#bitfighter IRC Log

Index Search ←Prev date Next date→

IRC Log for 2012-02-05

Timestamps are in GMT/BST.

01:21:34raptor has joined
01:21:36ChanServ sets mode +o raptor
01:23:04Watusimoto Quit (Ping timeout: 260 seconds)
01:23:16raptorhello!
01:23:21raptori missed him!
01:26:08raptorhi sam686, any luck tracing down that evil virtual memory leak?
01:26:18raptori'm looking at that txt file
01:26:27raptornot sure i understand what you're showing..
01:32:56raptorexplain what you do with packetRewind
01:49:26raptori think all things Mac are moving away from gcc to llvm
01:54:44sam686the packetrewind stuff was there, becasue tnl won't know how big that event / packUpdate will be, until after seeing the size after running packUpdate
01:55:32raptorso after a packet is built it is analyzed to see if it needs to be sent, if not, rolls back?
01:55:48sam686yes
01:56:03raptorcould that analysis not be done during the packet creation?
01:57:14sam686there is many things that affect the size of sending, including TableStringEntry (send full string once, send small bits of index after)
01:58:29raptorok
01:58:45raptordo you think you know what that memory leak from last night may b e?
02:18:38karamazovapymy server's been holding steady memory-wise since I relaunched it five hours ago
02:18:59raptorthe new level server?
02:19:14karamazovapyyeah, it was crashed earlier today, though
02:19:25raptorhmm...
02:20:29raptorwell when we figure it out we should apply the patch directly to the 016 revision
02:34:08karamazovapyI'm sending a mass e-mail for FREE BAIT, but it's moving awfully slowly...
02:41:30karamazovapyI'm not sure it's sending
02:49:12sam686the memory leak seems to only happens on linux, and only seem to happen when trying to connect to master...
02:49:36raptoryou really think it's because of connecting to master?
02:50:47sam686when i run a server, with a different, non-existant master address, it starts to increase VIRT size my 9m everytime it tries to connect
02:51:04raptorok, mine does that too
02:51:14raptorsoo...
02:51:20raptorwe need to find out why
02:58:54sam686can you maybe test this? http://sam686.maxhushahn.com/upload/test.cpp and see if there is any memory leak there when you run that test...
02:59:16sam686as in, using the same compile settings as /zap
02:59:25raptorok
03:01:12sam686but, you might need to temperory get rid of main.cpp, and maybe rename test.cpp into main.cpp (or in make file, clear all .0 files and add test.o)
03:04:42raptorok
03:04:46raptorwant the output?
03:05:11raptorhttp://pastie.org/3319140
03:05:24sam686i don't care what the output said, did the VIRT size go way up? (as if you are using "top")?
03:05:36raptorit hit a trap/breakpoint
03:06:31sam686i think your gcc simply make the program hit a breakpoint when the program ends..
03:07:56raptor#3 0x0000000000417e72 in TNL::Assert::processAssert (filename=0x45ad93 "netBase.cpp", lineNumber=194,
03:07:57raptor message=0x45af80 "Error! Object deleted with non-zero reference count!")
03:08:43raptoror better: http://pastie.org/3319152
03:09:07sam686did my text.cpp have errors for you?
03:09:10raptoryes
03:09:36raptor1. tnlUDP.h <-- wrong case
03:09:49raptor2. it keeps hitting an assert and doesn't finish
03:09:57raptorso i can see the memory usage
03:09:59raptor*can't
03:10:20raptorproblem in RefPtrDatra
03:11:05sam686looks like your linux seems to not like this line "NetConnection net = NetConnection();"
03:11:41sam686you can get rid of that test...
03:12:05raptorok commenting out
03:12:15sam686i will try writing a multiple thread test, too..
03:14:17raptorvirt: 12252K mem: 188K shared: 1008K
03:16:19sam686thats a lot of "virt" memory being used...
03:16:36raptori guess i don't fully understand 'virt' memory
03:17:54raptorbecause it's huge for every application that I have running
03:20:56karamazovapymy stupid mass e-mail is still sending, supposedly
03:21:28karamazovapyand I've received it twice
03:21:41raptorwhat system are yo uusing?
03:22:31karamazovapythe ACP
03:22:41raptori didn't know that was afeature
03:23:21karamazovapyoh yeah, I've used it for every BBB
03:24:27karamazovapyI'm gonna give it ten more minutes of sitting there "Waiting for bitfighter.org..." and then I'm gonna close the window
03:26:29raptorthey reversed the mouse wheel for mac os x 10.7
03:26:34raptorpunks
03:28:33sam686` has joined
03:28:33sam686 Quit (Read error: Connection reset by peer)
03:28:53karamazovapyI think that's part of apple's plan to be rid of scroll bars
03:28:53sam686`http://sam686.maxhushahn.com/upload/test_multi_thread.cpp i have a new test you can try
03:29:58karamazovapysince they're moving in the direction of a unified apple platform, they want desktop dragging to conform to touch control standards where down is up and up is down
03:30:23raptorpunks
03:31:52sam686`it does kind of seem like the problem is with the memory leak somewhere inside Address.set
03:32:08sam686`but, i don't see any memory leak in windows
03:32:20sam686` is now known as sam686
03:32:22ChanServ sets mode +v sam686
03:32:24raptorcan you build dedicated on windows?
03:32:30sam686yes
03:33:15sam686i don't think dedicated not dedicated make a differenct, but rather it happens on linux, and not windows
03:36:35karamazovapyI've given up on my mass e-mail. here's hoping it got to anyone who wanted it.
03:38:12raptoroh wow - i'm testing 015, sam686, and virtual memory is not increasing at all
03:38:26raptorstays at 20MB for dedicated
03:39:41sam686try my test again, after changing all "bitfighter.org:25955" to "127.0.0.1:25955"
03:39:52raptorthat's what i did already
03:39:55raptorstays constant
03:40:18sam686so, is there a memory leak problem when using "bitfighter.org:25955" ?
03:41:22raptorok trying again
03:43:05raptorwith 015a, you mean?
03:43:26sam686you can try "bitfighter.org:43443" and look for memory leak..
03:43:53sam686or better yet, try /connect "bitfighter.org" in server list, repeatedly
03:44:13raptorno leak in 015
03:44:15raptor015a
03:44:18raptorthis is nutes
03:44:20raptornuts
03:44:47sam686so, on my old test, or new test, which one had bigger VIRT mem?
