Show enters and exits. Hide enters and exits.
| 00:10:00 | antares_ leaves the room. | |
| 00:15:31 | lopex leaves the room. | |
| 00:16:02 | headius enters the room. | |
| 00:26:24 | headius_ enters the room. | |
| 00:27:04 | headius_ leaves the room. | |
| 00:27:44 | headius_ enters the room. | |
| 00:27:45 | headius leaves the room. | |
| 00:36:41 | headius enters the room. | |
| 00:37:27 | imajes enters the room. | |
| 00:38:13 | headius_ leaves the room. | |
| 00:50:37 | rue leaves the room. | |
| 00:51:05 | headius leaves the room. | |
| 00:56:50 | rue enters the room. | |
| 01:09:16 | dysinger leaves the room. | |
| 01:17:53 | ezmobius leaves the room. | |
| 01:18:18 | ezmobius enters the room. | |
| 01:23:04 | wmoxam enters the room. | |
| 01:25:29 | Fullmoon leaves the room. | |
| 01:29:41 | benburkert leaves the room. | |
| 01:40:25 | benburkert enters the room. | |
| 01:50:38 | rue leaves the room. | |
| 01:51:20 | rue enters the room. | |
| 01:54:03 | imajes leaves the room. | |
| 01:56:43 | ezmobius leaves the room. | |
| 01:59:05 | evan | ok |
| 01:59:10 | evan | i've pushed the LLVM work into the cpp branch |
| 01:59:20 | evan | you need to install LLVM first |
| 02:09:03 | jero5 enters the room. | |
| 02:11:33 | benburkert leaves the room. | |
| 02:22:08 | mutle_ leaves the room. | |
| 02:25:14 | VVSiz_ enters the room. | |
| 02:43:16 | VVSiz leaves the room. | |
| 02:51:46 | benburkert enters the room. | |
| 02:55:14 | mutle enters the room. | |
| 02:59:41 | wyhaines_ enters the room. | |
| 03:18:14 | wyhaines leaves the room. | |
| 03:22:48 | wyhaines_ leaves the room. | |
| 03:26:45 | benburkert leaves the room. | |
| 03:50:43 | rue leaves the room. | |
| 03:51:56 | rue enters the room. | |
| 04:14:29 | wmoxam leaves the room. | |
| 04:22:57 | ryanlowe leaves the room. | |
| 04:39:54 | wyhaines_ enters the room. | |
| 04:50:47 | rue leaves the room. | |
| 04:51:26 | rue enters the room. | |
| 04:52:34 | benny leaves the room. | |
| 04:55:08 | headius enters the room. | |
| 05:01:46 | headius leaves the room. | |
| 05:01:59 | headius enters the room. | |
| 05:03:12 | benburkert enters the room. | |
| 05:04:42 | headius leaves the room. | |
| 05:16:26 | seydar enters the room. | |
| 05:18:20 | headius enters the room. | |
| 05:20:20 | ezmobius enters the room. | |
| 05:28:40 | wmoxam enters the room. | |
| 05:37:42 | enebo enters the room. | |
| 05:37:45 | enebo leaves the room. | |
| 05:38:06 | enebo enters the room. | |
| 05:42:36 | headius leaves the room. | |
| 05:42:45 | jbarnette enters the room. | |
| 05:50:25 | rue leaves the room. | |
| 05:50:58 | rue enters the room. | |
| 06:04:37 | seydar leaves the room. | |
| 06:13:28 | blakewatters enters the room. | |
| 06:27:33 | binary42 enters the room. | |
| 06:29:16 | enebo_ enters the room. | |
| 06:29:17 | enebo leaves the room. | |
| 06:30:54 | evan | dbussink: you around? |
| 06:32:20 | blakewatters leaves the room. | |
| 06:34:17 | ezmobius | evan: nice changeset ;) |
| 06:34:25 | evan | :D |
| 06:34:30 | evan | it's rough still |
| 06:34:37 | evan | but it gives people something to play with |
| 06:34:44 | ezmobius | coolness |
| 06:48:51 | evan | been thinking about ways to improve it today |
| 06:50:25 | ezmobius | looks pretty clean for c++ code |
| 06:50:25 | rue leaves the room. | |
| 06:50:33 | evan | been trying to keep it that way |
| 06:51:29 | rue enters the room. | |
| 06:51:38 | ezmobius | you flying to portland on monday? |
| 06:51:59 | evan | yep |
| 06:52:07 | evan | seriously early. |
| 06:52:12 | evan | like 6:15am |
| 06:52:13 | ezmobius | :/ |
| 06:52:23 | evan | i'm making a habit of it :) |
| 06:52:26 | ezmobius | ;) |
| 06:52:33 | ezmobius | gotta walk the dogs, bbiab |
| 07:00:12 | binary42 leaves the room. | |
| 07:00:21 | benburkert leaves the room. | |
| 07:17:29 | wmoxam leaves the room. | |
| 07:23:10 | jbarnette | gah, autotest is borked again |
| 07:26:37 | jbarnette | ...and git bisect claims that it never worked. awesome |
