»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
timotimo <3 00:00
how's the memory usage?
like, maxrss or something?
00:00 hoverboard joined 00:02 denisboyun left
jnthn supernovus: For tests on JVM, you're mostly paying startup time... 00:03
timotimo most probably
supernovus jnthn: That's what I was assuming, as 'prove' spawns a separate process for each test.
jnthn aye 00:04
supernovus Well, I'm certainly impressed with the performance of MoarVM so far. I haven't tested any of my socket/networking libraries yet, or anything using Buf/pack/unpack. 00:05
timotimo do we have pack and unpack on moarvm yet?
jnthn We've lots more work to do on MoarVM yet... :) 00:06
supernovus Not a clue, I haven't tried it yet, as the FastCGI library is one of the few of mine that uses it, and given how horrifically slow it was the last time I ran it, I stuck with SCGI for any web apps I was building in Perl 6. 00:07
timotimo pack and unpack seem like a prime target to get the specializer really going, no? 00:08
that was not english. 00:09
jnthn timotimo: Probably not with the way they're written at the moment...
timotimo the specializer should get a good speed boost out of that; if the pack specification is constant, it *could* be specialized away
jnthn Well, but the VM has no clue it's a pack/unpack 00:10
00:10 eternaleye joined
timotimo mhh 00:10
the pypy tracing jit can just do that :P
jnthn I know they do stuff with formats, yes. 00:11
supernovus Well, I'll test some more of my libraries out on the various backends another day. Have a good * #perl6 :-) 00:12
00:13 supernovus left
timotimo it's not that they "do stuff with formats" 00:13
it kind of mostly falls out of the tracing jit without much extra stuff
jnthn Ah, found the article 00:14
Yes, I see how it can work. 00:15
timotimo oh, i might be proven wrong :)
jnthn If the string formatter is not opaque to the tracing JIT it can figure out the paths taken.
timotimo i'm kind of sad we don't get a tracing jit like they have easily, because i'm really, really, really impressed by their stuff
jnthn Tracing isn't everything.
timotimo aye, but traces seem to be pretty good for getting turned into super efficient low-level code 00:16
you obviously have to do good work with the guards and bridges, of course 00:17
00:17 masak left, PerlJam left
jnthn Yes; if you can't figure out enough to do sensible method JIT then tracing can really come into its own. 00:17
timotimo well, all my knowledge of the tracing jit stuff is from reading what cool stuff they suddenly made faster with hardly any work :P
00:18 masak joined, PerlJam joined
jnthn It wouldn't surprise me if Moar ends up doing traces on hot paths also. 00:18
timotimo fair enough
raydiak daydreams about rakudo-lua running Pray under LuaJIT 00:25
00:26 eternaleye left 00:28 skids joined 00:32 eternaleye joined 00:33 lizmat joined 00:35 kivutar_ left 00:38 berekuk left 00:41 bjz_ joined, dmol joined
jnthn sleep time...should have tuits tomorrow :) 00:42
&
raydiak gnight jnthn
00:45 bjz left, cognominal left 00:47 woolfy joined 00:51 dmol left 00:53 cognominal joined 01:09 hoverboard left 01:10 skids left 01:11 pecastro left 01:15 pecastro joined 01:19 skids joined 01:29 cognominal left, cognominal joined, fridim_ joined 01:33 hoverboard joined 01:51 bjz_ left
raydiak does not understand why tests which pass with perl6, prove, and panda-test, fail with panda install 01:56
