Show enters and exits. Hide enters and exits.
| 00:00:40 | vruz leaves the room. | |
| 00:03:40 | GMFlash leaves the room. | |
| 00:03:48 | GMFlash enters the room. | |
| 00:05:38 | vruz enters the room. | |
| 00:12:05 | wycats leaves the room. | |
| 00:12:48 | rubymaverick_ leaves the room. | |
| 00:12:53 | Arjen_ leaves the room. | |
| 00:21:45 | graza_ enters the room. | |
| 00:21:51 | graza leaves the room. | |
| 00:27:57 | rue | graza_: Let me know if it works |
| 00:28:00 | rue | crafterm: Morning |
| 00:28:11 | crafterm | rue: morning mate |
| 00:28:19 | crafterm | hows it going? |
| 00:30:09 | rue | Alright, been busy |
| 00:30:21 | rue | Getting everything ready for the conf? |
| 00:30:28 | crafterm | same here, flat out but good |
| 00:30:38 | crafterm | yep, pretty much all set, we're all booked up now |
| 00:30:52 | crafterm | for minor things left but looking forward to it |
| 00:31:56 | rue | Coolness. Were you guys able to accommodate everyone? |
| 00:32:34 | crafterm | there's still some demand, but the room is at capacity now |
| 00:33:33 | crafterm | so we might have to run a couple of them :) |
| 00:35:15 | crafterm leaves the room. | |
| 00:37:36 | rutlov enters the room. | |
| 00:37:40 | jeremydurham leaves the room. | |
| 00:39:06 | technomancy enters the room. | |
| 00:39:19 | graza | rue: are you in Australia? |
| 00:42:23 | wvl_ leaves the room. | |
| 00:44:32 | wycats_ leaves the room. | |
| 00:46:33 | wycats enters the room. | |
| 00:48:46 | rue | graza_: Not I |
| 00:49:24 | ragnard leaves the room. | |
| 00:49:27 | wycats_ enters the room. | |
| 00:52:08 | agardiner leaves the room. | |
| 00:52:09 | wvl enters the room. | |
| 00:57:25 | macournoyer enters the room. | |
| 01:00:18 | AndrewO leaves the room. | |
| 01:02:15 | agardiner enters the room. | |
| 01:02:25 | oleganza enters the room. | |
| 01:06:11 | wycats__ enters the room. | |
| 01:08:21 | wycats | evan: ping |
| 01:08:41 | wycats leaves the room. | |
| 01:12:36 | jeremydurham enters the room. | |
| 01:20:26 | wycats_ leaves the room. | |
| 01:22:20 | MenTaLguY enters the room. | |
| 01:22:30 | MenTaLguY | howdy |
| 01:23:58 | rue | Aloha |
| 01:26:28 | radarek leaves the room. | |
| 01:31:59 | radarek enters the room. | |
| 01:32:35 | dancroak enters the room. | |
| 01:33:08 | radarek leaves the room. | |
| 01:34:21 | eventualbuddha leaves the room. | |
| 01:35:05 | crafterm enters the room. | |
| 01:44:42 | mernen leaves the room. | |
| 01:57:40 | graza enters the room. | |
| 01:57:41 | graza_ leaves the room. | |
| 01:57:44 | ataiar enters the room. | |
| 01:58:34 | ataiar leaves the room. | |
| 02:00:45 | benburkert enters the room. | |
| 02:04:22 | wycats_ enters the room. | |
| 02:07:45 | lvmc leaves the room. | |
| 02:08:33 | MenTaLguY leaves the room. | |
| 02:21:11 | graza_ enters the room. | |
| 02:21:12 | graza leaves the room. | |
| 02:23:11 | jeremydurham leaves the room. | |
| 02:26:15 | wycats leaves the room. | |
| 02:30:06 | mutle leaves the room. | |
| 02:32:41 | wycats_ leaves the room. | |
| 02:35:28 | kofno enters the room. | |
| 02:36:34 | radarek enters the room. | |
| 02:36:47 | radarek leaves the room. | |
| 02:39:42 | mutle enters the room. | |
| 02:46:02 | VVSiz_ enters the room. | |
| 02:49:36 | AndrewO enters the room. | |
| 02:50:12 | _mutle enters the room. | |
| 02:52:38 | graza enters the room. | |
| 02:52:39 | graza_ leaves the room. | |
| 02:53:37 | tbmcmullen enters the room. | |
| 02:58:37 | noclue123 enters the room. | |
| 03:00:21 | xmlhacker leaves the room. | |
| 03:02:20 | jtoy enters the room. | |
| 03:03:53 | xmlhacker enters the room. | |
| 03:04:10 | VVSiz leaves the room. | |
| 03:04:33 | jtoy leaves the room. | |
| 03:05:51 | jtoy enters the room. | |
| 03:05:54 | mutle leaves the room. | |
| 03:11:17 | hairyfeet enters the room. | |
| 03:19:14 | graza_ enters the room. | |
| 03:19:15 | graza leaves the room. | |
| 03:21:17 | rubymaverick enters the room. | |
| 03:23:15 | aotearoa leaves the room. | |
| 03:23:33 | aotearoa enters the room. | |
| 03:25:36 | aotearoa leaves the room. | |
| 03:25:59 | aotearoa enters the room. | |
| 03:26:57 | antares leaves the room. | |
| 03:32:30 | imajes leaves the room. | |
| 03:37:23 | xmlhacker_ enters the room. | |
| 03:44:40 | graza enters the room. | |
| 03:44:41 | graza_ leaves the room. | |
| 03:46:12 | jtoy leaves the room. | |
| 03:48:38 | ttmrichter leaves the room. | |
| 03:53:32 | xmlhacker leaves the room. | |
| 03:57:21 | benburkert leaves the room. | |
| 03:59:53 | wycats enters the room. | |
| 04:02:00 | graza_ enters the room. | |
| 04:02:01 | graza leaves the room. | |
| 04:02:43 | benburkert enters the room. | |
| 04:04:10 | twshelton enters the room. | |
| 04:05:05 | ttmrichter enters the room. | |
| 04:09:11 | benburkert_ enters the room. | |
| 04:09:11 | benburkert leaves the room. | |
| 04:09:43 | benburkert_ leaves the room. | |
| 04:10:17 | benburkert enters the room. | |
| 04:12:21 | nicksieger leaves the room. | |
| 04:13:59 | ezmobius leaves the room. | |
| 04:16:31 | be9 enters the room. | |
| 04:20:29 | benburkert leaves the room. | |
| 04:20:50 | benburkert enters the room. | |
| 04:26:02 | graza enters the room. | |
| 04:26:03 | graza_ leaves the room. | |
| 04:29:22 | rutlov enters the room. | |
| 04:38:56 | graza_ enters the room. | |
| 04:38:57 | graza leaves the room. | |
| 04:51:32 | graza enters the room. | |
| 04:51:33 | graza_ leaves the room. | |
| 04:56:49 | xmlhacker_ enters the room. | |
| 05:06:06 | RyanTM leaves the room. | |
| 05:10:29 | brixen | sweet, just ran mspec-ci on array with the new tagged excludes |
| 05:15:10 | xmlhacker leaves the room. | |
| 05:16:07 | technomancy leaves the room. | |
| 05:16:40 | rue | brixen: Cool |
| 05:16:49 | rue | evan: You there? |
| 05:17:23 | brixen | rue: I'm removing the #fails_on guard, since that can be done with excludes |
| 05:17:31 | brixen | rue: any case you can think of that I'm missing? |
| 05:25:56 | mae enters the room. | |
| 05:26:05 | crafterm leaves the room. | |
| 05:26:28 | graza leaves the room. | |
| 05:29:54 | graza enters the room. | |
| 05:32:51 | macournoyer leaves the room. | |
| 05:37:39 | dancroak leaves the room. | |
| 05:37:49 | binary42 leaves the room. | |
| 05:39:05 | xmlhacker_ leaves the room. | |
| 05:46:15 | mae leaves the room. | |
| 05:55:22 | graza leaves the room. | |
| 05:55:33 | _ADS enters the room. | |
| 06:05:36 | rue | brixen: I would imagine it is fine |
| 06:06:07 | agardiner leaves the room. | |
| 06:13:40 | defunkt enters the room. | |
| 06:14:16 | wycats_ enters the room. | |
| 06:14:39 | xmlhacker enters the room. | |
| 06:15:33 | kofno leaves the room. | |
| 06:15:48 | mae enters the room. | |
| 06:16:36 | defunkt leaves the room. | |
| 06:21:15 | benburkert leaves the room. | |
| 06:25:04 | dctanner enters the room. | |
| 06:26:41 | wycats leaves the room. | |
| 06:30:01 | dctanner leaves the room. | |
| 06:32:39 | mae leaves the room. | |
| 06:37:33 | benburkert enters the room. | |
| 06:46:06 | benburkert enters the room. | |
| 06:48:57 | headius enters the room. | |
| 06:49:57 | bitbckt enters the room. | |
| 07:10:30 | lstoll leaves the room. | |
| 07:13:18 | xmlhacker_ enters the room. | |
| 07:15:32 | mae enters the room. | |