03:44:59raptorhaven't done new test...
03:45:01raptorwill do now
03:46:05raptor3.4 GB!!!
03:46:13raptorthat's the virtual size
03:46:18raptorof the threaded one
03:46:38sam686haha, the problem is threading it?
03:46:55raptori don't know...
03:47:16raptornot sure what you did; split up the object creation into threads?
03:47:48sam686you can try blanking the NameToAddressThread::run() (leave the "return 0" there) and re run the test
03:47:58raptorok
03:47:59sam686or commenting out
03:48:45raptor3.3GB!
03:49:14sam686looks like thread problems...
03:49:47sam686how much memory does your computer have, by the way?
03:49:53raptor4.0GB
03:49:58raptorwait
03:50:00raptornot
03:50:01raptor8GB
03:50:51sam686i am guessing it could slow down your computer by a lot if the thread count went much higher
03:50:57raptorprobably
03:51:51raptori don't see any important changes to thread.cpp or tnlThread.h that could have caused this..
03:53:56raptordo we even know if the threads are getting cleaned up?
03:55:52sam686the threads is cleaned up after it is done running (at least the classes is deleted)
03:59:07raptori bet it has something to do with posix threads: http://www.cs.cf.ac.uk/Dave/C/node30.html#SECTION003040000000000000000
04:00:37sam686each thread have it's own memory stack, which i think is somehow not cleaned up after that thread is done..
04:00:48raptorso i bet TNL implementation of threading on linux isn't good
04:01:18raptorand just the fact that you moved to a thread for connection to master was what introduced the problem
04:01:20sam686and, maybe the linux stack size might be too big, maybe some linux default to 10MB stack size
04:01:26raptoryeah
04:01:29raptorit can be big
04:01:59raptordid you read the note:
04:02:00raptorNote - When there is no explicit synchronization to prevent it, a newly created, detached thread can die and have its thread ID reassigned to another new thread before its creator returns from pthread_create(). For nondetached (PTHREAD_CREATE_JOINABLE) threads, it is very important that some thread join with it after it terminates -- otherwise the resources of that thread are not released for use by new threads. This commonly results in a
04:02:02raptormemory leak. So when you do not want a thread to be joined, create it as a detached thread.
04:10:59sam686 Quit (Read error: Connection reset by peer)
04:20:07sam686 has joined
04:20:07ChanServ sets mode +v sam686
04:24:53raptori may know how to fix memory leak...
04:26:56raptorcouple more tests.. i think i may have fixed it..
04:28:16raptorFIXED IT
04:28:17raptoryay
04:36:30raptorok karamazovapy
04:37:05raptorif you pull from hg on your server and update to the new 'bitfighter-016' it will have the memory leak fixed
04:39:42BFLogBot - Commit 871e87047a7c | Author: watusim...@bitfighter.org | Log: Merge
04:39:43BFLogBot - Commit 4dcc1eb89ea3 | Author: watusim...@bitfighter.org | Log: Fix to specifying leveldir on the cmd line
04:39:45BFLogBot - Commit 6acda1ffb7a6 | Author: watusim...@bitfighter.org | Log: whitespace
04:39:46BFLogBot - Commit e76924cf9f2b | Author: watusim...@bitfighter.org | Log: add k to credits for his level contributions
04:39:49BFLogBot - Commit d813fc204606 | Author: buckyballreaction | Log: Remove old classic levels. Add in new levels by karamazovapy
04:39:50BFLogBot - Commit 7a643e0aff33 | Author: sam8641 | Log: Fix a problem with getmap
04:39:52BFLogBot - Commit d87435f7051f | Author: watusim...@bitfighter.org | Log: New version number
04:39:54BFLogBot - Commit 8b27c24262dd | Author: buckyballreaction | Log: Fix TNL::Thread. Linux pthread was never actually being clean-up, thus introducing a memory leak
04:39:55BFLogBot - Commit b5ffd0fc78b1 | Author: buckyballreaction | Log: Change tag of 016 because of nasty linux memory leak
04:39:57BFLogBot - Commit 28b72de3fe41 | Author: buckyballreaction | Log: Merge in new 016 tag and Linux pthread memory leak fix
04:39:59raptoruhoh, i messed up the logbot
04:43:59-adams.freenode.net- *** Looking up your hostname...
04:43:59-adams.freenode.net- *** Checking Ident
04:43:59-adams.freenode.net- *** No Ident response
04:43:59-adams.freenode.net- *** Couldn't look up your hostname
04:44:03BFLogBot has joined
04:44:03Topic is 'Latest release 016 http://bitfighter.org | Forums: http://bitfighter.org/forums/ | GC Project: http://code.google.com/p/bitfighter/'
04:44:03Set by raptor!~raptor@unaffiliated/greenmachine on Wed Feb 01 04:14:39 GMT 2012
04:44:03-ChanServ- [#bitfighter] Welcome to #bitfighter. This is an IRC channel, many or all of the users may not be paying attention. Please have patience when waiting for a response.
04:46:33sam686your linux fix doesn't work on server saving game stats, the VIRT memory keeps going up after the end of each game
04:46:45raptorwhat??
04:47:31sam686it might be due to a "delete this" in a thread to erase the current class..
04:47:48raptorbefore teh destructor is called, you mean?
04:48:03raptorhmmm...
04:48:20raptorso the destructor is not being called?
04:50:10sam686InsertStatsToDatabase in gameType.cpp have a "delete this" in run()
04:50:23raptordo we need that?
04:50:59sam686well, what are we going to do without "delete this"? ... let teh "InsertStatsToDatabase" memory leak?
04:51:10raptorof course not! :)
04:51:40raptorso... how to call the TNL::Thread destructor?
04:53:52sam686maybe the pthread_join might be failing when mThread is the currect thread running the destructor..
04:55:05raptori don't think we need the delete
04:55:25sam686it may memory leak without the delete...
04:55:36sam686you could try, but it won't help..
04:56:05raptorthen that means i found a bug in TNL
04:56:23raptorlook at ThreadQueue constructor in thread.cpp:246
04:57:12raptorbut i guess we don't use ThreadQueue anywhere?
