#bitfighter IRC Log

Index Search ←Prev date Next date→

IRC Log for 2012-04-23

Timestamps are in GMT/BST.

00:12:05sam686 has joined
00:12:05ChanServ sets mode +v sam686
00:17:12koda has joined
00:27:03raptorwhat happened to 'Bitfighter sam686'?
00:29:35sam686not sure..
00:29:59sam686the server is still there, maybe bitfighter might have just crash...
00:30:26raptoralso, cracatoa's supernova server is running 3 times again
00:33:36Heyublol
00:37:43sam686i should try building with -o2 -g , not sure if that work, if it does, at least i could get something when it does crash (with -g) and without big slowdown (by using -O2)
00:38:09raptorit works
00:38:25raptoryou get decent stack traces, but not variable values passed into methods
00:39:04sam686it won't be full debug information due to some variables being optimized out, and some function being inlined...
00:39:10raptoryep
00:39:31sam686but at least thats better then slowdown caused by not using -O2
00:44:08sam686i should try the latest changes, we are still at the code that is compatible with 017, right?
00:44:14raptorumm...
00:44:21raptorthe latest has lua problems
00:44:29raptorbut i think still protocol-compatible...
00:44:59sam686any LUA stuff is server side (and editor side), not client side..
00:45:24raptoryes - is it stable for you to run bots?
00:45:46sam686i will find out when i do run the latest code in server
00:45:50raptorhehe
00:45:54raptorgood luck...
00:46:10raptoryou'll have to change your levelgens i think
00:46:31sam686yes, i just need to upload my levelgen file
01:10:07sam686i guess everything works with the latest code (so far)... (using "wget http://sam686.maxhushahn.com/upload/6357.levelgen " for my levelgen )
01:10:15raptorhooray!
01:11:28sam686it does appear to fit 80 robots with maybe little slowdowns... in bitmatch
01:11:42raptorthat's much better though isn't it
01:12:17sam686PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
01:12:17sam68622523 sam686 16 0 23432 9.9m 3104 R 49.7 1.9 0:47.09 bitfighterd
01:12:25sam686thats while doing 80 robots
01:12:37raptorthat's way better!
01:13:02BFLogBot - Commit a72ea8fe0794 | Author: buckyballreaction | Log: Revert BirdOfPrey ship design and add flames; disable bots using new shapes for now
02:46:54raptordoes s_bot shoot worse now?
02:47:10raptorseems like the aim is off when moving in sync with it at an angle
02:47:23raptormust have to do with the timing..
02:47:31raptorerr, the control tick
03:06:22sam686maybe the robots having only 30 ticks per second (instead of 100, the usual server fps) is making robot aim a little worse?
03:06:44raptori think that is the case - but i'm not sure why...
03:07:05raptorit was even worse before, but i fixed it by not using getTime
03:07:37raptorhave you noticed it being worse?
03:08:28raptorwait!
03:08:31raptorit's not worse
03:08:41raptorit's the same - i just tested with 017a...
03:08:44raptorso never mind
03:08:51sam686that means, one game tick, robot calculates aiming, moving, and such, the next tick, the robot moved, but doesn't calculate anything (like in LUA), and doesn't change aiming either.. (eventually it recalculates on the next robot tick of running LUA)
03:09:23sam686as in, every game ticks all robots move (and all bouncy objects, all asteroids movements)
03:10:03sam686but, now we are limiting the LUA robot run time, which does the aiming calculation to about 30 per second, producing somewhat worse aiming for robot
03:10:33raptorit is actually the same... 30 per second is still more than how many bullets can get shot per second
03:10:43raptorbut i'm going to compare with 015a...
03:12:07raptoryep, same as 015a, too
03:12:19raptorok nevermind
03:12:25raptors_bot is still ok
03:12:51sam686maybe its the robots random aiming to prevent "too hard" (as in the robot aiming shakes back in forth)
03:13:08raptoryeah - i'm thinking it is other minor factors like that
03:38:08koda Quit (Quit: I used to be chatting like you. Then I took an arrow in the knee)
04:43:21raptorwell goo dnight
04:43:40raptor Quit ()
04:57:43Zoomber Quit (*.net *.split)
05:04:09Heyub Quit (Ping timeout: 246 seconds)
05:14:20Zoomber has joined
07:05:59watusimoto has joined
07:05:59ChanServ sets mode +o watusimoto
07:23:29sam686 Quit (Ping timeout: 245 seconds)
07:23:58sam686 has joined
07:23:58ChanServ sets mode +v sam686
07:40:15sam686 Quit ()
07:49:21kodaws has joined
11:18:56watusimoto Quit (Remote host closed the connection)