02:01 bjz joined 02:13 slavik left 02:16 telex left 02:18 telex joined 02:20 FROGGS left 02:21 Sqirrel_ joined 02:22 Sqirrel left, hoverboard left 02:23 eternaleye left 02:28 hoverboard joined, eternaleye joined 02:34 slavik joined 02:35 FROGGS joined 02:36 dayangkun joined 03:40 jnap left 04:10 jnap joined 04:15 jnap left
colomon raydiak: might be something wrong with serializing the compiled code during the install... 04:15
I know in the (somewhat distant) past I saw situations where ABC could not be precompiled, but would run just fine if the .pms were called directly. 04:17
04:26 araujo left, SamuraiJack__ joined 04:41 rurban left 04:42 SamuraiJack__ left 04:49 fridim_ left 05:02 kaleem joined 05:11 jnap joined 05:16 jnap left 05:26 hypolin joined, hypolin left 05:41 zby_home_ joined 05:52 BenGoldberg left 06:08 SamuraiJack__ joined 06:39 dmol joined 06:40 dmol left
tadzik FROGGS: Ooh, it should definitely be absolute 07:00
moritz s/definitely/absolutely/ # :-) 07:03
good morning everybody
tadzik :) 07:04
07:08 hoverboard is now known as moistcherry 07:20 FROGGS left 07:21 moistcherry is now known as hoverboard 07:25 dmol joined 07:36 darutoko joined 07:40 Vlavv left 07:43 dayangkun left, FROGGS joined 07:55 zby_home_ left, dayangkun joined 07:56 Vlavv joined, zakharyas joined
FROGGS timotimo: about PackFile, I think we'd need to change quite a bit because we are generating pir code in some places in nqp, and my guess is that this will need to go away 07:58
(in case it actually is PIR and not PASM) 07:59
morning o/ 08:00
08:03 flussence left 08:05 Mouq joined
Mouq o/ 08:05
FROGGS o/
08:06 flussence joined
dalek kudo/pod-code-allow: 0bc8fb7 | Mouq++ | src/Perl6/Grammar.nqp:
Separate =code etc. from =comment in grammar
 08:07
kudo/pod-code-allow: 39308e5 | Mouq++ | src/Perl6/ (3 files):
Mostly working Pod code block refactor
Mouq ^^ this is meant to eventually fix =begin code :allow<X Y Z> 08:10
08:13 hoverboard is now known as water
Mouq Whoa 08:14
It already fixes it! :D
There are just some issues with the grammar that need to be fixed, but the :allow<X Y Z> thing works 08:15
And then there's some things to be fixed with Pod::To::Text, because it tries to stringify Pod structures in code 08:16
08:20 virtualsue joined
moritz Mouq++ # awesome grammar and pod hacking 08:23
08:24 berekuk joined 08:34 pdcawley joined
nwc10 good *, #perl6 08:40
FROGGS hi nwc10++ 08:41
moritz \o
08:42 Nei left 08:43 water left 08:44 kaare_ left 08:54 thou left 08:57 berekuk left 08:59 rindolf joined 09:12 erkan joined 09:22 berekuk joined
jnthn morning, #perl6 09:26
moritz mrnng, jnthn
JimmyZ morning, jnthn 09:28
09:31 dakkar joined 09:33 bjz left
dalek kudo/pod-code-allow: 5accaed | Mouq++ | src/Perl6/Grammar.nqp:
$*ALLOW_CODE –> $*ALLOW_INLINE_CODE

  $*ALLOW_INLINE_CODE is more descriptive of what it's actually for		 09:39
kudo/pod-code-allow: e2b8379 | Mouq++ | src/Perl6/ (2 files):
Save text verbatim when in a code block
tadzik Mouq++
sergot o/ 09:40
tadzik \o
09:47 bjz joined 09:48 dakkar left 09:49 crab2313_ joined 09:53 dakkar joined 09:56 darutoko- joined 09:58 kaare_ joined 09:59 darutoko left 10:01 berekuk left 10:04 kaleem left 10:05 darutoko joined, kaleem joined 10:07 darutoko- left 10:12 dakkar left, kivutar joined 10:13 dakkar joined 10:16 dalek left 10:17 dalek joined, ChanServ sets mode: +v dalek 10:34 darutoko- joined 10:36 darutoko left 10:41 cognominal left
dalek kudo/pod-code-allow: f3b5470 | Mouq++ | src/Perl6/ (2 files):
Parse empty lines in a =begin code block
 10:55
kudo/pod-code-allow: efce067 | Mouq++ | src/Perl6/Actions.nqp:
Accept empty code block + unshift recursive stucture

Prior, "=begin code\n=end code" would die. As well, it's more effecient to start with the already made ast and unshift new elements constructing the tail-recursive delimited_code_content
Mouq Only big issue now that I've noticed (besides having to modify the Pod -> $format code) is that indented =begin code blocks aren't parsed correctly, fairly inexplicably 10:59
jnthn Did anybody have any r-m bugs they especially wanted me to look into before the monthly? 11:04
11:04 rurban joined 11:10 rurban left
Mouq r: grammar Foo {token TOP {$<f>=[a+] \h+ <again($<f>)>}; token again($what = '') {$what}}; say(Foo.parse('aa aa')) 11:11
camelia rakudo-parrot 9025c3, rakudo-jvm 9025c3, rakudo-moar 9025c3: OUTPUT«#<failed match>␤»
Mouq Yeah
Should that work? 11:12
jnthn Mouq: Try a {} before the <again(...)>
Mouq r: grammar Foo {token TOP {$<f>=[a+] \h+ {} <again($<f>)>}; token again($what = '') {$what}}; say(Foo.parse('aa aa'))
camelia rakudo-parrot 9025c3, rakudo-jvm 9025c3, rakudo-moar 9025c3: OUTPUT«｢aa aa｣␤ f => ｢aa｣␤ again => ｢aa｣␤␤»
Mouq ...
11:12 darutoko joined
Mouq I don't understand 11:12
But that should fix some issues
jnthn We generate the Match object $/ on demand.
And <rule(args)> is currently not triggering its creation. 11:13
{} on the other hand does.
Mouq jnthn: Ok, thank you. This is a bug, though, right? 11:14
jnthn In full-blown Perl 6, yes. In NQP...we get to choose.
Mouq jnthn++
ok
11:14 darutoko- left
jnthn It's just that if we make it always sync the $/ before <rule(args)> then things like <pblock(1)> would trigger it needlessly 11:15
Unless we implement more sophisticated analysis.
Mouq :D it works 11:16
jnthn \o/
11:17 darutoko- joined 11:21 darutoko left, tgt joined 11:22 darutoko joined
dalek kudo/pod-code-allow: 354419d | Mouq++ | src/Perl6/Pod.nqp:
Remove unused sub in Perl6/Pod.nqp
 11:24
kudo/pod-code-allow: 24a2945 | Mouq++ | src/Perl6/Grammar.nqp:
Make sure $/ is generated before calling Pod rules that need it

Fixes issues with extended configure lines and ending delimiters when these things are indented
11:25 darutoko- left 11:28 tgt left
dalek p: 321e13b | jnthn++ | tools/build/MOAR_REVISION:
Bump to latest MoarVM, for macro -> inline.
 11:29
Mouq –> school 11:30
o/
11:34 tgt joined 11:37 crab2313_ left 11:38 Mouq left, tgt left 11:44 darutoko- joined 11:48 darutoko left, darutoko joined 11:49 darutoko- left 11:53 darutoko- joined 11:55 kaleem left, kaleem joined, darutoko left 11:56 darutoko joined 11:57 virtualsue left 11:59 darutoko- left 12:02 denis_boyun joined, darutoko- joined 12:05 darutoko left
timotimo Mouq++ # pod hacking that i've been meaning to do for a looooong time, but never made it 12:06
also, o/ 12:07
tadzik Mouq++ # I like what I see :)
timotimo jnthn: should i do a full three-backend-spectest-run of my partially implemented lex-to-local code and merge that for the nqp release or should i rather wait until next month when it'll hopefully do the cascading inlining, too? 12:10
moritz timotimo: if you have to ask, it's probably better to wait until after the release 12:11
jnthn timotimo: It may be better to land such a large thing after a release.
timotimo heh. yeah. 12:12
i think i'll write a few words about MoarVM in the release guide for nqp 12:13
12:15 darutoko joined
timotimo and i suppose about jvm, too 12:15
12:17 darutoko- left 12:18 kivutar left
dalek p: 7fd93b7 | (Timo Paulssen)++ | docs/release_guide.pod:
mention MoarVM and JVM in the release guide.