04:58:23sam686there isn't a ThreadQueue in /zap/
04:58:28raptornope
04:58:32raptorso it's a bug, but not used..
05:00:52raptorwhat about moving the 'delete this' into the destructor?
05:01:24sam686not sure if that will help, and, delete this doesn't work in destructor, because it is already being deleted
05:01:35raptorhmmm...
05:04:17raptorcan we avoid calling 'new'?
05:07:48sam686pthread_join seem to return a number of 35 when it fails to clear the memory
05:07:58sam686but returns zero when it works ok
05:11:22raptoris it returning a 35 when doing gamestats?
05:13:22sam686yes
05:13:58sam686maybe a pthread_exit might do? i can test as i have linux up now..
05:14:30raptormaybe exit is better...
05:18:10raptorwell actually
05:18:24raptorpthread_exit would have to be called in that thread..
05:20:30raptor Quit (Remote host closed the connection)
05:22:25raptor has joined
05:22:38ChanServ sets mode +o raptor
05:23:12raptorpthread_exit doesn't work...
05:23:45raptormaybe an extra pthread_join in that thread...
05:27:23sam686http://stackoverflow.com/questions/1688424/can-a-pthread-perform-clean-up-on-its-own
05:28:02sam686maybe a pthread_detach() in line 203 (ThreadProc) thread.cpp
05:28:35raptorok, will try that..
05:31:39raptortesting...
05:31:56raptori've never done threads in anything but java... it's a whole new world!
05:32:59sam686never mind, there is already a pthread_detach
05:33:08sam686or not
05:33:18raptornot
05:33:27raptorso far it's working with the master connection thread
05:34:45raptorso far it's working with the stats thread, too
05:34:46raptorcool
05:34:52raptorok, i'll recommit...
05:34:55raptorand re-tag
05:36:01raptori notice that when logging stats for the first time, virtual memory goes from about 30MB to 95MB
05:41:33sam686ok, i got it working now on my side (i haven't seen your commit yet)
05:41:44raptori'm retagging for 016
05:41:51raptori am commiting..
05:43:18sam686it looks like pthread_detach is the way to go (it needs to be called while on the new thread itself)
05:43:30raptoryes, i put it in the constructor
05:43:35raptorsorry
05:43:38raptornot in constructor
05:43:41raptorin start()
05:44:56raptorok pushed
05:45:00raptora little messy
05:45:02sam686oh, i tried putting it n Thread::start and it segfault, i guess detach first, then trying to run will crash...
05:45:09raptorhaha
05:46:45sam686it works ok with your changes now..
05:46:49raptoryay!
05:47:16raptorkaramazovapy: you'll have to update to bitfighter-016 again if you did already since my message about it...
05:48:37sam686i think i should stop doing /settime 0.01 , it was putting game reports here: http://bitfighter.org/gamereports
05:48:43raptorhaha
05:49:44BFLogBot - Commit 2f775c312a86 | Author: buckyballreaction | Log: Better Linux thread fix - now the thread just cleans itself up automatically
05:49:45BFLogBot - Commit 84c61918b58c | Author: buckyballreaction | Log: Re-tag bitfighter-016 again with better thread memory leak fix
05:49:47BFLogBot - Commit 8cfe1a5ed6b5 | Author: buckyballreaction | Log: Merge new 016 tag again
06:12:12sam686i should fix dedicated build of latest "tip" revision, but i need to get to sleep first, good night..
06:12:29raptorgood night
06:14:32sam686 Quit (Remote host closed the connection)
06:18:05raptor Quit (Remote host closed the connection)
09:09:13Watusimoto has joined
10:02:17LordDVG has joined
10:42:29Watusimoto Quit (Ping timeout: 240 seconds)
10:54:38Watusimoto has joined
11:23:14Watusimoto!log
11:23:14BFLogBotThis channel is logged - http://199.192.229.168/irclogs/
11:23:19Watusimoto!bug
11:23:19BFLogBotTo enter a bug, please make sure it is reproducible and then go to http://code.google.com/p/bitfighter/issues/list | Also, see current running bug list: http://bitfighter.org/wiki/index.php/Running_Bug_List
11:25:35BFLogBot - Commit bd61a9f7b652 | Author: watusim...@bitfighter.org | Log: First pass at core with shootable panels. Not perfect yet, but seems stable.
11:25:36BFLogBot - Commit 2427bd173ade | Author: watusim...@bitfighter.org | Log: Merge
11:57:37Zoomber Quit (Ping timeout: 252 seconds)
12:03:59Zoomber has joined
12:03:59ChanServ sets mode +v Zoomber
12:05:41BFLogBot - Commit a6e31487de66 | Author: watusim...@bitfighter.org | Log: Fix bug with using such low resolution for health transmission -- low panel healths get transmitted as 0, making it hard to figure out which panels are still alive
12:05:43BFLogBot - Commit b6c4c0c5d3e3 | Author: watusim...@bitfighter.org | Log: Comment
12:10:19koda has joined
12:29:25koda Quit (Quit: I used to be chatting like you. Then I took an arrow in the knee)
12:45:49Watusimoto_ has joined
12:48:18Watusimoto Quit (Ping timeout: 255 seconds)
13:57:03koda has joined
14:47:46raptor has joined
14:47:46ChanServ sets mode +o raptor
14:50:43raptorgood morning!
14:52:55Watusimoto_hi
14:53:14Watusimoto_sounds like you guys have had a pleasant 48 hours
14:53:19raptorhi
14:53:24raptorthat was an interesting bug
14:53:43raptorfirst time i've worked with threads in a non-jvm environment
14:53:53Watusimoto_glad you found it -- I never would have
14:54:01raptorit was luck
14:54:02Watusimoto_why do we use threads?
14:54:31raptorsam686 introduced them for writing stats to the local sqlite database and for running connections to master
14:54:44raptorit was done because they would make the game hang
14:55:46Watusimoto_oh, right
14:55:56Watusimoto_if db is non responsive, we can't wait for a response
14:56:10raptorwhich brings up the question: do we want to logStats locally by default?
14:56:37raptori looked at it and loading up the sqlite stuff uses an addition 60MB of virtial address space
14:56:42Watusimoto_except sqlite shouldn't hang or get hung up, should it? or does it just get bogged down?
14:56:54Watusimoto_do we log stats localy by default now?
