#bitfighter IRC Log

Index Search ←Prev date Next date→

IRC Log for 2012-11-22

Timestamps are in GMT/BST.

01:06:28Wuzzy Quit (Quit: Wuzzy)
01:20:58BFLogBot Commit: c9ad108e7962 | Author: buckyballreaction | Message: Update FSF addresses
01:22:42raptorjust made a test RPM of bitfighter 018 - 5.8 MB
02:10:44koda Quit (Quit: koda)
02:41:57fordcars has joined
02:45:45fordcarshello
02:46:36sam686hi
02:47:17fordcarshi
02:47:45sam686hi
02:47:50fordcarshi
02:47:52sam686hi
02:47:55fordcarshi
02:47:58sam686hi
02:48:01fordcarshi
02:48:03sam686hi
02:48:07fordcarshi
02:48:08sam686hi
02:48:10fordcarshi
02:48:11sam686hi
02:48:12fordcarshi
02:48:22sam686hello
02:48:29fordcarshola
02:48:45sam686happy Wednesday night
02:49:19sam686happy bitfighter
02:49:32fordcarseuh happy 10th hour of the 21 of novemeber
02:49:47fordcarsor the 22nd hour
02:50:15sam686umm, time zone problem, my clock is 8:50 PM
02:50:58fordcarswell happy 9th hour of the 21 of november then
02:51:39sam686hint, /ctcp sam686 time that will tell you my clock..
02:52:12fordcarshaha ok
02:55:14raptorhi sam686
02:55:19sam686hi
02:55:37sam686missed hundreds of "hi"?
02:56:00raptori see them
02:56:14raptorhow have you been?
02:56:20sam686good
02:58:06sam686I seem to really been playing Marathon Aleph One first person shooting games in a past 2 weeks.. those single player games are really long..
02:58:20raptoryeah they are
02:58:35raptorand games are designed now to keep you hooked for a long time
03:01:11raptoroh wow, that's an open source game??
03:01:37sam686i guess, i was just looking for another first person shooter game..
03:02:24fordcarsever tried a 2nd person shooter game?
03:02:38fordcarsyou are the one getting shot lol
03:02:58sam686AlephOne Marathon may be an online game but mostly there is hardly no one online. Also with my own testing, I realizes it only transmit the controls, relying on making both client in sync to each other..
03:08:13sam686whats kindof funny is, when both clients disconnect from each other, they become 2 seperate gameplay of playing by themself instead of dropping to main menu.
03:08:39fordcarshaha
03:21:08fordcars Quit (Ping timeout: 245 seconds)
03:22:46raptorsam686: did you see that editor crash?
03:24:58raptorwatusimoto said to do the following: <Watusimoto> I create a u shaped wall, add a repair, resize, undo redo undo crash
03:25:13sam686Taking time recompiling newest code (after a week of non-bitfighter)
03:25:33raptori've only ever got it to crash once on my linux system, but he says it does it every time in windows
03:59:27sam686it always crashes when creating multiple point barrier, adding repairable, moving repairable, repeatedly undo / redo
03:59:43raptoryes
03:59:45sam686some memory corruption it appears
03:59:59raptorsomething is not using a smart pointer?
04:01:27sam686I tried converting every Shared_Ptr into RefPtr, and getting a different problem..
04:02:15raptorwatusimoto added some comments about the crash: https://code.google.com/p/bitfighter/source/detail?r=b0f2608fd43db4ffdcee5dd5c1675fa16c178412
05:37:42kodaws has joined
05:44:12kodaws Quit (Ping timeout: 244 seconds)
05:47:36CrazyLinuxNerd Quit (Quit: Leaving)
05:47:58CrazyLinuxNerd has joined
06:35:38BFLogBot Commit: c6caf6aee6dd | Author: sam8641 | Message: Appears to fix editor undo/redo crash.
06:38:43sam686raptor fell asleep? I am falling asleep!
06:38:48raptorhi
06:38:52raptoryou fix?
06:38:59raptorinteresting - how did you find that?
06:39:19sam686looks like it fix the problem...
06:40:00sam686somehow trying to run a destructor, in which the destructor removed from database in a process of removing object from database may be a cause for corruption
06:40:43raptorweird
06:40:51raptorwatusimoto will be glad...
06:41:00sam686the problem started with c000c91e9e3b
06:41:27raptorso you did a bisect?
06:41:59sam686kindof did (although not with the bysect plugin if that exist)
06:42:15sam686more like using notepad and taking notes on which revision works or not..