| 07:29:36 | Arjen_ leaves the room. | |
| 07:45:08 | enebo_ leaves the room. | |
| 07:45:46 | wyhaines leaves the room. | |
| 07:50:15 | rue leaves the room. | |
| 07:50:59 | rue enters the room. | |
| 07:57:08 | headius enters the room. | |
| 07:58:16 | headius leaves the room. | |
| 08:01:09 | gnufied leaves the room. | |
| 08:02:10 | Maledictus enters the room. | |
| 08:02:34 | ezmobius_ enters the room. | |
| 08:24:05 | qwert666 enters the room. | |
| 08:32:09 | qertoip enters the room. | |
| 08:33:12 | sholden enters the room. | |
| 08:42:01 | VVSiz | whooops |
| 08:42:27 | VVSiz | rubinius just killed my Linux box (kernel panic due to insufficient memory) |
| 08:43:31 | sholden_ enters the room. | |
| 08:44:09 | sholden leaves the room. | |
| 08:46:15 | twbray enters the room. | |
| 08:48:11 | VVSiz | OK, don't run this with rubinius or your system is going down! ;) at least mine does |
| 08:48:18 | VVSiz | http://pastie.org/228454 |
| 08:49:49 | jbarnette | VVSiz: that's a hefty chunk of strings there :) |
| 08:49:59 | VVSiz | yeah |
| 08:50:02 | rue leaves the room. | |
| 08:51:14 | VVSiz | MRI can handle that but consumes quite some memory |
| 08:51:20 | VVSiz | (up to 1 gig) |
| 08:51:25 | sholden_ leaves the room. | |
| 08:51:29 | rue enters the room. | |
| 08:51:35 | VVSiz | Jruby handles with ease (less than 20mb needed) |
| 08:51:52 | VVSiz | rbx consumes entire 2 gigs of memory available to my system and then it goes down |
| 08:52:20 | ezmobius | this makes rbx consume all system memory in a few seconds too ;) http://pastie.org/228457 |
| 08:52:41 | jbarnette | ezmobius_: that's just plain mean |
| 08:52:45 | VVSiz | heh :) |
| 08:52:51 | krsh enters the room. | |
| 08:53:11 | VVSiz | but check the output from MRI on that pastie I gave above!! Funnnny |
| 08:53:28 | VVSiz | http://pastie.org/228459 |
| 08:53:48 | VVSiz | and it *does* respond to all those weird methods ;) |
| 08:56:00 | VVSiz | http://pastie.org/228462 -- even more evil code. It SIGSEGVs MRI, Crashes JVM(!), kills linux box if run with rbx :) |
| 08:56:13 | ezmobius | haha |
| 08:56:37 | ezmobius | wow it even kills the jvm |
| 08:57:44 | VVSiz | yeah, with sacry output |
| 08:57:44 | VVSiz | http://pastie.org/228463 |
| 09:03:49 | jbarnette | VVSiz: my laptop fan hates you right now |
| 09:03:57 | VVSiz | :) |
| 09:04:12 | VVSiz | my linux box fills the same (after 3 reboots) |
| 09:04:21 | VVSiz | feeeeeels |
| 09:10:03 | jbarnette leaves the room. | |
| 09:28:24 | radarek enters the room. | |
| 09:36:31 | twbray leaves the room. | |
| 09:42:57 | NoKarma enters the room. | |
| 09:49:58 | rue leaves the room. | |
| 09:56:01 | rue enters the room. | |
| 10:21:04 | yugui enters the room. | |
| 10:27:19 | rue | Morning |
| 10:27:24 | Yurik leaves the room. | |
| 10:31:15 | thehcdreamer enters the room. | |
| 10:32:08 | benny enters the room. | |
| 10:33:44 | antares enters the room. | |
| 10:45:37 | krsh leaves the room. | |
| 10:49:46 | rue leaves the room. | |
| 10:50:36 | rue enters the room. | |
| 10:52:12 | antares leaves the room. | |
| 10:54:58 | NoKarma | rue: morning |
| 10:57:03 | octopod enters the room. | |
| 10:57:21 | ezmobius_ leaves the room. | |
| 10:57:22 | ezmobius leaves the room. | |
| 11:00:15 | qertoip | On clean clone, I invoke 'rake spec'. There are 27 failures and 1 error. Is that ok? |
| 11:03:21 | dbussink | qertoip: what platform are you on? |
| 11:04:21 | dbussink | rue: evan's push almost didn't touch the rakefile, so it's still there to recover from insanity ;) |
| 11:05:30 | qertoip | Ubuntu |
| 11:05:51 | qertoip | Intel Core Duo 32 bit |
| 11:06:02 | dbussink | qertoip: can you pastie the problems? |
| 11:08:04 | qertoip | http://pastie.org/228490 |
| 11:09:58 | qertoip | Perhaps it has something to do with the way I have cloned the repository. I cloned as root. |
| 11:10:52 | thehcdreamer leaves the room. | |
| 11:13:34 | antares_ enters the room. | |
| 11:18:54 | michalw enters the room. | |