 12:20
kudo/nom: 1a5cb47 | nicholas++ | src/vm/moar/ops/ (2 files):
Pass an MVMCollectable* to MVM_ASSIGN_REF(), without using casts.

All derivatives of MVMObject and MVMSTable contain a nested MVMCollectable struct. By using structure member access, we can get a pointer to it without using a cast. This encourages type safety, and reduces the number of casts and hence places that might conceal strict aliasing violations.		 12:21
kudo/nom: 9660042 | jnthn++ | tools/build/NQP_REVISION:
Get latest NQP, for Moar bump/NativeCall bits.
kudo/nom: 23ecfa4 | jnthn++ | tools/build/ (2 files):
Add debugger build for JVM backend.
 12:22
12:25 darutoko- joined 12:27 woolfy1 joined, woolfy left 12:28 lizmat_ joined, darutoko left
dalek kudo/nom: 3bbd8fd | jnthn++ | tools/build/Makefile-JVM.in:
Correct a debugger dependency in Makefile.
 12:29
kudo/nom: 500e609 | jnthn++ | src/Perl6/Actions.nqp:
Make for compilation robust to debug hook addition
12:29 darutoko- left 12:30 lizmat left 12:31 darutoko- joined
dalek kudo/nom: 8c22767 | (Tobias Leich)++ | docs/ChangeLog:
add note about IO::Path.ACCEPTS
 12:33
12:33 woolfy joined
moritz Makefile:726: warning: overriding commands for target `perl6-debug-m' 12:33
Makefile:366: warning: ignoring old commands for target `perl6-debug-m'
12:33 lizmat joined
jnthn wtf 12:34
12:34 woolfy1 left
moritz tools/build/Makefile-JVM.in:J_DEBUG_RUNNER = perl6-debug-m@runner_suffix@ 12:34
should be -j
12:34 lizmat_ left
timotimo do we also build the debug runner for parrot? 12:34
jnthn oh...
timotimo: No, I'm half way through writing that patch. 12:35
timotimo good :)
timotimo is reviewing changes done to spice up the changelog
jnthn :)
I did a ChangeLog for Moar this month and it's quite long :) 12:36
12:36 lizmat left
timotimo saw it :) 12:36
12:36 lizmat joined
timotimo i think i'll not write moar-related changes into the rakudo changelog until that whole situation has calmed down a bit 12:36
jnthn Well, I think we could have a line in there along the lines of "Many improvements to Rakudo on MoarVM, including sockets support" or so. 12:37
dalek kudo/nom: d1610ac | moritz++ | tools/build/Makefile-JVM.in:
[build] fix copy&paste-o
jnthn And any others that seem noteworthy
moritz: oh, that's why my push failed... 12:38
moritz jnthn: sorry :-)
timotimo yes, but only one entry :D 12:39
jnthn timotimo: no, moar! :P 12:41
.oO( <timotimo> no, no moar! )
 12:42
12:43 mtk left 12:44 colomon left 12:46 mtk joined 12:47 mtk left 12:50 Sqirrel_ left 12:51 fhelmberger joined, fhelmberger left 12:52 fhelmberger joined 12:55 skids left 12:57 jnap joined 13:00 mtk joined 13:03 kaleem_ joined
dalek kudo/nom: b23cbc6 | jnthn++ | tools/build/Makefile-JVM.in:
Fix JVM make install on Windows.