| 07:24:25 | hairyfeet leaves the room. | |
| 07:24:44 | mae_ enters the room. | |
| 07:27:41 | ezmobius enters the room. | |
| 07:29:19 | xmlhacker leaves the room. | |
| 07:29:59 | be9 leaves the room. | |
| 07:34:46 | TheVoice enters the room. | |
| 07:35:25 | mae leaves the room. | |
| 07:35:51 | TheVoice leaves the room. | |
| 07:41:58 | dewd leaves the room. | |
| 07:45:50 | mae_ leaves the room. | |
| 07:56:16 | AndrewO leaves the room. | |
| 08:01:26 | radarek enters the room. | |
| 08:02:33 | headius leaves the room. | |
| 08:03:06 | dbussink leaves the room. | |
| 08:03:36 | dbussink enters the room. | |
| 08:04:36 | dbussink | morning! |
| 08:06:38 | lstoll enters the room. | |
| 08:08:26 | brixen | morning dbussink |
| 08:09:51 | dbussink | looks like removing the gcc optimizations also fix building on openbsd |
| 08:09:56 | rue | Morning |
| 08:11:02 | _ADS leaves the room. | |
| 08:11:36 | dbussink | hmm, ./bin/ci failed on openbsd with Too many open files (EMFILE) error |
| 08:13:29 | rue | brixen: You know if there is a conf tomorrow? |
| 08:14:54 | brixen | rue: conf? |
| 08:15:02 | rue | ~Call |
| 08:15:37 | brixen | no idea, no one has mentioned anything to me |
| 08:15:58 | brixen | at 10am pst I'll be at a talk on correct incremental gc at psu |
| 08:16:29 | brixen | rue: I'm thinking of pushing the mspec rewrite, but mspec-run isn't complete yet, what do you think? |
| 08:16:35 | brixen | mspec-ci runs fine |
| 08:16:50 | brixen | mspec-tag isn't complete either (for adding/removing tags) |
| 08:17:11 | rue | Manual add for now? |
| 08:17:41 | brixen | yeah |
| 08:17:57 | brixen | if you do bin/mspec tag it will output a mesg to that effect |
| 08:18:06 | brixen | I should finish the runners tomorrow |
| 08:18:29 | brixen | this is mspec-ci: 1131 files, 5161 examples, 17514 expectations, 0 failures, 0 errors |
| 08:18:32 | brixen | 21 sec |
| 08:18:39 | brixen | not too bad, since it was running in 19 |
| 08:18:55 | brixen | oh, and tips 5k :D |
| 08:19:53 | thehcdreamer enters the room. | |
| 08:19:59 | rue | Cool |
| 08:20:50 | brixen | well, I guess I'll wait since I'll be out part of the morning |
| 08:21:00 | brixen | I should probably be here when the bitching starts in earnest :P |
| 08:21:20 | dbussink | guess the number has gone up because library specs are ran too |
| 08:21:21 | dbussink | ? |
| 08:21:32 | brixen | dbussink: haven't added library specs yet |
| 08:21:39 | brixen | one of the next things on the agenda |
| 08:22:27 | tbmcmullen | quick question; whats the usual timeframe for getting patches accepted? |
| 08:22:47 | brixen | tbmcmullen: totally depends, do you have a hot one waiting? :) |
| 08:22:48 | aotearoa leaves the room. | |
| 08:23:08 | dbussink | tbmcmullen: it usually helps to nag people here :P |
| 08:23:09 | tbmcmullen | brixen: yessir. 331. |
| 08:23:34 | tbmcmullen | I was hoping to get ruby-openid working. For which I need bits of CGI. For which I need that patch in. Heh. |
| 08:24:09 | brixen | looks |
| 08:24:59 | dbussink | tbmcmullen: cgi isn't tested at all afaik |
| 08:25:11 | dbussink | tbmcmullen: or specced for that matter |
| 08:25:40 | tbmcmullen | dbussink: Yeah, I talked to someone about it on here. I was planning on spec'ing at least part of it. |
| 08:25:45 | brixen | tbmcmullen: ahh, Array#pack |
| 08:25:49 | mae_ enters the room. | |
| 08:25:56 | tbmcmullen | dbussink: Would the whole thing need to come over at once or could it come over one piece at a time? |
| 08:26:26 | dbussink | you could partially spec it first |
| 08:26:38 | dbussink | something is always better than nothing |
| 08:26:58 | brixen | tbmcmullen: I'll do it now |
| 08:27:00 | tbmcmullen | cool. I really just need the parse and escaping bits at first anyway. |
| 08:27:08 | tbmcmullen | brixen: Great! Thanks. |
| 08:29:03 | headius enters the room. | |
| 08:29:54 | brixen | tbmcmullen: no problem, thanks for submitting |
| 08:30:34 | dbussink | tbmcmullen: well, getting those bits specced is already a nice addition to have :) |
| 08:30:44 | brixen | headius: do you know why there are fails_on :jruby in the specs? are you unable to use excludes for any of those? |
| 08:31:03 | imajes enters the room. | |
| 08:31:13 | headius | probably old |
| 08:31:28 | boyscout | 1 commit by Tyler McMullen |
| 08:31:29 | boyscout | * Add support for H and h to Array#pack.; a683dd7 |
| 08:31:39 | brixen | headius: ok cool |
| 08:31:59 | brixen | headius: I've removed them, I should push the new tagged stuff tomorrow |
| 08:32:15 | brixen | headius: then VVSiz_ can add bug #'s to all the jruby excludes |
| 08:32:38 | headius | okeedoke |
| 08:36:46 | jtoy enters the room. | |
| 08:38:17 | VVSiz | brixen, headius: indeed, no need for fails_on :jruby anymore. I was going to remove them sooner or later (there are not that many left) |
| 08:38:27 | brixen | tbmcmullen: next time evan is around, pastie him your public ssh key and point to #331 |
| 08:38:44 | tbmcmullen | brixen: Will do. Thanks. |
| 08:38:48 | brixen | VVSiz: no problem, I've removed them locally, should push the stuff tomorrow |
| 08:39:04 | VVSiz | thanks! |
| 08:41:38 | brixen | VVSiz: np |
| 08:42:03 | brixen | ok, sleep for me, catch y'all in a few hours |
| 08:42:17 | dbussink | nite! |
| 08:43:08 | brixen | night |
| 08:43:57 | rubuildius | Tyler McMullen: a683dd757; 4844 examples, 17581 expectations, 0 failures, 0 errors |
| 08:44:03 | dbussink | looks like i found another symbol collision... |
| 08:45:54 | mae_ leaves the room. | |
| 08:54:35 | dbussink | anyone in with some knowledge on the parser? |
| 08:57:02 | octopod enters the room. | |
| 08:58:05 | tbmcmullen | Have you ever seen a seg fault when trying to run shotgun/rubinius with "Unknown primitive hash_redistribute"? |
| 08:59:32 | dbussink | are you sure you have the latest version? |
| 08:59:40 | dbussink | try rake rebuild |
| 08:59:47 | dbussink | or rake distclean build if that doesn't help |
| 09:00:15 | tbmcmullen | I actually deleted my local code and pulled a fresh copy, after it failed the first time. |
| 09:00:43 | tbmcmullen | Trying the rebuild now... |
| 09:00:43 | dbussink | did you ever do rake install? |
| 09:00:50 | tbmcmullen | I did. |
| 09:01:00 | dbussink | ah, that's the problem then |
| 09:01:04 | dbussink | remove everything that it installed |
| 09:01:27 | dbussink | installing always results in a lot of trouble, i never do it |
| 09:01:37 | dbussink | we're moving too fast to keep that working |
| 09:01:48 | tbmcmullen | Ahh. Excellent. That helps. |
| 09:04:54 | dbussink | rue: still there? |
| 09:05:12 | rue | Yeah, what is up? |
| 09:05:30 | dbussink | do you know how i can get the entire parse tree for a certain file? |
| 09:06:23 | rue | `shotgun/rubinius describe` or `shotgun/rubinius lib/compiler/describe.rb` |
| 09:06:31 | rue | Both slightly different |
| 09:07:14 | Arjen_ enters the room. | |
| 09:07:21 | dbussink | rue: ah thnx! |
| 09:09:30 | langenberg enters the room. | |
| 09:09:35 | dbussink | rue: hmm, looks like i found a bug in the parser on freebsd / openbsd |
| 09:09:51 | dbussink | rue: a symbol collision |
| 09:09:56 | crossblaim enters the room. | |
| 09:10:10 | rue | Hm, can you give me the test case? |