| 11:19:39 | qwert666 leaves the room. | |
| 11:26:45 | dbussink | qertoip: i think that explains then :) |
| 11:26:55 | dbussink | qertoip: why did you clone as root then? |
| 11:28:00 | rue | Only daffodils clone as root :) |
| 11:28:23 | dbussink | rue: already axing the rakefile? :P |
| 11:28:46 | dbussink | Defiler was eager to do so too |
| 11:29:20 | Defiler | it needs some axin' |
| 11:29:37 | qertoip | Well, it was an accident ;) |
| 11:29:42 | dbussink | Defiler: i actually just got out of bed :P |
| 11:30:07 | dbussink | those late train trips are not really invigorating |
| 11:31:20 | Defiler | No doubt |
| 11:32:48 | rue | I just created :build and :build_tests tasks locally, but it could use a bit more radical uprooting, I suppose |
| 11:33:48 | Defiler | Yeah, it also needs a generate_glue task |
| 11:40:04 | mitchellvriley__ enters the room. | |
| 11:40:22 | rue | breed_horses |
| 11:44:52 | rue | I broke down and ordered one of those mobile broadband thingies. 2Mib connection with unlimited data came to ~25e |
| 11:46:29 | Defiler | worth it, in my opinion |
| 11:46:35 | Defiler | I use mine all the time |
| 11:46:56 | rue | Yeah, should be nice |
| 11:47:01 | dbussink | yeah, really easy in the train too |
| 11:47:25 | Defiler | Of course mine cost well over 50% more than that since in the US only rich people are allowed to have data =( |
| 11:47:27 | dbussink | rue: sounds a lot like the 3g iphone subscription conditions here :p |
| 11:48:55 | rue | Defiler: Heh, I had forgotten what it was like in the civilized world ;) |
| 11:49:48 | rue leaves the room. | |
| 11:51:07 | rue enters the room. | |
| 11:51:40 | scoopr | seems to work really well ;) |
| 11:51:56 | dbussink | rue: where are you now then? |
| 11:51:57 | antares enters the room. | |
| 11:54:13 | yasuhito enters the room. | |
| 11:54:54 | qertoip | Bad news: after clean clone + rake build as ordinary (non-root) user, the problem is still present. |
| 11:54:58 | qertoip | http://pastie.org/228496 |
| 11:55:18 | qertoip | There are 24 errors: |
| 11:55:21 | qertoip | 1301 files, 5685 examples, 20957 expectations, 0 failures, 24 errors |
| 11:55:32 | dbussink | all related to access rights? |
| 11:55:33 | qertoip | (after running rake spec) |
| 11:55:47 | dbussink | be sure to check whether there is not some stale stuff in /tmp |
| 11:56:06 | rue leaves the room. | |
| 11:56:16 | qertoip | dbussing: yes, all related to access rights |
| 11:56:40 | rue enters the room. | |
| 11:57:14 | thehcdreamer enters the room. | |
| 11:59:27 | qertoip | dbussing: thanks, cleaning /tmp helped! |
| 12:07:32 | botanicus enters the room. | |
| 12:10:54 | dbussink | qertoip: not using nickname auto completion? :) |
| 12:12:36 | rue | dbussink: hel.fi |
| 12:13:19 | dbussink | rue: ah, so now i can expect you to not be here during my day time hours? ;) |
| 12:13:30 | rue | scoopr: This is just Elisa's ass-backwards DHCP nonsense. Yet to figure out how to work around it |
| 12:13:34 | rue | dbussink: Hehe |
| 12:13:44 | rue | I have actually tried to normalise my hours somewhat |
| 12:18:09 | dbussink | ah, that's a good thing then :) |
| 12:19:20 | olabini enters the room. | |
| 12:21:20 | thehcdreamer leaves the room. | |
| 12:23:22 | qertoip | dbussink: thanks for a tip :) |
| 12:23:46 | thehcdreamer enters the room. | |
| 12:24:35 | dbussink | qertoip: everything to minimize typing :) |
| 12:26:48 | lopex enters the room. | |
| 12:41:32 | mitchellvriley leaves the room. | |
| 12:48:50 | botanicus leaves the room. | |
| 12:50:50 | lopex leaves the room. | |
| 12:52:29 | olabini leaves the room. | |
| 12:54:19 | octopod leaves the room. | |
| 13:06:10 | chris2 enters the room. | |
| 13:07:36 | botanicus enters the room. | |
| 13:37:45 | botanicus leaves the room. | |
| 13:43:21 | neelance enters the room. | |
| 13:43:50 | neelance | hi there |
| 13:45:34 | botanicus enters the room. | |
| 13:48:14 | dbussink | afternoon |
| 13:48:37 | neelance | it is alive! xD |