 13:04
kudo/nom: 1f690c7 | jnthn++ | tools/build/Makefile-Parrot.in:
Build debugger on Parrot also.
kudo/nom: f30f805 | jnthn++ | .gitignore:
Update .gitignore for debugger build additions.
timotimo notes that in the changelog
13:06 kaleem left
timotimo "allow definition of custom postcircumfixes" warrants a changelog entry also, don't you think? 13:06
jnthn I've tagged 2014.01 in rakudo-debugger for the "worked with that release" version 13:07
Going to remove the Build.pm since it's now never needed/wanted against HEAD
timotimo sounds good 13:08
dalek kudo/nom: fafd39e | (Timo Paulssen)++ | docs/ChangeLog:
add a bunch more changelog entries
 13:09
jnthn wiki.perl6.org/Create%20and%20Distr...%20Modules currently doesn't say anything about "provides" 13:10
Shuld it?
timotimo i suppose so 13:11
dalek kudo-debugger: 7082cb8 | jnthn++ | META.info:
Add "provides" to META.info.
 13:13
kudo-debugger: c27659f | jnthn++ | / (2 files):
perl6-debug now in Rakudo; remove it from here.

This module is now just a command-line front end, meaning it needs no Build.pm either, and should install/work with Parrot, JVM and Moar.
timotimo so ... i forgot again ... parrot on moar works how well now?
dalek kudo-debugger: 7fe0967 | jnthn++ | META.info:
Tweak module name to reflect what it is now.
timotimo ... panda on moar*
13:18 SamuraiJack joined 13:19 SamuraiJack__ left
jnthn tries it on Windows 13:21
Hopefully perl6-m bootstrap.pl dos it...
*does 13:22
==> Bootstrapping Panda
==> Fetching File::Find
===SORRY!===
Start of substr out of range. Is: 40, should be in 0..38
:(
timotimo oh, oops 13:23
hm. i just tried translating S26-documentation with latest perl6 and the allowed formatting codes don't work there yet; maybe support is still missing from ::To::HTML?
13:24 rurban joined
timotimo oh, that's because it's in a branch %) 13:26
dalek kudo/nom: 7f4e733 | (Timo Paulssen)++ | lib/Pod/To/Text.pm:
eval -> EVAL in Pod::To::Text
13:28 daniel-s_ joined 13:29 benabik left, tgt joined 13:32 ajr joined, kbaker joined, ajr is now known as Guest70241 13:33 Guest70241 is now known as ajr_ 13:40 ggoebel1117 joined
timotimo there's still something quite b0rked regarding :allow in code blocks 13:40
"=begin R<BLOCK_TYPE> R<OPTIONAL CONFIG INFO>" renders out as "OPTIONAL CONFIG INFO BLOCK_TYPE =begin" 13:41
dalek kudo/nom: ecdb527 | (Timo Paulssen)++ | src/Perl6/Compiler.nqp:
document --doc flag in usage
 13:42
kudo/nom: ebd80a1 | (Timo Paulssen)++ | docs/ChangeLog:
:allow wasn't actually merged into nom yet
13:42 ggoebel1116 left
dalek kudo/nom: 0d367e2 | (Timo Paulssen)++ | docs/ChangeLog:
wording tweaks
 13:44
jnthn huh...somehow my make spectest on Moar has suddenly got slower... 13:45
FROGGS uhh 13:47
moritz jnthn: are you using 6 cores again? :-)
tadzik only 6? :P
FROGGS jnthn: the IO-Socket-INET should not have a huge impact...
timotimo Perl6 cores? :) 13:48
jnthn moritz: No, still set to 8
No, I mean slower compared to earlier today
FROGGS ohh
tadzik CORE.setting, RESTRICTED.setting, FORGOTTENREALMS.setting...
FROGGS is there a process dangling around?
I think I had to kill a server+client from that said test once that was still there after hours 13:49
jnthn FROGGS: Doesn't seem like it
Hmm. Startup seems longer. 13:50
dalek kudo/nom: 1d2c436 | (Timo Paulssen)++ | docs/ChangeLog:
typo, dedup, merge, ...