11:26:53watusimoto has joined
11:26:53ChanServ sets mode +o watusimoto
13:24:28LordDVG has joined
13:45:13LordDVG Quit (Read error: Connection reset by peer)
13:49:45Zoomber_ has joined
13:49:45Zoomber Quit (Read error: Connection reset by peer)
13:49:45ChanServ sets mode +v Zoomber_
13:49:46Zoomber_ is now known as Zoomber
14:14:18Heyub has joined
14:21:24LordDVG has joined
14:43:18LordDVG Quit (Remote host closed the connection)
14:51:11Watusimoto_ has joined
14:59:54raptor has joined
14:59:54ChanServ sets mode +o raptor
15:00:16raptorgood morning!
15:40:02HeyubGood morning!
15:41:02raptorhi Heyub
15:41:18HeyubHello
15:42:39raptorwatusimoto: sam686 decided to put the latest LUA changes in 'production', so to say: he recompiled his 'Bitfighter sam686' server to run on it
15:43:35raptorso far so good, with drastic memory and CPU usage reductions
16:15:20watusimotogood
16:15:44watusimotoso the smart pointer thing isn't going to work
16:15:53raptor:(
16:16:20watusimotoyeah, but there may yet be a solution that doens't involve manually creating proxies for all objects
16:16:31watusimotopossibiliites include templates and/or macros
16:16:36raptorah templates
16:16:43raptori still need to wrap my head around those
16:17:01watusimotoon the plus side, I have test items working with luawrapper, and it seems quite stable
16:17:12raptorany advantages?
16:17:23raptor(over Lunar)
16:17:24watusimotoI'm going to be away on a trip until Friday, btw, so I may be scarce
16:17:33raptorok
16:17:40watusimotomain advantage at this point is the built in support for inheritance
16:17:50watusimotowhich will reduce duplicate code
16:17:59watusimotoand no dynamic_casts
16:18:13watusimotoand no thunk method :-)
16:18:20raptoroooo
16:18:24watusimotoit is a lot like lunar, but feels cleaner
16:18:39raptori can get behind that...
16:18:52raptorcleanliness helps the mind be at peace
16:18:58watusimotoso I will probably check in my test item code tonight, so I can access it if I get some time while I'm on the road
16:19:06raptorok
16:19:19watusimotoit is stable, but I wouldn't put it on a server without further testing
16:19:27raptorok i'll just stay away from it
16:19:39watusimotoin any case, it will only affect testitems, which barely come into play with bots
16:20:19raptori've been thinking about putting up a post in the forums about bot body shapes (then I can close a bug!)
16:20:19watusimotoI wrote a test item proxy class, and am using that as a model for figuring out how to generalize the proxy creation
16:20:27watusimotosounds good
16:20:56watusimotoyou can look at the way the test item interface is written for luawrapper
16:20:58raptori was thinking i'd show pictures and mention that the code will be disabled until a valid gameplay usage, other than novelty, is thought up for them
16:20:59watusimotoit is pretty simple
16:21:09raptorok
16:21:18watusimotomuch more so than other classes that use proxies (projectiles and ships)
16:21:35raptorthe proxy code bugs me a bit
16:21:36kodaws Quit (Ping timeout: 260 seconds)
16:21:53watusimotoone thing we could do is allow users to display bots as other body shapes on their own machines
16:21:55raptorbecause of the cleanliness factor..
16:21:58watusimotoan INI setting or something
16:22:07watusimotowell, you can look at the code; they're not too bad
16:22:34raptorwell, i'll make the post, close the bug, and see the reaction...
16:22:37raptor:)
16:22:46watusimotosmart pointers won't work because objects are passed to lua by pointer, and the smartpointer has no way of reference counting that
16:22:53raptorah
16:22:58watusimotohence the need for proxy
16:23:29watusimotothe proxy is basically a pointer to the original data object, with a boolean reporting wither the original has been deleted
16:24:04watusimotolua controls the life cycle of the proxy, c++ controls the life cycle of the data object
16:24:12watusimotoeither can exist without the other
16:24:29raptoryuk
16:25:09watusimotoso when lua wants to know the position of an object, it asks the proxy, the proxy checks if the object has been deleted. if so, it returns nil, if not, it passes the request on to the object and returns the result
16:25:40watusimotothe alternative is lua making calls ona deleted object, which causes crashes very quickly
16:27:05raptorlots o' crash
16:28:11raptorso that means to get around the proxy, you need a 'smart' templated object of sorts that can return nil if it detects the c++ object is deleted?
16:30:03watusimotoI don't think you can get around the need for the proxy -- the key is how simple can we make them and how automatic?
16:30:24raptoryes, i think that's what I meant
16:30:25watusimotoor can we find a way to use a single proxy class for all our objects
16:30:41watusimotoright now we need one proxy class per object class
16:30:53raptorthe logical idea of a proxy is still important, but we have so much intertwined code
16:30:54watusimotoif I can get it to one class for any object, that would be great
16:31:28watusimotothen we could get rid of luarobot and luaprojectile classes which are heavy proxy classes
16:32:02watusimotoluawrapper uses static methods for a lot of stuff, which I don't like, but does simplify the proxy issue
16:32:18watusimotobecause things lunar wants on the proxy can be statics anywhere
16:32:34raptorhave you seen this?: http://codereview.stackexchange.com/questions/10421/luaproxy-semi-automatic-exportation-of-c-classes-to-lua
16:34:56watusimotonot sure they are using the proxy for the same thing as I'm describing
16:35:03raptorseems like several people have done work on different things...
16:35:17raptorbut no one great project to rull them all
16:35:21raptor*rule
16:37:53raptorthis is the second time i've come across OOLUA, but now I have a different outlook since I understand a bit more: http://code.google.com/p/oolua/
16:41:11LordDVG has joined
16:55:39watusimotoI'm not sure it would improve our lot a lot
16:55:47watusimotoespecially as we're using luavec
16:55:53watusimotoand I'm not sure they are compatible
16:56:05raptoryeah, looking through it again, it seems like it isn't much different than Lunar
16:59:00raptorwere you ever in contact with the lua-vec developers?
17:37:36LordDVG Quit (Quit: Leaving)
17:49:25watusimotono
17:49:35watusimotoI think it's left for dead
17:49:42watusimotoI may be able to upgrade it though
17:51:17raptoroh neat - you can download lua-vec as a patch
18:02:40watusimotothough we patched the patch
18:02:47watusimotook, heading home... 8PM here
18:02:48watusimotobye
18:02:53raptorbye
18:07:09watusimoto Quit (Ping timeout: 240 seconds)
18:19:16Watusimoto_ Quit (Ping timeout: 244 seconds)
19:29:09raptorhttp://bitfighter.org/forums/viewtopic.php?f=4&t=1447
19:29:14raptornow to see the backlash...
19:35:27HeyubTo the chopping block!
19:35:57raptori fear for my head
19:36:05Heyublol
19:37:43HeyubI like the bird of prey, the box is... well... a box lol
19:37:58raptorhaha, yep
19:38:08raptorbut it was the pinnacle of my creation!
19:38:47raptorthe trick is making a shape simple enough to not be distracting and easy enough to tell where the front is..
19:39:05HeyubYeah
19:39:35HeyubThey could be handy for dungeon maps though.
19:40:10HeyubIf each shape had a certain weapon it used (burst, phaser, triple) would be interesting. But thats just dongeon maps.
19:40:36HeyubErr, if they were bots, would be annoying to limit humans to a specific weapon haha.
19:40:39raptorthat's the 'functional-type' idea we'd want for justifying their use
19:41:00raptormake a specific bot a specific shape is a good idea
20:16:40sam686 has joined
20:16:42ChanServ sets mode +v sam686
20:33:30sam686 Quit (Ping timeout: 245 seconds)
20:33:51sam686 has joined
20:33:51ChanServ sets mode +v sam686
20:42:34Watusimoto has joined
20:46:37Heyub|2 has joined
20:48:30sam686 Quit (Ping timeout: 245 seconds)
20:48:55Heyub|2 Quit (Read error: Connection reset by peer)
20:49:07sam686 has joined
20:49:07ChanServ sets mode +v sam686
20:49:09Heyub Quit (Ping timeout: 246 seconds)
20:49:16Heyub|2 has joined
20:53:42Heyub|2 Quit (Ping timeout: 246 seconds)
22:00:34BFLogBot - Commit b34747cc11fe | Author: watusim...@bitfighter.org | Log: Initial exploration of LuaWrapper
22:00:35BFLogBot - Commit f7b588973bda | Author: watusim...@bitfighter.org | Log: Merge
22:00:37BFLogBot - Commit 49e91e1420d3 | Author: watusim...@bitfighter.org | Log: Introduce LuaWrapper code, sample testItem lua proxy. Still very much a work in progress.
22:00:38BFLogBot - Commit 18664c6f25ab | Author: watusim...@bitfighter.org | Log: Merge
22:03:13raptor Quit ()
22:21:18Watusimoto Quit (Ping timeout: 245 seconds)
23:31:50sam686 Quit (Ping timeout: 245 seconds)
23:32:20sam686 has joined
23:32:20ChanServ sets mode +v sam686

Index Search ←Prev date Next date→

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