| 13:48:57 | neelance | I just got the first ruby code running on the Wii ^^ |
| 13:49:15 | dbussink | ah, cool :) |
| 13:49:39 | botanicus leaves the room. | |
| 13:50:07 | botanicus enters the room. | |
| 13:50:26 | dbussink | how hard did it turn out to be then? |
| 13:51:19 | neelance | is wasn't easy but i think the good code structure of rubinius made it possible |
| 13:54:00 | botanicus leaves the room. | |
| 13:55:33 | dbussink | ah, that's good to know |
| 13:55:41 | dbussink | how did you work around the libev issues then> |
| 13:55:42 | dbussink | ? |
| 13:56:11 | neelance | i completely removed libev for now |
| 13:56:11 | rue leaves the room. | |
| 13:56:24 | neelance | so no threading atm |
| 13:56:50 | neelance | but theoretically it should be possible to also port libev |
| 13:57:05 | neelance | but i decided to leave it out at first |
| 13:57:10 | rue enters the room. | |
| 13:57:49 | Arjen_ enters the room. | |
| 14:01:03 | dbussink | ah, but cool that the code actually runs :) |
| 14:11:54 | neelance | yeah, funny thing to be happy to see a 7 on the screen (my test case was -e "puts 3 + 4" ^^) |
| 14:13:39 | neelance | but also every more complex ruby code works |
| 14:16:08 | dbussink | hehe, only things depending on libev don't work then i guess |
| 14:16:10 | dbussink | ? |
| 14:30:37 | sholden enters the room. | |
| 14:34:34 | Yurik enters the room. | |
| 14:35:28 | Yurik leaves the room. | |
| 14:42:50 | sholden leaves the room. | |
| 14:43:04 | sholden enters the room. | |
| 14:44:35 | michalw leaves the room. | |
| 14:44:54 | michalw enters the room. | |
| 14:53:00 | Defiler | neelance: You have Rubinius running on the Wii? |
| 14:55:52 | rue leaves the room. | |
| 14:56:40 | rue enters the room. | |
| 14:57:24 | neelance | yep |
| 14:59:06 | neelance | dbussink: there are also no dynamic libraries so if you like to use extensions you have to somehow include them at compile time (haven't tried yet) |
| 14:59:29 | dbussink | neelance: ah, true of course |
| 15:00:37 | neelance | Defiler: yep |
| 15:01:34 | robin_dewd enters the room. | |
| 15:05:27 | sholden leaves the room. | |
| 15:05:30 | sholden enters the room. | |
| 15:08:58 | Defiler | neelance: That is pretty awesome |
| 15:09:26 | Defiler | Do you have a blog post about it? if not, you should :) |
| 15:20:51 | mutle_ enters the room. | |
| 15:23:23 | mutle__ enters the room. | |
| 15:28:13 | mutle leaves the room. | |
| 15:32:30 | yugui leaves the room. | |
| 15:34:51 | gnufied enters the room. | |
| 15:36:17 | octopod enters the room. | |
| 15:36:57 | mutle enters the room. | |
| 15:38:48 | mutle_ leaves the room. | |
| 15:39:16 | rue | neelance: Nice work |
| 15:39:49 | rue | neelance: Do you have a fork on Github for it or anything? |
| 15:41:19 | dbussink | if there are changes that make rubinius more cross platform, we would welcome those in the main repo :) |
| 15:42:40 | yasuhito leaves the room. | |
| 15:44:57 | rue | What, we do? |
| 15:45:07 | dbussink | of course :P |
| 15:45:23 | dbussink | if it doesn't intervene with other stuff though |
| 15:45:34 | robin_dewd leaves the room. | |
| 15:48:39 | robin_dewd enters the room. | |
| 15:53:28 | binary42 enters the room. | |
| 15:54:16 | binary42_ enters the room. | |
| 15:54:17 | binary42 leaves the room. | |
| 15:57:37 | sholden_ enters the room. | |
| 15:57:51 | twbray enters the room. | |
| 15:58:32 | mutle__ leaves the room. | |
| 15:58:46 | michalw leaves the room. | |
| 15:58:51 | sholden leaves the room. | |
| 16:02:20 | twbray leaves the room. | |
| 16:05:30 | benburkert enters the room. | |
| 16:09:08 | neelance | back again |
| 16:09:35 | neelance | i will take a look a github now (only used my local git repo atm) |
| 16:09:37 | twbray enters the room. | |
| 16:11:03 | twbray leaves the room. | |
| 16:13:47 | dbussink | neelance: did you have to change a lot to get it running? |
| 16:14:13 | rue | neelance: Yeah, you can make it a "fork" there or just push it in standalone |