14:57:01raptorwe always have, i think..
14:57:16Watusimoto_the only reason to disable it is to reduce memory?
14:57:32raptoryep - well, and I've heard of no one actually using the feature
14:57:44Watusimoto_what if we wrote all the stats to a text file in the form of sql statements that could be inserted into a db later?
14:58:19Watusimoto_but yes, perahabs we could disable by default
14:58:28raptorthat would be easy enough
14:58:28Watusimoto_after all, al lthe info is available on the master, theoretically
14:58:33raptoryeah...
14:58:40Watusimoto_and if no one cares...
14:58:48raptorand i think it would only be useful for those who run dedicated servers
14:58:50Watusimoto_only sam and k would really care
14:58:54raptorexactly
14:59:04Watusimoto_sure, let's turn it off by default
14:59:14raptorok
14:59:58raptorI hope the merge went OK for you...
15:00:20Watusimoto_seemed to
15:00:28raptoran 8MB leak ever time a level ended was serious enough that i thought I needed to re-tag...
15:00:33Watusimoto_looked kind of gross
15:00:40raptoryeah, it was messy
15:00:42Watusimoto_only on dedicated server?
15:00:50raptorand if you hosted a game on mac/linux
15:01:09raptorany system that uses pthreads
15:01:29Watusimoto_ah, ok
15:01:39raptorwhich is basically non-Windows
15:01:48Watusimoto_well, good. I'm trying to figure out why the bullets don't show up on cmdrs map
15:01:55raptori know why
15:01:58Watusimoto_oh, and my latest core checkin has the panels get thinner
15:02:00Watusimoto_you do?
15:02:16raptoryeah - it's the gameobject TestFunc
15:02:24Watusimoto_ah yes, I know why now too
15:02:31Watusimoto_and that's not it :-)
15:02:35raptorok
15:02:36Watusimoto_but it is related
15:02:43Watusimoto_It's this:
15:02:44Watusimoto_ if(ship->isModulePrimaryActive(ModuleSensor))
15:02:45Watusimoto_ mGameObjDatabase->findObjects((TestFunc)isVisibleOnCmdrsMapWithSensorType, rawRenderObjects);
15:02:45Watusimoto_ else
15:02:45Watusimoto_ mGameObjDatabase->findObjects((TestFunc)isVisibleOnCmdrsMapType, rawRenderObjects);
15:02:56raptorah yes
15:03:03Watusimoto_but what is the fix?
15:03:06raptori know
15:03:18raptorneed to test for the equip state
15:03:40raptorlet me find the proper way to do that...
15:03:47Watusimoto_yes, but what i don;t know is whether there is any cmdrs map bonus to activating sensor
15:03:56Watusimoto_ok, you want to handle this case?
15:03:59raptorsure
15:04:06Watusimoto_clientGame 1501
15:04:07raptori was going to as my next one anyways
15:04:18Watusimoto_well, ok then. I'll move on
15:05:06Watusimoto_i slightly reorged the buglist
15:05:20raptoryes - i have lots of mac work to do
15:05:34raptoralso, i can't believe they made 10.7 Lion ever worse
15:05:43raptorscroll wheel is reversed
15:05:56raptoryou can't select you install location with installers...
15:06:06BFLogBot - Commit 25a05918bf2b | Author: watusim...@bitfighter.org | Log: Formatting, comments
15:06:08BFLogBot - Commit 3b148b7f27ce | Author: watusim...@bitfighter.org | Log: Formatting, comments
15:06:09BFLogBot - Commit 6b5190198d2f | Author: watusim...@bitfighter.org | Log: Formatting, comments
15:06:11BFLogBot - Commit e8d24d980ee1 | Author: watusim...@bitfighter.org | Log: Formatting
15:06:12BFLogBot - Commit 8a5bff4682da | Author: watusim...@bitfighter.org | Log: Formatting
15:06:14BFLogBot - Commit 538b8d95902a | Author: watusim...@bitfighter.org | Log: Formatting
15:06:15BFLogBot - Commit 0e2e49a846a6 | Author: watusim...@bitfighter.org | Log: trivial
15:06:17BFLogBot - Commit fbe2ec6df4c9 | Author: watusim...@bitfighter.org | Log: Replace rect drawing code with call to drawRect()
15:06:18BFLogBot - Commit 87b12d2e6042 | Author: watusim...@bitfighter.org | Log: Comment
15:07:27Watusimoto_all on the way to making it a heavy iphone
15:09:17raptorhehe - energy was not loose
15:09:24raptori'm sure i made that mistake..
15:11:20BFLogBot - Commit a472873c8b40 | Author: watusim...@bitfighter.org | Log: Energy was not loose
15:11:32Watusimoto_back in a while; son wants attention
15:11:37raptorcool
15:22:08raptorCores are invisible
15:22:32raptoralso upon map start-up it there are many sparks - as though they've exploded
15:26:24BFLogBot - Commit a03e861b7138 | Author: buckyballreaction | Log: Fix bullets not being shown on commander's map if Sensor module is equipped
15:26:25BFLogBot - Commit f5ce7d986755 | Author: buckyballreaction | Log: Default don't log game statistics to local sqlite database
15:41:28BFLogBot - Commit 7851b8af7b6b | Author: buckyballreaction | Log: Hostile loadout zones drain energy no matter what
15:54:08Zoomber Quit (Ping timeout: 240 seconds)
15:55:37raptor Quit (Remote host closed the connection)
15:56:01raptor has joined
15:56:02ChanServ sets mode +o raptor
15:56:10BlackBird has joined
15:56:17BlackBirdhi
15:56:22raptorhi
15:56:36BlackBirdwhere can i download the scorch code for bit fighter?
15:56:40BlackBirdarg
15:56:47BlackBirdwhy did i put a space?
15:57:01raptoron the download page
15:57:08BlackBirdok
15:57:22BlackBird Quit (Client Quit)
16:00:35Zoomber has joined
16:00:36ChanServ sets mode +v Zoomber
16:11:45Watusimoto_>> Hostile load-out zones should drain energy no matter what (not just when not moving/shooting) <<
16:11:53raptorfixed
16:11:58Watusimoto_argh... what if you're in both zones simulatneously?
16:12:00Watusimoto_really?
