|
github.com/moarvm/moarvm | IRC logs at irclog.perlgeek.de/moarvm/today Set by moderator on 17 September 2013. |
|||
|
00:49
cognominal joined
01:02
benabik joined
01:14
FROGGS_ joined
|
|||
| dalek | arVM/static-repr2: 934aa2e | jimmy++ | src/6model/ (3 files): added Container spec deserialization |
02:03 | |
| arVM/static-repr2: 8b5455a | (Gerhard R)++ | src/strings/ascii. (2 files): Add missing const to MVM_string_ascii_decode*() clean up MVM_proc_spawn |
|||
|
02:03
dalek joined
02:18
benabik joined
02:21
FROGGS joined
03:05
cognominal joined
|
|||
|
03:13
JimmyZ_ joined
03:18
masak joined,
Util joined
|
|||
| dalek | arVM/static-repr2: 2eac62f | jimmy++ | src/6model/serialization.c: better error msg for 'missing serialize REPR function' |
03:37 | |
| diakopter | TiTping | 05:03 | |
| er | |||
| TimToady: ping | |||
|
05:14
FROGGS joined
|
|||
| TimToady | pung | 05:26 | |
| TimToady is in the middle of upgrading ubunto to mint, so may come and go... | 05:27 | ||
| *untu | |||
| johnny5_ | This is interesting: mailman.cs.huji.ac.il/pipermail/lin...10649.html | 06:00 | |
| so MoarVM could run on this soon? :D | 06:02 | ||
| JimmyZ | only if it's posix compatibility | 06:03 | |
| TimToady | well, rakudo runs on the JVM, even if moarvm doesn't | 06:04 | |
| and it says OSv already runs JVM stuff | |||
| johnny5_ | ahh :) | 06:06 | |
| TimToady | but don't let me stop you from porting MoarVM to it :) | 06:07 | |
| johnny5_ | yeah I couldn't port my way out of a box. | ||
| I can barely write perl5! | |||
| I am progressing though. | 06:08 | ||
| TimToady | obviously you need more input | ||
| johnny5_ | haha wait is this Larry Wall? TimToady? :D | ||
| yeah more input. | |||
| diakopter | haha port out of a box | 06:10 | |
| johnny5_ | Googling my suspicion I was correct. | 06:13 | |
| It is Larry. | |||
| TimToady | .oO(There goes my chance of getting any kind of honest opinion out of johnny5_...) |
06:14 | |
| johnny5_ | haha | ||
| I am a huge fan of perl and the ecosystem. The last year has been nice since technically it is my first Perl Job. | 06:21 | ||
|
06:33
FROGGS joined
06:51
FROGGS joined
|
|||
| FROGGS | o/ | 06:54 | |
|
07:01
woolfy joined
07:09
not_gerd joined
|
|||
| not_gerd | o/ | 07:09 | |
| yoleaux | 17 Sep 2013 11:43Z <JimmyZ> not_gerd: I switched to static-repr2 branch, and removed static-repr branch | ||
| 17 Sep 2013 14:05Z <diakopter> not_gerd: I thought previously msvc was statically linking to its microsoft .dll/.lib dependencies...? could you fix it (also if it wasn't previously, could you make it do it by default, but optional?) | |||
| not_gerd | I'm going to add support for registering new reprs at runtime | 07:11 | |
| as we cannot guarantee insertion order accross instances, I'm going to get rid of the ID member of MVMREPROps | 07:12 | ||
| they will still be accessible via the instance-specific registry alongside the MVMString* name | |||
|
07:20
odc joined
|
|||
| JimmyZ | not_gerd: there are some code like if (REPR(obj)->ID == MVM_REPR_ID_MVMHash ) { ... } | 07:27 | |
| not_gerd | JimmyZ: for core reprs, that will still work | 07:28 | |
| JimmyZ | not_gerd: yeah | ||
| not_gerd | I'll have to think about how to best fix that... | 07:30 | |
| JimmyZ | That could be great | 07:31 | |
| not_gerd | (some ideas: new STable entry, keeping the ID field and requiring dynamically added reprs to be unique, a hash mapping MVMREPROps* to IDs, ...) | 07:32 | |
| JimmyZ | I think core repr could keep the ID, but dyn repr maybe not | ||
| I mean core repr ID is hardcode :P | 07:34 | ||
| s/is/could be/ | 07:36 | ||
| not_gerd | on 64-bit systems, we actually can get a 32-bit STable member for free thanks to padding | ||