| 09:10:12 | dbussink | rue: the file contains the correct symbol names, but the parser makes another symbol out of it |
| 09:10:21 | dbussink | rue: do you have a bsd system? |
| 09:10:34 | dbussink | ./shotgun/rubinius describe lib/irb/ruby-token.rb |
| 09:10:45 | dbussink | line 482 of that output |
| 09:11:38 | dbussink | says [:lit, :TkUNLESS] but that should be [:lit, :TkIF_MOD], see line 118 of lib/irb/ruby-token.rb |
| 09:11:38 | crossblaim | hi |
| 09:12:05 | dbussink | crossblaim: hi |
| 09:19:58 | jtoy leaves the room. | |
| 09:20:14 | dbussink | rue: are you able to reproduce it? |
| 09:20:57 | rue | Yeah, I see it too |
| 09:24:55 | wycats_ leaves the room. | |
| 09:25:17 | ezmobius leaves the room. | |
| 09:26:48 | mae_ enters the room. | |
| 09:28:27 | dbussink | rue: i suspect that it could occur on other platforms, but that it also depends on the number of symbols already defined |
| 09:28:44 | dbussink | rue: because that number differs for each platform, because of the types generator |
| 09:31:07 | warren_s enters the room. | |
| 09:32:19 | rue | Yeah |
| 09:32:36 | rue | Probably need to revisit the algo |
| 09:32:42 | dbussink | any idea where this problem could originate? |
| 09:45:53 | rue | Sorry, wandering in and out |
| 09:46:01 | dbussink | hehe, no problem |
| 09:46:18 | rue | Is it libmquark that is used for it now? |
| 09:46:48 | mae_ leaves the room. | |
| 09:48:12 | dbussink | yeah, but that is pretty dumb afaik |
| 09:48:26 | dbussink | but it might be a bug there, dunno |
| 09:48:50 | imajes leaves the room. | |
| 09:49:36 | rue | Hm, actually looks like string.c defines its own hash function |
| 09:51:04 | dbussink | yeah, that one can cause collisions, but evan fixed that afaik |
| 09:51:11 | dbussink | maybe the fix is incomplete |
| 09:51:19 | dbussink | or not properly used at compile time |
| 09:51:33 | dbussink | because the previous collision was at runtime, not parse / compile time |
| 09:52:51 | ragnard enters the room. | |
| 09:53:25 | rue | Grammar uses quark |
| 09:53:47 | rue | See shotgun/external_libs/libmquark/quark.c |
| 09:54:20 | rue | And the hash function there is not great |
| 09:54:39 | rue | eyes mass menacingly |
| 09:54:56 | w1rele55 enters the room. | |
| 09:56:04 | rue | Also libcchash, I think |
| 09:58:42 | dbussink | did you test with the quark hash function whether they collide? |
| 09:59:30 | rue | No, sorry, trying to puzzle out an arch |
| 10:08:25 | dbussink | rue: hmm, no collision for the quark hash algo |
| 10:10:14 | cavalle enters the room. | |
| 10:13:31 | dbussink | rue: but that would have been strange, because that algorithm doesn't seem to depend on the number of existing symbols |
| 10:27:18 | headius leaves the room. | |
| 10:27:48 | mae_ enters the room. | |
| 10:35:04 | ragnard | dbussink: hello! I saw the removal of the specific inline optimization flags made the build work on openbsd... are you still getting random segfaults when compiling? |
| 10:35:16 | dbussink | ragnard: nope |
| 10:35:19 | dbussink | ragnard: works fine now |
| 10:35:36 | dbussink | ragnard: those optimizations broke stuff on os x ppc too |
| 10:35:50 | ragnard | dbussink: hm. I have built without optimizations and I still get the segfaults. |
| 10:36:03 | dbussink | ragnard: hmm, are you sure you did a distclean first |
| 10:36:03 | dbussink | ? |
| 10:36:09 | langenberg leaves the room. | |
| 10:36:18 | ragnard | dbussink: tracked them down, seems like there is a problem with pthread_getspecific in environment.c |
| 10:36:40 | ragnard | dbussink: it simply returns NULL |
| 10:37:28 | ragnard | dbussink: on openbsd, that is. It feels a bit scary... I have asserted that pthread_setspecific succeeds, the key is properly set |
| 10:38:10 | ragnard | dbussink: but as I said, sometimes pthread_getspecific returns 0 |
| 10:38:19 | dbussink | ragnard: that's really strange |
| 10:39:09 | dbussink | ragnard: i'm gonna try building again |
| 10:39:43 | ragnard | dbussink: yep. I'm afraid it might have something to do with the pthread impl. and signals... because the call is to environment_current_machin, from _cpu_task_preempt, which is the callback for a SIGVTALRM |
| 10:40:42 | dbussink | hmmm, if i remember correctly, there were also problems on solaris with SIGVTALRM when i tried it there |
| 10:40:43 | ragnard | (i built with DEV=1 though... so no optimizations) |
| 10:42:36 | dbussink | ragnard: hmm, i built it without DEV |
| 10:45:41 | ragnard | dbussink: I'll try that, do some stresstesting with the asserts still in and see what happens... |
| 10:46:19 | ragnard | dbussink: but it doesn't feel real good when a sigsegv is optimized away :) |
| 10:46:35 | dbussink | ragnard: nope, but i've seen it happen before |
| 10:47:02 | dbussink | ragnard: on os x ppc i got a sigbus, and it looked like the heap pointer was 0x0 |
| 10:47:32 | dbussink | we've had it with -funroll-loops too |
| 10:47:47 | dbussink | couldn't find the problem, but it seemed like that optimization triggered it |
| 10:47:54 | ragnard | dbussink: hm. scary. |
| 10:49:08 | dbussink | ragnard: yeah, but i've used gentoo and if you look at that, there a lot of packages that have excludes for specific compiler flags |
| 10:51:03 | mae_ leaves the room. | |
| 10:51:46 | ragnard | dbussink: you mean excludes to avoid certain known 'hazardous' optimizations? |
| 10:52:08 | dbussink | ragnard: yeah, sometimes even -O2 is known to break certain applications |
| 10:53:16 | ragnard | dbussink: well... I have the great respect for compiler writers... guess it is really hard to predicit side effects, but it would be nice with a flags to avoid possibly hazardous optimizations. but i guess that is -O0 :) |
| 10:54:34 | dbussink | ragnard: i'm planning on doing some benchmarking on rubinius with various flags |
| 10:54:42 | dbussink | to see whether they make a difference |
| 10:56:34 | ragnard | dbussink: regarding the EMEMFILE failure on openbsd, have you bumbed the openfiles limit? it's quite conservatively set by default. |
| 10:56:47 | dbussink | ragnard: no, not yet |
| 10:57:01 | dbussink | ragnard: but that was my idea too, but didn't get to it yet |
| 10:57:16 | ragnard | dbussink: sure... just catching up on the irc logs :) |
| 10:58:08 | antares enters the room. | |
| 11:11:23 | imajes enters the room. | |
| 11:11:23 | _mutle leaves the room. | |
| 11:12:43 | mutle enters the room. | |
| 11:16:12 | cjheath leaves the room. | |
| 11:18:02 | lstoll leaves the room. | |
| 11:19:11 | lstoll enters the room. | |
| 11:28:53 | mae_ enters the room. | |
| 11:50:45 | mae_ leaves the room. | |
| 11:57:38 | rue | Nite |
| 12:01:34 | dbussink | nite |
| 12:03:28 | kofno enters the room. | |
| 12:03:44 | GMFlash leaves the room. | |
| 12:03:52 | GMFlash enters the room. | |
| 12:04:01 | Fullmoon enters the room. | |
| 12:07:14 | random8r enters the room. | |
| 12:07:49 | random8r enters the room. | |
| 12:29:27 | mae enters the room. | |
| 12:40:33 | mae_ enters the room. | |
| 12:50:13 | mae leaves the room. | |
| 12:56:03 | fork leaves the room. | |
| 12:56:16 | fork_ enters the room. | |
| 13:01:45 | mae enters the room. | |
| 13:08:21 | rubymaverick leaves the room. | |
| 13:09:18 | dbussink | i've been running ./bin/ci benchmarks, and removing the inline optimizations actually make it a bit faster |
