00:01 pyrimidine joined 01:25 pyrimidine joined 01:30 gk_-1wm_- joined, gk_-1wm_- left 02:48 ilbot3 joined 02:56 pyrimidine joined 05:14 nine_ joined, diakopte1 joined, sivoais_ joined 05:15 harrow` joined 05:24 sivoais joined 06:28 pyrimidine joined 06:40 pyrimidine joined 08:21 pyrimidine joined 08:38 pyrimidine joined 10:41 pyrimidine joined
dogbert17 marvels at the total silence 10:46
dogbert17 is contemplating github.com/MoarVM/MoarVM/blob/mast...ream.c#L68 wondering if we could ever run into a situation where 'ds->bytes_head' is NULL and 'pos' not being equal to zero ... 10:49
dogbert17 suspects not 11:04
jnthn I'd hope not, but can always stick an assert in 11:38
Geth MoarVM: 0d0e2140f5 | (Stefan Seifert)++ | build/probe.pm
Yet another attempt at silencing the pthread_yield warnings

According to commit d47dec1c1c229e22037b9e2bf8429a1a713b1c80 sched_yield should be used when _POSIX_PRIORITY_SCHEDULING is defined. We lost that with commit ddb1bf9e47715c03209153fa7fe06f6a9cc61bd3.
According to sched_yield(2) POSIX systems on which sched_yield() is ... (5 more lines)
12:09
MoarVM: 979ec4d8e3 | (Jonathan Worthington)++ | build/probe.pm
Merge pull request #541 from niner/master

Yet another attempt at silencing the pthread_yield warnings
MoarVM: ab426c1477 | (Daniel Green)++ | src/strings/ops.c
Change MVM_string_equal_at_ignore_case to use fc
MoarVM: 772f60da30 | (Jonathan Worthington)++ | src/strings/ops.c
Merge pull request #540 from MasterDuke17/change_MVM_string_equal_at_ignore_case_to_use_fc