| there's a 32-bit hole between size and HOW | |||
| I'll choose the least invasive option for now (keeping the ID field in REPROps) | 07:43 | ||
| it can always be refactored later | |||
| JimmyZ | +1 | 07:54 | |
| dalek | arVM: 2a66495 | (Tobias Leich)++ | src/mast/driver.c: write bytecode to disk |
08:34 | |
| jnthn | I think dyn repr cna just malloc its memory :) | 08:50 | |
| Then it can have different IDs in different instances. | |||
| Problem solved! | |||
| :) | |||
| But yeah, please keep ->ID field :) | |||
| not_gerd | jnthn: I've actually implemented it that way | 08:52 | |
| JimmyZ | jnthn: How about merging static-repr2 branch? | ||
| not_gerd | I'll just need to figure out why every test segfaults with my changes ;) | ||
| I think I just forgot to root some strings | 08:53 | ||
| jnthn | JimmyZ: um, given not_gerd++ still seems to be working on it...? | ||
| JimmyZ | jnthn: that's another feature :P | ||
| and we can keep the branch after merging | 08:54 | ||
| Well, I just hate 'Merge Conflict' ;) | |||
| not_gerd: push it and maybe I can find missing root | 09:03 | ||
| not_gerd | JimmyZ: I'm still fiddling with it | 09:04 | |
| dalek | arVM/static-repr3: 62c8b47 | (Gerhard R)++ | / (10 files): Support repr registration at runtime (still buggy) |
09:19 | |
| not_gerd | if no one else fixes it, I'll take another look in the evening | ||
| jnthn | not_gerd: This branch is based off static_repr2? | ||
| not_gerd | jnthn: yes | ||
| jnthn | k, thanks. | ||
| I'll try and get to the branch reviewing today :) | 09:20 | ||
| JimmyZ | jnthn++ | ||
| not_gerd | it might be something really silly, but I don't have the time to look into it right now | ||
| jnthn | Yeah, I got various things to attend to today also, but hopefully will get a moment later... | ||
| nwc10 | the moarvm threads test still SEGVs occasionally | 09:23 | |
| jnthn | That may be dealt with in diakopter's gcorch branch | ||
| nwc10 | ah OK | ||
| not_gerd | .tell diakopter could you clarify what you want me to do re static linking - we already use /MT, or is something missing from LDLIBS? | 09:42 | |
| yoleaux | not_gerd: I'll pass your message to diakopter. | ||
| not_gerd needs to run | |||
| o/ | |||
| FROGGS | o/ | ||
|
09:42
not_gerd left,
cognominal__ joined
10:10
woosley left
11:59
jnap joined
12:07
FROGGS joined
13:03
lizmat joined
13:36
flussence joined
13:59
lizmat joined
|
|||
| FROGGS | jnthn: how do you want it factored? should bigint to the hard work or serialization.c do the hard work? gist.github.com/FROGGS/a4f263bdea31d553be84 | 14:03 | |
| jnthn | FROGGS: serialize/deserialize, and please use the same approach that other backends do. | 14:05 | |
| (which is to just turn the number into a string and write_str it) | |||
| FROGGS | ahh, okay | ||
| jnthn | Then we needn't leak expand_storage_if_needed | ||
| FROGGS | yeah, that is why I was asking :o) | 14:06 | |
| jnthn | :) | ||
| yes, the "if it feels wrong it probalby is" rule often works well :) | |||
| FROGGS | when I "need" to expose some stuff then already a flag raises in me internally | ||
| jnthn | .tell not_gerd Looked over validation. Please document the new mark syntax in oplist (maybe just with comments at the top of the oplist file) so it's clear what they do. Other than that, +1 to merge. | 14:18 | |
| yoleaux | jnthn: I'll pass your message to not_gerd. | ||
| jnthn | .tell not_gerd (provided tests pass, etc.) | ||
| yoleaux | jnthn: I'll pass your message to not_gerd. | ||
| diakopter | . | 14:36 | |
| yoleaux | 09:42Z <not_gerd> diakopter: could you clarify what you want me to do re static linking - we already use /MT, or is something missing from LDLIBS? | ||