| 13:09:22 | dbussink | not significant though |
| 13:10:56 | mae__ enters the room. | |
| 13:11:44 | graza enters the room. | |
| 13:11:56 | dbussink | within a 1% margin over 50 ./bin/ci runs |
| 13:14:54 | mae_ leaves the room. | |
| 13:15:34 | mae leaves the room. | |
| 13:16:00 | mae enters the room. | |
| 13:20:02 | mae leaves the room. | |
| 13:20:16 | mae enters the room. | |
| 13:20:26 | mae__ leaves the room. | |
| 13:20:35 | random8r leaves the room. | |
| 13:23:07 | Fullmoon leaves the room. | |
| 13:28:46 | lstoll leaves the room. | |
| 13:29:23 | graza leaves the room. | |
| 13:29:26 | graza_ enters the room. | |
| 13:29:43 | lstoll enters the room. | |
| 13:30:35 | skaa1 enters the room. | |
| 13:31:43 | mae_ enters the room. | |
| 13:33:26 | mae leaves the room. | |
| 13:35:29 | mae_ leaves the room. | |
| 13:43:46 | graza enters the room. | |
| 13:43:47 | graza_ leaves the room. | |
| 13:52:52 | jtoy enters the room. | |
| 13:53:29 | hornbeck leaves the room. | |
| 13:55:52 | binary42 enters the room. | |
| 13:58:19 | dbussink | for whoever is left here: http://pastie.caboo.se/pastes/153689 |
| 13:58:31 | crossblaim leaves the room. | |
| 13:58:45 | dbussink | looks like all those inline optimizations are basically useless at the moment |
| 13:59:13 | Arjen | dbussink, agreed. |
| 14:02:08 | rubymaverick enters the room. | |
| 14:03:17 | crossblaim enters the room. | |
| 14:04:47 | langenberg enters the room. | |
| 14:07:06 | langenberg leaves the room. | |
| 14:07:59 | langenberg enters the room. | |
| 14:08:45 | langenberg leaves the room. | |
| 14:08:57 | RyanTM enters the room. | |
| 14:09:44 | langenberg enters the room. | |
| 14:09:47 | graza leaves the room. | |
| 14:11:49 | random8r enters the room. | |
| 14:12:26 | ragnard | dbussink: the evidence is clear, seems like an easy case... :) |
| 14:12:37 | dbussink | ragnard: you found it? |
| 14:12:40 | ragnard | dbussink: openbsd still segfaults for me though |
| 14:12:45 | dbussink | ragnard: or you mean the benchmark? |
| 14:12:48 | ragnard | dbussink: no, i mean the inline benchmarks |
| 14:13:04 | dbussink | ragnard: yeah, i've seen the segfaults too |
| 14:13:11 | dbussink | ragnard: do you also get them after a build? |
| 14:14:00 | ragnard | dbussink: i seem to get them randomly... i trigger them by doing a rake clean:rbc, then rake... eventually something blows |
| 14:14:15 | ragnard | dbussink: sometimes i have rebuilt shotgun, sometimes not... |
| 14:14:41 | ragnard | dbussink: this is where it dies for me: http://pastie.org/153225 |
| 14:15:08 | dbussink | ragnard: hmm, that build without DEV ? |
| 14:15:13 | ragnard | dbussink: yep. |
| 14:15:31 | ragnard | dbussink: no, sorry... |
| 14:15:37 | cavalle leaves the room. | |
| 14:15:39 | ragnard | dbussink: with DEV=1, that is |
| 14:16:01 | mae enters the room. | |
| 14:16:13 | ragnard | tried most recent tree both with, and without DEV... same old. |
| 14:16:27 | dbussink | ah ok, but no idea on specific causes? |
| 14:17:00 | dbussink | could probably be a ucontext problem, openbsd is the only system using it right now afaik |
| 14:17:02 | random8r | I reckon there should be three parts to development: 1. Writing Code 2. Testing 3. Documentation. They should be inseparable. |
| 14:17:07 | ragnard | well no... it crashes 'randomly', ie. not when compiling any specific file... |
| 14:17:41 | ragnard | as you say, its probably something hairy with ucontext/signals/pthread... |
| 14:18:39 | ragnard | I am sure that pthread_setspecific succeeds, and sets the correct value (ie. not NULL)... but when the value is retrieved using pthread_getspecific, it is somehow NULL. |
| 14:18:56 | kAworu leaves the room. | |
| 14:19:08 | ragnard | and this happens, as I said, in a signal handler... |
| 14:19:43 | cavalle enters the room. | |
| 14:22:23 | langenberg leaves the room. | |
| 14:23:16 | langenberg enters the room. | |
| 14:23:17 | langenberg leaves the room. | |
| 14:24:09 | langenberg enters the room. | |
| 14:24:21 | langenberg leaves the room. | |
| 14:24:35 | dbussink | ragnard: hmm, why is the cast to a rubinius_globals in there? |
| 14:25:09 | dbussink | because that's a totally different thing |
| 14:25:53 | langenberg enters the room. | |
| 14:26:35 | ragnard | dbussink: i don't understand... different to what? it still returns global->m? |
| 14:27:02 | ragnard | it gets set in environment_setup_thread. am i mistaken? |
| 14:29:18 | dbussink | ragnard: nvm, i misread it |
| 14:29:28 | dbussink | there is a rubinius_global and a rubinius_globals (mind the final s) |
| 14:29:38 | langenberg leaves the room. | |
| 14:29:39 | ragnard | aha, didn't know that |
| 14:30:33 | langenberg enters the room. | |
| 14:30:39 | langenberg leaves the room. | |
| 14:31:04 | random8r | wow rubinius has come along heaps since 4 weeks ago |
| 14:31:12 | mae_ enters the room. | |
| 14:31:13 | kAworu enters the room. | |
| 14:31:25 | mae_ leaves the room. | |
| 14:31:28 | langenberg enters the room. | |
| 14:31:30 | langenberg leaves the room. | |
| 14:31:40 | random8r | 83%! |
| 14:32:08 | twshelton leaves the room. | |
| 14:32:10 | mae leaves the room. | |
| 14:32:45 | dbussink | 83% of what? |
| 14:32:46 | random8r leaves the room. | |
| 14:32:59 | langenberg enters the room. | |
| 14:33:17 | langenberg leaves the room. | |
| 14:34:05 | langenberg enters the room. | |
| 14:34:25 | dbussink | langenberg: what's up with your connection? |
| 14:35:48 | langenberg leaves the room. | |
| 14:36:38 | langenberg enters the room. | |
| 14:37:19 | langenberg leaves the room. | |
| 14:38:10 | langenberg enters the room. | |
| 14:38:52 | dbussink | langenberg: you there? |
| 14:39:15 | langenberg leaves the room. | |
| 14:40:11 | langenberg enters the room. | |
| 14:40:21 | langenberg leaves the room. | |
| 14:41:12 | langenberg enters the room. | |
| 14:43:22 | langenberg leaves the room. | |
| 14:43:51 | anonuser enters the room. | |
| 14:44:17 | langenberg enters the room. | |
| 14:44:17 | pietia enters the room. | |
| 14:44:48 | langenberg leaves the room. | |
| 14:46:31 | langenberg enters the room. | |
| 14:47:14 | moofbong enters the room. | |
| 14:47:27 | langenberg | dbussink: bad wireless reception at my desk. |
| 14:48:56 | dbussink | langenberg: ah, well you were filling up my chat window :P |
| 14:49:13 | langenberg | dbussink: my bad :) |
| 14:49:38 | dbussink | langenberg: btw, did you read my comment on your ticket? |
| 14:49:56 | langenberg | dbussink: Yeah, just checked my mail. Thanks for your reply. |
| 14:50:26 | mae enters the room. | |
| 14:50:41 | mae leaves the room. | |
| 14:50:46 | langenberg | dbussink: I haven't updated the stdlib version in that branch. |
| 14:51:05 | dbussink | but it adds a version to lib/ |
| 14:51:13 | langenberg | dbussink: Is that stdlib just an exact copy of ruby's lib? |
| 14:51:18 | dbussink | yeah |
| 14:51:30 | dbussink | so if stuff goes into lib/ it should be removed from stdlib |
| 14:51:32 | langenberg | Isn't it a good idea to update all files? |
| 14:51:41 | langenberg | Oh, I didn't know |
| 14:52:02 | dbussink | lib/ is what is tested and known to work in rubinius |
| 14:52:08 | dbussink | stdlib is sort of a todo |