06:42:20raptoroh wow
06:42:27raptoryeah, there is 'hg bisect'
06:43:10sam686the easy GUI hg is really easy to switch revision to somewhere between 2 revision
06:43:57raptorok
06:44:47sam686looks like, there isn't a GUI version of bysect, is there?
06:45:03raptori'm not sure
06:45:06raptori've never used the gui
06:46:33sam686I don't see GUI version of bisect, i guess I am fine with turtoiseHG and quickly switching revision that way to guess and narrow down which one may be the problem..
07:04:43raptorwell i think we're really close to releasing
07:20:30raptorthe windows updater probably needs to be tested..
07:37:38amgine1234567890 has joined
07:37:48amgine1234567890checking up real fast
07:38:57raptorsam686 thinks he fixed the crash
07:39:11raptorso hopefully watusimoto shows up this weekend and we can release!
07:39:17raptorbut i'm going to bed
07:39:19raptorgood night!
07:41:49raptor Quit ()
07:54:30amgine1234567890one minor thing that maybe could be added that option al is putting me in the credits =)
07:59:08amgine1234567890but if i havent earned it yet ill help find bugs in as many verisons as it takes
08:08:27kodaws has joined
08:55:41amgine1234567890 Quit (Quit: Page closed)
14:49:47raptor has joined
14:49:49ChanServ sets mode +o raptor
14:56:25raptorgood morning!
15:14:22raptorkodaws: hello
15:14:49raptorI find that mac seems to send a command line parameter like -psn_0_....
15:14:56kodawshello raptor
15:15:13raptoram I supposed to do something with that? I found out that it's a 'Process serial number'
15:15:21kodawswell
15:15:32kodawsi guess it is something for cocoa libs
15:15:35kodawsto be used privately
15:16:20raptorhmm, ok
15:16:33raptorit is only added when launching from the .app
15:16:56raptorand we currently exit the game if we find unknown command line parameters..
15:17:12raptorso i should probably code in a case to ignore that one..
15:21:06kodawsyeah
15:21:22kodawshaven't noticed that on hedgewars
15:38:33BFLogBot Commit: 116ba14fb2ce | Author: buckyballreaction | Message: Ignore the process serial number command line option on OSX
15:38:39raptorwell after that change, mac works great!
16:54:25watusimoto has joined
16:54:28ChanServ sets mode +o watusimoto
16:54:36watusimotohello
16:54:42raptorhello!
16:55:26watusimotohow are things going?
16:55:29raptorgood
16:55:44raptori'm refining all the build systems now (on Mac...)
16:56:38watusimotowhat do you think? does it feel "right"?
16:57:04raptorthe release?
16:57:15watusimotono, your new code
16:57:20raptoroh
16:57:26raptoryes, it's actually great
16:57:28raptori think
16:57:29watusimotothe installer stuff
16:57:43raptoralthough I'm still not sure how to organize it...
16:58:09raptorwait, you mean my new in-code resource copying, etc?
16:59:20watusimotoyes
16:59:46raptoryeah, it's pretty well done - i still need to remove some debugging statements
17:00:05raptorbut i wouldn't mind more testing, of course
17:00:50watusimotoI saw the commit logs, and it looks like sam686 may have fixed the bug that was kicking my butt before my trip
17:00:59watusimotoI need to check that tonight, and see how he did it
17:01:08raptoryes - i think you should double check it
17:01:13raptori could never reliably duplicate
17:01:17watusimotoI really hope it was the same bug
17:06:22BFLogBot Commit: 7c417e387c24 | Author: buckyballreaction | Message: Point Sparkle to the proper feed URL
17:06:24BFLogBot Commit: 0042160ebbcc | Author: buckyballreaction | Message: Automatically update the Info.plist file with the proper versions when building the bundle
17:06:51raptorgood, mac is really good - except i need to figure out how to serve up the proper download by architecture
17:08:22watusimotoso what happens on windows?
17:08:37raptorso basically, it:
17:08:56raptor1. detects if standalone (right now this is true only for a debug build)
17:09:22raptor2. checks to see if the user data dir exists, if not, then we know it is a 'first run'
17:09:47raptor3. if first launch, create user data dir and copy resources
17:10:20raptor4. set up rootdatadir and sfxdir accordingly
17:10:38raptoroh: 0. change working directory to executable path
17:11:54raptorso you now don't need to force the -rootdatadir option when debugging :)
17:12:40watusimotook, a couple of questions
17:12:52watusimotoin step 1, how do you know?
