|
github.com/moarvm/moarvm | IRC logs at irclog.perlgeek.de/moarvm/today Set by moderator on 25 September 2013. |
|||
|
00:58
FROGGS_ joined
01:58
FROGGS__ joined
02:15
diakopter joined
|
|||
| diakopter | . | 02:15 | |
| dalek | arVM: d0c48cb | jimmy++ | src/io/procops.c: Small clean up to MVM_proc_spawn |
03:25 | |
|
04:11
cognominal joined
04:55
not_gerd joined
|
|||
| not_gerd | o/ | 04:55 | |
| yoleaux | 24 Sep 2013 16:52Z <jnthn> not_gerd: in b431a11, I suspect the code following "/* Identify REPRs by address of name string */" is fragile. For example, two different libraries trying to register the same REPR name would have the string constants at different addresses. I'd just do a strcmp there. | ||
| JimmyZ | \\o | 04:56 | |
| TimToady | o/ | 05:01 | |
| not_gerd | oo( IndyBootstrap.java - I wonder if that's named after the dog ) | 05:09 | |
| hm... | 05:15 | ||
| CallFrame.java uses the same (broken?) algorithm to find he outer frame that MVM_frame_invoke() uses | 05:16 | ||
| JimmyZ | broken? | 05:17 | |
| not_gerd | the 56-role.t failure | 05:18 | |
| role_meth is lexically enclosed by frames 34, 30, 0 | 05:19 | ||
| if there's no explicit outer set, we start looking for the outer (frame 34) at cur_frame, which, in this case, is frame 30 | |||
| which of course means we miss frame 34 and fail with 'Cannot locate an outer frame for the call' | 05:20 | ||
| JimmyZ | I see there is autoClose in the CallFrame.java | 05:22 | |
| not_gerd | that... doesn't look right to me | 05:26 | |
| JimmyZ | a cheat? :P | 05:27 | |
| dalek | arVM/ctypes2: 9b7e6af | (Gerhard R)++ | / (9 files): Add VMBlob and VMPtr representations |
06:56 | |
| MoarVM/ctypes2: 3ad7f44 | (Gerhard R)++ | .gitignore: | |||
| MoarVM/ctypes2: Ignore files with leading __ | |||
|
06:57
dalek joined
|
|||
| not_gerd | bye, #moarvm | 06:59 | |
|
06:59
not_gerd left
07:17
FROGGS joined
|
|||
| FROGGS | o/ | 07:22 | |
| dalek | arVM/cleanup: c531595 | jimmy++ | src/6model/reprs/ (5 files): Code clean up |
07:24 | |
| JimmyZ | \\o | ||
| .tell jnthn are you fine to github.com/MoarVM/MoarVM/commit/c531595dd7 ? | 07:26 | ||
| yoleaux | JimmyZ: I'll pass your message to jnthn. | ||
| FROGGS | hmmm, then you don't see which repr it was :o( | 07:29 | |
| JimmyZ | FROGGS: you will, see reprs.c | 07:30 | |
| FROGGS | ahh, yeah, I see now :o) | ||
| JimmyZ++ | |||
| JimmyZ | ;) | ||
| FROGGS | static void die_no_ass() *lol* | 07:31 | |
| tadzik | what? :D | 07:40 | |
| arnsholt | tadzik: Remember that Python-on-NQP thing we talked about in Oslo a bit back? =) | 08:06 | |
| tadzik | arnsholt: yep :) | ||
| arnsholt | I accidentally one: github.com/arnsholt/nqpy | 08:12 | |
| Oh, this isn't #perl6. Whoops | 08:13 | ||
| FROGGS | this isnt #python either :o) | ||
| arnsholt | I think NQP things are relevant to #perl6 at least =) | ||
| FROGGS | gah, arguments | 08:14 | |
| tadzik | arnsholt: yay :) | 08:16 | |
|
08:26
cognominal joined
08:28
odc joined
09:09
foo_bar_baz joined
|
|||
| dalek | arVM: 9160004 | jimmy++ | src/6model/reprs/ (5 files): Code clean up |
09:11 | |
|
09:59
donaldh joined
12:08
jnap joined
12:54
donaldh joined
13:12
cognominal joined
14:29
cognominal joined
15:05
jnap joined
15:43
FROGGS[mobile] joined
15:53
not_gerd joined
|
|||
| not_gerd | o/ | 15:53 | |
| dalek | arVM/ctypes3: b734303 | (Gerhard R)++ | src/native/ (3 files): Bits and pieces of CArray |
||
| arVM/ctypes3: 010f6a4 | (Gerhard R)++ | / (28 files): Refactor C scalars |
|||
| arVM/ctypes3: 05cdd18 | (Gerhard R)++ | src/native/CArray.c: (Partly) implement CArray at_pos/bind_pos |
|||
| not_gerd | using C types now looks like this: gist.github.com/gerdr/627e5508d69920e6e3aa | 15:56 | |