| 14:52:12 | langenberg | Ah :) |
| 14:52:19 | langenberg | Thought it was just a reference |
| 14:57:20 | langenberg | I'll add another patch okay? |
| 14:59:13 | dbussink | yeah, that's ok |
| 14:59:25 | dbussink | does everything work on rubinius? |
| 15:01:16 | langenberg | yeah |
| 15:01:32 | langenberg | examples pass |
| 15:01:44 | langenberg | Except the code that relies on kconv |
| 15:01:56 | twshelton enters the room. | |
| 15:05:00 | anonuser | Okay, anyone know where to get quality parts for a new computer on the cheap? My home computer died last weekend ;_; |
| 15:05:29 | dbussink | langenberg: if it's not too much trouble, could you use git mv to move the file? |
| 15:05:53 | langenberg | dbussink: mv instead of rm? |
| 15:06:11 | dbussink | langenberg: it keeps history over the moved file |
| 15:06:12 | langenberg | I did make some minor changes to it. |
| 15:06:42 | dbussink | langenberg: ah ok, well, we would like to be able to trace those changes even over renames or moves of the file |
| 15:07:10 | macournoyer enters the room. | |
| 15:07:32 | langenberg | sure |
| 15:08:01 | dbussink | my usual approach would be to fix it up in stdlib and finally move it if it's ready |
| 15:08:10 | Arjen | langenberg, Mathijs, congrats on the patch! :) |
| 15:08:21 | langenberg | Hah, it's a little start. |
| 15:08:36 | langenberg | Glad I can make a tiny contribution to Ruby's future. |
| 15:08:45 | Arjen | It's the small things that make a difference (bad attempt at translating dutch to english) |
| 15:08:52 | dbussink | don't let my nitpicking hold you back ;) |
| 15:09:23 | jtoy leaves the room. | |
| 15:09:50 | Arjen | I tried building rubinius at home to do make a little contribution of my own, but couldn't get it to work. I'll try again later... |
| 15:09:56 | dancroak enters the room. | |
| 15:10:05 | rubuildius_ppc | Tyler McMullen: a683dd757; build failed! |
| 15:10:25 | dbussink | Arjen_: what's your system? |
| 15:10:38 | Arjen | dbussink, Ubuntu. |
| 15:10:49 | dbussink | Arjen_: hmm, shouldn't be a problem |
| 15:11:04 | Arjen | dbussink, I don't think it is, I'm just missing stuff and I can't figure out what. |
| 15:11:32 | Arjen | I'm a linux newbie, so I'm learning while I go... |
| 15:12:07 | dbussink | Arjen_: ah ok, well, make sure to check the dependencies that are listed on the lighthouse page about installing it |
| 15:12:20 | d2dchat enters the room. | |
| 15:12:22 | dbussink | cremes: that's your bot? |
| 15:12:25 | Arjen | dbussink, I got all those, and it still didn't work. |
| 15:12:27 | cremes | benny: need to ask a question about rubuildius |
| 15:12:28 | nikolasco enters the room. | |
| 15:12:36 | cremes | dbussink: yep |
| 15:13:20 | dbussink | cremes: ah well, got it working on your system |
| 15:13:21 | chris2 enters the room. | |
| 15:13:35 | dbussink | cremes: didn't fix the truncate errors though |
| 15:13:53 | cremes | dbussink: yeah, the latest builds have been running nicely except for the File#truncate failures |
| 15:13:59 | dbussink | cremes: there is another problem that could cause it, that also causes irb to fail |
| 15:14:27 | cremes | dbussink: the rubuildius buildlog is complaining it can't find a rakefile |
| 15:14:36 | langenberg | dbussink: I'll do a mv from stdlib to lib, and re-apply changes, okay? |
| 15:14:48 | dbussink | langenberg: yeah, that's ok |
| 15:15:07 | dbussink | langenberg: do the move in a separate commit, makes it more clear |
| 15:16:04 | mae enters the room. | |
| 15:16:31 | kAworu leaves the room. | |
| 15:19:52 | hornbeck enters the room. | |
| 15:21:29 | kAworu enters the room. | |
| 15:22:03 | tbmcmullen leaves the room. | |
| 15:24:36 | rubyconsumer leaves the room. | |
| 15:32:54 | mae leaves the room. | |
| 15:34:32 | langenberg leaves the room. | |
| 15:46:29 | d2dchat leaves the room. | |
| 15:53:39 | benburkert enters the room. | |
| 15:55:28 | benny | cremes: hey |
| 15:55:51 | cremes | benny: getting rubuildius running on osx ppc and have a question... |
| 15:56:15 | cremes | benny: i need to "seed" the repo/rubinius/HEAD dir with a manual clone before starting, yes? |
| 15:56:23 | benny | yes |
| 15:56:42 | cremes | okay, doing that now... it wasn't obvious to me before |
| 15:56:57 | cremes | anything else not in the readme that i should be on the lookout for? |
| 15:57:06 | d2dchat enters the room. | |
| 15:57:37 | benny | it's in the readme :-) |
| 15:58:04 | cremes | d'oh! i need to read more carefully! |
| 15:58:22 | cremes | thanks for your help |
| 15:58:46 | benny | hehe, no problem :-) |
| 15:59:23 | cremes | oh, one more thing... you commented out pastie.rb in your version. i added it back using the pastie code from pastie.caboo.se |
| 15:59:36 | cremes | any problem with that? should it "just work?" |
| 16:02:39 | benny | oh I didn't add either pastie.rb or nopaste |
| 16:02:52 | benny | in bin/ |
| 16:04:05 | benny | but pastie was down when I committed it, that's why I switched to rafb |
| 16:04:30 | cremes | i see, thanks |
| 16:04:59 | benny | do you have a pastie.rb? |
| 16:05:16 | cremes | i created one from the code available at pastie.caboo.se |
| 16:05:31 | cremes | i put it into ~/continuous/bin |
| 16:05:37 | benny | oh there is one... do you mind pasting it somewhere so I can commit it? |
| 16:05:52 | cremes | sure... hold on a sec |
| 16:06:16 | ttmrichter_ enters the room. | |
| 16:08:00 | cremes | you need both of these: http://pastie.caboo.se/153747 and http://pastie.caboo.se/153748 |
| 16:11:05 | benny | thanks |
| 16:15:47 | mae enters the room. | |
| 16:16:10 | mae leaves the room. | |
| 16:18:42 | boyscout | 1 commit by Benjamin Andresen |
| 16:18:43 | boyscout | * added nopaste & pastie to rubuildius (thanks to Chuck Remes); 8309fe8 |
| 16:20:43 | cavalle leaves the room. | |
| 16:24:40 | cavalle enters the room. | |
| 16:25:21 | ttmrichter leaves the room. | |
| 16:27:25 | agile leaves the room. | |
| 16:28:21 | rubuildius_ppc | Benjamin Andresen: 8309fe877; 4844 examples, 17568 expectations, 2 failures, 10 errors; |
| 16:28:53 | rubuildius | Benjamin Andresen: 8309fe877; 4844 examples, 17581 expectations, 0 failures, 0 errors |
| 16:29:04 | benny | cremes: is pastie.rb accepting stdin to paste? |
| 16:33:48 | bitbckt leaves the room. | |
| 16:33:53 | cyndis_ enters the room. | |
| 16:34:11 | cyndis leaves the room. | |
| 16:38:55 | cavalle leaves the room. | |
| 16:44:51 | cremes | benny: looks like I need to test that out. gotta run right now but will look at it later |
| 16:47:06 | AndrewO enters the room. | |
| 16:50:53 | lopex enters the room. | |
| 16:51:25 | enebo enters the room. | |
| 16:56:01 | benburkert enters the room. | |
| 17:02:11 | d2dchat leaves the room. | |
| 17:02:28 | d2dchat enters the room. | |
| 17:03:15 | langenberg enters the room. | |
| 17:07:18 | nikolasco leaves the room. | |
| 17:19:35 | benburkert leaves the room. | |
| 17:25:15 | Silex enters the room. | |
| 17:25:37 | eventualbuddha enters the room. | |
| 17:29:24 | agile enters the room. | |
| 17:30:36 | mae enters the room. | |
| 17:30:54 | mae leaves the room. | |
| 17:31:03 | enebo leaves the room. | |
| 17:31:53 | dewd enters the room. | |
| 17:32:58 | benburkert enters the room. | |
| 17:33:01 | yipstar enters the room. | |
