01:42:55 -!- Zaba has quit [Ping timeout: 276 seconds] 01:50:32 -!- Zaba has joined ##crawl-dev 01:59:30 -!- stabwound has quit [Ping timeout: 276 seconds] 02:02:40 Good job on the release candidate, doy :) 02:03:23 -!- stabwound has joined ##crawl-dev 02:10:51 -!- ortoslon has joined ##crawl-dev 02:35:53 -!- ogaz has quit [Remote host closed the connection] 02:40:16 -!- ortoslon has quit [Read error: Connection reset by peer] 02:51:45 -!- Spads has joined ##crawl-dev 03:05:50 -!- hotsun has joined ##crawl-dev 03:22:15 -!- hotsun has quit [Quit: HydraIRC -> http://www.hydrairc.com <- Wibbly Wobbly IRC] 03:47:20 -!- purge has joined ##crawl-dev 03:48:13 greensnark? 03:48:16 Are you in? 04:01:34 Morning Napking :) 04:01:46 !coffee 04:01:47 * Henzell hands greensnark a cup of black coffee, brewed by Crazy Yiuf. 04:01:53 Hey man :) how's it going? 04:01:55 !coffee 04:01:55 * Henzell hands Napkin a cup of irish coffee, brewed by Snorg. 04:02:00 \o/ 04:02:08 You like Irish too? :) 04:02:24 fits great for today :D 04:02:41 Finally found the problem why my new linux desktop kept crashing.. 04:03:00 cdrom on ide-controller! modprobe netconsole helped me read the call trace :) 04:03:26 You still use drives on IDE? 04:03:42 Or do SATA drives show up in ide-controller, no idea :) 04:03:45 My hardware fu is low 04:03:46 cdrom, yes 04:04:02 no, cdrom on ide, disks on sata 04:04:06 Strange, all DVD drives I find these days are SATA 04:04:32 I remembered maybe bood cds having problem with sata cdrom.. so i went for ide - same price 04:04:39 seems to have been a mistake ;) 04:04:43 Hehe 04:04:51 Most shops here don't have IDE DVD drives in stock 04:05:06 I haven't had problems with SATA DVD 04:05:29 Core i7? :P 04:05:41 i remember i had to switch the sata conrtoller to ide-mode for the knoppix cd to book 04:05:58 and ide-mode for disks sucks.. no ncq 04:06:11 i5 - cheaper ;) 04:06:17 especially the mainboard 04:06:18 Ahh :) 04:06:25 What mainboard? 04:06:36 -!- stabwound has quit [Ping timeout: 240 seconds] 04:06:39 anyways... these mysterious crashs kept me busy for 3 f***ing weeks! 04:06:46 ASUS? Gigabyte? Model number? :) 04:06:57 some gigabyte - P55A-UD4 04:07:27 So the problem was a driver issue? 04:07:28 i wouldn't suggest it though... much trouble with the automatically selected voltages.. as my million google searches reported 04:07:51 cpu_idle call was trying to get interrupt from cdrom 04:08:54 cdrom_newpc_intr -> ide_complete_rq -> oops -> trying to switch away from X with nvidia close-source driver -> panic 04:09:29 but the problem only occured in the idle calls! 04:10:08 can you imagine? running 720p video in full screen on one screen... glxgears & some heavy flash on the other tft... runs fine.. even for 50 minutes 04:10:30 then I decide to reboot.. and after closing everything.. and moving the mouse ontop of the logout button -> panic 04:11:13 anyways.. I hope this really is the problem that was causing the crashes.. 04:11:22 I wanted to talk to you... about: 04:11:28 -!- stabwound has joined ##crawl-dev 04:11:51 Sounds like an exciting problem to debug :) 04:11:59 > 0.6-rc users who don't upgrade to latest don't get logfile entries on end of game. Can we fix this? 04:12:08 I didn't understand that 04:12:27 Someone started a game on 0.6 alpha, refused to upgrade when CDO offered new versions 04:12:29 since all versions use a link to the logfile/milestone/score as their 04:12:34 Napkin: trying to switch away from X on the nvidia close-source driver crashed like hell 10 years ago and crashes like hell now 04:12:38 Then when they completed their game, Gretell did not report it 04:12:49 And Henzell did not find it in the logfile either 04:12:52 same for trying to hibernate or something 04:12:53 must me some matching problem, greensnark 04:13:03 about the version= strings maybe? 