16:12:18raptorundefined behavior
16:12:27Watusimoto_ha
16:12:34raptori'm learning from the c++ specs...
16:12:37Watusimoto_because that's lame?
16:13:08Watusimoto_just run with whatever isInZone returns?
16:13:19raptoryep
16:13:35raptorbecause... i don't consider that special case to be anything but, um... 'special'
16:14:09Watusimoto_I guess not -- we don;t really haev a way to retrieve mulitple zones in this case
16:15:44raptorworking on getting lion to work...
16:16:21raptorso now they make you download an Xcode DMG, then when you install the package, it puts an .app in /Applications called 'Install Xcode'
16:16:34raptorso you get to install again
16:16:45raptor8GB later you finally get Xcode installed
16:19:51Watusimoto_ah, does your loadout zone fix actually work?
16:19:56raptoryes
16:20:11raptori tested it on all types of zones (non-overlapping)
16:20:17Watusimoto_everything is wrapped in this condition:
16:20:18Watusimoto_if(!anyActive && mSpawnShield.getCurrent() == 0)
16:20:30raptoryes
16:20:31Watusimoto_if you are active, the hostile code never gets run, does it?
16:20:43raptoranyActive is misleading, i think
16:20:50raptoralso that is an &&
16:21:05Watusimoto_what does anyactive mean?
16:21:50Watusimoto_mean's a module's being used
16:22:02raptorwait wait - what did i do..
16:22:04Watusimoto_so !anyactive means a module's not being used
16:22:54Watusimoto_so if any module is active, !anyactive will be false and that blcok won't run
16:22:59raptorbasically you drain unless you use a module - in which case you still drain
16:23:23Watusimoto_why do you still drain?
16:23:35raptorbecaue a module is being used..
16:24:00raptorok, so
16:24:02Watusimoto_but loadout zone drain stops when you use a module
16:24:07raptoryes
16:24:10Watusimoto_so basically, that module becomse cheaper
16:24:26Watusimoto_because you'd be losing some of that energy anyway
16:24:40raptorit's the same as it always is - it just isn't accumulating with the hostile drain
16:24:56raptorso if you use cloak over hostile zone, you'll not drain as fast :)
16:25:13raptorso yes, that's a bug
16:25:16raptori'll fix...
16:25:24Watusimoto_right -- but if I,m losing 10% of my energy per second for doing nothing, and cloak takes 15%, I might as well cloak, since it really only coss me ... ok, you get it
16:25:31Watusimoto_I can fix it -- I'm in the right place
16:25:35raptorok
16:27:18Watusimoto_even hostile drain should be stopped when spawnshield is on?
16:27:34raptoryes - otherwise it isn't a spawn shield
16:27:46raptorright? it's supposed to be invulnerability
16:27:53Watusimoto_I agree with the conculsion, but not with the logic
16:28:01Watusimoto_but no matter
16:28:04raptorhehe
16:28:43raptori guess technically energy loss doesn't have anything to do with health..
16:40:31Watusimoto_do you get regualr recharge while spawnshield si on?
16:40:41raptori don't remember
16:40:44raptori don't think so
16:40:47Watusimoto_ok
16:41:26Watusimoto_when idle, do you get double normal?
16:41:38Watusimoto_oh, wait I see
16:42:27raptorno, like +33%
16:46:15Watusimoto_ok... what happens en enemy loadout zone?
16:46:19raptornothing
16:46:23raptorwe never decided
16:46:26Watusimoto_no recharge?
16:46:42Watusimoto_currently, I have no recharge, but this is more accidental than intentional
16:46:49raptorhostile -> negative; team/neutral -> positive
16:46:54raptorenemy -> nothing
16:48:08Watusimoto_no change, or no recharge?
16:48:15Watusimoto_no recharge
16:48:21raptorif you're in an enemy load-out zone?
16:48:22Watusimoto_energy just sits there
16:48:24Watusimoto_yes
16:48:31raptorjust like if you were not in a zone
16:48:35Watusimoto_no effect
16:48:38raptorcorrect
16:48:40Watusimoto_ok
16:54:24Watusimoto_thomas thinks there should be some penalty for blue ship in red zone
16:54:59Watusimoto_I kind of agree -- why should hostile zone be so punative (and that;s a high discharge rate we have), but an enemy zone has no bad effects at all?
16:55:04raptori don't know - people make levels with huge loadout zones
16:55:12Watusimoto_that's true
16:55:20Watusimoto_they use them for decoration
16:55:57Watusimoto_he thinks it would make the game more fun
16:56:31raptormaybe a slower recharge?
16:56:34Watusimoto_ok, well, I;ve basically got the case done. going to go work on dinner, will decide about the enemy zones. will be congnizant of not wanting to break levels
16:56:40Watusimoto_that;s what I'm thinking
16:56:45raptorinstead of 8000, maybe 4000
16:56:49Watusimoto_I'll code it up with a new constant
16:56:55raptorok
16:56:59Watusimoto_and we can tinker easily
16:57:02Watusimoto_later!
16:57:05raptorbye
17:39:09Watusimoto_ Quit (Ping timeout: 240 seconds)
17:58:39Watusimoto has joined
18:02:40karamazovapyhostile zones are punative because they have no other function of any kind
18:03:45karamazovapywhen I was initially thinking about this, I considered making enemy zones punitive, but I came to a similar conclusion about large or unavoidable enemy zones
18:05:26raptoryes
18:05:40raptorwhat about a reduction in positive recharge?
18:06:21raptori still think they should probably have no effect, but maybe a slight reduction in normal recharge might be OK
18:08:55sam686 has joined
18:08:56ChanServ sets mode +v sam686
18:12:28raptorthanks karamazovapy for clarifying my clarify... I guess I use too much jargon sometimes..
18:31:43sam686there is a problem with #define min(a,b) ((a) <= (b) ? (a) : (b)), if a is stream->readFloat(3), it may call readFloat twice..
18:37:25raptori'm thinking we shoudl remove every 'define min' in the code
18:37:34raptorbecause it exists in cstdlib
18:37:39raptoror cmath
18:37:43raptorthe c++ std libraries
18:49:49Watusimoto Quit (Ping timeout: 240 seconds)
19:35:38raptor Quit (Remote host closed the connection)
19:36:43raptor has joined
19:36:43ChanServ sets mode +o raptor
19:37:50Watusimoto has joined
19:51:46raptori have to redo my OS X 10.7 vm...