Change MVM_string_equal_at_ignore_case to use fc
12:31 pyrimidine joined 12:42 pyrimidine joined
MasterDuke what sets the version/commit/tag of libuv to use? 13:30
timotimo that's done through a git submodule 13:35
i.e. cd into the libuv folder, "git reset --hard some_tag", cd out, git add libuv-folder, git commit -m 'update libuv to some_tag', done
otherwise if you --has-libuv, you can use the system-provided one
MasterDuke timotimo: thanks, thought i'd see what happens when updating to current stable (1.11.0 vs 1.8.0 we're using now) 13:37
timotimo cool!
regularly updating libuv should be a thing we do 13:38
MasterDuke nqp built and passed `make m-test`. rakudo built but didn't run t/spec/S11-modules/nested.t, however did a whole bunch of `make t/spec/S11-modules/nested.t` manually and they all passed 13:53
and i think i've seen that test fail before during a `make m-spectest`
should i PR the update? 13:55
13:57 dalek joined
IOninja I see nested.t flap once in a blue moon. 13:58
MasterDuke irclog.perlgeek.de/perl6-dev/2017-...i_14068163 13:59
oh, ha, you beat me to it
what's a bash invocation to run something in an infinite loop? 14:01
14:01 pyrimidine joined
MasterDuke nm. running `./perl6-valgrind-m t/spec/S11-modules/nested.t` in a loop, maybe something will turn up 14:03
dogbert17 for i in {1..500}; do ./perl6-valgrind-m t/spec/S11-modules/nested.t; done 14:08
MasterDuke i have three of them running. nothing so far 14:09
dogbert17 IOninja: any other flappers you know about? 14:11
MasterDuke ~430 runs so far, no failures 14:37
timotimo might have to be run as part of the harness for it to blow up 14:39
MasterDuke yeah. does `make <test>` run it under the harness? 14:40
timotimo i don't think it does, but what do i know :)
IOninja while make t/spec/S11-modules/nested.t; do true; done # runs in a loop until it blows up 14:42
Yes, make some-test uses the harness. You can see the command run at the start of output: /home/zoffix/perl5/perlbrew/perls/perl-5.24.0/bin/perl t/harness5 --fudge --moar --keep-exit-code --verbosity=1 t/spec/S11-modules/nested.t 14:43
MasterDuke yeah, i'm running a couple those in a loop with valgrind now 14:44
IOninja dogbert17: I frequently see a TODO pass in S17-procasync/kill.t
timotimo nice
IOninja dogbert17: also this test is fudged on windows due to flapping there: github.com/rakudo/rakudo/blob/nom/....t#L10-L13 14:45
(dunno if that's still the case tho)
dogbert17 does anyone know the return value of 'nqp::seekfh' ? 14:55
timotimo that op doesn't return anything at the moarvm level 14:56
dogbert17 timotimo: interesting in IO::Handle we have this: 14:58
proto method seek(|) { * }
multi method seek(IO::Handle:D: Int:D $offset, SeekType:D $whence = SeekFromBeginning) {
nqp::seekfh($!PIO, $offset, +$whence);
}
the docs on the other hand say 'method seek(IO::Handle:D: Int:D $offset, SeekType:D $whence --> True)'
MasterDuke it doesn't return anything 14:59
dogbert17 should one then assume that if seek does not throw then all has gone according to plan?
MasterDuke src/io/io.h:113:void MVM_io_seek(MVMThreadContext *tc, MVMObject *oshandle, MVMint64 offset, MVMint64 flag); 15:00
dogbert17 so how do one know if a seek has succeeded?
MasterDuke `else MVM_exception_throw_adhoc(tc, "Cannot seek this kind of handle");` 15:01
dogbert17 MasterDuke: cool, so it seems as if the docs are in error then 15:02
MasterDuke that's the only explicit error. it doesn't seem to catch if the seek itself failed 15:05
whoops, the function MVM_io_seek calls does 15:07
github.com/MoarVM/MoarVM/blob/mast...file.c#L81
Geth MoarVM: MasterDuke17++ created pull request #542:
Update libuv to v1.11.0
15:12
IOninja dogbert17: you can leave the docs as is. This will be fixed as part of the IO grant and one of the notes I have concerns the current uselessness of .seeks' return value 15:16
MasterDuke man, these three `valgrind make t/spec/S11-modules/nested.t` have been running in a loop for a while now, no errors
dogbert17 IOninja, cool, do you have fixing 'unlink' and 'tell' on your list as well? 15:22
IOninja dogbert17: don't think so. What's broke with them? 15:23
dogbert17 I believe that calling 'tell' after opening a file for append returns 0
instead of the 'end of the file' 15:24
with unlink, there's some controversy about the order of the args
IOninja dogbert17: it seems to take just the files to delete tho? 15:25
confirmed the .tell one; thanks \o/
MasterDuke IOninja: did you see my comment on github.com/rakudo/rakudo/pull/775? 15:29
dogbert17 IOninja: wrt unlink, if you get an exception (failed unlink) the error message confuses 'src' and 'dest' if I remember correctly
IOninja MasterDuke: nope. Now I did and yeah, I will 15:30
MasterDuke IOninja++
IOninja dogbert17: which src and dest?
Do you mean move or rename and not unlink? 15:31
15:32 MasterDuke joined
dogbert17 hmm, trying to find the discussion on irclog, you might be right though it could have been move or rename 15:33
IOninja neither of them confuses src and dest in error though 15:35
dogbert17 hmm, let me look around a bit :)
MasterDuke timotimo, jnthn: what's involved with fixing the "overflow to negative" boundary detection mentioned here irclog.perlgeek.de/perl6-dev/2016-0...i_13173728 ? 15:42
dogbert17 IOninja: I believe it was 'link' that I found confusing, a failed link looks like thisĀ§:
dogbert@dogbert-VirtualBox ~ $ perl6 -e 'say link("newdata2.txt", "data2.txt")'
Failed to create link called 'data2.txt' on target '/home/dogbert/newdata2.txt': Failed to link file: no such file or directory
the new file I want to create here is newdata2.txt 15:43
the docs say 'Create a new link named as $target (or the name of the invocant in the method form) to the existing file named $name.' 15:46
IOninja dogbert17: thanks. Added to my notes. Will fix the issue and the docs. 15:47
... and write tests -_- not a single one for `link` 15:48
dogbert17 IOninja++
IOninja: you recently commented on RT #123838, someone had added some tests to that case earlier, might possibly come in handy 15:50
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=123838
IOninja will look for them, thanks 15:51
16:23 zakharyas joined 16:43 pyrimidine joined 18:27 zakharyas joined
TimToady IOninja: so I guess with RatStr.FatRat we'll just need to tear apart the string part ourownselves; maybe we can decide that at compile time, and maybe not 18:33
maybe add a third sekrit field that says "here's what you want if you want a FatRat"
IOninja Alright. 18:43
18:43 pyrimidi_ joined 18:48 pyrimidine joined 19:02 pyrimidine joined 20:44 BenGoldberg joined