04:13:21 or a bug in his version not writing the entry 04:13:22 I forgot the person's name, or we could check the logfile 04:13:28 adrock? 04:13:31 i think - wait 04:13:36 It's possible it's that weird wizmode bug again 04:13:41 "I was the guy who didn't save a logfile. Believe I started the save about a month ago, and ascended 04:13:44 with an rc1 save on 3/18 on CDO." 04:13:46 Where the game spontaneously kicks itself into wizmode 04:13:49 !lg adrock 04:13:50 49. adrock the Shooter (L2 HEAM), slain by Terence (a +1,+2 hand axe of freezing) on D:2 on 2010-02-13, with 94 points after 1162 turns and 0:06:52. 04:13:50 yes, "adrock" 04:13:53 !lg adrock win 04:13:53 1. adrock the Slayer (L25 MiBe), worshipper of Trog, escaped with the Orb and 3 runes on 2010-01-16, with 1410627 points after 93640 turns and 10:39:20. 04:14:12 I think the win was earlier this month 04:14:19 he messages me 4d 11h 11m 32s ago 04:14:27 says Henzell 04:14:32 !lm adrock cdo 04:14:33 90. [2010-03-18] adrock the Archmage (L27 DSFE) found the Orb of Zot! (Zot:5) 04:14:37 That game 04:14:40 That game -game 04:14:43 !lm adrock cdo -game 04:14:43 adrock, XL27 DSFE, T:156490 (milestone) has no matching game. 04:15:05 Let me grep the logfile 04:15:15 !lm adrock cdo x=wiz 04:15:16 90. [2010-03-18] [wiz=] adrock the Archmage (L27 DSFE) found the Orb of Zot! (Zot:5) 04:16:37 It's not in the logfile 04:16:56 kilobyte - it works ok at the moment.. but the release_console_sem() function kills it 04:17:01 wizard then, i guess, greensnark 04:17:23 It's possible, but that would mean some nasty bug still around 04:17:40 Napkin: Your game that auto-wizmoded -- did you meet Mara? 04:18:35 oh.. not sure anymore - i believe so, yes 04:18:57 but.. that bug is not necessarily still around 04:19:06 considering he was playing a version from November last year 04:19:13 Oh, really? 04:19:25 Ok, then let's not worry about it :) 04:19:33 if i remember correctly he was one of the very old chars still kept around 04:19:59 If it's that old it would have been very buggy 04:19:59 > Have you noticed any incidents of Crawl 0.6 eating too much RAM? 04:20:08 Ah yes, have you? :) 04:20:08 no :) let's check for cores? :D 04:20:42 root@ipx14623:/var/lib/dgamelaunch/cores> ls -1 | wc -l 04:20:42 69 04:20:48 Exciting 04:21:11 We need a quick way to filter out the cores involving Kyrris using command-repeat to attack bushes 04:21:20 hehehe 04:21:41 how can I help? :) 04:21:43 Did someone fix the command-repeat crashes ? 04:22:11 That is a lot of cores 04:22:14 i chmod-644'ed the cores 04:22:29 Excellent 04:22:29 do a ls -latr 04:22:32 Where are the binaries again? 04:22:44 ~dgl/../bin/ 04:23:30 Ah, and last thing: Lordsloth told me, that the ruby/screen/dtach rl-server is up for public: http://rl.heh.fi 04:24:17 -!- by has joined ##crawl-dev 04:24:33 Interesting 04:24:46 ttyrec filenames include term size <3 04:24:51 Morning by 04:25:06 Moin Rob! :) 04:25:11 hi 04:25:41 Napkin: This is joosa's custom server, right? 04:25:56 Does he have logfile/milestone links? 04:25:58 I believe so, yes 04:26:00 yes 04:26:23 we should talk with him and announce him in the wordpress maybe? 