| jnthn: you'll take care of fixing prior_invocation/outer handling on MoarVM? | 16:03 | ||
| jnthn | not_gerd: Yes, prior_invocation must die | 16:04 | |
| And MoarVM has little enough atop of it so far (just NQP) that it's the least painful place for me to get a new way in place. | |||
| May get to it this weekend. | |||
| not_gerd | ++jnthn | 16:05 | |
| that will hopefully fix t/nqp/56-role.t | |||
| jnthn | yeah. Hope so :) | 16:06 | |
| not_gerd | another question: what's your plan with containers re native types? | 16:09 | |
| a kind argument à at_pos/bind_pos | |||
| and assign_i/decont_i ops? | |||
| or something completely different? | |||
| jnthn | Probably assign_i/decont_i style | ||
| jnthn really should gets his grant app done for doing native arrays and fixing native is rw in Rakudo (amongst other things) in... | 16:10 | ||
| TimToady | no, paint our bikesheds!!! | 16:12 | |
| jnthn | not_gerd: We'll probably need 3 kinds of native container: lexical, array and attribute | 16:14 | |
| [Coke] | as a member of the GC (which has nothing to do with hague grants), I agree, finish your grant. :P | ||
| jnthn | not_gerd: Of course, I hope that the *majority* of these can be optimized away before Moar ever gets 'em. | ||
| not_gerd: But we can maybe kill even more once we get to inlining... | |||
| [Coke] | as the next release manager, let's get moarnqp ready for that release! | ||
| jnthn | [Coke]: Well, we're also aiming at first JVM Star from the next release too :) | 16:15 | |
| TimToady | RJar | 16:16 | |
| [Coke] | jnthn: DO ALL THE THINGS | ||
| ... btw, this is my directive at work. | |||
| TimToady | you have to do a few things that aren't things too | ||
| jnthn also has to DO ALL THE THINGS at $dayjob too, for a certain set of things. :) | |||
| TimToady | .oO(Trogdor Burnoutator) |
16:17 | |
| jnthn | Well, I had a long weekend break just now and I've got a week of vacation time on my schedule at the end of October, so hopefully I'll stay sane-ish :) | 16:19 | |
| [Coke] was just inspired to draw a "code all the things" guy on his whiteboard. jnthn++ | 16:20 | ||
| I had the mini vacay to come to the workshop... first day back at work I felt awesome, motivated, ready to... next day, RAWR. | 16:21 | ||
| it's been a week and I'm considering job hunting. :P | |||
| FROGGS[mobile] | I know that feeling | 16:39 | |
|
16:54
FROGGS joined
|
|||
| dalek | arVM/ctypes3: e1111b1 | (Gerhard R)++ | src/native/ (3 files): Push stuff from CArray instances into REPR_data, I originally wanted to avoid having to create new type objects for each array of different size. However, this is closer to C semantics and will make multi-dimensional arrays easier to implement. |
17:38 | |
| TimToady | most of the time an application will only use a few shaped sizes anyway; applications with variable shapes will generally just use a shape of * instead of a bunch of specific sizes | 17:43 | |
| jnthn | Buf does type obj per size also | 17:44 | |
| Fits in well with having a parametric role | |||
| TimToady | so basically every Blob gets its own type object? | 17:45 | |
| that won't scale down very well | |||
| jnthn | Not every *blob* | ||
| Every parameterization of Blob | |||
| Buf[int8], Buf[int16], etc. | 17:46 | ||
| TimToady | okay, I grok now | ||
| Bufs/Blobs are still shape(*) in the .elems direction | 17:47 | ||
| you confused me by using an overloaded term like "size", which is almost as bad as "length" :) | 17:48 | ||
| not_gerd | gist.github.com/gerdr/627e5508d69920e6e3aa # 4x4 matrix of long double | 17:52 | |
|
19:06
foo_bar_baz joined
20:48
BenGoldberg joined
21:07
donaldh joined
|
|||