17:13:09watusimotoa debug run is considered standalone
17:13:27raptorthe logic is basically: 1. if TNL_DEBUG, then standalone; else not standalone
17:13:43raptorbut i have a plan to expand it
17:13:46watusimotook. so we could add some additional mecahnism later for creating
17:13:48watusimotoyes, ok
17:13:49raptoryes
17:14:06raptorso from now on, release build are considered non-standalone
17:14:11watusimotoactually... there may be a better way
17:14:19raptori'm open to better ways
17:14:25watusimotowhen we install via the installer, a registry key gets written
17:14:28watusimotowe could check for that
17:14:34raptoruh
17:14:35watusimotoif absent, it's a portable install
17:14:37raptorwait
17:14:42raptorwe already use a registry key?
17:14:48watusimotoin windows, yes
17:14:58watusimotojust one, to make uninstaller work
17:15:09raptori was just thinking that if the INI was detected in the executable folder, then we'd turn standalone on
17:15:13watusimotono registry key == no uninstaller
17:15:54watusimotowhat happens the first time you run if you are running a portable install?
17:15:57watusimotothere will be no ini
17:16:30raptoryou mean if the INI is already present? hmm...
17:16:34raptormaybe that wouldn't work
17:17:06raptorif it detects standalone right now, it will just build out the default INI
17:17:46raptori'm thinking that standalone would be available on linux or windows
17:17:52watusimotoif you download a portable copy, and run it, there will be no INI, and you'll get your resources copied somewhere you don't want
17:17:55raptorlinux = no registry
17:17:59watusimotoright
17:18:13watusimotothis is clearly a solution for the win version only
17:18:26watusimotoin linux it seems less of a problem
17:18:32watusimotoyou don't really have portable installs
17:18:40raptortrue
17:18:45watusimotoeven programs you "install" by copying to hur home folder write stuff in .settings
17:18:56watusimotohur-your
17:18:56raptortrue true
17:19:02raptorok, so i guess windows-only
17:19:04watusimotobut windows is a bit different
17:19:08raptorso we can get on more school computers...
17:19:16watusimotoportable should be leave no trace
17:19:41watusimotook, so in windows, if the reg key is missing, we'll consider it a portable standalone copy
17:19:49raptorthe registry idea has a flaw...
17:19:55watusimotoif it's present, its a real copy
17:19:55watusimotoyes
17:20:05raptorif there is already a copy installed to the system, portable won't work
17:20:12watusimototrue
17:20:25raptorthat's why i was thinking of having some detection of a file alongside the executable
17:20:29watusimotoa second possibility would be to package portable with a .portable file
17:20:34watusimotoand we could check for that
17:20:36raptoryes that
17:20:57watusimotothat would also work
17:21:04watusimotobut would require duplicate packaging
17:21:10watusimotobut maybe that's ok
17:21:24watusimotomaybe the nsi script can produce portable and installer versions
17:21:32watusimotook, next question
17:22:01raptorok
17:22:33watusimotoRight now when I'm debugging, I build to an exe folder
17:22:43watusimotoI guess all my resources will just live in there for
17:22:43raptoryep
17:23:10watusimotoand I'll just get rid of the bitfighter_debug folder where I keep them now
17:23:21watusimotoso that should probably work
17:23:33raptorwell, you can still specify a different rootdatadir, and it'll override the defaults
17:23:52watusimotook
17:23:55raptorbut
17:23:56watusimotomakes sense
17:24:14raptorthat's funny - i've always just worked out of that exe folder on windows, and i think sam686 does too
17:24:23raptori didn't know you had a completely different setup
17:24:34watusimotonow you do!
17:24:54watusimotoso where will the windows rootdatadir for a user be?
17:24:57watusimotoin appdata?
17:25:05raptoryes
17:25:13raptorso that will be different for people now
17:25:21watusimotoappdata/roaming/bitfighter
17:25:25raptori was thinking we may need to do some migration stuff
17:25:27watusimotoor something like that
17:25:27raptoryes
17:25:37watusimotowe probably will have to do some migration
17:25:41raptori figure it'd be good to make things standard...
17:25:46watusimotoyes, probably it is
17:39:53raptorall packages include music appropriately..
17:40:50raptoroh rats, credits music is broken for some reason...
17:43:36raptorremoved done bugs from the running list..
18:13:53kodaws Quit (Ping timeout: 244 seconds)
19:18:06raptor Quit ()
19:18:22raptor has joined
19:18:22ChanServ sets mode +o raptor
19:39:57LordDVG has joined
20:14:50BFLogBot Commit: 9ae508b18fdf | Author: buckyballreaction | Message: Send profile information to the Sparkle appcast URL. Why doesn't this send info every time an update is called?