04:26:28 Yes 04:27:20 And if we have logfile/milestone links and ttyrecs are retained, I can make Henzell and FooTV keep tabs on the server 04:27:21 Cool - I'll try to catch him next time he's around 04:27:33 joosa! he's nicely idling here :D 04:27:48 These ttyrec times are UTC, I trust :) 04:28:23 can we please make all the servers sizestamp ttyrecs? heuristically determining ttyrec sizes from data is the kind of problem that gives me nightmares 04:29:12 !tell Joosa Hey there! http://rl.heh.fi is yours, right? How about adding its stats in Henzell's database and making tty-records available in FooTV? I'd also like to announce it on Crawl's wordpress, if you don't mind? :) 04:29:13 Napkin: OK, I'll let Joosa know. 04:29:34 about the memory issue, valgrind indeed shows leaks. For ./crawl -seed 1 -arena 'cerebov v test spawner delay:0', there's 1.2MB of leaks 04:29:50 !tell joosa If you have http URLs for logfile and milestones, I can make changes to Henzell and FooTV 04:29:51 greensnark: OK, I'll let joosa know. 04:29:52 sorear: what do you want? The stream of ANSI codes without timing data? 04:29:57 kilobyte: Where are the leaks? 04:30:32 lemme check, I just ran valgrind for the first time 04:30:37 heh, this is nice - now that there is action here I can go back to work ;-P 04:31:33 kilobyte: just the size 04:32:26 not sure what would you want the exact size of -- isn't a set percentage good enough? 04:33:04 kilobyte: not that kind of size - rows and columns 04:33:37 heuristically extracting coordinate range data from ttyrecs is made of nightmares in the presense of scrolling and automargins 04:33:49 aah, yeah 04:34:00 Hi 04:34:00 joosa: You have 2 messages. Use !messages to read them. 04:34:08 I read them already :P 04:34:12 Hello :) 04:34:30 Crawl doesn't use automargins I think, nor scrolling, so you can check for the max coord written to -- but that's still a dirty hack 04:35:29 Napkin: that would be great, we need more player ghosts! 04:35:59 my termrec and termproxy write that as a \e[8 ; rows ; cols t -- perhaps we can add that to DGL 04:36:01 greensnark: I've made an announder bot for the server, maybe I should bring it to ##crawl? 04:36:13 joosa: Of course 04:36:24 hey joosa :) 04:36:31 But I'm referring to Henzell stats, not just announcements 04:36:33 Nice to see your server up and running! :) 04:36:35 Also FooTV playback 04:36:39 although it uses notice instead of privmsg per IRC RFC guidelines :) 04:37:04 Henzell and Gretell just spam the channel :P 04:37:12 Or haven't you visited ##crawl? :) 04:37:14 you broke all reasonable guidelines already with ruby/screen/dtach :-P 04:37:19 Napkin: haha 04:37:50 I'll have to change the command char, there may be overlapping stuff 04:37:52 could you limit it to crawl announcements? I see nethack is running on your server too 04:38:04 Yes, we don't want to hear about NH games :) 04:38:11 Of course, just needs a bit of modification 04:38:13 joosa: how do you solve unknown charset? screen doesn't work if you don't it the charset. 04:38:34 I force UTF-8 locale on connect 04:38:53 I haven't seen anything go wrong yet 04:39:00 you are using kilobyte's termrec, right? the ttys are compatible with FooTV, greensnark? 04:39:08 joosa: it's possible to detect that (kind of, just UTF-8 vs 8-bit, plus some CJK stuff we don't care ab out) 04:39:32 Napkin: I guess so. I just use Term::TtyRec::Plus for playback 04:39:44 great :) 04:40:10 so, what would be the rl.heh.fi command char (for getting log links etc) 04:40:10 FooTV will need changes to locate the ttyrec correctly since joosa's server puts them in different dirs 04:40:18 Whereas dgamelaunch drops all ttyrecs in one massive dir 04:40:34 rumor has it that sending NOTICEs to channels causes some minor client to explode in a ball of hellfire 04:40:39 I think it was mIRC 04:40:53 using NOTICE isn't following rfc guidelines anyway 04:40:55 The world has so many amusing irc clients 04:40:57 that's intended behaviour, sorear ;> 04:40:58 we've been over this issue 04:41:04 Like that one that explodes when someone types in runes 04:41:13 joosa: none. This is 2010, Henzell handles queries for all servers 04:41:15 I thought NOTICE is for automated messages that should never be replied to 04:41:39 !lg * s=src 04:41:46 566916 games for *: 475173x cao, 91743x cdo 04:41:47 joosa: it's for automated replies, so that things know not to reply back 04:41:53 says nothing at all about messages in general 04:42:03 regardless of whether they come from a human or a bot 04:42:06 So as not to cause loops 04:42:07 03by * r2d6bd84ec830 10/crawl-ref/source/beam.cc: bolt::draw clean-up. 04:42:08 right 04:42:18 03by * rd098134e2229 10/crawl-ref/source/message.cc: Call tiles.set_need_redraw() from message_window::show(). 