FROGGS this is the bad libuv commit: github.com/joyent/libuv/commit/b19...b306f9474e 00:23
tomorrow I'll try to revert it locally on top of libuv latest... we'll see 00:24
gnight
00:52 avuserow joined 00:55 cognome joined
leedo FROGGS: wow i never would have found that, nice job :o 00:55
[Coke] FROGGS++ FROGGS++ # That's dedication. 01:11
diakopter FROGGS++ # impressive as all get out :) 01:15
01:29 FROGGS_ joined
japhb So since the description says that it is a retry of a revert, after fixing a "select trick" for OS X ... what is different for them such that this trick works for them and not us? 01:47
02:04 cognome joined 02:49 vendethiel- joined
[Coke] older version of OS X? 02:52
03:09 cognome joined 03:50 JimmyZ joined 05:44 Ven joined 05:46 JimmyZ_ joined 06:28 cognome joined 06:30 JimmyZ_ joined
sergot hi o/ 06:31
06:37 FROGGS[mobile] joined 06:50 FROGGS joined 07:26 zakharyas joined 07:39 JimmyZ_ joined 07:47 JimmyZ_ joined 08:29 itz_ joined 08:30 JimmyZ_ joined, cognome joined 08:47 colomon joined
dalek arVM/libuv-upgrade: 2af01a9 | (Tobias Leich)++ | src/io/procops.c:
we are not supposed to open a children's pipe to our stdin

We hand back the other side of the pipe in order to read from it, so it does not make any sense to ground it to stdin. Same goes for the other direction, where we want to print to the children.		 09:54
FROGGS leedo: you've got now a working rakudo on you mac :o) 10:16
your*
dalek arVM/smoke-me-libuv-upgrade: 20eb200 | (Tobias Leich)++ | / (6 files):
Revert "Revert "update libuv from 0.11.18 to 0.11.29""

This reverts commit b24c94bf32fc312d6793aab2cad115ebfe908921.		 10:20
arVM/smoke-me-libuv-upgrade: 6a1b339 | (Tobias Leich)++ | src/io/syncfile.c:
Revert "Revert "uv_fs_write also takes an uv_buf_t instead of char*""

This reverts commit 93d47ddcef3e3fb308dd158b6273fa997038ba24.
arVM/smoke-me-libuv-upgrade: 8f70d1a | (Tobias Leich)++ | src/io/procops.c:
we are not supposed to open a children's pipe to our stdin

We hand back the other side of the pipe in order to read from it, so it does not make any sense to ground it to stdin. 3519a6d | (Tobias Leich)++ | src/io/procops.c: remove more troublemaking calls to uv_pipe_open, JimmyZ++
10:27 travis-ci joined
travis-ci [travis-ci] MoarVM build passed. Tobias Leich 'we are not supposed to open a children's pipe to our stdin 10:27
[travis-ci] travis-ci.org/MoarVM/MoarVM/builds/35000245 github.com/MoarVM/MoarVM/compare/2...70d1a32ff6
10:27 travis-ci left
jnthn FROGGS++ # outstanding work 10:33
nwc10 good UGT heresy, jnthn
jnthn o/
FROGGS jnthn: it wasnt me after all :o) 10:35
saghul++ # really
11:30 oetiker joined
nwc10 FROGGS: origin/libuv-upgrade passes all* spectests on x86_64 when built with ASAN 11:57
previously things hung
but on Power all things involving shelling out hang
* except already failing spectests. Naughty #perl6 devs
FROGGS nwc10: what about smoke-me-libuv-upgrade? 11:59
nwc10 tretries
retries
also, previously new libuv failed assertions on Power 12:00
that's gone
FROGGS yeah, these #perl6ers are just loonies :o) 12:01
jnthn: smoke-me-libuv-upgrade is purrfect AFAICT... 12:09
of course, I'd like to know if it works on lizmat++'s box before merging anything :o)
lizmat building now 12:16
FROGGS lizmat: in case rakudo builds, please also test a 'say qx/sw_vers/' 12:19
lizmat yes, that was the first thing I was going to test 12:20
if that didn't work, the problem still exists
$ perl6 -e 'say qx/sw_vers/' 12:22
ProductName:Mac OS X
ProductVersion:10.9.4
BuildVersion:13E28
aka Success!
FROGGS \o/
lizmat running spectest now
FROGGS okay...
I'm still curious why shell() hangs on powerpc... 12:23
nwc10 I doubt that I have time to dig into it and figure out the answer 12:25
lizmat came out of spectest with no unexpected errors
JimmyZ FROGGS: I see there still is some uv_pipe_open , which is in stderr but not stdout, is it right? 12:28
FROGGS JimmyZ: where? 12:30
lizmat: awesome! thank you :o) 12:31
lizmat FROGGS: no, thank you! :-)
12:31 JimmyZ_ joined
JimmyZ_ FROGGS: github.com/MoarVM/MoarVM/blob/8f70...#L834-L845 12:31
FROGGS JimmyZ: ohh, yeah, that is most likely wrong too 12:32
FROGGS damn
wrong lines *g*
dalek arVM/libuv-upgrade: d0f7340 | (Tobias Leich)++ | src/io/procops.c:
remove more troublemaking calls to uv_pipe_open, JimmyZ++
 12:36
FROGGS so libuv-upgrade will be the branch I'm going to merge
nwc10 tests 12:37
FROGGS nwc10: and shell() succeeded on powerpc before, is that right?
or... it just did not hang
nwc10 yes, it worked on the previous version of libuv that moar bundled 12:38
lizmat builds with latest changes
JimmyZ github.com/MoarVM/MoarVM/compare/libuv-upgrade # I didn't see any libuv upgrade?
FROGGS dang 12:39
lizmat brb
FROGGS JimmyZ: that's because there are reverted commits on master that did not get reverted in that branch 12:41
JimmyZ oh
dalek arVM/libuv-upgrade: c924cb0 | (Tobias Leich)++ | Configure.pl:
Revert "forcing a realclean to get rid of old libuv"