20:14:53raptori'm done banging my head on mac for now
20:16:55raptori th ink i need koda's help..
20:53:05koda has joined
20:57:55raptorkoda! just the person I've been wanting to see...
20:58:13kodai feel summoned!
20:58:23raptortwo questions about Sparkle
20:58:29koda throws in some gas bombs
20:59:07raptor1. my latest commit sends profile information, but only seemed to do it once... do you know why?
20:59:16raptor suffocates
20:59:33kodadesign decision
21:00:01kodai order not to skew the stats from clients that connects too often
21:00:06koda-s
21:00:51raptorwell, I just need to know which architecture connects so i can serve up the correct download link in the appcase
21:00:54raptor*appcast
21:01:08kodaumh
21:01:16kodano universal binary?
21:01:40raptorwe have a 32bit bundle and a 64bit one
21:01:49raptorso one for i386/ppc, the other just for s86_64
21:01:55raptor*x86_64
21:02:20kodaah just like hedge :)
21:02:30koda2 ways
21:02:33raptorthe 32bit one is compiled for 10.4
21:02:34kodato solve this
21:02:40raptorand the 64bit for 10l6+
21:02:42koda1. compile sparkle yourself
21:02:43raptor10.6
21:02:56kodain this way you'll get access to <maxSystemVersion> tag
21:03:13kodathat will put a lower limit to the prompt
21:03:34koda2. use a different appcast
21:03:44kodadiscriminating legacy and 64bit
21:03:53kodathe latter is done in hedgewars
21:04:08raptori have a php file that reads the GET parameters and serves up the appcast dynamically
21:04:22raptorcould you show me where #2 is for hedgewars?
21:04:36kodait's in the info.plist
21:04:42kodasuurl iirc
21:04:55kodai manually set it different at release time
21:05:05kodahm might want to automate it...
21:05:12raptorheh
21:05:13koda puts it in his todo list
21:06:19raptori found out how to auto update the version numbers (i recently made this commit): https://code.google.com/p/bitfighter/source/detail?r=0042160ebbccf46b0ab5387f8de683f100811822
21:06:43raptor(just as an FYI)
21:06:54kodaoh nice
21:06:55raptorwhere are the hedgewars' appcasts?
21:07:21kodahttp://www.hedgewars.org/download/appcast.xml & http://www.hedgewars.org/download/appcast_legacy.xml
21:07:54raptorhmmm
21:09:04kodathe solution n1 would be cleaner i guess
21:09:34raptorso for #1 - is there some sort of platform info delay?
21:09:51kodadelay?
21:09:56kodathere is just a check
21:10:19kodathat if your os is between minSystemVersion and maxSystemVersion one xml item is displayed
21:10:24kodaotherwise it is hidden
21:10:37raptorsorry,i mean for the profile info sending
21:10:41raptori made this change: https://code.google.com/p/bitfighter/source/detail?r=9ae508b18fdf7de1d31955efdb5f4f69c0b5e119
21:11:01raptorit only sent it once when i launched the application, then stopped sending it
21:11:05kodayes
21:11:11kodait sends 1 once a week
21:11:24raptorand i'd have to compile sparkle myself to change that?
21:11:27kodahttps://github.com/andymatuschak/Sparkle/wiki/System-Profiling
21:11:35kodaNote: In order to standardize the statistics across a userbase with varying update check intervals, Sparkle submits profiling information only once per week.
21:11:42raptorargh
21:11:53raptori read that whole page...
21:12:42raptorso
21:12:54raptormaybe setting SUURL is what i need
21:13:20raptoris there a reason you set that in the Directory.mm file instead of the plist?
21:13:33raptor(can it be done in both places?)
21:13:45kodain code will override the one in the plist
21:14:00kodai prefer consistency so i put everything in plist
21:14:35raptorOK, maybe it's easiest to just auto-change the URL to have a parameter like 'legacy=true'
21:15:35raptorexcept, you coded in that URL? but you prefer plist?
21:21:09kodawell
21:21:18kodait's a nice solution as well
21:21:40kodathe question remains only for how long you want to support legacy
21:23:16raptoruntil mac moves to ARM
21:23:53koda(ios?) :p
21:24:17kodaby the way! if release is close i can start destroying the ui \o/
21:24:41raptorheh
21:24:43raptorwe're really, really close (for real this time!)
21:24:53kodaliiiiies
21:24:58raptortha't why i'm doing all this platform deployment stuff...