| 17:36:03 | thehcdreamer leaves the room. | |
| 17:44:44 | cavalle enters the room. | |
| 17:49:17 | technomancy enters the room. | |
| 17:50:00 | xmlhacker enters the room. | |
| 17:52:55 | tmornini enters the room. | |
| 17:55:11 | AndrewO leaves the room. | |
| 18:01:00 | octopod leaves the room. | |
| 18:04:50 | xmlhacker_ leaves the room. | |
| 18:07:59 | tbmcmullen enters the room. | |
| 18:08:56 | wifelette enters the room. | |
| 18:11:26 | srbaker leaves the room. | |
| 18:12:47 | w1rele55 leaves the room. | |
| 18:16:14 | GMFlash leaves the room. | |
| 18:16:15 | nemerle_afk enters the room. | |
| 18:16:22 | GMFlash enters the room. | |
| 18:16:24 | langenberg | dbussink: I've updated the ticket's attachments. |
| 18:16:25 | mae enters the room. | |
| 18:18:15 | srbaker enters the room. | |
| 18:18:36 | Defiler | Power outages are fun |
| 18:21:58 | pauldix enters the room. | |
| 18:23:40 | Silex | I'm a bit new to this, ruby 1.9 will have a vm right ? |
| 18:23:50 | Silex | (could it be that rubinius will be used for it?) |
| 18:23:58 | Defiler | It has its own VM, unrelated to Rubinius |
| 18:24:04 | Silex | ok |
| 18:24:28 | Defiler | It doesn't store the compiled files in between runs, though.. it just processes them and runs them 'internally' |
| 18:24:46 | Defiler | ..but it is still a VM (called YARV) |
| 18:24:47 | Silex | I see |
| 18:24:56 | Silex | ty for the infos |
| 18:25:08 | Defiler | No problem |
| 18:26:44 | wycats enters the room. | |
| 18:27:50 | rue | Morning |
| 18:28:04 | dbussink | morning |
| 18:28:23 | dbussink | for who's interested, i've got some stats on the gcc optimizations |
| 18:28:38 | dbussink | http://pastie.caboo.se/pastes/153689 |
| 18:28:55 | dbussink | i say, we remove all those inline things, they don't help and only break stuff |
| 18:32:19 | mae leaves the room. | |
| 18:32:39 | Defiler | dbussink: Interesting. |
| 18:33:07 | dbussink | Defiler: really no significant differences there |
| 18:33:51 | Defiler | Yeah |
| 18:34:28 | nemerle leaves the room. | |
| 18:35:40 | dbussink | which actually confirms my experience with gentoo and that is that uber 1337 compiler flags usually don't have any noticeable effect |
| 18:36:45 | rue | You forgot -fquantum-pipe |
| 18:37:30 | scoopr | inlining is a double-edged sword |
| 18:37:31 | langenberg | --omg-optimized |
| 18:37:34 | dbussink | hehe, yeah |
| 18:37:52 | Defiler | --fcharles-barkley-shut-up-and-jam-gaiden |
| 18:37:55 | scoopr | but stuff like -mcpu=exact-cpu-model does help somewhat |
| 18:38:26 | scoopr | at least, if the baseline is 386 |
| 18:40:30 | dbussink | yeah, could be useful, but we need better platform detection for that |
| 18:40:44 | dbussink | i could benchmark with that some more |
| 18:41:21 | djwhitt | dbussink: yeah, I've had the same experience with compiler flags in Gentoo |
| 18:41:44 | djwhitt | all I ever use is -O2 -mtune=my-cpu-type -march=my-cpu-type |
| 18:42:03 | dbussink | and since we still have a lot to gain with just being smarter inside rubinius, i don't we think we should do exotic gcc things |
| 18:42:03 | djwhitt | omitting frame pointer helps too |
| 18:42:12 | djwhitt | but at the expense of being able to debug stuff |
| 18:42:13 | dbussink | yeah, but we need that for debugging afaik |
| 18:42:15 | Defiler | GCC's -O2 is already so smart that the rest just seem like a waste of effort |
| 18:42:15 | Defiler | heh |
| 18:42:39 | djwhitt | some stuff like unrolling loops actually slows stuff down |
| 18:42:47 | djwhitt | because the loops no longer fit inside the cache |
| 18:42:51 | dbussink | djwhitt: well, that actually broke stuff on linux |
| 18:43:08 | dbussink | we had funrolls, but alas, no more |
| 18:43:17 | djwhitt | fast-math can legitimately speed stuff up too |
| 18:43:33 | tmornini leaves the room. | |
| 18:43:34 | djwhitt | but I think it make some math stuff wrong |
| 18:43:36 | dbussink | yeah, but you have to be really careful with that |
| 18:43:43 | djwhitt | don't know the details about that one though |
| 18:43:48 | TheVoice enters the room. | |
| 18:43:48 | dbussink | because it breaks floating point stuff |
| 18:43:56 | rue | Getting a SIGSEGV on HEAD with DEV=1 |
| 18:44:05 | dbussink | hmm, platform? |
| 18:44:09 | rue | Trying sans DEV |
| 18:44:21 | enebo enters the room. | |
| 18:44:24 | rue | FreeBSD meow.kittensoft.org 6.2-RELEASE FreeBSD 6.2-RELEASE |
| 18:44:51 | dbussink | ah, strange, i haven't had a sigsegv there, only the parser / compiler issue |
| 18:45:25 | dbussink | i have 6.3 btw |
| 18:46:20 | rue | This is new as of my last commits |
| 18:46:33 | rue | I think anyway, still need to check straightline |
| 18:46:33 | oleganza leaves the room. | |
| 18:47:02 | oleganza enters the room. | |
| 18:48:54 | rue | dbussink: Only fails with DEV=1 |
| 18:49:04 | dbussink | rue: that's really strange |
| 18:49:17 | rue | So this may be a week or two old even |
| 18:50:11 | Defiler | Anyone feel like doing battle with what I think is a lightning/ffi problem? |
| 18:50:55 | dbussink | is there any way to properly debug ffi atm? |
| 18:51:12 | rue | Sure, with GDB :) |
| 18:51:12 | dbussink | would love to be able so simply look at what gets passed down to the c functions |
| 18:51:19 | Defiler | I haven't figured out a way, no |
| 18:51:19 | dbussink | yeah, i know |
| 18:51:42 | Defiler | As far as I can tell, this signature is broken |
| 18:51:43 | Defiler | [:state, :string, :string, :int, :int], :object |
| 18:52:03 | dbussink | what goes wrong? |
| 18:52:04 | wycats leaves the room. | |
| 18:52:10 | headius enters the room. | |
| 18:52:13 | rue | Broken how? Incorrect for the function, passed wrong, converted wrong? |
| 18:53:08 | Defiler | SIGILL |
| 18:53:20 | Defiler | (on Linux 64 and MacOS 32, thus far) |
| 18:53:55 | Defiler | slightly different failures.. on MacOS, you can get in far enough to see it mangle up the first 'int' argument.. on Linux, it just does SIGILL before that |
| 18:54:05 | Defiler | Which makes sense, given that we use different ffi techniques in the two places |
| 18:54:53 | imajes leaves the room. | |
| 18:56:58 | Defiler | I might be diagnosing it wrong, of course |
| 18:59:19 | dbussink | Defiler: might also be the cause of the problems on os x ppc |
| 18:59:20 | antares leaves the room. | |
| 18:59:45 | dbussink | but what about the problems with symbols and the parser? maybe it extracts the wrong value from the ffi typedefs |
| 19:01:16 | Defiler | I am working on reproducing it with a more trivial setup here |
| 19:02:22 | antares enters the room. | |
| 19:02:51 | dbussink | but i'm off, time for some social time spending, back later |
| 19:03:13 | antares leaves the room. | |
| 19:04:02 | Arjen_ leaves the room. | |
| 19:10:23 | wifelette leaves the room. | |
| 19:10:29 | rue | What is this 'social' time? |
| 19:12:10 | Defiler | Sounds subversive |
| 19:14:36 | headius | hiya hiya |
| 19:14:45 | headius | how are things |
| 19:15:44 | mae enters the room. | |
| 19:15:52 | mae leaves the room. | |
| 19:16:14 | wycats enters the room. | |
| 19:17:20 | wifelette enters the room. | |
| 19:25:02 | cavalle leaves the room. | |
| 19:26:25 | pauldix leaves the room. | |
| 19:26:35 | pietia leaves the room. | |
| 19:26:41 | defunkt enters the room. | |