This reverts commit cf8088a28e894610309b0b7327de0d5b7dfcafb2.		 12:42
FROGGS sorry dalek /o\
12:43 dalek joined
nwc10 FROGGS: seems that problem exists between the keyboard, chair and git submodules 12:43
FROGGS nwc10: that's what I wanted to hear! :D
I am going to merge now 12:47
dalek Heuristic branch merge: pushed 16 commits to MoarVM by FROGGS 12:48
12:50 FROGGS[mobile] joined
lizmat FROGGS[mobile]: I guess we need REVISION bumps again? 13:03
FROGGS[mobile] we can do that, yes 13:04
lizmat ok, I'll take care of it
FROGGS[mobile] lizmat++ 13:06
13:25 JimmyZ_ joined 13:41 JimmyZ_ joined 13:57 JimmyZ_ joined 14:15 JimmyZ_ joined
jnthn back :) 14:20
14:21 JimmyZ_ joined 14:22 JimmyZ__ joined
leedo FROGGS++ awesome work! 14:23
jnthn Indeed :) 14:28
nine jnthn: how is MoarVM supposed to know when I'm not using NativeCall callbacks anymore so it can free the memory (and objects referenced from the closures)? 14:29
jnthn It doesn't. 14:30
For now, don't use closures there unless they have program lifetime.
14:30 JimmyZ_ joined
jnthn Just use a non-closure sub instead. 14:30
We should probably have a mechanism for letting you release them. 14:31
JimmyZ_ good night, moarvm 14:32
nine jnthn: the whole point of those callbacks is that they are closures. I use them to get back at the Perl 6 object from the Perl 5 wrapper.
jnthn: if you have another idea on how I can get back the Perl 6 object I'm all ears! 14:33
jnthn OK, then (uness you want to maintain a closure pool yourself) it'll need some work to let them be free-able somehow 14:34
14:35 xiaomiao joined, brrt joined
nine Actually using closures for this is not my favourite solution anyway. It's just the only solution I found. 14:35
I'd much rather just store a pointer in my Perl 5 object and use that to get back at the Perl 6 object from some (non-closure) callback 14:36
jnthn: head -> table. The solution is so obvious and simple it's really embarassing. 14:49
jnthn ;)
jnthn knows that feeling :)
nine jnthn: without closures I have to reference the wrapped Perl 6 objects manually from Perl 6 code. I'd just push them to an array. Now I can use the index in the array and store it in my Perl 5 wrapper object and use it to get back to the object in Perl 6. 14:50
jnthn Sounds workable :) 14:59
[Coke] jnthn: JNTHN! 15:01
(nothing, just hi. :)
jnthn :)
Hi from...a hotel which is having a power outage
Oddly, the wifi network is still up. :) 15:02
15:09 brrt joined
[Coke] priorities 15:14
15:34 JimmyZ joined
JimmyZ jnthn: I reported a bug in #perl6 about thread :) 15:34
jnthn Since when was #perl6 a bug tracker? :P 15:35
Please pop it in RT, or Moar's issue tracker if it's VM-specific :)
JimmyZ jnthn: irclog.perlgeek.de/perl6/2014-09-11#i_9335568
I don't know whether is MoarVM or rakudo ...
really.. 15:36
jnthn Well, segv is suggestive of VM... 15:37
JimmyZ github.com/MoarVM/MoarVM/issues/140 15:39
'night
jnthn Thanks :) 15:41
Fun fact: Hong Kong's underground has a station called Long Ping. Guess that's *not* a place to find fast internets... :) 16:08
TimToady unless that refers to the size of the packet, not the latency 16:09
jnthn heh, true :) 16:10
Well, that's my route planning for tomorrow evening done. 16:19
jnthn is having a weekend in HK :)
Time for sleep...'night
brrt 'night jnthn 16:36
nwc10 still continues to be pleasantly surprised how (relatively) unbroken Power is. 18:33
timotimo FROGGS[mobile]: why is there a travis.yml and a travis.yaml? 18:58
19:04 FROGGS joined
FROGGS[mobile] timotimo: the .travis.yaml should go 19:08
timotimo OK 19:09
FROGGS[mobile] from reading the backlog it seems like there was no complaints until now? 19:18
19:22 brrt joined
dalek arVM: 095bb8b | (Tobias Leich)++ | .travis.yaml:
remove duplicate travis config file
 19:23
timotimo ah, good 19:26
19:37 TimToady_ joined 19:58 cognome joined 20:08 Ven joined 20:34 brrt joined 22:26 colomon joined 22:52 carlin joined

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!