 13:51
jnthn oh...
13:52 daniel-s_ left
jnthn Yeah. I turned on the ConEmu thingy that injects an extra DLL into every process that re-writes ANSI escape codes 13:52
Which adds a rather noticable amount to eveyr process start. 13:53
timotimo bleeeeeh, i can't reproduce the spectest results from moritz easily, i still get the "no subtests run" thing after a while 13:57
jnthn timotimo: How many test jobs are you using? 13:58
timotimo 4
jnthn Sometimes reducing that helps, if the problem is it's hitting the RAM limit 13:59
timotimo mhh 14:00
i still get occasional "no subtests run" even with only 1 test job 14:06
14:07 rurban left 14:08 dalek left 14:09 dalek joined, ChanServ sets mode: +v dalek 14:11 johnmilton joined 14:15 colomon joined 14:16 jnap left
timotimo moritz: i could not reproduce any fail in perl.rakudo.jvm 14:16
and the roll.t fail is a pass on my machine as well 14:17
the exceptions/misc.rakudo.jvm and tc.rakudo.jvm failures seem legit, though 14:18
14:21 jnap joined, tgt left 14:24 thou joined, rurban joined 14:26 skids joined 14:31 virtualsue joined 14:36 tgt joined, kaleem_ left 14:37 kaleem joined 14:38 Alula left 14:41 zakharyas left 14:42 zakharyas joined 14:47 btyler joined 14:48 myfree joined, virtualsue left 14:49 myfree left 14:50 Alula joined 14:54 kaleem left 15:05 xinming_ joined, kaare_ left 15:07 SamuraiJack left 15:09 xinming left, crab2313 left 15:13 tgt left 15:16 Vlavv left 15:18 salv0 left 15:20 benabik joined 15:21 tgt joined, sivoais left 15:23 sivoais joined 15:29 araujo joined 15:36 tgt left 15:37 kurahaupo_mobile joined, tgt joined 15:58 kivutar joined 15:59 Psyche^ joined 16:03 Psyche^_ left, tgt left 16:06 kbaker left
btyler hi all. I've been using vim-perl for my perl 6 highlighting needs -- it does a great job with syntax coloring, but it is really sluggish - I get noticable lag when moving around the text, and it makes editing a bit of a pain. do most just write p6 sans syntax highlighting? 16:07
16:08 tgt joined 16:09 kbaker joined
FROGGS btyler: I use SciTE with P5 highlighting 16:12
and SciTE is always astonishing fast
16:15 ronaldo joined 16:16 ronaldo left, denis_boyun left 16:17 kurahaupo_mobile left 16:18 kurahaupo_mobile joined
btyler hm, maybe just forcing p5 highlighting would be good enough. thanks FROGGS 16:19
16:20 hoverboard joined 16:21 tgt left
dalek p/moar-conc: 58e1b65 | jnthn++ | src/vm/jvm/ (8 files):
Add VMThread REPR and thread-related nqp:: ops.
 16:22
kudo/moar-conc: 2097d53 | jnthn++ | / (11 files):
Move non-JVM-specific conc code out of src/vm/jvm.

This makes it clearer what bits are JVM-specific and need abstracting or porting.		 16:25
kudo/moar-conc: 9a717ab | jnthn++ | src/vm/jvm/core/Thread.pm:
Re-work Thread class using new nqp::ops.
kudo/moar-conc: a2204bc | jnthn++ | / (3 files):
Move Thread.pm to src/core, now it uses nqp:: ops.