| 19:27:01 | skaa1 leaves the room. | |
| 19:28:32 | pietia enters the room. | |
| 19:29:51 | Defiler | headius: Happy President's Day |
| 19:30:02 | headius | it is, isn't it! |
| 19:30:10 | headius | we're at the sun office today and there's not a soul around |
| 19:30:17 | headius | we didn't plan this trip very well |
| 19:30:21 | Defiler | I just found out that this is an Engine Yard holiday, which explains a lot ha |
| 19:30:36 | headius | so it's been pretty quiet in here then? |
| 19:30:45 | Defiler | Yeah |
| 19:31:06 | headius | I have forgotten what a day off is like |
| 19:34:13 | srbaker leaves the room. | |
| 19:37:13 | Silex | about the gcc discussion, maybe there is a slight diff between -O2 and -O3 |
| 19:37:15 | Silex | (fwiw) |
| 19:37:48 | Defiler | What are you running that shows a difference? |
| 19:37:53 | jeremydurham enters the room. | |
| 19:38:32 | Silex | no I'm talking theory |
| 19:38:40 | Silex | I have nothing to back that up :) |
| 19:38:48 | oleganza leaves the room. | |
| 19:39:02 | Defiler | Well, if there is a benchmark that shows it being worth doing, we will switch to -O3. |
| 19:39:03 | Silex | but theorically -O3 is more agressive about optimisation, but considered less stable (logical) |
| 19:39:08 | jeremydurham | tbmcmullen: hey. Congrats :) |
| 19:39:21 | rue | -O3 is largely useless |
| 19:39:26 | Defiler | If it never crashes, it is stable enough :) |
| 19:39:32 | Silex | oh I think you'd stay with O3, except if performance becomes an issue maybe |
| 19:39:39 | Silex | *O2 |
| 19:39:58 | Silex | it's not like you'll gain much |
| 19:40:29 | Defiler | I thought you were just suggesting that we would gain something? |
| 19:40:46 | Silex | you'd probably gain but not much |
| 19:41:05 | defunkt leaves the room. | |
| 19:41:18 | Silex | the best optimisations are done in the code anyway :) |
| 19:41:26 | srbaker_ enters the room. | |
| 19:42:09 | djwhitt | my experience with O3 on Gentoo has been that it makes things rather unstable |
| 19:42:16 | Defiler | I haven't worked myself up to watching it yet, but.. http://www.infoq.com/interviews/wilson-bilkovich-discusses-rubinius |
| 19:42:41 | djwhitt | Silex: which I guess basically agrees with what you're saying |
| 19:43:21 | Silex | djwhitt: yeah, I use O3 for personal stuffs when I need a little extra speed and O2 everywhere else |
| 19:43:59 | Silex | but usually I gain less than 2% :P |
| 19:44:03 | defunkt enters the room. | |
| 19:45:54 | scoopr | I took a bad habit writing -O9 everywhere in my dos days =) (that just puts -O3, but there ever happens to be -O4 it would use that, etc.) |
| 19:46:45 | Defiler | -Ox does something as well, right? |
| 19:46:46 | Silex | I think optimisation should be done by choosing the right design/algorithm and data structures instead of expecting some compiler magic :) |
| 19:47:27 | Defiler | Then the real world shows up at your door =( |
| 19:47:32 | Silex | \o/ |
| 19:47:59 | Silex | it's already amazing what -O2 does |
| 19:48:52 | scoopr | highlevel opts should definetly be design/algo stuff, but lower level optimization should definately be all compiler magic |
| 19:49:04 | Silex | scoopr: good point |
| 19:49:27 | Defiler | I think compiler magic needs to come last, but that doesn't mean it isn't worth looking at |
| 19:49:32 | langenberg | still talking about those GCC optimizations? |
| 19:50:35 | scoopr | on that note, what ever the optimization level, it should still yield a working program =) |
| 19:50:57 | scoopr | assuming at least somewhat conformant code |
| 19:53:36 | defunkt leaves the room. | |
| 19:56:06 | rue | Noo |
| 19:56:09 | rue | Wait yes |
| 19:56:12 | rue | No |
| 19:56:19 | rue | What? |
| 19:56:29 | langenberg | Does this VM makes a java-like performance possible? |
| 19:56:45 | Defiler | Eventually, yes |
| 19:57:02 | Defiler | We don't do JIT yet, though we (as of the last couple weeks) now have the infrastructure for it |
| 19:57:09 | defunkt enters the room. | |
| 19:57:14 | djwhitt | this VM + lots of smart people + lots of time I think ;) |
| 19:57:36 | ragnard leaves the room. | |
| 19:57:54 | Defiler | Well, sure.. that's how everything challenging works |
| 19:58:09 | djwhitt | yeah, not knocking it or anything |
| 19:58:24 | langenberg | Good to know the infrastructure is at least ready for it |
| 19:58:36 | langenberg | What steps need to be taken? |
| 19:59:02 | Defiler | Lots of analysis about when and where it makes sense to do JIT, to begin with |
| 19:59:48 | Defiler | Then wiring up a runtime bytecode -> machine code translation process |
| 20:00:10 | Defiler | After that, it just needs to swap in a new pointer in the SendSite that points to the native code, rather than the CompiledMethod object |
| 20:00:36 | Defiler | We have a ton of easier optimizations to do first, though, so I doubt JIT is anywhere on the horizon |
| 20:01:19 | thehcdreamer enters the room. | |
| 20:02:34 | langenberg | So if I understand this right, the actual code is compiled into bytecode. And with JIT, the actual assembler code is getting stored somewhere in the memory, waiting to be called? |
| 20:05:30 | ragnard enters the room. | |
| 20:05:58 | rue | Basically, 1) Think 2) Implement JIT 3) ??? 4) PROFIT! |
| 20:06:16 | therealadam | you forgot "chase skirt" |
| 20:06:32 | imajes enters the room. | |
| 20:08:44 | Defiler | langenberg: Not exactly.. compiling to native code is expensive |
| 20:08:55 | Defiler | langenberg: So most JIT systems only do it when a method is determined to be 'hot' enough |
| 20:09:11 | defunkt leaves the room. | |
| 20:09:13 | kofno_ enters the room. | |
| 20:09:44 | Defiler | langenberg: hence the 'just in time' (lame) terminology |
| 20:10:14 | kofno leaves the room. | |
| 20:16:16 | mae enters the room. | |
| 20:16:32 | cored enters the room. | |
| 20:23:03 | srbaker_ leaves the room. | |
| 20:24:26 | rue | Hmm? I thought it just meant that it compiles 'just in time' for execution |
| 20:24:37 | Defiler | Right |
| 20:25:05 | rue | Does not imply hotness in any way though--not that I dispute that that is probably how most systems handle it |
| 20:25:54 | Defiler | What I took langenberg's "So if I understand this" line to mean was that the assembly code was always there, waiting |
| 20:26:05 | Defiler | ..when in reality it does not get generated until 'just' before execution |
| 20:26:12 | Defiler | So that is what I was addressing |
| 20:29:57 | rue | Ah, sure |
| 20:30:38 | mae leaves the room. | |
| 20:33:23 | jeremydurham leaves the room. | |
| 20:37:49 | Silex | writing a VM sounds tough business |
| 20:38:18 | Defiler | Fun work if you can get it, though :) |
| 20:38:34 | langenberg leaves the room. | |
| 20:39:27 | ezmobius enters the room. | |
| 20:39:28 | Silex | well, you do everything from parsing ruby to "executing" it I guess ? |
| 20:39:56 | Defiler | Yep |
| 20:40:06 | Defiler | It is a big job. Ruby is a complex language. |
| 20:40:19 | Defiler | We would be so done by now if this were Lua. Hah |
| 20:41:12 | Silex | hehe |
| 20:41:32 | Silex | yeah I guess parsing is more or less the "easy" work that needs to be done |
| 20:41:37 | Silex | but once you have the trees all set |
| 20:41:44 | Defiler | parsing ruby is ridiculously hard, actually |