04:42:19 joosa: Napkin makes cdo logfile available here: http://crawl.develz.org/allgames-svn.txt 04:42:36 Just make your logfile + milestones available by http and Henzell can do the rest 04:42:46 That's easy enough 04:43:07 it's wget -c - so not transfering everything all the time - but make sure you don't modify it ;) 04:43:11 (i don't really care either way, i just know this because people were whining a few months back that my bot didn't use NOTICE) 04:43:15 (: 04:43:19 doy: d098134e should fix some more tiles lag issues... 04:44:02 phew - did we just manage to prevent an RFC crusade? :D 04:45:13 That was too easy 04:45:30 I'll make it use privmsg on ##crawl :) 04:45:55 but fist, those logfiles and milestones 04:46:15 The fist of first men 04:46:48 also, it's not like anyone really follows the irc rfc these days 04:46:55 so it's kind of a moot point 04:46:56 (: 04:47:01 doy: indeed :P 04:47:21 doy: But I also thought it's nice to have the bot spam be different and not "hilight" the channel 04:47:43 logfiles and milestones for each version, respectively http://rl.heh.fi/crawl/ http://rl.heh.fi/crawl-0.6/ http://rl.heh.fi/trunk/ 04:48:19 joosa: Thanks 04:48:29 I'll make the changes to the bot 04:48:29 joosa: Those files are append-only, right? 04:48:36 Do you rotate those files at any point? 04:48:51 Not, yet, I should at some point perhaps 04:48:57 one, comma, too much 04:49:14 It's easier for Henzell if they're append-only :) 04:49:47 does Crawl write them append only? 04:50:14 I think so, only scores file is sorted 04:50:14 Yes 04:50:21 Hmm, your server seems to have noticeably slow redraw, unlike cdo or cao 04:50:36 it's located in Finland 04:50:45 I'm in Poland 04:50:50 I see 04:52:20 Most players have been from Finland to this point 04:52:27 joosa: do not rotate the files... 04:52:37 it looks like it's a fault of screen, I see the same delay locally when comparing running directly vs via screen 04:52:44 sorear: ok, won't do. 04:52:54 kilobyte: Oh, that's kind of sad. 04:53:11 if you rotate the files, wget -c will get very confused 04:53:28 I bet 04:54:15 Nobody else has commented on delays though 04:54:50 joosa: nano asks me to confirm path to save rc file :) 04:54:54 Can I uh overwrite files :P 04:55:03 greensnark: You won't be able to change it 04:55:09 Oh 04:55:24 nano has a nice restricted mode 04:55:47 I'm having trouble building crawl from master atm... is this known? 04:56:09 ixtli: It might be known if you specify your trouble :P 04:56:14 im getting "SQLITE_HAS_CODEC" has not been defined then it fails to link 04:56:21 Heh i was getting to that :D 04:56:21 I think by fixed that 04:56:33 :o 04:56:38 80x25, eh? 04:56:57 by: Who, me? :P 04:57:02 yes :) 04:57:07 Damn, busted 04:57:14 hah 04:57:19 I run 80x31 :D 04:57:32 it's almost optimal use of message space :) 04:57:39 ixtli: I made a patch for that, but don't know if neunon applied it yet 04:58:05 you can look for my message to neunon in yesterday's logs, should include a link 04:58:06 Sounds good, i'll just fix it here ^^ 04:58:36 SQLITE_HAS_CODEC may be an independent issue to the linking error 04:58:40 Redraw does feel somewhat slower than cdo here 04:58:55 Playable though 04:59:00 But I am used to a bit of lag :P 04:59:20 by: I will see. Give me a moment. 