| diakopter | nm | ||
| dalek | arVM: 059ad08 | (Tobias Leich)++ | src/6model/reprs/P6bigint.c: serialize/deserialize for bigint |
14:46 | |
|
14:57
jnap joined
|
|||
| dalek | arVM/static-repr2: 2a66495 | (Tobias Leich)++ | src/mast/driver.c: write bytecode to disk |
15:08 | |
| arVM/static-repr2: 059ad08 | (Tobias Leich)++ | src/6model/reprs/P6bigint.c: serialize/deserialize for bigint |
|||
| arVM/static-repr2: e5d1bfe | jimmy++ | src/ (2 files): Merge branch 'master' into static-repr Conflicts: \tsrc/6model/reprs/P6bigint.c |
|||
|
15:11
jnap joined
15:15
d^_^b joined
|
|||
| dalek | arVM: 8fbec06 | jnthn++ | src/6model/reprs/P6opaque.c: First pass at P6opaque.serialize. |
15:25 | |
| arVM: 62d80c9 | jnthn++ | src/6model/reprs/MVMArray.c: serialize/deserialize for MVMArray. |
|||
| arVM: ca206be | jnthn++ | src/6model/serialization.c: Improve missing serialization function error. |
|||
| timotimo | serialisation is the major roadblock to self-hosting compilation now? | 15:26 | |
| jnthn | yeah | ||
| timotimo | is there lots and lots missing or only bits and pieces? | 15:27 | |
| jnthn | Well, major as in "needs most work" | ||
| Lots of small pieces. None of it is especially difficult. | |||
| Or at least, I hope not :) | |||
| timotimo | sounds good :D | ||
| i'll tell my local .pm group "rakudo on moarvm will be ready in a couple of weeks" :P | 15:28 | ||
| tadzik | just tell them we'll have Perl 6 ready | 15:29 | |
| timotimo | hehehe | ||
| dalek | arVM: 6d587c2 | jnthn++ | src/6model/reprs/P6str.c: Poke serialize function into table for P6str. |
||
| timotimo | so, it shouldn't be much work to get the threading and async io stuff up to the same point where the jvm stuff is after we get rakudo on moarvm to run 90+% spectests, yes? | 15:30 | |
| jnthn | It'll be work :) | 15:31 | |
| But hopefully not too terrible. | |||
| timotimo | easier than on parrot anyway | ||
| dalek | arVM/static-repr2: 8fbec06 | jnthn++ | src/6model/reprs/P6opaque.c: First pass at P6opaque.serialize. |
15:52 | |
| arVM/static-repr2: 62d80c9 | jnthn++ | src/6model/reprs/MVMArray.c: serialize/deserialize for MVMArray. |
|||
| arVM/static-repr2: ca206be | jnthn++ | src/6model/serialization.c: Improve missing serialization function error. |
|||
| arVM/static-repr2: 6d587c2 | jnthn++ | src/6model/reprs/P6str.c: Poke serialize function into table for P6str. |
|||
| MoarVM/static-repr2: 04c56fd | jimmy++ | src/6model/ (3 files): | |||
| MoarVM/static-repr2: Merge branch 'master' into static-repr | |||
| timotimo | dalek hung again? | 15:56 | |
| benabik | That'll happen with a lot of commits. | 15:58 | |
| timotimo | oh, i think i'm ignoring parts, joins and quits in this channel | ||
| that would explain why i saw 'em stop posting, but not quit | |||
| benabik | Hm. I reenabled display of those and still don't see a dalek quit. Eh? Maybe an internal limit to stop those kicks. *shrug* | 15:59 | |
| JimmyZ | Good night | 16:27 | |
| lizmat | gnight, JimmyZ! | 16:28 | |
| arnsholt | Don't think dalek quit | 16:35 | |
| Still listed in /names | |||
| jnthn | yeah, still here | ||
| It sometimes does this too | |||
| diakopter | it doesn't get kicked; it gets temp klined or whatever | 16:36 | |
| benabik | Kline bottled? | 16:37 | |
| diakopter | commutestrianizing & | ||
|
16:53
foo_bar_baz joined
17:13
jnap joined
18:04
benabik joined
|
|||
| dalek | arVM/dynamic-repr: a8afbba | (Gerhard R)++ | / (10 files): Support repr registration at runtime (still buggy) |
18:24 | |
| arVM/dynamic-repr: 5172c20 | (Gerhard R)++ | src/moarvm.c: Fix double-free bug |
|||
|
18:25
not_gerd joined
|
|||
| not_gerd | o/ | 18:25 | |