| 16:15:02 | NoKarma | rue: making a fork is preferred by the github guys, as it wastes less space :) |
| 16:15:19 | twbray enters the room. | |
| 16:15:50 | twbray leaves the room. | |
| 16:16:28 | rue | Psham |
| 16:29:18 | Fullmoon enters the room. | |
| 16:34:55 | sholden_ leaves the room. | |
| 16:36:26 | wmoxam enters the room. | |
| 17:02:01 | radarek leaves the room. | |
| 17:16:58 | evan | morning |
| 17:18:45 | antares leaves the room. | |
| 17:19:14 | headius enters the room. | |
| 17:23:53 | dbussink | evan: howdy |
| 17:24:05 | dbussink | i was wondering, what version of llvm does it depend on? |
| 17:24:08 | dbussink | 2.3? |
| 17:24:28 | dbussink | cause i have 2.2 installed through macports and it complains about certain headers that seem to be missing |
| 17:25:48 | evan | hm |
| 17:25:50 | evan | i used 2.3 |
| 17:25:50 | evan | yeah |
| 17:25:56 | evan | what headers doesn't it like? |
| 17:26:00 | evan | ah crap. |
| 17:26:07 | dbussink | StandardPasses.h isn't there |
| 17:26:08 | dbussink | in 2.2. |
| 17:26:14 | evan | ack! |
| 17:26:18 | evan | it's not in 2.3 either |
| 17:26:22 | evan | it's my custom addition |
| 17:26:25 | evan | remove it |
| 17:26:28 | evan | i'm not using it anymore anyway |
| 17:26:54 | Yurik enters the room. | |
| 17:27:47 | dbussink | on to the next one ;) |
| 17:27:47 | dbussink | error: ‘UnwindTablesMandatory’ is not a member of ‘llvm’ |
| 17:28:08 | evan | that must be 2.2 versus 2.3 |
| 17:28:11 | evan | comment that line out |
| 17:29:38 | dbussink | evan: http://pastie.org/228562 |
| 17:29:49 | headius | evan: what do you use as a target for break-in-closure |
| 17:30:07 | headius | to know where it gets back to |
| 17:30:14 | evan | currently, a pre-allocated exception |
| 17:30:43 | evan | saved in a special local slot called @lre |
| 17:30:48 | headius | so the surrounding methodcontext has a pre-allocated exception object, and it just bubbles out until they match? |
| 17:30:55 | evan | 'special' in that it has a name that can't be accessed |
| 17:31:16 | headius | ok, I remember discussing this before |
| 17:31:22 | headius | I'm rewiring our break logic a bit |
| 17:31:27 | evan | well, yeah, the methodcontext has in so far as the MC has locals |
| 17:31:36 | gnufied | headius, can i have a moment? |
| 17:31:38 | evan | it's not wired in a magic internal field of an MC |
| 17:31:47 | headius leaves the room. | |
| 17:31:51 | gnufied | oops |
| 17:31:51 | evan | dbussink: hrm. |
| 17:31:59 | evan | dbussink: guess you need 2.3 |
| 17:32:00 | evan | :) |
| 17:32:07 | headius enters the room. | |
| 17:32:24 | dbussink | is there a better way to install it except download and compile it oneself? |
| 17:32:30 | evan | no |
| 17:32:33 | evan | thats how I did it. |
| 17:32:51 | evan | if we need to, I can pull llvm into the external_libs build process |
| 17:33:04 | evan | i didn't because it's quite a bit bigger than the others |
| 17:33:08 | evan | and takes a while to compile |
| 17:33:16 | headius | oops |
| 17:33:18 | headius | flaky cabin internet |
| 17:33:20 | headius | ok, I get it |
| 17:33:22 | headius | thanks |
| 17:33:37 | evan | headius: np. |
| 17:34:23 | benny leaves the room. | |
| 17:34:28 | evan | headius: the pre-allocated exception is raised, then checked against in exception handling code just outside the call |
| 17:34:46 | neelance leaves the room. | |
| 17:35:06 | headius | got it...seems like about the same approach as here |
| 17:35:26 | headius | except I'll be just attaching a marker payload to the exception |
| 17:35:26 | Fullmoon leaves the room. | |
| 17:37:43 | evan | headius: gotcha |
| 17:38:01 | evan | headius: the one thing that needs to be done is knowing when to emit the pre-allocated exception and rescue code |
| 17:38:06 | evan | it happens too often now |
| 17:38:12 | evan | and imposes an un-needed burdon |
| 17:38:22 | headius | right |
| 17:38:34 | headius | in our case I'll probably just use the frame/method context itself as the target |