04:59:27 by: Actually, where can I find your patch? 04:59:39 Ok, maybe not that much lag :) 04:59:50 Yeah, all this dtach/screen stuff is experimental anyway, I may go back to ttyrec tail 05:00:31 by: My linker issue is that "_gethostuuid" is undefined, referenced from libsqlite3.a 05:00:48 ixtli: yesterday's log, as I said above 05:01:00 Ah, sorry. Missed that message. 05:01:42 I'll name the bot... any suggestions? 05:01:56 Hehzell 05:01:59 haha 05:02:02 joosa: what do you need? The current screen of the ttyrec? 05:02:41 kilobyte: initially I had watching implemented exactly like in dgamelaunch 05:02:47 Hmm, what's the name of the witch who wanted to eat Hänsel and Gretel? 05:03:29 Hexell? 05:03:38 dgl will blast all data since the last screen clear, which can be megabytes long 05:04:01 Hehfiel 05:05:10 kilobyte: Napkin and I were considering using your libtty to maintain complete frames in dgl to make it easier to jump in and watch someone 05:06:48 if you have problems with something in the library, or just comments, tell me -- no one else used it yet 05:07:11 I should get to it one of these weekends 05:07:13 Complete frames sounds nice 05:07:22 Spectating is the most awful aspect of current dgl 05:07:28 I agree 05:07:41 But my dtach kludge is way uglier :P 05:08:32 i kinda like the screen/dtach idea though :) 05:08:37 you'd need to tell the library what the encoding is, though -- but it's better to detect that than to force people to toggle it by hand anyway 05:15:58 03kilobyte * r77cce521a61b 10/crawl-ref/source/misc.cc: Fix a nasty memory leak. 05:18:03 Does that bug exist in 0.6 as well? 05:18:04 kilobyte: thanks; not sure how that happened 05:18:10 -!- Mu_ has joined ##crawl-dev 05:18:19 We should take that fix for 0.6 if necessary 05:18:37 likely to be in 0.6, also 05:19:00 Yes it does 05:19:11 !tell doy Please pick 77cce521a61b for 0.6 05:19:11 greensnark: OK, I'll let doy know. 05:28:44 Perhaps I should filter some of the announcements like Henzell does 05:29:06 it's been ok with so few players 05:29:29 Filtering out startscumming helps once your server gets popular :) 05:29:49 I imagine a player like Meow will startscum another 20k games in 0.6 before he notices that stats are fixed now :P 05:29:58 !gamesby Meow 05:30:01 Meow has played 21422 games, between 20071031 and 20091129, won 2 (0.0%), high score 7203896, total score 19667603, total turns 6493533, total time 638:51:11. 05:30:09 !gamesby Meow ktyp=quitting 05:30:10 Meow (ktyp=quitting) has played 20631 games, between 20071031 and 20091129, won 0, high score 42133, total score 573436, total turns 724628, total time 86:49:06. 05:43:46 It's nice the stats are fixed now 05:58:51 greensnark: did you see the comment requesting a mac build of rc4? 05:59:19 by: Yes, but wrapped up in work atm 06:13:49 -!- stabwound has quit [Ping timeout: 264 seconds] 06:17:49 03by * r79d8622b89cd 10/crawl-ref/source/los.cc: Make LOS radius change invalidate global LOS 06:19:01 -!- stabwound has joined ##crawl-dev 06:21:52 -!- by has quit [Quit: leaving] 06:39:41 joosa: "Couldn't open settings/KiloByte-macro.txt for writing!" 06:39:56 oh noes. 06:41:32 I guess you have macro_dir = settings 06:41:41 I didn't know it overrides command line 06:45:13 -!- hotsun has joined ##crawl-dev 06:45:50 yeah, I pasted in my config 06:48:21 -!- Spads has quit [Ping timeout: 268 seconds] 06:53:24 hmm... if it overrides the command line, that's bad shit on DGL 07:06:24 helloooo 07:20:11 I agree, that's bad... 07:25:08 I don't recall this happening in 0.5.2 or earlier though 07:25:44 kilobyte: sure enough, I was able to save macros as a different file using the macro_dir option 07:26:04 baaad. 