| yoleaux | 14:18Z <jnthn> not_gerd: Looked over validation. Please document the new mark syntax in oplist (maybe just with comments at the top of the oplist file) so it's clear what they do. Other than that, +1 to merge. | ||
| 14:18Z <jnthn> not_gerd: (provided tests pass, etc.) | |||
| FROGGS | hi not_gerd | ||
| not_gerd | jnthn: static-repr2 is now known as dynamic-repr and should be fixed | 18:30 | |
| jnthn | not_gerd: What should I review? :) | ||
| jnthn figures he should do this this evening, before he goes for vacation :) | 18:31 | ||
| not_gerd | dynamic-repr should be mergeable, but isn't used yet | ||
| dalek | arVM: 67138b1 | (Tobias Leich)++ | src/6model/reprs/KnowHOWREPR.c: serialize for KnowHOWREPR |
||
| not_gerd | (as in, we do not add reprs at runtime, so no guarantees that it actually works ;)) | 18:32 | |
|
19:44
jnap joined
20:17
flussence joined
20:21
flussence joined
|
|||
| dalek | arVM: b764c1c | (Tobias Leich)++ | src/6model/serialization.h: mapped write_int32 and write_int16 |
20:44 | |
| arVM: a06272b | (Tobias Leich)++ | src/6model/reprs/KnowHOWAttributeREPR.c: added ->serialize for KnowHOWAttributeREPR |
20:47 | ||
| FROGGS | this is a wip of serialize_repr_data for P6opaque: gist.github.com/FROGGS/e907c786f7b18ff0c345 | 20:51 | |
| it makes a few tests pass but breaks at least one | |||
| diakopter | whoa | ||
| serious stuff | |||
| :D | |||
| timotimo | \\o/ | ||
| FROGGS | I need to figure out two things, num_classes and descrim | 20:52 | |
| the commented out stuff for num_classes is definitely wrong (as it explodes), the descrim stuff is just copied over as a note | 20:53 | ||
| not_gerd | FROGGS: should I wait with merging the validation branch? | 20:55 | |
| FROGGS | not_gerd: no need to wait for me :o) | 20:56 | |
| my brain is already empty for today I guess | |||
| not_gerd | then I'll push after a final compile/selftest | 20:57 | |
| jnthn | not_gerd: "+ /* FIXME: flattening necessary? */" - yes, it is before a hash lookup, so these comments in repr branch can go | ||
| *repr_static | 20:58 | ||
| not_gerd | jnthn: just wan't sure if we needed to flatten just after creating the string | ||
| diakopter is reminded again to debug that | |||
| jnthn | not_gerd: It's probably a good idea. | 21:00 | |
| I wish this branch hadn't done a bunch of MVMROOT cleanups along the way, making it harder to review... | |||
| Anyways, I like it. | 21:03 | ||
| Saves some allocation, some startup work, and a level of indirection on a bunch of hot paths. | |||
| FROGGS | uhh, that sounds nice | ||
| jnthn | So, +1 from me on merging. | ||
| not_gerd++, JimmyZ++ | 21:04 | ||
| timotimo listens up | |||
| dalek | arVM: 933b810 | (Gerhard R)++ | / (8 files): Reimplement bytecode validation |
||
| MoarVM: d79593b | (Gerhard R)++ | tools/update_ops.p6: | |||
| MoarVM: Fix comment fail | |||
| not_gerd | I expect JimmyZ will be a happy camper tomorrow ;) | 21:05 | |
| diakopter | why | ||
|
21:05
dalek joined
|
|||
| not_gerd | MERGES! | 21:05 | |
| jnthn | 'cus he's been asking approximately 10,000 times a day for me to review those branches so we can merge them. :) | ||
| Probably not gonna be online that much in the next several days. | 21:14 | ||
| diakopter | have a nice vacation | 21:15 | |
| jnthn | And may not have chance ot backlog, so .tell anything :) | ||
| Thanks :) | |||
| sleep...early flight... & | 21:16 | ||
| not_gerd | o/ | ||
|
21:17
donaldh joined
|
|||
| FROGGS | have fun! | 21:17 | |
| lizmat | gnight jnthn, safe journeys! | 21:19 | |
| dalek | Heuristic branch merge: pushed 32 commits to MoarVM by gerdr | 21:21 | |
| not_gerd | good night, #moarvm | 21:48 | |
|
21:48
FROGGS joined
21:49
not_gerd left
22:22
FROGGS joined
|
|||