Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2013-12-18

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:07 dalek MoarVM: 41836eb | jnthn++ | src/6model/s (3 files):
00:07 dalek MoarVM: Implement deserialize bit of object repo.
00:07 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/41836eb4c2
00:11 dalek MoarVM: 9345cbe | jnthn++ | src/core/interp.c:
00:11 dalek MoarVM: Scatter some SC object write barriers.
00:11 dalek MoarVM:
00:11 dalek MoarVM: Taken from all the places the JVM backend puts them.
00:11 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/9345cbef31
00:12 jnthn Note that you'll need to rebuild Moar, NQP and Rakudo with the latest stuff I've done.
00:14 jnthn If you managed to reproduce what I have, Test.pm now pre-compiles and 4 sanity tests fail.
00:14 jnthn *manage
00:35 dalek MoarVM: 576042c | jnthn++ | src/6model/reprs/P6opaque.c:
00:35 dalek MoarVM: Fix clone/null sentinel interaction bug.
00:35 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/576042c7cd
01:02 [Coke] I just built rakudo-moar... "make test" is trying to run perl6-p ?
01:04 jnthn [Coke]: Likely just makefile copy-pasto
01:04 jnthn [Coke]: I've been invoking prove directly
01:08 jnthn [Coke]: Feel free to commit a fix, of course :)
01:31 seaker joined #moarvm
01:59 jnap joined #moarvm
02:25 [Coke] https://gist.github.com/coke/8016363 looks like a good start, but I get a makefile error when I rebuild.
02:30 moritz [Coke]: the rebuild error looks unrelated to your patch
02:35 [Coke] if the patch looks good, feel free to apply. sleep here soon. (and why are you awake?)
02:35 [Coke] oh, you did already. :)
02:59 jnap joined #moarvm
03:03 moritz why am I awake? $daughter is to blame :-)
04:00 jnap joined #moarvm
04:10 colomon joined #moarvm
04:20 harrow joined #moarvm
05:01 jnap joined #moarvm
06:02 jnap joined #moarvm
07:02 jnap joined #moarvm
07:30 FROGGS joined #moarvm
08:03 jnap joined #moarvm
09:04 jnap joined #moarvm
09:12 flussence joined #moarvm
09:12 _sri joined #moarvm
09:12 TimToady joined #moarvm
09:21 flussence joined #moarvm
09:21 _sri joined #moarvm
09:21 TimToady joined #moarvm
10:05 jnap joined #moarvm
10:11 dalek MoarVM: 0005ecc | jnthn++ | src/core/exceptions.c:
10:11 dalek MoarVM: Handle CONTROL category correctly.
10:11 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/0005ecce28
10:12 jnthn That was at least half of why warnings ended up fatal.
10:15 FROGGS cool
10:15 timotimo \o/
10:15 * FROGGS .oO( or rather (Cool) ? )
10:16 jnthn And fixed another issue in NQP related to it
10:16 jnthn Now they still end up fatal. grr.
10:16 jnthn But the handler actually runs.
10:17 jnthn So, suspect nqp::resume is not doing quite what it needs to...
10:17 * FROGGS builds everything
10:19 jnthn oh hmm, how on earth does resume work...
10:19 FROGGS I did that me thinks
10:21 jnthn I don't quite understand wha tit's doing.
10:21 FROGGS so at least you don't need to blame the weatherman :o)
10:21 jnthn Of note, I don't see it updating tc->cur_frame anywhere...
10:21 jnthn Also I think it's trying to make non-resumable things into resumable ones...somehow ;)
10:22 FROGGS :S
10:22 FROGGS id does some magic here too: src/core/interp.c:3541
10:23 FROGGS but I don't see an update on tc->cur_frame there either
10:23 jnthn Hmm
10:23 jnthn Yeah, this looks...fishy.
10:24 FROGGS I'm sorry
10:25 jnthn Don't be. It was good enough for the NQP tests for resume at least ;)
10:25 jnthn And it's *not* a nice thing to implement...
10:28 FROGGS I still think I learned something at that time, which is always nice :o)
10:30 tgt joined #moarvm
10:55 FROGGS joined #moarvm
11:05 jnap joined #moarvm
11:06 jnthn ooh, I may just have something that works
11:11 * timotimo is eager to pull before leaving towards his train
11:14 dalek MoarVM: 743bc1b | jnthn++ | src/ (3 files):
11:14 dalek MoarVM: Fix resumable exceptions.
11:14 dalek MoarVM:
11:14 dalek MoarVM: Now warnings in Rakudo are not fatal.
11:14 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/743bc1b077
11:15 timotimo \o/
11:22 jnthn Down to 3 failing sanity tests with that, btw.
11:22 timotimo wow, that's awesome!