07:32:23 joosa: Are you not using DATA_DIR_PATH? 07:33:31 I think I'm using pretty much the standard DGL configuration 07:33:46 has nothing to do with dgl itself :P 07:33:49 macro_dir is supposed to be ignored if DATA_DIR_PATH is set 07:34:09 I see 07:34:23 Then I must have had it defined in 0.5 but not in 0.6 07:44:10 -!- Spads has joined ##crawl-dev 07:45:06 -!- timecircuits1 has quit [Quit: Leaving.] 07:52:28 * joosa kicks self in the shin 08:23:39 -!- stabwound has quit [Ping timeout: 276 seconds] 08:26:41 -!- Twinge has quit [Read error: Connection reset by peer] 08:26:51 -!- Twinge has joined ##crawl-dev 08:27:11 -!- stabwound has joined ##crawl-dev 08:44:25 -!- ortoslon has joined ##crawl-dev 08:58:09 -!- Blitz82 has joined ##crawl-dev 08:59:55 -!- Madtrixr has quit [Ping timeout: 240 seconds] 08:59:56 -!- Blitz82 is now known as Madtrixr 09:06:40 -!- ais523 has joined ##crawl-dev 09:09:35 Modified skills (m) screen (https://crawl.develz.org/mantis/view.php?id=1138) by dpeg 09:24:13 Name, title, god, place, level (https://crawl.develz.org/mantis/view.php?id=1139) by dpeg 09:24:13 -!- Henzell has quit [Ping timeout: 240 seconds] 09:24:13 -!- Ashenzari has quit [Ping timeout: 252 seconds] 09:24:56 -!- Ashenzari has joined ##crawl-dev 09:24:56 -!- The topic of ##crawl-dev is: DCSS development channel! | Logs: http://tozt.net/crawl | People with +v have commit access. | Please keep general crawl-related chit-chat to ##crawl. | Dev wiki: http://crawl.develz.org/wiki | Ashenzari source: http://github.com/doy/crawlbot 09:25:26 -!- Henzell has joined ##crawl-dev 09:26:45 -!- Cryp71c_ has joined ##crawl-dev 09:26:57 -!- Cryp71c has quit [Disconnected by services] 09:26:58 -!- Cryp71c_ is now known as Cryp71c 09:27:06 Morning. 09:31:11 Cryp71c: Hey :) 09:31:20 game ended when i reached level 2 as a DSCr on 0.7 09:31:26 no save game to be seen 09:31:33 mutation related crash bug perhaps? 09:31:51 just thought i'd mention it here 09:32:46 wait a sec... seems the build is quite broken 09:32:59 yeah i just crashed again, no ding this time 09:33:04 seems unrelated 09:36:55 -!- purge has quit [Ping timeout: 240 seconds] 09:38:46 I'm getting contrib/install/include/sqlite3.h:3692:5: warning: "SQLITE_HAS_CODEC" is not defined 09:39:27 i think that was discussed a bit ago 09:39:29 git submodules are upto date 09:39:38 I'll scroll a bit 09:39:39 its not just you 09:39:48 ctrl+f sqlite 09:40:05 Yeah, ok. 0.7 will be offline until it's fixed 09:40:22 I'm rebuilding 0.6 now with fixed data dir 09:40:39 -!- eith has joined ##crawl-dev 10:22:20 -!- syllogism has joined ##crawl-dev 10:39:18 -!- hotsun has quit [Read error: Connection reset by peer] 10:39:41 -!- hotsun has joined ##crawl-dev 10:56:26 -!- ogaz has joined ##crawl-dev 10:57:36 -!- by has joined ##crawl-dev 11:05:38 sqlite build/linking problem (https://crawl.develz.org/mantis/view.php?id=1140) by rob 11:08:27 by: my fix to that memory leak is wrong, apparently lua holds on to those monster_info structures beyond the ch_mon_is_safe call 11:08:47 but somehow it appears to crash only during saving 11:09:07 ah, that would explain why I put the "new" in; certainly didn't put too much thought into that however 11:13:53 by: instead of using monster_info* as the userdata type, maybe use monster_info as the userdata? That way Lua will manage the memory there 11:14:14 Only catch is that if monster_info needs C++ constructor/destructor you have to mess with placement new/delete 11:14:39 kilobyte: can you check if the findray test also leaks memory? 11:15:18 Or you can bypass that by using monster_info* and newing up the instance there and using the Lua metatable gc to delete the instance 11:15:43 monster_info doesn't seem to have constructor/destructor 11:16:04 Aha, you're using lua_object_gc 11:16:17 Which will call delete on the pointer when Lua gc runs 11:16:34 So you must allocate with new 11:16:45 -!