19:51:49raptorargh
19:52:11karamazovapyI say we just make all the mac users buy actual computers
19:52:43raptoryou mean instead of their false idols and golden images? never!
19:53:31raptori guess BlackBird and family just got a brand new shiny Mac
19:53:35karamazovapywith any luck, built-in mac limitations and censorship will move all apple acolytes to their own interweb
19:53:42raptorhahaha
19:53:47raptorthat's a great idea!
19:53:55karamazovapymaybe by 2015
19:57:08raptorWatusimoto: do you have a launchpad ID?
19:57:29raptorwe need to open a bug to get the ubuntu folks to upgrade their package
19:58:45raptoroh hey look, this site just pulls from our RSS feed: http://www.lgdb.org/aggregator/sources/32
20:01:59Watusimotothis has become totally ridiculous
20:02:08Watusimoto EnergyRechargeRateWhenInNoZoneWhenIdle = 8000,
20:02:08Watusimoto EnergyRechargeRateWhenInNoZoneWhenActive = 6000,
20:02:08Watusimoto EnergyRechargeRateInFriendlyLoadoutZoneWhenIdle = 12000,
20:02:08Watusimoto EnergyRechargeRateInFriendlyLoadoutZoneWhenActive = 6000,
20:02:08Watusimoto EnergyRechargeRateInEnemyLoadoutZoneWhenIdle = 12000,
20:02:09Watusimoto EnergyRechargeRateInEnemyLoadoutZoneWhenActive = 6000,
20:02:10Watusimoto EnergyRechargeRateInNeutralLoadoutZoneWhenIdle = 12000,
20:02:13Watusimoto EnergyRechargeRateInNeutralLoadoutZoneWhenActive = 6000,
20:02:15Watusimoto EnergyRechargeRateInHostileLoadoutZoneWhenIdle = -12000,
20:02:17Watusimoto EnergyRechargeRateInHostileLoadoutZoneWhenActive = -12000,
20:02:22raptorhahahahaha
20:02:33Watusimoto5 has... that;s a record!
20:03:07Watusimotohaving all those contants makes the code logic much clearer
20:03:15raptorok
20:03:58Watusimotoso yes, I think I have a launchpad id
20:05:25Watusimotoso which package are you referring to?
20:08:15raptorthey say if we want them to update a package (bitfighter to 016) we need to open a bug through launchpad
20:08:28raptor"If you want to request a new package/package upgrade please use the "Report a bug" function at launchpad."
20:08:35raptoron this page: http://www.getdeb.net/contact/
20:08:41Watusimotoreally?
20:10:21Watusimotodone... that's not very automated
20:13:03raptoryeah
20:13:23raptori wonder about linux mint..
20:13:41raptorwill ubuntu packages work fine with mint?
20:18:09Watusimotoyes
20:18:25Watusimotosent a message to playdeb as well. Submission form looked suspiciously familliar
20:18:37raptorlaunchpad is playdeb
20:18:43Watusimotooh
20:18:47Watusimotooh well
20:18:49raptor:)
20:18:56raptorstill wondering about linux mint..
20:19:04WatusimotoI thought that was all supposed to be automated
20:19:10Watusimotowonder away
20:19:19raptoryou still use mint, no?
20:21:27Watusimotoyes
20:21:40WatusimotoI installed the playdeb pkg for 015a
20:21:40raptordo normal ubuntu packages work seamlessly?
20:21:44Watusimotoyes
20:21:49raptorok, cool
20:22:05Watusimotoor at least they have so far!
20:24:53Watusimotoyou said you saw 4K downloads on google code... must have been a glitch
20:25:08raptori didn't say that...
20:25:11WatusimotoI;m seeing much more believable number s now
20:25:13Watusimotothought you did
20:25:15raptorwas that for 015a
20:25:16raptor?
20:25:30Watusimotooh
20:25:33Watusimotook then
20:27:25raptoryay, i can reverse the scroll wheel in Lion
20:29:43sam686there, made changes to CoreItem pack / unpackUpdate, so that individual panel can be updated instead of the whole panels.
20:30:03sam686also, i got it to work perfectly well with existing 016 servers
20:31:19Watusimotoso what do you think about the idea of core panels?
20:31:34sam686seems like a good idea, so far..
20:31:54BFLogBot - Commit 2056efd4c1d5 | Author: sam8641 | Log: Made big changes to CoreItem's pack / unpackUpdate, now works on existing 016. Better place to delete Game's GameType
20:32:57WatusimotoI don't like the current implementation of partially damaged panes; I've tried colors and thickensses, and I don;t like either
20:36:12WatusimotoI don't see how panel core and panel-less core can be compatible
20:36:52Watusimotodo you still need to shoot all the panels on the older version?
20:39:18sam686yes, as thats how the server have it
20:39:29raptor Quit (Ping timeout: 240 seconds)
20:40:11sam686new server and old clients can make a little confusing though as old clients don't have the ability to render panels..
20:41:01sam686new clients and old server, the new clients simply uses the old method of shrinking core on old servers
20:42:35raptor has joined
20:42:37ChanServ sets mode +o raptor
20:42:52Watusimotoit's very likely that the next release will not be compatible with 016, so I wouldn;t spend too much effort keeping them so
20:43:42raptori hereby declare next release be 017
20:43:50Watusimotofor one thing, the only way to trigger the alert messages on mac and linux is to change sc protocol versions, which automatically makees things incompatible
20:44:05Watusimotoand with that memory leak fix, we want everyone to upgrade
20:45:18sam686see this? TNL_IMPLEMENT_RPC(..... , 0) you can add more RPC by changing it to 1 or 2, just like master way,
20:45:34sam686older server / clients will ignore unsupported RPC
20:46:27sam686I have actually added one, with RPC 1, in gameConnection, for s2rVoiceChatEnable
20:52:45sam686good news, because a way to show an upgrade message to older clietns is through m2cSendUpdgradeStatus(bool), the master side can be adjusted to return true for outdated 016, and false on new 016 clients
20:53:49sam686the master can see mClientBuild from all clients, and determine if it is outdated 016
21:04:15Watusimotoyes; we need to do that, but the fact is the game becomes unplayable if everyone is working with different rules, contant values, objects, etc.
