moritz irclog.perlgeek.de/perl6-dev/today works now 07:14
I had some negative caching of channel existence 07:15
nine psch: this is very odd. 2 EVALs actually work just fine before the third, which should be just the same, breaks. The third does seem to be compiled differently, like not properly detecting that it's run during precompilation. 09:29
psch nine: that does indeed sound very odd 09:35
nine At least it would seem like we're quite close to have it working 09:36
psch well, i'd guess that depends on why it the third EVAL breaks 09:37
bartolin psch: I think I figured out what went wrong with $fh.slurp-rest(:bin) after $fh.get. looks like we read all the data (it's a small file) into readBuffer when calling readlinechompfh. Later we only looked in the channel -- which had no more data, since it was read to readBuffer already. 09:45
with the following patch I get the expected result: gist.github.com/usev6/788d5ee38489...f2a83aab22
psch bartolin: neat! can you add that to your PR? 09:46
bartolin btw, did you figure out how to create an VMArrayInstance_u8? I think, I managed to do that. 09:47
psch: I'm not really sure if my patch could break other cases. nqp's 'make test' had no regressions at least ...
but, yes, I'll add that to my PR 09:48
psch bartolin: well, nqp 'make test' is somewhat basic... i'd run spectest, too fwiw
bartolin only that 'make spectest' doesn't work at the moment :-/
psch huh, it doesn't? 09:49
bartolin I'll try it with an older version of rakudo
psch is that because Test doesn't work..?
bartolin does it? the precomp problem is in the way 09:50
psch ah
yeah i think that's cause nine++ didn't push his rakudo-level fix there
well, or an nqp bump either
although i suppose the precomp EVAL fix still belongs in the same area, so that's understandable 09:51
bartolin I tried to build with --gen-nqp=master, but that was not enough
psch bartolin: gist.github.com/niner/48daf2fef50e...1e4b3eef03 09:52
i think gist.github.com/niner/44ab8449ab5b...9e525a59f3 should be upstream already..?
well, minus the debug output :)
bartolin maybe I'll try my patch on d7698f3 -- I had a clean spectest run with rakudo-j on that commit 09:53
bartolin is looking at the gists
hmm, no, looks like that patches are not in rakudo/nqp, yet 10:03
psch bartolin: fwiw, i'm not sure how much of that is still needed with f6670c88e 10:16
nine If you have a fix for slurp-rest(:bin), I won't need to push my rakudo level workaround :) 10:43
bartolin nine: i'm running a spectest on an older commit now (it'll take some time) 10:52
pmurias should dalek report commits here rathern then to #p6dev? 13:35
timotimo yeah
for all intents and purposes, #p6dev doesn't exist 13:36
bartolin well, spectest for rakudo-j (on commit d7698f3) was fine (only two failures -- one known and the other for a newly added test) 14:11
on the other hand slurp-rest seems undertested
I updated my PR (github.com/perl6/nqp/pull/286) 14:12
timotimo whoa 14:14
bartolin oops, it fails for nqp-m. looking ... 14:18
aha, pmurias++ renamed the test files. I updated the failing tests accordingly: github.com/perl6/nqp/commit/10f2b4...da12317c5d 14:35
bartolin looks around for dalek
RabidGravy wonders why dalek isn't announcing the right commits here
bartolin is it possible to activate another run of travis for a PR? 14:36
RabidGravy if it failed then there will be a little "refresh" button, but not if it passed 14:39
bartolin yeah, it failed because of the renamed test files: github.com/perl6/nqp/pull/286 14:40
hmm, can't find that "refresh" button 14:41
RabidGravy maybe you need to be an "owner", it should be on the right hand side of the build summary near the commit id 14:43
bartolin well, it's not there (or I'm blind). anyway, I think the build failure was not related to my patches 14:50
just saw that dalek still reports roast and nqp commits to #p6dev 15:25
timotimo ah 15:29
fixed roast 15:30
and nqp
i thought i had already fixed nqp
[Coke] news.perlfoundation.org/2016/05/yap...dates.html # craigslist++ 15:31
timotimo craigslist++ # also a fantastic song by weird al yankovic 15:32
cognominal is TimTaody still the "artist in residence" at craiglist++ ? 16:03
pmurias anyone figured out the travis failure? 16:19
pmurias fixes it 16:20
ahh, it's already fixed, usev6++ 16:23
nine Can I use "echo 1 > ..." in a Makefile without screwing up Windows? 16:33
[Coke] nine - works here in a command prompt 16:37
dalek p: 912b502 | (Pawel Murias)++ | src/vm/js/Operations.nqp:
[js] Implement nqp::chain.
16:40
p: b3a1014 | (Pawel Murias)++ | t/qast/01-qast.t:
Test nqp::chain.
[Tux] This is Rakudo version 2016.04-200-gad82657 built on MoarVM version 2016.04-134-g9879233 16:57
test 20.140
test-t 13.034
csv-parser 34.652
RabidGravy so it's creeping back down again :) 16:59
timotimo yeah, seembs like we've stabilized 17:00
RabidGravy well I don't quite mean to quit the whole server, but hey 17:02
timotimo %)
nine Can I call the parent class' BUILD submethod? 17:10
timotimo i think you can self.ClassName::BUILD(...) 17:11
jnthn nine: Won't that have already happened anyway as part of normal object construction? 17:17
m: class P { submethod BUILD() { say "p" } }; class C is P { submethod BUILD() { say "c" } }; C.new
timotimo what if nine is implementing some optimization that can skip BUILD for us if it's not necessary? :D 17:18
jnthn /o\
:)
Food time; bbl :)
timotimo i was considering forcing buildplans to have their opcode 14 (the one that's only relevant in BUILD_LEAST_DERIVED) all at the end, so that when the BUILDALL runner encounters one it can immediately exit 17:19
nine Somehow it just doesn't seem to work with CompUnit::Repository::Installation 17:45
Err... No concretization found for CompUnit::Repository::Locally? 18:21
RabidGravy If anyone wants to take a look at github.com/rakudo/rakudo/pull/769 that would be cool :) 19:27
dalek kudo/nom: 5bff97c | timotimo++ | src/Perl6/Compiler.nqp:
elaborate on different kinds of profiles for --profile
20:33
kudo/nom: 9db9102 | moritz++ | src/Perl6/Compiler.nqp:
Merge pull request #756 from rakudo/update_profile_usage

elaborate on different kinds of profiles for --profile
timotimo oh, accepted it unmodified? 20:34
dalek kudo/nom: 0ee8331 | Xliff++ | src/core/Match.pm:
Match infix eqv (#739)

Adds infix:<eqv> for Match objects
20:35
ast: 99e759c | (Zoffix Znet)++ | S06-advanced/wrap.t:
Rewrite test to avoid throws-like()

The use of throws-like here makes the pass/fail of the test depend on the specifics of implementation of throws-like. See github.com/rakudo/rakudo/pull/743
20:37
ast: 9482e54 | moritz++ | S06-advanced/wrap.t:
Merge pull request #113 from zoffixznet/rework-nextsame

Rewrite test to avoid throws-like()
kudo/nom: e239f6e | (Zoffix Znet)++ | / (2 files):
Implement alternate ways to call `subtest` from Test.pm6 (#743)

  * Implement alternate calling methods for `subtest`
  * Fix incorrect sub names in test
  * Add diag() on start of subtest
This makes it easier to see where the subset starts in the output