21:25:15raptorbut, you never know when watusimoto decides to pull a refactor.. :)
21:25:27kodawhat happened to the gci participation?
21:26:24raptori'm not sure - he was talking with some folks on IRC i last heard
21:26:36raptor(watusimoto, that is..)
21:35:02raptorwhen is the deadline for entries?
21:36:49raptoroh yikes it starts monday
21:46:03kodano deadlines
21:46:08kodabut ends in mid jan :p
21:49:22LordDVG Quit (Remote host closed the connection)
22:01:49Watusimoto_ has joined
22:05:01Watusimoto_hi
22:09:16raptorhi
22:13:09Watusimoto_initial tests suggest sam686 fixed the memory error I was seeing... but now I'm seing a different memory corruption error in the editor. I think I'd better do a full rebuild before pressing on
22:13:42raptorfull rebuild!
22:31:22Watusimoto_sigh... still evident
22:31:34raptoraww man
22:31:39raptorwell that stinks
22:33:03Watusimoto_indeed
22:34:58raptordid you see sam686's notes on the matter (in IRC)?
22:35:16Watusimoto_no
22:35:38Watusimoto_I think his fix is wrong, but reverting it makes things crash even sooner
22:35:56Watusimoto_do you know when he made notes on the issue?
22:35:59raptorhe basically did a bisect, but said that it started with this rev: c000c91e9e3b
22:36:12Watusimoto_that was smart
22:36:38raptorhe also said the following: [6:40:00] <sam686> somehow trying to run a destructor, in which the destructor removed from database in a process of removing object from database may be a cause for corruption
22:36:59Watusimoto_yes, maybe. we agree on what's happening
22:37:16Watusimoto_I think it is a problem with a double delete
22:37:23Watusimoto_but i really don't see how it cuold be happening
22:37:41Watusimoto_unless that's a symptom of memory corruption elsewhere
22:37:46raptorvalgrind?
22:38:26Watusimoto_yes, but if you can;t reproduce, then that tells me it may be hard to get something to happen thre
22:38:38Watusimoto_let me look at that version
22:38:44raptoryeah, i can't reproduce...
22:38:53Watusimoto_because I can, reliably
22:39:04raptori was suggesting you do valgrind, but i admittedly don't know how on windows...
22:39:20Watusimoto_moreso whn I undo sam's fix, but still pretty reliably with his fix
22:40:47Watusimoto_any idea how I can translate c000c91e9e3b into a revision number?
22:41:34raptoryes
22:42:19raptor5694
22:42:52Watusimoto_Fix memory leak in editor with ctrl-R functionality
22:42:54Watusimoto_Hmmmm
22:43:02Watusimoto_sounds suspicious
22:43:11Watusimoto_so how did you do the translation
22:43:20Watusimoto_I was looking on google code, and couldn't find it there
22:43:52Watusimoto_found it there
22:44:15raptori changed revisions real fast, then typed 'hg sum'
22:44:19raptorand it gives both revisions
22:44:37raptorhg up <hash_rev>; hg sum
22:44:44Watusimoto_ah, interesting
22:44:52raptorsum = summary
22:45:03Watusimoto_ok, so he said that that revision introduced the problem. that was the first revision that had the crash.
22:45:16raptoryes
22:45:49Watusimoto_I remember this edit
22:45:52raptoryou can see our conversation here: http://bitfighter.org/irclogs/index.php?date=2012-11-22
22:45:56raptorstarting at 3:22
22:47:06raptoroh hey - that was last night!
22:47:39Watusimoto_ha
22:48:06Watusimoto_there is a gui bisect tool in tortoise
22:48:11Watusimoto_btw
22:48:23Watusimoto_interestingly, we use git at work and I only use cmd line for that
22:48:47Watusimoto_we also use svn, and I've started ont eh cmd line there too, even though at my last job I totally eschewed it
22:49:00raptorhooray!
22:49:15Watusimoto_I prefer the gui though -- I think it is faster
22:49:30Watusimoto_especially if you are picking files to include in a commit
22:51:15raptori would agree
22:51:50raptorbut sadly my platform of choice has few (adhered-to) standards
23:14:40BFLogBot Commit: b5d6c7656737 | Author: buckyballreaction | Message: Comment out some debug messages. Also disallow standalone on mobile platforms
23:30:25raptorwell, it's almost dinner time!
23:34:48kodawhat's standalone?
23:36:42koda Quit (Quit: you can't say 'hello' without saying 'hell')
23:43:21raptorkoda:... is gone

Index Search ←Prev date Next date→

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