11:22 timotimo and now i'll go train
11:40 FROGGS joined #moarvm
11:52 dalek MoarVM: 006f6f6 | jnthn++ | src/ (10 files):
11:52 dalek MoarVM: Correctly provide invocant to invocation handlers.
11:52 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/006f6f6644
11:52 jnthn And then there were 2...
11:56 diakopter O_O
11:58 jnthn 07-op-string.t and 99-test-basic.t
12:06 jnap joined #moarvm
12:07 woolfy joined #moarvm
12:24 nwc10 I assume that the build shouldn't fail with:
12:24 nwc10 src/vm/moar/ops/perl6_ops.c: In function ‘p6capturelex’:
12:24 nwc10 src/vm/moar/ops/perl6_ops.c:333: error: too few arguments to function ‘MVM_frame_find_invokee’
12:24 nwc10 src/vm/moar/ops/perl6_ops.c: In function ‘p6captureouters’:
12:24 nwc10 src/vm/moar/ops/perl6_ops.c:350: error: too few arguments to function ‘MVM_frame_find_invokee’
12:25 jnthn Uh, not if you have the local patch I do... :)
12:27 timotimo d'oh :)
12:40 timotimo okay, i built a fresh nqp-jvm, now i can try going at the boot interop code again
12:41 tgt joined #moarvm
12:44 tgt joined #moarvm
12:57 jnthn timotimo: yay :)
12:57 lizmat joined #moarvm
13:00 FROGGS joined #moarvm
13:02 timotimo (in reality, i was still missing rakudo)
13:02 timotimo (so i'm compiling that now, while playing a bit of Race The Sun in the mean time :P)
13:02 FROGGS "Rakudo has been built and installed."
13:02 FROGGS I like that :o)
13:02 timotimo .his is a no-cellphone-area with "psst" icons on the walls but still people are talking to each other or on the phone :\
13:03 dalek MoarVM: 67d0815 | jnthn++ | src/strings/ops.c:
13:03 dalek MoarVM: Fix bitwise string and.
13:03 dalek MoarVM:
13:03 dalek MoarVM: Should take length of the shortest string.
13:03 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/67d081594f
13:03 FROGGS ohh, the runner seems to be wrong
13:04 jnthn FROGGS: While you're looking at that stuff, "make test" doesn't work for me, but running prove --exec=perl6-m t/01-sanity does...
13:05 FROGGS k, I see what is up with the runner at least
13:06 jnthn I'm now down to just one sanity test fail
13:06 FROGGS will looks at `make test` in a sec
13:06 FROGGS jnthn++
13:06 * FROGGS .oO( but the he took a sanity test to the knee )
13:07 moritz jnthn: does prove --exec=perl6-m.bit t/01-sanity   work?
13:07 jnap joined #moarvm
13:07 FROGGS bit?
13:08 FROGGS that would be the problem I am about to commit
13:08 jnthn bat
13:08 FROGGS err, I would commit the fix of course
13:08 jnthn But yes, it does with .bat
13:14 FROGGS hmpf, it tries to recompile the ops now when running `make test`
13:23 moritz moar-support setting build segfaults again :(
13:24 timotimo 66
13:24 timotimo ~j.
13:24 timotimo ~j.
13:24 timotimo whoops.
13:24 moritz better without the j :-)
13:25 timotimo yeah, works a lot better :D
13:25 timotimo the most frustrating part is when you *think* the connection is dead
13:25 timotimo and you <cr>~. and whatever line you've partially typed before makes it out m)
13:28 FROGGS jnthn: if you set "merge => 1" to 0,  in t/harness, you get:
13:28 FROGGS Unhandled exception: Cannot find method 'postcircumfix:<( )>'
13:28 FROGGS at <unknown>:1  (./CORE.setting.moarvm:print_exception:4294967295)
13:28 * moritz is familiar with that
13:28 FROGGS ... for all tests
13:34 FROGGS ahh, I think I need to wait for jnthn's local patch...
13:34 FROGGS src/vm/moar/ops/perl6_ops.c: In function ‘p6captureouters’:
13:34 FROGGS src/vm/moar/ops/perl6_ops.c:362:9: error: too many arguments to function ‘MVM_frame_find_invokee’
13:36 nwc10 git pull, as it's already there
13:36 FROGGS ahh, have missed that... thanks!
13:36 FROGGS that seems to help :o)
13:41 FROGGS jnthn: `make test` does work for me now
13:42 FROGGS four test files fail
13:42 FROGGS 11, 14, 15 and 99
13:43 jnthn You're likely missing some latest commits from Moar or NQP.
13:43 * moritz missed a moar commit
13:44 FROGGS k
13:45 jnthn wtf, now I always get a segfault building Test.pm...