| 17:38:37 | dbussink | evan: i was wondering, stuff like meta_send_op_plus is an optimization isn't it? |
| 17:38:39 | headius | since it's allocated anyway |
| 17:38:42 | dbussink | or how should i read that? |
| 17:38:55 | evan | dbussink: yeah, it's an optimization |
| 17:39:19 | evan | headius: yeah, i had a reason for not using the MC directly before... but i don't recall what it was... |
| 17:39:30 | headius | well, perhaps I'll find out :) |
| 17:39:34 | dbussink | basically because it eliminates method dispatching is it? |
| 17:39:44 | evan | headius: even so, i wonder if break can be detected staticly |
| 17:39:54 | evan | to know whether to insert the rescue code around the call |
| 17:40:12 | evan | dbussink: exactly |
| 17:40:27 | evan | dbussink: it's a very specialized version of 'def +' that knows how to handle Fixnum's builtin |
| 17:40:27 | headius | I only insert it if there's a block attached to a call |
| 17:40:35 | evan | right, me too. |
| 17:40:36 | evan | but still |
| 17:40:38 | dbussink | couldn't large parts be auto generated like that? |
| 17:40:45 | headius | you're saying if there's no break in the block |
| 17:40:56 | headius | I've thought about doing that too |
| 17:41:04 | evan | headius: right. being able to see "into" the block to see if a break was used |
| 17:41:05 | headius | with an obvious fallback for eval |
| 17:41:08 | evan | and thus if you need to emit the code |
| 17:41:19 | headius | I already do some of that inspection, but I stop at the block right now |
| 17:41:22 | evan | well, if eval is thrown in, i guess you have to leave it |
| 17:41:27 | headius | right |
| 17:41:33 | headius | eval is always a full deopt for us |
| 17:41:49 | evan | that eval("break") works I consider to be a bug. :) |
| 17:42:19 | evan | dbussink: not sure what ya mean |
| 17:42:56 | headius | evan: quite |
| 17:43:09 | dbussink | well, because it seems now like there is only one for adding values |
| 17:43:30 | dbussink | but what about subtraction, and others etc. |
| 17:43:41 | dbussink | i kinda miss where you would put the line |
| 17:43:57 | evan | oh oh |
| 17:44:02 | evan | where you "draw the line" |
| 17:44:03 | evan | gotcha |
| 17:44:06 | evan | agreed |
| 17:44:14 | dbussink | draw the line yeah |
| 17:44:19 | Fullmoon enters the room. | |
| 17:44:50 | evan | well, it's drawn just where we think it should be :) |
| 17:44:58 | evan | ie, having those speeds up the system nicely |
| 17:45:09 | evan | agreed though, it's not a solution that scales. |
| 17:45:18 | dbussink | hence the question of auto generating it for some things perhaps |
| 17:45:26 | evan | nah. |
| 17:45:37 | evan | because what would you be auto generating? instructions? |
| 17:45:54 | evan | the better solution is, now that we've got LLVM, is to go the direction of IronRuby on this |
| 17:46:06 | dbussink | what exactly did they do? |
| 17:46:16 | evan | basically, generate stubs based on learned types |
| 17:46:42 | evan | they'd try to dispatch a method quickly, and let some code patch the sendsite |
| 17:46:47 | evan | similar to how we have now |
| 17:46:54 | evan | but the sites can be patched with any code you want |
| 17:47:13 | evan | so you just teach the system how to generate math code for Fixnums |
| 17:47:25 | evan | and let the system generate the small stubs as needed |
| 17:47:28 | evan | they call that... |
| 17:47:29 | evan | um... |
| 17:47:29 | dbussink | so basically patch in the current primitive implementation into the sendsite? |
| 17:47:36 | evan | oh, thats builtin into the DLR |
| 17:47:36 | dbussink | roughly speaking |
| 17:47:41 | evan | thats basically the 'core' of the DLR |
| 17:47:53 | evan | dbussink: yeah, sorta. |
| 17:48:14 | evan | saw you're going to send method :+ to 3, with an argument of 4 |
| 17:48:37 | evan | the first time, the sendsite automatically misses, so it lets the system fill it in |