21:04:30Watusimotoand the code becomes difficult to manage as well
21:04:48Watusimotowe have enough trouble finding the bugs without worrying about mixed versions
21:05:15raptorwe can break compatibility as much as we want - we're still not at 1.0 :)
21:05:19Watusimotobesides -- our users are usually very happy to upgrade
21:06:50Watusimotook; is anyone working on the number keypad bug? if not, I'll take care ofit
21:08:34BlackBird has joined
21:08:37BlackBirdhi
21:08:51BlackBirdHOW THE HECK DO I USE INKSCAPE?!?!?!?
21:08:52Watusimotohi
21:09:48Watusimotohttp://www.google.lu/url?sa=t&rct=j&q=inkscape+tutorial&source=web&cd=4&ved=0CEQQFjAD&url=http%3A%2F%2Finkscape.org%2Fdoc%2Fbasic%2Ftutorial-basic.html&ei=EPAuT7v5KcKe8gP6vv2TCQ&usg=AFQjCNEhtJPrp0vLNvjRIzjsXYHVEtoT3A
21:10:18BlackBirdugh
21:10:26BlackBirdI'm trying to tracy an image
21:10:30BlackBird*trace
21:11:59BFLogBot - Commit 8c5b9a70b021 | Author: watusim...@bitfighter.org | Log: Comment
21:12:01BlackBirdhello?
21:12:01BFLogBot - Commit e2c9e763005c | Author: watusim...@bitfighter.org | Log: Merge
21:12:02BFLogBot - Commit d9e08b58f5a2 | Author: watusim...@bitfighter.org | Log: it's simple boolean logic, people!
21:12:04BFLogBot - Commit 6459a772ec2e | Author: watusim...@bitfighter.org | Log: Clarify logic for energy recharge in loadout zones of various flavors
21:12:05BFLogBot - Commit a9769aaa5e7a | Author: watusim...@bitfighter.org | Log: Merge
21:12:08raptorBlackBird: look at second comment here: http://www.inkscapeforum.com/viewtopic.php?t=584&f=22
21:12:39raptorthere's a tracing thing built in - i've used it once... it's ok
21:12:47raptorbut i don't know the program well
21:14:15raptorawww, Watusimoto you reduced my karnaugh map optimizations... :)
21:14:40Watusimoto???
21:15:02raptorremember karnaugh maps for logic simplification?
21:15:20BlackBirdi don't think that was much help
21:15:23Watusimotono; did you mention something about this a week of so ago?
21:15:53raptorBlackBird: it points you in the right direction, you may have to figure out the inkscape settings yourself
21:16:08BlackBirdugh
21:16:19raptorhttps://en.wikipedia.org/wiki/Karnaugh_map
21:16:26raptorjust something i remember from engineering classes
21:16:59raptoruse the minumum amount of logic statements to get the job done - you've done the opposite with all your new Energy constants :)
21:21:12WatusimotoClarity is more imporatnt than minimization
21:21:41raptornowadays, i believe so, too
21:22:08BFLogBot - Commit 2b5933bd5e44 | Author: sam8641 | Log: added build_version as master.cfg option to tell older build clients to show upgrade message
21:23:27WatusimotoBesides, now changing the rules is trivial; no need to alter (and therefore test) the logic
21:25:24raptor(i was just poking fun...)
21:25:41WatusimotoI know
21:26:25Watusimotothe code could be optimized to look for the most common cases first; actually it does; most common case (only common case) is player is not on zone
21:34:33koda Quit (Quit: koda)
21:41:00Watusimotoflomo is complaining that cmd-q does not quit on tiger, nor does cmd-h (supposed to hide game?)
21:41:08Watusimotoare those on our list?
21:41:09raptoryes
21:41:15Watusimotook
21:41:20raptorthey work perfectly fine in my VMs :)
21:41:28raptori guess real hardware is needed...
21:41:30Watusimotoexcellent!
21:41:48Watusimotodo we have independent confirmation?
21:41:57Watusimotoie other than flomo?
21:42:01raptoryes
21:42:08raptorBB/LA confirmed
21:42:16raptorand I did on some old ppc hardware at work
21:42:24raptori think it's the mac keyboards
21:42:27Watusimotostupid mac
21:43:06sam686so what are the options, now that the CoreItem have changes? 1. change CS_protocol... 2. keep as it is (older client have trouble with new server's CoreItem) 3. auto disconnect outdated clients when ever a nwe CoreItem is needed
21:43:50Watusimoto1
21:43:54raptorsam686: I think we should break compatibility and go with 017
21:44:16sam686ok..
21:44:27Watusimotopeople are really excited about upgrading
21:44:35Watusimotothey'll all do it ASAP
21:44:48Watusimotoexcept for that one computer lab somewhere that is still running 012 or somesuch
21:44:59Watusimotobut they're hopeless
21:45:18Watusimotothe only flaw with my argument is if the ubuntu builds continue to lag
21:45:55raptoreh
21:46:03raptor1% of users are linux players anyways, right?
21:46:23Watusimotoprobably
21:46:29Watusimotook, here's my dilemma
21:47:11sam686not sure, as i dont think we keep track of who is on linux / mac / windows, as no operating system name is sent to master..
21:47:12Watusimotoas discussed, I'm going to automatically create synonms -- if a user maps the 4 key to something, keypad_4 will also work, unless mapped to somthing different
21:47:38Watusimoto[[[maybe we should send os to server? ]]]
21:47:43raptori'm going by pure # of downloads from google code
21:48:21Watusimotoso the question is if the user has mapped keys 1,2,3 to soemthing, and numeric_keypad 3 to something else, shoudl numeric_keypad 1 & 2 be synonyms for regular 1 & 2?
21:48:37raptorno
21:48:39raptorbut
21:48:44Watusimotoor should the presence of one synonym conflict be enough to disable the whole thing
21:48:59raptorwe only keep track of certain bindings
21:49:13Watusimotomost bindings, but true
21:49:46raptorcan you just create a method that searches the bindings for the specific numpad keys, if they are in use, don't allow them to duplicate the originals
21:50:03raptorotherwise duplicate
21:50:34Watusimotoso, in my example, if keypad_5 were bound, you'd disable keypad synonyms?
21:51:04raptoryes
21:51:17raptorsynonym - that's a good work, i will use it (if i can spell it)
21:51:32WatusimotoI think I agree...