- Textmode has joined ##crawl-dev 11:17:18 but then where was the memory leak? 11:17:50 -!- karmatic has joined ##crawl-dev 11:17:57 I don't know :) 11:18:28 morning all 11:18:33 hi 11:21:58 by: The leak could be if ch_mon_is_safe is not defined 11:22:18 The new should be in lua_push_moninf, not in the user of clua.callfn() 11:22:28 ah 11:22:55 i.e. *miref = new monster_info(*mi) would be correct 11:23:42 If the caller does new monster_info, but ch_mon_is_safe is not defined, the Lua VM will never see the monster_info userdata, so it cannot gc it 11:23:49 yes 11:29:26 03by * rbe3b7cc44ae1 10/crawl-ref/source/l_moninf.cc: Properly fix ch_mon_is_safe memory leak (greensnark). 11:29:32 kilobyte: can you test if this helps? 11:32:29 03by * r465f2585162a 10/crawl-ref/source/l_moninf.cc: Adapt mi_get_monster_at to lua_push_moninf-change. 11:42:52 ??git 11:42:53 git[1/1]: git clone git://crawl-ref.git.sourceforge.net/gitroot/crawl-ref/crawl-ref 11:46:49 by: it works! No crash on saving, too. 11:49:34 yay 11:50:26 !tell doy be3b7cc4 and 465f2585 should go with 77cce521 11:50:26 by: OK, I'll let doy know. 11:54:48 I'm running some stress tests, but with valgrind being ~200 times slower than native code, they seem to take weeks 11:55:12 but basic in-game commands appears to work ok, same for short arena tests 11:55:46 could you try the hit_closest macro, also? that also uses monster info through lua (different path) 11:56:09 that should have had a memory leak between be3b and 465f 11:56:18 how do you modify a patch file without corrupting it? 11:57:01 depends on the modification; if you delete some lines or change +/ /-, you need to update the numbers at the top 11:57:06 Cryp71c: You need to keep the hunk line numbers in sync with your changes but you should ask yourself why you're hand-editing diffs 11:57:15 well, specifically. I've developed several patches that - when applied to HEAD, now - get a "patch does not apply" trying to circumvent that. 11:57:30 greensnark, is there an easy way to update patch files when they no longer apply to the head revision? 11:57:34 you can try to apply with 'patch' instead of some git invocation 11:57:41 Yes, use patch directly 11:58:03 If it can't apply a patch it'll leave you .rej files that you can apply manually 12:00:29 :( havn't seen dpeg today 12:02:32 alternatively, you can go back to the revision that the patch was created at, apply the patch and commit it, and then rebase it on top of master 12:02:32 doy: You have 2 messages. Use !messages to read them. 12:02:36 but that's a bit more advanced 12:02:37 (: 12:03:00 And you might have to resolve conflicts anyway :) 12:03:21 But yeah, if you already have commits in git you should just rebase 12:03:27 sure, but resolving conflicts against one commit is easier than resolving them against all the intervening commits 12:03:29 And if you don't have commits in git, what are you doing?! 12:05:30 -!- by has quit [Quit: leaving] 12:06:15 Heh, so NH now has a native OSX port 12:06:30 I wonder if the devteam will assimilate Dirk Zimmerman 12:07:00 no leaks with hit_closest, too 12:07:32 devteam? 12:07:35 what devteam? 12:08:14 I think they've finished NH 3.5 and are just waiting for a suitable release date 12:08:21 Such as the reappearance of Halley's comet 12:12:12 greensnark: unlikely, his code isn't 20 years backwards compatible 12:12:48 we even had to complain for a 10.5 version :) 12:14:05 Oh, he's a Snow Leopard user? :0 12:14:24 !messages 12:14:25 (1/2) greensnark said (6h 55m 13s ago): Please pick 77cce521a61b for 0.6 12:14:27 !messages 12:14:28 (1/1) by said (24m 1s ago): be3b7cc4 and 465f2585 should go with 77cce521 12:21:39