| 17:48:49 | evan | some code detects it's a math method with a Fixnum |
| 17:49:22 | evan | it then writes a LLVM function that does basically what meta_send_op_plus does |
| 17:49:30 | evan | and install that function into the sendsite |
| 17:49:41 | evan | so the next time, this faster version is used |
| 17:49:45 | dbussink | cool |
| 17:49:47 | evan | doing that is a lot more generic |
| 17:49:50 | evan | but more heavy weight |
| 17:50:03 | dbussink | seems like a smart thing to do ;) |
| 17:50:07 | evan | yeah |
| 17:50:11 | evan | it can be done smartly too |
| 17:50:25 | evan | you could use the same function over and over again in multiple sendsites |
| 17:50:27 | evan | for Fixnum math |
| 17:50:42 | evan | so you don't end up having 1000 versions of the same LLVM function in memory |
| 17:50:52 | evan | that just try and add 2 numbers and fail otherwise |
| 17:51:37 | dbussink | yeah, you don't want to have that explode |
| 17:51:44 | dbussink | on memory usage, did you see those numbers? |
| 17:51:52 | evan | the meta_'s could all be replaced, soon, with sendsites that use a pre-populated initial case |
| 17:52:48 | robin_dewd leaves the room. | |
| 17:53:53 | evan | i think last night, i figured out a way to generate LLVM functions that can optionally allocate a context |
| 17:54:00 | evan | so that accessors and such are very cheap |
| 17:55:34 | rue leaves the room. | |
| 17:55:43 | dbussink | evan: 2.3 indeed fixed it |
| 17:56:12 | rue enters the room. | |
| 17:58:38 | twbray enters the room. | |
| 18:06:19 | evan | well, I think i'm going to watch some tennis |
| 18:20:20 | anonuser leaves the room. | |
| 18:20:22 | twbray leaves the room. | |
| 18:28:53 | rubuildius_amd64 leaves the room. | |
| 18:30:01 | rubuildius_amd64 enters the room. | |
| 18:33:59 | dbussink | evan: are the tests failing for you too? |
| 18:50:41 | wyhaines enters the room. | |
| 18:51:27 | imajes enters the room. | |
| 19:00:02 | twbray enters the room. | |
| 19:01:17 | twbray leaves the room. | |
| 19:01:57 | imajes leaves the room. | |
| 19:01:58 | cremes leaves the room. | |
| 19:10:53 | twbray enters the room. | |
| 19:18:14 | twbray leaves the room. | |
| 19:26:21 | radarek enters the room. | |
| 19:27:43 | lopex enters the room. | |
| 19:37:23 | gnufied leaves the room. | |
| 19:42:00 | cremes enters the room. | |
| 19:44:49 | twbray enters the room. | |
| 19:50:04 | twbray leaves the room. | |
| 19:51:12 | roo enters the room. | |
| 19:51:31 | headius leaves the room. | |
| 20:03:37 | benny enters the room. | |
| 20:05:30 | roo | Ah, the LLVM issue was already addressed |
| 20:05:48 | rue leaves the room. | |
| 20:11:48 | weepy enters the room. | |
| 20:11:52 | weepy | lo |
| 20:11:58 | rue | Hallo |
| 20:12:04 | weepy | I just wondered how far off the C++ VM was ? ? ? : ) |
| 20:12:26 | chris2 leaves the room. | |
| 20:14:32 | weepy | or perhaps its a secret ? |
| 20:16:24 | rue | Hard to say in any exact way but it is fairly complete. You can take a look at the cpp branch |
| 20:17:29 | weepy | ah kk - thanks - i'm just a lurker really - but very interested in the outcome - Rubinius seemed to go very quiet after Railsconf |
| 20:21:22 | rue | I told them to do it nude but did they listen to me? No |
| 20:22:14 | headius enters the room. | |
| 20:22:40 | weepy | nude |
| 20:22:41 | weepy | lol |
| 20:23:12 | weepy | would have been a pretty surprising talk |
| 20:37:22 | enebo enters the room. | |
| 20:39:27 | Defiler | weepy: It is coming along. A couple of weeks maybe? |
| 20:39:34 | Defiler | weepy: Definitely before 1.0 |
| 20:39:54 | weepy | awesome. i understand that it will be using LLVM ? |
| 20:40:09 | Defiler | Yep |
| 20:40:17 | Defiler | Though initially it will only be using it lightly |
| 20:40:29 | Defiler | Eventually we want to use it to replace libffi as well |
| 20:40:54 | Defiler | The 'cpp' branch has the LLVM code in it now, if you want to look directly |
| 20:41:01 | Defiler | Or, some LLVM code :) |