21:51:46Watusimotofor the feature to be enabled, no synonyms in a class can be bound
21:51:53raptorcorrect
21:52:12raptorhowever
21:52:15Watusimoto(and I can only think of one class, nums, +-*/enter) to keypad
21:52:29raptorexcept for us, numbers are only 1-9
21:52:33raptorwe don't use 0
21:52:43Watusimotowhat do you mean?
21:52:43raptori know Footloose like to map 0 to cmap
21:53:06Watusimotowhy would 0 be somehow privleged?
21:53:10raptoror for module select we only use 1-3
21:53:53raptoror for arrows we use the even numbers
21:53:54Watusimotobut if I bind 0 to cmdrs map, then it should work on keypad or num key
21:54:17raptoryes, but what if they bind the numpad number instead?
21:54:27Watusimotothen synonyms get disabled
21:54:38Watusimotowell, if they bind numpad # and a regular #
21:54:39raptoryes (i'm just playing devil's advocate)
21:54:50Watusimotosure, but that was exactly the case I posed above
21:55:09raptorso we have by default bound 1, 2, 3 to module select
21:55:18Watusimotoyes
21:55:37raptoractually, never mind - just do it and we can see what happens :)
21:55:44Watusimotook
21:56:04raptorbecause it's a step forward
21:56:50Watusimotoor
21:57:03Watusimotowe could scrap this idea now and let you bind two keys in the keybinding screen
21:57:38Max has joined
21:57:55raptorhello Max
21:57:57MaxAnyone know how to use this on mac os lion?
21:58:08Max is now known as Guest54078
21:58:19raptorhi Max - I'm currently in the process of testing what the problem may be....
21:58:23sam686what is "this"?
21:58:35Guest54078oh sorry bitfighter, to play the games
21:58:39raptorwe don't really have Mac hardware on our development team
21:58:47BlackBird Quit (Quit: Page closed)
21:59:13Guest54078you have downloads for OS X, is that not the same?
21:59:34sam686zoomber might know, but haven't seen zoomber in bitfighter in a long time now..
21:59:42raptoryes - well... OK, the problem is we don't have Lion set-up yet... :)
22:00:02raptorI had access to a Snow leopard machine to make the builds
22:00:12Guest54078Ahh ok. So it's not really possible to play without serious lag?
22:00:31raptoroh - actually, it runs fine on Lion, at least the client part
22:00:44raptoryour lag problem is probably your internet connection
22:00:54raptorbut i don't know if hosting games will work
22:01:22raptoryeah, BlackBird just confirmed he's playing with no problems on Lion
22:01:32Guest54078Hmm ok.
22:02:12raptorlots of things cause lag: playing over wireless with many other people connected; downloading too much
22:03:14Guest54078OK thanks for the help!
22:03:21Guest54078 Quit (Quit: Page closed)
22:03:32raptorha
22:05:06max678 has joined
22:05:14max678hello
22:05:33raptorhi
22:07:14BFLogBot - Commit 1da46ff435eb | Author: sam8641 | Log: changed CS_PROTOCOL_VERSION, and some clean up
22:08:52LordDVG Quit (Remote host closed the connection)
22:09:27max678 Quit (Client Quit)
22:17:05raptorit's amazing how an OS can only further limit your control and then require double the RAM to run...
22:34:02raptorso Mac Lion hosting does work
22:34:06raptorbut bots are not found
22:34:21raptoractually scripts are not found, bots are
22:34:27raptorlooks like a pathing issue
22:36:22sam686the new cores have a rotating problem, especially with /settime 0
22:36:34raptoronly with settime 0?
22:36:47sam686and, the center part goes backwards
22:37:53raptori haven't seen them yet...
22:39:34raptori keep getting this annoying assert when testing any level in editor...
22:40:48sam686what is "this"?
22:40:56Watusimoto@sam -- I figured that would be an issue, should be a straightforward fix
22:41:02Watusimoto@raptor -- what is the assert?
22:41:06Watusimotoone of ours?
22:41:40raptor#3 0x000000000070f102 in TNL::Assert::processAssert (filename=0x7c1bdd "gameConnection.cpp", lineNumber=57,
22:41:42raptor message=0x7c1bb0 "Client should not be using this constructor!") at assert.cpp:68
22:41:57raptori get it everytime i load up a level in the editor and test it
22:42:14raptorhowever, if I host a game first, then it works ok in the editor
22:42:32raptorand i don't have a nice 'continue' pop-up like you windows folks do
22:43:15sam686gameConnection.cpp line 57? i guess i mistakenly changed gServerGame into mServerGame
22:43:44sam686you can remove that assert
22:43:47raptorok
22:44:39sam686i guess the reason it doesn't assert for me is, mServerGame points to 0xCCCCCCCC, which won't assert as it is not zero
22:44:52raptorok removed
22:45:52Watusimotowhy would hosting first fix the problem?
22:46:03raptorbecause we have goofy code
22:46:52Watusimotothat's not an answer I like to hear
22:46:59raptor:)
22:47:19BFLogBot - Commit 861abd6fc0ed | Author: buckyballreaction | Log: Remove annoying assert
22:50:23sam686the way mServerGame is gotten is in GameConnection::readConnectRequest
22:52:56raptori guess i was expecting Core wall width to be much wider
22:53:00raptorwider than 2px
22:53:03raptorat least
22:54:17Watusimotomaybe we could design a thicker wall; just making the stroke thick looks awful
22:56:15raptorif the difference is only 1-2 px, then there's not much reason to have thickness
22:56:33raptormaybe color? slowly change the alpha?
22:58:22raptorxcode 4 looks like the HUD of a flight sim
22:58:36Watusimototried that
22:58:45Watusimotounhappy with result
22:58:58raptorok
22:59:12Watusimotowe could perahps construct a hollow rectabgle as panel, and make that thinner
22:59:21Watusimotorather thicker
23:01:58raptorhmm..
23:02:54raptori'm taking a break for a while - i think mac has hurt my head
23:03:01raptor(for today)
23:03:09raptor Quit (Remote host closed the connection)
23:47:59karamazovapy...so are we talking about skipping 016a and going straight to 017, or are core changes just going to wait a version?

Index Search ←Prev date Next date→

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