Parrot 4.5.0 "Buff-faced Pygmy Parrot" | parrot.org/ | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC
Set by moderator on 27 June 2012.
whiteknight these socket semantics make me so angry 00:02
00:18 Timbus joined
dalek sella: d770c23 | Whiteknight++ | / (2 files):
Add in a test file for some new xml features, from some ad hoc tests I was using.
 00:35
sella: 4897e48 | Whiteknight++ | src/net/protocol/Http.winxed:
Set a fixed_8 encoding on the socket, since we need to be able to read something without throwing exceptions. Add in some loop logic to the chunked encoding to make sure we can read large chunks
sella: d02a703 | Whiteknight++ | src/unstable/xml/ (3 files):
[Xml] Add in a strict flag to the parser, on by default. If the flag is off, we can attempt to work around some errors, like tags which are not properly closed (common in HTML docs). This allows us to parse XML docs with unbalanced tags
01:11 kid51 joined 01:23 kid51_ joined
dalek rrot/whiteknight/io_cleanup1: c841732 | Whiteknight++ | / (7 files):
Introduce a new PIO_READ_SIZE_ANY flag that instructs the read path to grab a large chunk of data (CHUNK_SIZE or however much the buffer can hold).
 01:37
01:48 mtk joined
kid51 whiteknight/io_cleanup1: now passes 'make test' with --cc=g++|--link=g++|--ld=g++ ... 01:57
... but it's still failing test 14 in t/pmc/filehandle.t when I add --optimize to that
See smolder.parrot.org/app/projects/rep...ails/29941 01:58
nopaste "kid51" at 70.85.31.226 pasted "iocleanup1 branch: optimized all g++ build still has 1 test failure" (57 lines) at nopaste.snit.ch/145485 01:59
"kid51" at 70.85.31.226 pasted "iocleanup1 branch: warnings during all-g++ 'make'" (43 lines) at nopaste.snit.ch/145486 02:07
kid51 With all optimized gcc build, io_cleanup1 branch still fails tests 14 and 28 in t/pmc/filehandle.t (linux/i386) 02:08
And io_cleanup1 fails those same two tests with non-optimized, all-gcc build on linux/i386 02:10
So, a curious case, where a non-optimized all-g++ build is getting the best results.
io_cleanup1 branch on Darwin/PPC with --cc=gcc --link=g++ --ld=g++, not optimized, fails test 28 in t/pmc/filehandle.t 02:26
Now wait, this is curious.
The output of that failure includes: Write error: No space left on device 02:27
... even though df shows me at 93% full.
But during the course of this 'make test', I got a popup warning that my "startup disk was nearly full" 02:28
And I got that same warning on this machine a couple of days ago.
probably also while running make test in io_cleanup1 branch (never saw this warning previously)
I don't really know what is meant by my "startup disk"
nopaste "kid51" at 71.255.50.7 pasted "io_cleanup1 branch on Darwin/PPC: t/pmc/filehandle.t" (12 lines) at nopaste.snit.ch/145490 02:29
02:32 contingencyplan joined
nopaste "kid51" at 71.255.50.7 pasted "'df' output on box where t/pmc/filehandle.t test 28 is failing with 'No space left on device'" (9 lines) at nopaste.snit.ch/145491 02:32
benabik kid51: Startup disk is whatever your OS is stored on, and is generally the one mounted as / 02:43
02:48 contingencyplan joined
kid51 benabik: But I know that / never gets more than 95% full 02:49
And I've run 'make test' on this machine hundreds of times without seeing that popup.
So I think something particular to this branch/test is causing the problem.
hopefully whiteknight will figure it out 02:50
kid51 must sleep
04:42 aloha joined 06:59 brrt joined 08:10 lucian joined 11:12 JimmyZ joined
brrt i just had an idea, i'm going to cheat and look how mod_perl makes a persistent interpreter 11:20
JimmyZ cheat was always a good idea when no idea 11:23
*is
moritz especially when it's not cheating, but looking for prior art :-) 11:27
brrt :-) however you call it 11:28
11:49 mtk joined 12:12 alvis joined
brrt a pool, is how they do it 12:16
moritz ooh, a pool would be greatly appreciated in this warmth
brrt it is quite waqrm 12:19
warm
13:00 PacoAir joined 13:10 Psyche^ joined 13:13 kjs joined 13:37 bluescreen joined 13:45 JimmyZ joined 14:03 alvis joined 14:15 alester joined
brrt terribly complex pool, that, too 14:20
but nevertheless 14:21
brrt will try something simpler obviously
14:57 whiteknight joined
whiteknight good morning, #parrot 14:57
15:08 bluescreen joined
brrt good morning, afternoon, whiteknight 15:19
whiteknight hello brrt
brrt i've looked at how mod_perl gets an persistent interpreter
they use an interpreter pool
their implementation is slightly obscure 15:20
whiteknight great! sounds straightforward and easy to do
brrt functions name ATKDN
sCFBGG
or something :-)
but, yes, straightforward
15:22 brambles joined 15:24 dmalcolm joined 15:28 brambles_ joined 16:22 kjs joined
dalek nxed: cab36aa | NotFound++ | winxedst1.winxed:
simplify new operator in stage 1
 16:59
17:54 birdwindupbird joined
dalek kudo/nom: cc780c2 | moritz++ | src/core/IO.pm:
move file tests to IO::FileTestable
 18:13
kudo/nom: 39087ad | moritz++ | src/core/IO.pm:
Merge branch 'io' into nom
18:22 kjs joined 18:56 lucian joined 18:58 brambles joined 19:15 tuxit joined
dalek : 951d8f6 | kjs++ | / (5 files):
move array bounds check to semcheck.
 19:39
p/sigspace: b49ec74 | pmichaud++ | src/ (2 files):
Leading whitespace at beginning of regex isn't significant.
 19:52
p/sigspace: 5249e92 | pmichaud++ | src/QRegex/P6Regex/Grammar.nqp:
Allow <.ws> at the beginning of an arglist.
: e69f6fb | kjs++ | src/gencode.c:
fix 2 bugs in gencode for new INS infrastructure.
 20:20
20:30 contingencyplan joined 20:44 Liz joined 20:49 kjs joined
dalek : 4af3a22 | kjs++ | / (2 files):
fix 2 more bugs.
 20:51
21:49 bluescreen joined 21:51 perlite_ joined 22:14 kjs joined
dalek : 4a29348 | kjs++ | src/ (3 files):
more corrections for INS infrastructure.
 22:16
: 4b1c328 | kjs++ | src/gencode.c:
a few more fixes.
: ff30348 | kjs++ | src/gencode.c:
fix last bug to make test pass with new infrastructure.
23:17 whiteknight joined
whiteknight good evening, #parrot 23:18
benabik o/ 23:20
cotto: I blogged: www.parrot.org/content/pact-time-fl...ting-code.
whiteknight benabik++ 23:21
benabik For a more general audience, this is the current output of the PACT disassembler: gist.github.com/3014648
Far less pretty than it used to be, but fixing that is in the plans.
cotto benabik++
benabik dinner & 23:24

Powered by Cro and the Raku® Programming Language.

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

Thank you!