| 20:41:49 | Silex | I have no clue how you do for optimising what needs to be done |
| 20:41:55 | Defiler | We got this far because we were able to reuse the grammar file from ruby 1.8 |
| 20:42:01 | Silex | oh ? okay |
| 20:42:06 | Defiler | ..but zenspider is working on a pure-Ruby replacement |
| 20:42:13 | Defiler | So that we can change it more readily |
| 20:42:31 | Silex | I come from the c++ world and I always thougth c++'s syntax was ridiculously hard |
| 20:42:41 | Defiler | Yeah, C++ has some fun ambiguity |
| 20:42:42 | Silex | but I guess Ruby has so many little syntax hacks that it makes it difficult |
| 20:43:04 | Defiler | Ruby lets you do stuff like some_method(<<-ARGUMENT_ONE, <<-ARGUMENT_TWO) |
| 20:43:21 | Defiler | ..and then you have two back-to-back literal string blocks that end with ARGUMENT_ONE and ARGUMENT_TWO respectively |
| 20:43:27 | Defiler | whee |
| 20:43:32 | Silex | wow |
| 20:44:06 | Defiler | It is interesting working on a VM and compiler for a language that was designed to run on an interpreter |
| 20:44:22 | Defiler | I think Ruby is much more powerful and useful because of that heritage.. but it doesn't make writing a new implementation easy |
| 20:44:30 | drbrain | Defiler: "change it more readily" hahaha |
| 20:44:32 | Silex | I bet |
| 20:44:41 | Defiler | drbrain: Hey I had to think of a reason ha ha |
| 20:45:00 | Defiler | I wish I had had my awesome Robot Jox t-shirt in time for this interview |
| 20:45:07 | Defiler | We can live. We can both live. |
| 20:45:46 | zenspider | C++'s syntax IS ridiculously hard... but so is ruby's. :) |
| 20:48:00 | pauldix enters the room. | |
| 20:48:04 | rue | Uh-oh, he has risen! |
| 20:49:48 | thehcdreamer leaves the room. | |
| 20:56:16 | tbmcmullen | jeremydurham: much delayed thanks |
| 20:58:48 | cored leaves the room. | |
| 21:00:43 | lstoll leaves the room. | |
| 21:01:22 | cjheath enters the room. | |
| 21:01:26 | jessop enters the room. | |
| 21:07:52 | rue | Ugh, I wrote a book |
| 21:11:19 | VVSiz | whoa! :) |
| 21:11:21 | ezmobius | what did you write a book on? |
| 21:15:59 | mae enters the room. | |
| 21:16:19 | mae leaves the room. | |
| 21:16:34 | defunkt enters the room. | |
| 21:20:39 | twshelton leaves the room. | |
| 21:21:24 | rue | ezmobius: See your e-mail :P |
| 21:22:09 | headius_ enters the room. | |
| 21:22:17 | enebo_ enters the room. | |
| 21:22:36 | ezmobius | k |
| 21:26:16 | ezmobius | rue: agreed |
| 21:26:28 | mae enters the room. | |
| 21:27:35 | kofno_ leaves the room. | |
| 21:29:37 | rue | I should try to curtail my writing |
| 21:31:23 | ezmobius | nah its all good, i went on a bit myself |
| 21:31:28 | rue | Heh |
| 21:31:36 | rue | mod_scgi is a really nice implementation |
| 21:31:40 | ezmobius | i think we're on the right track though |
| 21:31:51 | ezmobius | yeah scgi is probably the right protocol to use |
| 21:31:58 | rue | Far cry from the mod_wsgi 10000 lines of code |
| 21:32:08 | ezmobius | yeah totally |
| 21:37:36 | enebo leaves the room. | |
| 21:37:37 | cremes | i'm trying to get an instance of rubuildius going for osx_ppc; anyone have a working pastie script that takes stdin? |
| 21:38:01 | headius leaves the room. | |
| 21:41:57 | rue | Alright, be back in a bit |
| 21:43:54 | mae leaves the room. | |
| 21:47:01 | agardiner enters the room. | |
| 21:47:27 | agardiner | morning |
| 21:47:43 | twshelton enters the room. | |
| 21:48:00 | brixen | morning agardiner |
| 21:48:07 | agardiner | heya brixen |
| 21:51:00 | antares enters the room. | |
| 21:53:05 | Arjen_ enters the room. | |
| 21:53:27 | mae enters the room. | |
| 22:00:15 | sholden leaves the room. | |
| 22:03:26 | ctennis enters the room. | |
| 22:06:05 | TheVoice leaves the room. | |
| 22:06:23 | hornbeck leaves the room. | |
| 22:06:24 | TheVoice enters the room. | |
| 22:09:07 | headius_ leaves the room. | |
| 22:09:42 | Fullmoon enters the room. | |
| 22:09:50 | headius enters the room. | |
| 22:12:43 | mae leaves the room. | |
| 22:17:43 | graza enters the room. | |
| 22:20:50 | gnufied enters the room. | |
| 22:22:28 | tbmcmullen leaves the room. | |
| 22:23:08 | tbmcmullen enters the room. | |
| 22:23:54 | crossblaim leaves the room. | |
| 22:26:09 | pauldix leaves the room. | |
| 22:31:40 | mae enters the room. | |
| 22:31:46 | antares | Defiler, ping |
| 22:39:43 | macournoyer leaves the room. | |
| 22:40:22 | graza_ enters the room. | |
| 22:40:31 | graza leaves the room. | |
| 22:41:28 | Fullmoon_ enters the room. | |
| 22:47:36 | defunkt_ enters the room. | |
| 22:47:39 | rue | graza_: Any luck with git? |
| 22:48:32 | defunkt leaves the room. | |
| 22:49:37 | antares | rue, hey mind reviewing a small shotgun doc patch? |
| 22:49:52 | antares | rue, I want to get some feedback before I move on |
| 22:49:58 | graza | i got somewhere eventually... i think it's ok now. |
| 22:50:32 | graza | removing the git compiled from the source tarball was a pain |
| 22:52:36 | dbussink | package management for the win! |
| 22:52:41 | graza | Macports was the way to go in the end |
| 22:52:41 | graza | unfortunately, now it won't build |
| 22:52:56 | dbussink | zenspider: you're into the parser / compiler, right? |
| 22:53:02 | mae leaves the room. | |
| 22:55:01 | moofbong leaves the room. | |
| 22:55:06 | twshelton leaves the room. | |
| 22:55:23 | pietia leaves the room. | |
| 22:56:46 | rue | antares: Ah, sure, post it up |
| 22:57:28 | mae enters the room. | |
| 22:57:36 | Fullmoon leaves the room. | |
| 22:58:04 | antares | rue, pastie or LH ticket, what do you prefer? it is basically not yet ready to be merged anyway... probably too much stuff is got wrong |
| 22:58:12 | graza | OS X PPC 10.4.11 build failure: http://pastie.org/153961 |
| 22:58:17 | rue | Just pastie it |
| 22:58:55 | lstoll enters the room. | |
| 22:58:59 | chris2 leaves the room. | |
| 22:59:38 | dbussink | graza_: does it work if you remove that line 29? |
| 23:01:14 | graza | it gets further... until it reaches another instance of the same struct |
| 23:01:16 | binary42 leaves the room. | |
| 23:02:58 | aotearoa enters the room. | |
| 23:03:15 | dbussink | graza_: does 10.4 provide ucontext.h ? |
| 23:03:17 | graza | here: http://pastie.org/153963 and so kept removing those until I got this: http://pastie.org/153965 |
| 23:03:34 | rubymaverick leaves the room. | |
| 23:03:51 | rubymaverick enters the room. | |
| 23:05:02 | dbussink | graza_: yeah, looks like a ucontext problem |
| 23:05:30 | dbussink | graza_: is there a file /usr/include/ucontext.h ? |
| 23:09:14 | Arjen | Huzzah! I think it compiled... :) |
| 23:10:09 | dbussink | Arjen_: any specific issues that should be documented? |
| 23:10:37 | Arjen | Don't know. I installed more packages then the documentation mentioned. |
| 23:11:10 | Arjen | rubinius 0.8.0 (ruby 1.8.6 compatible) (f43383a15) (02/17/2008) [i686-pc-linux-gnu] |
| 23:11:17 | Arjen | Sweet. |
| 23:11:22 | mae_ enters the room. | |
| 23:12:57 | mae_ leaves the room. | |
| 23:14:37 | antares | rue, http://pastie.org/private/yultm4bbcem9ywhve2e4lw |
| 23:14:42 | Arjen | dbussink, I don't think anyone with a little more linux experience can't work it out though. |
| 23:14:57 | mae leaves the room. | |
| 23:15:03 | antares |