| 20:41:08 | dbussink | tests are broken though |
| 20:41:19 | dbussink | i did fix one case, but i don't know if it's a correct fix |
| 20:41:22 | dbussink | evan: back? |
| 20:45:43 | rue | dbussink: Which problem were you seeing? |
| 20:46:08 | dbussink | rue: did you already try running the tests? |
| 20:46:15 | rue | No, I am not at the house |
| 20:46:40 | rue | I only had time to fail with LLVM 2.2 and start building 2.3 :) |
| 20:46:58 | dbussink | ah ok, well, let me reproduce then |
| 20:48:10 | dbussink | http://pastie.org/228618 |
| 20:49:13 | dbussink | i've been able to fix that one, but now i have one that eludes me |
| 20:49:57 | dbussink | for who's interested: http://pastie.org/228620 |
| 20:52:34 | Defiler | That looks like a fun one |
| 20:53:15 | rue | Yep.. I will be home in a bit so I can shake a GDB at it |
| 20:53:15 | Yurik leaves the room. | |
| 20:53:54 | AndrewO enters the room. | |
| 20:56:34 | rue | Ah, `git pull --track` is handy |
| 20:56:46 | rue | Erm, `git checkout --track` even |
| 20:57:17 | Defiler | I'm pretty sure it is the default now |
| 20:57:33 | Defiler | when you do, say, git checkout -b local_foo origin/foo |
| 20:59:14 | rue | Is it? Coo' |
| 20:59:45 | Defiler | I think it is as of 1.5.3 or 1.5.4 |
| 21:01:50 | dbussink | Defiler: ready for the trip back? |
| 21:02:26 | Defiler | Yeah. I'm starting to look forward to being at home again |
| 21:02:30 | Defiler | though it has been a lot of fun |
| 21:03:11 | Defiler | but generally day 10 is when I really start to feel ready to end a trip, and this is day 11 |
| 21:03:34 | dbussink | well, day 10 was unline the others ;) |
| 21:03:38 | dbussink | unlike |
| 21:04:02 | Defiler | True :) |
| 21:05:54 | rue leaves the room. | |
| 21:11:23 | enebo leaves the room. | |
| 21:11:45 | rue enters the room. | |
| 21:15:15 | qertoip leaves the room. | |
| 21:25:12 | qrush enters the room. | |
| 21:32:46 | thehcdreamer leaves the room. | |
| 21:43:35 | antares_ leaves the room. | |
| 21:46:54 | benburkert leaves the room. | |
| 21:47:59 | imajes enters the room. | |
| 21:52:25 | twbray enters the room. | |
| 22:05:03 | rue leaves the room. | |
| 22:06:16 | rue enters the room. | |
| 22:07:13 | weepy leaves the room. | |
| 22:12:20 | octopod leaves the room. | |
| 22:17:43 | wmoxam leaves the room. | |
| 22:21:37 | ezmobius_ enters the room. | |
| 22:22:00 | enebo enters the room. | |
| 22:23:18 | wmoxam enters the room. | |
| 22:34:01 | benburkert enters the room. | |
| 22:43:33 | imajes leaves the room. | |
| 22:53:22 | ryanlowe enters the room. | |
| 22:53:24 | Fullmoon leaves the room. | |
| 22:56:38 | twbray leaves the room. | |
| 22:59:52 | binary42 enters the room. | |
| 23:01:32 | benburkert leaves the room. | |
| 23:04:07 | neelance enters the room. | |
| 23:04:57 | rue leaves the room. | |
| 23:09:23 | blakewatters enters the room. | |
| 23:10:48 | rue enters the room. | |
| 23:13:02 | binary42_ leaves the room. | |
| 23:15:20 | rue | Hm, this error is a bit worrisome |
| 23:15:45 | michalw enters the room. | |
| 23:20:41 | nexcastellan leaves the room. | |
| 23:22:16 | Fobax leaves the room. | |
| 23:23:07 | NoKarma | rue: what's up? |
| 23:26:04 | chopmo enters the room. | |
| 23:27:08 | chopmo leaves the room. | |
| 23:34:55 | radarek leaves the room. | |
| 23:35:20 | rue | There is an uninitialised variable being used in the cpp branch :P |
| 23:36:13 | Arjen_ leaves the room. | |
| 23:36:14 | rue | Fixeded |
| 23:41:10 | NoKarma | heh |
| 23:43:36 | twbray enters the room. | |
| 23:48:58 | evan | i'm back |
| 23:49:05 | lopex leaves the room. | |
| 23:49:28 | evan | yeah, the tests blow up currently. |
| 23:51:39 | benburkert enters the room. | |
| 23:52:03 | headius leaves the room. | |
| 23:52:05 | enebo leaves the room. | |
| 23:52:23 | shame enters the room. | |
| 23:52:43 | michalw leaves the room. | |
| 23:52:43 | enebo enters the room. | |
| 23:53:02 | headius enters the room. | |
| 23:56:43 | benburkert leaves the room. |