13:49 FROGGS :(
13:49 FROGGS I hope it is there under the debugger too
13:49 FROGGS see you later
13:49 FROGGS plain &
13:59 jnap joined #moarvm
15:20 eternaleye joined #moarvm
16:23 tgt joined #moarvm
17:01 cognominal joined #moarvm
17:06 eternaleye joined #moarvm
17:22 FROGGS joined #moarvm
18:08 FROGGS GatherIter is kinda pressing now, right?
18:13 _sri joined #moarvm
18:13 flussence joined #moarvm
18:13 TimToady joined #moarvm
18:37 jnthn Well, I'm gonna do a cheating one.
18:37 jnthn That ain't lazy
18:37 jnthn And then I'll do the continuations in the new year, unless somebody beats me to it :)
18:40 timotimo wouldn't likely be me :)
18:45 diakopter :D
18:45 timotimo jnthn: omg omg omg i think my crap works now
18:45 diakopter hopefully me
18:45 diakopter timotimo: better run to the .. oh nm
18:48 diakopter msg chanserv op #moarvm
18:48 diakopter gah
19:11 jnthn whoz op?
19:11 jnthn tiyay!
19:12 jnthn wtf
19:12 jnthn timotimo: yay :)
20:15 cognominal joined #moarvm
20:21 ssutch joined #moarvm
20:25 jnthn Urgh, current bug is taking quite some hunting... :/
20:25 diakopter talk it out to me :)
20:28 cognominal joined #moarvm
20:31 jnthn Well, given these lines:
20:31 jnthn nqp::say(nqp::elems($!rest));
20:31 jnthn nqp::say(nqp::elems(nqp::getattr($x, Parcel, '$!storage')));
20:32 jnthn nqp::splice($!rest, nqp::getattr($x, Parcel, '$!storage'), 0, 0);
20:32 jnthn nqp::say(nqp::elems($!rest));
20:32 jnthn The fact the says come out with
20:32 jnthn 0
20:32 jnthn 1
20:32 jnthn 2
20:32 jnthn Seems a bit odd
20:32 jnthn You have a list with 0 elems, you splice in a list of 1 elems, and suddenly you get a list with 2 elems.
20:32 diakopter istr splice was not thoroughly done/reviewed
20:33 jnthn I'd thought it was mostly a fairly direct port...
20:34 nwc10 jnthn: one for your later enjoyment - http://paste.scsys.co.uk/285849 - Can't build NQP with a MoarVM with a 2034K nursery
20:35 colomon joined #moarvm
20:40 eternaleye joined #moarvm
20:40 jnthn nwc10: Hm, if we try to dereference each thing as we add it to the worklist, it'll crash at a much more informative place, I'd guess.
20:41 nwc10 yes, that's an interesting idea that should work
20:45 nwc10 but tonight I am going to bed. And no guarantee if I have time tomorrow to figure out how many places need to be tweaked to do that
20:47 diakopter nwc10: I bet you'll have time to count to 1 :)
20:55 eternaleye joined #moarvm
20:57 eternaleye joined #moarvm
21:06 jnthn argh, found it
21:07 timotimo \o/
21:17 dalek MoarVM: ac15634 | jnthn++ | src/6model/reprs/MVMArray.c:
21:17 dalek MoarVM: Fix double-adding of start in splice.
21:17 dalek MoarVM:
21:17 dalek MoarVM: Led to incorrectly sized array and potentially looking at garbage.
21:17 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/ac15634a8e
22:29 colomon joined #moarvm
22:43 colomon joined #moarvm
23:23 lue I just pulled & compiled MVM and NQP, and now I get this with rakudo: src/vm/moar/ops/perl6_ops.c:350:5: error: too many arguments to function ‘MVM_frame_find_invokee’
23:23 lue (also pulled rakudo, I should mention)
23:24 jnthn lue: That sounds like your MVM didn't make install proprely, or it went in a different place then your NQP/Rakudo ended up pointed at.
23:24 jnthn That API change happened yesterday.
23:25 lue I'll run through the process again, then. Maybe manually delete some MVM stuff o'er in the install dir.
23:39 lue jnthn: this is weird. The timestamp of the offending header is today, but apparently it's still mismatched with— the last git log commit was from the 15th *headdesk*
23:40 jnthn That'd do it :P
23:42 * TimToady blows away all of install and nqp when he's serious about rebuilding
23:43 lue TimToady: detonating /usr/local is generally perceived as dangerous to your health :)
23:47 TimToady putting random junk in /usr/local is also generally perceived as suboptimal :)
23:50 lue /usr/local = non-package-manager programs, to me at least :)
23:51 TimToady /home/larry for me :)
23:51 lue /home feels too volatile for programs I want to use often.
23:52 lue jnthn: as predicted, updating my Moar repo worked, and it's moved on now :)
23:53 * TimToady doesn't like putting volatile stuff into /usr/local
23:54 lue I basically treat /usr/local as the location for programs that would go in /usr, had I installed them through portage.
23:55 * lue &

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary