Parrot 4.6.0 "Wild Parrots of Telegraph Hill" | parrot.org/ | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC
Set by moderator on 18 July 2012.
02:05 whiteknight left 02:40 contingencyplan joined 02:47 contingencyplan joined 03:01 l3l1p joined 03:19 contingencyplan joined 03:21 contingencyplan joined 03:36 contingencyplan joined 03:51 contingencyplan joined 04:03 contingencyplan joined 04:04 benabik joined 04:33 particle joined 05:33 fperrad joined 05:39 fperrad_ joined 05:48 benabik joined 06:08 JimmyZ joined 07:42 Khisanth joined
dalek kudo/parrot-iocleanup1-take2: 942e670 | moritz++ | src/core/IO/Socket/INET.pm:
explicitly set encoding on sockets
07:43
kudo/parrot-iocleanup1-take2: 7090070 | moritz++ | src/core/IO/Socket.pm:
Do not trans_encoding in IO::Socket.recv
moritz msg whiteknight I still have zero passing rakudo socket on io_cleanup1 commit 6d12b98f, still get "got: ''" on all tests. I see why you might get Unicode trouble if other stuff works for you, the parrot-iocleanup1-take2 branch of rakudo might or might not fix that for you -- no way for me to tell :( 07:44
aloha OK. I'll deliver the message.
dalek kudo/nom: 9db2d9b | jonathan++ | src/core/IO.pm:
Fix cwd() to transncode the result. Avoids a problem that gets in the way of Panda on Windows.
08:26
09:31 lucian joined 10:26 alvis joined 10:36 tuxit joined 12:22 brrt joined 12:27 whiteknight joined
brrt hi whiteknight 12:44
whiteknight good morning brrt 12:58
brrt hows life :-)
whiteknight msg moritz what platform are you on? I might be able to test this on a better platform to replicate your failures 12:59
aloha OK. I'll deliver the message.
whiteknight brrt: good enough. Low on sleep but good otherwise
brrt yeah, i know how that feels
moritz whiteknight: debian testing, amd64
brrt hey, i have a segfault in parrot with nqp on x86-64
actually, during the building of nqp 13:00
whiteknight moritz: Okay, that's pretty darn similar to my kit. I wonder why we get such different results
brrt: on master?
brrt yeah, i think so 13:01
address out of bounds
in hash
whiteknight brrt: And what platform are you on?
brrt linux (fedora 17), x86-64
i'll post a backtrace 13:02
13:04 kid51 joined
brrt its extremely hot out here.. and i'm quite sleepy though 13:04
whiteknight bleh 13:05
moritz: those sockets failures you're seeing, those are on nom, right?
moritz whiteknight: nom with one custom patch from the parrot-iocleanup1 rakudo branch 13:06
whiteknight what's the one custom patch?
moritz commit d0dc2d33d0a5eabc1a781b52c253d1f1d85d0a61, explicit set encoding on sockets
whiteknight ok
moritz or should i not need it?
brrt issue posted :-) 13:10
also
does not happen 'all the time'
so 13:11
not really the most repeatable of bugs
.. may not be a parrot bug 13:12
whiteknight moritz: I don't know. let me test around and see what's going on 13:15
brrt: backtrace would be great
dalek kudo/macros-d2: 0512eca | masak++ | src/ (3 files):
intermediate commit

Pushed to help jnthn debug a hang.
13:18
13:20 UncleFester6 joined
UncleFester6 dukeleto: ping 13:20
dalek p/jnthn-debug-macros-d2-on-this: 1be56d4 | masak++ | src/QAST/ (9 files):
intermediate code for debugging

For some reason, this hangs inside src/QAST/Unquote.nqp
13:21
kudo/nom: 2bbe527 | jonathan++ | src/core/IO.pm:
Fix a few more Windows/OS PMC/encoding issues for tadzik++.
13:25
brrt backtrace is posted with issue :-) 13:37
github.com/parrot/parrot/issues/815
dalek rrot/whiteknight/io_cleanup1: d10ef89 | Whiteknight++ | src/io/ (6 files):
Fix several build warnings in src/io/*. Several comparisons with older datatypes expect to be non-negative so cast things to size_t explicitly. Mark a few parameters unused
13:43
rrot/whiteknight/io_cleanup1: c9c0625 | Whiteknight++ | src/io/ (8 files):
silence several errors received with g++, mostly involving IO_VTABLE initialization where we play a little bit loose with const qualifiers.
whiteknight brrt: what compiler do you use? 13:44
brrt: And this is on unpatched master? 14:01
14:04 brrt joined
brrt msg whiteknight I use gcc 4.7 14:04
aloha OK. I'll deliver the message.
whiteknight okay 14:06
brrt: I can't reproduce your segfault. I've built that file about dozen times now 14:07
brrt i've totally forgotten what i was planning to do today w/regards to mod_parrot
whiteknight heh
brrt i'll see what checkout i have
UncleFester6 whiteknight: ping 14:08
brrt i have da9adfa03b21611da353a8627cd52cc6f4ed204d as my head 14:09
UncleFester6 whiteknight: I was hoping to ask a question about the parrot base64 library and character encodings ... 14:13
whiteknight UncleFester6: pong 14:16
UncleFester6: what's the question? 14:18
UncleFester6 I posted a bug report yesterday about Base64 and the way it does encoding to codepoints. I think the basic problem is that Base64 encoding should be on ByteBuffer with some string interfaces.
whiteknight yes, that makes sense 14:19
Rosella has a Base64 implementation which might be more correct. I'm testing that now
UncleFester6 thanks - I just wanted to verify I was headed in the right direction ... 14:21
whiteknight UncleFester6: yes, the Rosella implementation is correct if you want to use that 14:25
14:26 JimmyZ joined
UncleFester6 Rosella is winxed? Might there still be room for native pir or Perl6 implementations? 14:27
whiteknight oh yes, there's always room for other implementations too 14:28
dalek sella: 8f1783e | Whiteknight++ | src/test/Asserter.winxed:
[Test] Update Asserter.__default_message to format the string for more expressive error messages
14:32
sella: 9f92884 | Whiteknight++ | t/net/MimeBase64.t:
[Net] Add a test for MimeBase64 showing a UTF8 character encoded, which causes problems in other implementations
15:59 lucian joined 16:33 UncleFester6 joined
dukeleto ~~ 16:34
UncleFester6: howdy. I like your nick :)
UncleFester6: are you @donaldxs on github ? 16:35
UncleFester6 thanks did you see my discussion with whiteknight? - (ronaldxs)
dukeleto: I noticed you started a new parrot branch related to base64 - are you planning further work in the area? 16:37
dalek rrot/gh813_base64_utf8: 279b61b | dukeleto++ | t/library/mime_base64.t:
Improve and refactor failing #813 enc_sub test
16:39
dukeleto UncleFester6: i just ported your example code to a failing test 16:40
UncleFester6: anybody else is free to make the test pass :)
UncleFester6: i am not sure I know enough to make it pass
UncleFester6: do you think it is worth it to have tests for enc_sub on other utf8 strings? Or possibly different encodings? 16:41
UncleFester6 My essential point to whiteknight is that base64 is intended to ecode binary stuff
including images, audio as well as string encodings 16:42
So the base64 stuff should IMHO be encoding ByteBuffer instead of string 16:43
dukeleto UncleFester6: ah, interesting. If we made that change, what should happen when you try to encode("some string") ?
UncleFester6 but since string is a common usage there should also be an interface to utf8/utf16/ucs2 encode strings into ByteBuffers for base64 encode 16:44
dukeleto UncleFester6: should it throw an exception that only ByteBuffers can be encoded?
UncleFester6 did my last sentence make sense ?
dukeleto UncleFester6: i assume there is a way to convert strings -> ByteBuffers and back?
UncleFester6: yes, i think i understand what you are saying
UncleFester6 one moment ...
dukeleto hasn't used ByteBuffers much, but is familiar with what they do 16:45
UncleFester6 sample .pir gist here gist.github.com/3388310 16:47
I don't know parrot that well ... 16:48
dalek rrot/gh813_base64_utf8: 81f57c6 | dukeleto++ | t/library/mime_base64.t:
Improve the name and description of mime_base64.t
16:49
dukeleto UncleFester6: seems reasonable
UncleFester6 Any one last final point for whitknight if he comes across our chat ... his Rosella/winxed base64 still wants and string encoding for the "decode" function which I claim is off ... 16:51
You can decode pure binary data as well as strings so you might want a string encoding but you might just want a binary buff back too ...
16:51 l3l1p joined
UncleFester6 thanks for taking time to chat about this .. I am tempted to try to do a Pure Perl6 implementation of base64 with this approach for now and might try .pir later ... 16:53
dukeleto UncleFester6: that is a good point 16:55
UncleFester6: about wanting to decode to binary
UncleFester6: github.com/moritz/Perl6-MIME-Base64 16:56
UncleFester6: what about that?
UncleFester6: looks like that is the currently maintained fork from @snarkyboojum
UncleFester6 If you look at the code it depends on the current Parrot MIME::Base64 library so no help 16:57
dukeleto UncleFester6: blarg, that just calls down to Q:PIR { }
UncleFester6: just saw :)
UncleFester6 It once had a pure perl implementation which I might try to revive and update
s/perl/perl6 16:58
dukeleto UncleFester6: if you have time, you might want to send an email to parrot-dev with some of your comments above. Many parrot devs don't IRC and just live on the mailing list. You might get some good/helpful feedback
UncleFester6: in any case, good luck! 16:59
17:17 UncleFester6 left 17:28 UncleFester6 joined 17:29 UncleFester6 left 19:57 benabik_ joined 20:07 contingencyplan joined 20:12 contingencyplan joined 20:22 contingencyplan joined
benabik ~~ 21:14
22:04 contingencyplan joined 22:07 whiteknight joined 23:14 benabik joined
dalek p: 6255819 | pmichaud++ | src/QAST/Operations.nqp:
Enable parsing of :nohandler in QAST::Op while/etc loops.
23:50
p: b1d897d | pmichaud++ | src/QAST/Operations.nqp:
Add :nohandler option to PAST::Op loop nodes.
23:57 lucian joined