16:27 telex left 16:30 telex joined
timotimo \o/ 16:32
16:32 tgt joined
FROGGS ohh! 16:34
awesome times these are :o) 16:35
jnthn Well, abstracting what already works on JVM is the easy part :P 16:36
I'm hoping if I start on it now, though, then we'll be able to mostly pass S17-concurrency/ in r-m by the March release. 16:38
16:43 kurahaupo_mobile left, kurahaupo_mobile joined 16:44 molaf joined 16:49 zakharyas left 16:50 fhelmberger left 17:03 rindolf left 17:07 treehug88 joined 17:11 sqirrel joined, tgt left, tgt joined
colomon btyler: I use p5 highlighting for p6 too, but honestly it is frequently terrible looking 17:12
17:15 FROGGS left, sqirrel left 17:16 tgt left 17:20 kurahaupo_mobile left, kurahaupo_mobile joined 17:21 Rotwang joined 17:23 rurban left 17:24 isBEKaml joined
timotimo Byays 17:24
nwc10 Easter is the new Christmas?
isBEKaml btyler: If you're using vim-perl for p6 highlighting, turn off ft detection. Just use syntax highlighting. 17:26
17:27 zby_home joined, hoverboard is now known as gotpower 17:28 ajr_ left, gotpower is now known as hoverboard 17:29 kurahaupo_mobile left, [Sno]_ joined, kurahaupo_mobile joined, hoverboard is now known as gotpower, gotpower is now known as hoverboard 17:30 [Sno] left, [Sno]_ is now known as [Sno] 17:31 kurahaupo_mobile left 17:32 kurahaupo_mobile joined, rindolf joined 17:37 virtualsue joined, kurahaupo_mobile left, mtk left 17:38 kurahaupo_mobile joined, mtk joined
timotimo uses the perl6 syntax highlighting and knows what btyler is talking about 17:42
17:45 denis_boyun joined, Rotwang left 17:48 Rotwang joined 17:52 Vlavv joined 17:53 kurahaupo_mobile left 17:55 SamuraiJack joined 17:56 rurban joined 17:59 treehug88 left 18:00 hoverboard is now known as moistcherry, moistcherry is now known as hoverboard 18:01 treehug88 joined 18:02 rurban left, dakkar left 18:08 tgt joined
TimToady jnthn: arguably, using {} to force $/ is going to backfire in all those places where we're merely using it to stop LTM 18:09
jnthn TimToady: Well, it already does force it today
TimToady: I agree it should optimize that away in the end. 18:10
TimToady: By which point we need to either have done a fix for <foo($<bar>)> or found a different way to cheat... :)
18:13 tgt left
dalek p/moar-conc: 469bf84 | jnthn++ | src/vm/moar/QAST/QASTOperationsMAST.nqp:
Map thread related ops for MoarVM.
 18:15
kudo/moar-conc: f38d5ba | jnthn++ | tools/build/Makefile-Moar.in:
Add Thread.pm to MoarVM build.

No, don't expect it to work yet.		 18:16
18:17 ajr joined 18:18 ajr is now known as Guest74068, hoverboard is now known as water, Guest74068 is now known as ajr_
TimToady jnthn: the way STD cheats is by pulling $<foo> straight out of the cursor 18:21
it does go as far as to do 'my $M = $C;' to officially put the cursor into $/, but simply ignores the fact that there's extra stuff in $/ related to $¢ 18:22
masak evening, #perl6 18:23
jnthn TimToady: That feels...hacky :)
TimToady so {} really does translate to a P5 'do {}'
jnthn Finding dinner & 18:25
TimToady and I think the codegen knows whether a $M is needed, so it differentiages <foo($x)> from <foo($<x>)>
*ates
18:34 water is now known as hoverboard 18:52 FROGGS joined 18:53 isBEKaml left
timotimo would xterm-256color support go into Term::ANSIColor? Term::ANSIColor::256? 18:56
TimToady I'd think only if it were an ANSI standard... 18:57
timotimo okay. Term::256Color then? 18:58
and would 24bit color get its own module as well?
Term::ExtendedColors perhaps
FROGGS Term::Color? that hides the actual terminal used?
