:00karamazovapycholla
15:05:00watusimotohey there
16:03:00Flynnn has joined
16:26:00watusimotoyessir... I buggered things up pretty well with my last checkin
16:55:00zoomber_laptop has joined
17:15:00BitfighterLogBot - Commit 66dbf570d711 | Author: watusim...@bitfighter.org | Log: Got rid of gEditorGame
17:15:00BitfighterLogBot - Commit de9e650ec1e8 | Author: watusim...@bitfighter.org | Log: Fix some quirks resulting from last checkin; includes ugly hack
17:15:00BitfighterLogBot - Commit 1b037136df44 | Author: watusim...@bitfighter.org | Log: Merge
17:15:00[1]watusimoto has joined
17:15:00zoomber_laptophi [1]watusimoto
17:15:00Zoomber has joined
17:17:00watusimoto Quit (Ping timeout: 264 seconds)
17:17:00[1]watusimoto is now known as watusimoto
17:19:00zoomber_laptophey karamazovapy, are you present?
17:30:00[1]watusimoto has joined
17:31:00watusimoto Quit (Ping timeout: 252 seconds)
17:31:00[1]watusimoto is now known as watusimoto
17:55:00BitfighterLogBot - Commit 10fb08a39869 | Author: watusim...@bitfighter.org | Log: Fixed editor name prepopluation by instituting another crappy hack... there is a flaw with the UIManager design.
18:12:00watusimoto Quit (Ping timeout: 246 seconds)
19:20:00watusimoto has joined
19:49:00watusimoto Quit (Ping timeout: 240 seconds)
20:00:00watusimoto has joined
20:00:00zoomber_laptop Quit (Quit: zoomber_laptop)
20:11:00zoomber_laptop has joined
20:30:00raptor has joined
20:30:00ChanServ sets mode +o raptor
20:30:00zoomber_laptophi raptor
20:30:00raptorhi
20:42:00raptorsam686: did you do any more test with forward decs?
20:46:00raptorso karamazovapy, how many contests would I have to win to get the cheapest dual action USB joystick on the market? :)
21:04:00watusimotodo you have one in mind?
21:04:00raptornegatory
21:07:00watusimotoI'll buy you one
21:07:00raptorno no no no
21:07:00raptorbut thanks
21:07:00raptorI have to earn one...
21:10:00watusimotodo you mow lawns?
21:10:00raptori used to... lots and lots
21:11:00watusimotowell, great. Come to Portland, I'll put you to work
21:11:00raptori would seriously take you up on that if I lived nearby :)
21:14:00watusimotoso would I!
21:28:00zoomber_laptopyou could come to los angeles, which is a bit closer, but then..I have no lawn..
21:29:00zoomber_laptopI'm making the biggest longest levelgen I have ever made
21:55:00BitfighterLogBot - Commit eff1731ae6d9 | Author: watusim...@bitfighter.org | Log: Minor refactor -- fewer globals
22:34:00zoomber_laptop Quit (Quit: zoomber_laptop)
22:46:00zoomber_laptop has joined
23:17:00raptorok. so. falling asleep in the middle of the afternoon amounts to dreams about your teeth falling out in pieces...
23:24:00zoomber_laptoplol
23:24:00zoomber_laptopive had one of those before
23:24:00raptorstill slightly shaken up..
23:24:00zoomber_laptopanyone good at lua stuff?
23:25:00zoomber_laptopim trying to make a levelgen, and its all good, but every occasionall time i load one, it fails
23:25:00zoomber_laptopmost of the times it works, but once in a while, the map is bare
23:25:00raptori was just thinking I should learn to use LUA...
23:26:00zoomber_laptopatually, i think i might have found the error
23:26:00zoomber_laptopmaybe*
23:27:00zoomber_laptopVariable 'LooadoutImportant' cannot be used if it is not first declared
23:27:00zoomber_laptopthat cant be it, because I defined that; it must be an old error
23:28:00zoomber_laptopoh no
23:28:00zoomber_laptopthat is it
23:28:00zoomber_laptopnow tht message is at the top of the log
23:28:00zoomber_laptopah
23:28:00zoomber_laptopaha!
23:29:00zoomber_laptopfunny how you can spell one word wrong and the whole levelgen crashes
23:32:00raptorwell - i supposed I should get PNG screenshots to work in Mac...
23:34:00zoomber_laptopno need to?
23:35:00raptori was thinking that..
23:35:00raptorhow do you do it now?
23:35:00zoomber_laptopcommand + option + 3 = full screen shot automatically
23:35:00zoomber_laptopcommand + option + 4 = Select area of screen to capture
23:35:00zoomber_laptopif using windows, command will be the windows icon button
23:35:00zoomber_laptopor the apple key, if you refer to it that way
23:41:00raptoris wiki fully functional again?
23:42:00zoomber_laptopi dont know
23:42:00zoomber_laptopreadable\
23:47:00raptorgreat - having a hard time finding the binary for 10.4
23:47:00raptorsince lion came out - i think everyone is dropping 10.4 quickly
23:51:00zoomber_laptopyes
23:51:00zoomber_laptopi plan to get it just for my macbook
23:51:00zoomber_laptopbut keep my desktop with 10.6
23:52:00zoomber_laptopPerfect for bitfighter, as I will have xcode 4 on one machine, and xcode 3 with the old sdks on the toher
23:52:00zoomber_laptopas soon as i get it, i will compile a bitfighter-for lion
23:53:00zoomber_laptopI did some edits to the code, (some not on my clone yet). I think It will work, but who knows

Index Search ←Prev date Next date→

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