18:59 rurban joined
timotimo you mean it would do things like probe for 256color support and 24bit color support and then do approximation if needed? 18:59
FROGGS I'd make it Term::Color::XTerm256
yes
timotimo mhm 19:00
FROGGS and if you are on a terminal without coloring support it should just print the text
timotimo and then Term::Color::ANSI as an alias for Term::ANSIColor?
FROGGS so the "end user" does not have to care about the terminal at all
as one of the backends, yes
timotimo what exactly is the interface supposed to look like then? just give it rgb color triplets and it tries to do its best? 19:01
FROGGS I mean, Term::Color could do some diagnostics and pass these information to the user's script
so you can ask Term::Color for coloring support and the actual palette if available 19:02
rgb triplets, color names, palette indexes
timotimo okay. this is about colors; would these modules also handle underline/blink/reverse video/...? 19:03
FROGGS hmmmm, dunno
I think I would let it do one thing only
19:04 rurban left
jnthn There's also Term::WindowsColor that we maybe want at some point 19:06
timotimo yeah, sure 19:09
because ConEmu is not so nice
19:10 tgt joined, Sqirrel joined
jnthn No, emus are already bad enough when they aren't trying to con you... 19:11
timotimo aye!
jnthn nom ready...bbiab 19:13
19:14 rindolf left 19:15 tgt left 19:17 telex left 19:18 telex joined 19:19 darutoko- left 19:23 SamuraiJack left
[Coke] wonders if anyone else is excited for Guardians of the Galaxy this summer. :) 19:28
dalek kudo-star-daily: 7ac6940 | coke++ | log/ (5 files):
today (automated commit)
benabik [Coke]: Saw the trailer today. It definitely looks like something...
dalek rl6-roast-data: f189a44 | coke++ | / (6 files):
today (automated commit)
[Coke] rakudo.parrot has the fewest failing tests today with 1. 19:29
star is clean, no failures or deprecations. 19:30
19:30 vendethiel left
benabik \o/ 19:30
moritz \o/ 19:31
[Coke] ^^ s/today/yesterday/ 19:32
today is still running.
planeteria updates are through.
19:32 kbaker left
[Coke] parrot 6.1.0 out. 19:32
19:33 kbaker joined
colomon is cautiously optimistic about Guardians of the Galaxy. 19:36
[Coke] did you see the trailer? 19:40
19:44 logie__ left 19:45 logie joined 19:47 logie left
rurban_ I'm now working to update the nqp new-packfile-api branch (for newer parrots) 19:48
19:49 berekuk joined 20:00 rurban joined
rurban_ Oh, I missed patrick's better pm-packfile-api branch on nqp. Trying that now 20:00
20:03 benabik left 20:05 virtualsue left 20:06 rurban left 20:11 tgt joined
timotimo rurban_: thank you! if you get it to work, ping me a quick message and i'll see to it that it can get into tomorrow's rakudo release 20:14
20:15 tgt left
rurban_ timotimo: there are some merge conflicts I don't really understand yet. Should be easier for a nqp guy 20:15
timotimo i'll have a look 20:16
FROGGS I don't we can release it even if it does work... 20:17
timotimo the only merge conflicts seem to be the stage0 files 20:18
FROGGS as I understand it the perl6 binary will support --target=pbc but not --target=pir afterwards
timotimo oh
so that will change the way we stage0 on parrot
FROGGS I mean, I can be totally off 20:19
rurban_ No, both should be possible
20:19 arnsholt joined
arnsholt o/ 20:19
rurban_ method is_precomp_stage($stage) { 20:20
$stage eq 'pir' || $stage eq 'pbc'
}
so you can still dump to pir
FROGGS that sounds very good 20:23
hi arnsholt
masak arnsholt: \o 20:27
20:35 flussence left 20:37 flussence joined
rurban_ I was wrong. nqp-p already works with latest parrot. I did a fresh reinstall. pmichaud already merged pm-packfile-api 4 months ago 20:50
20:51 ggoebel1117 left
timotimo oh! 20:54
21:03 flussence left, rurban joined 21:06 flussence joined
[Coke] yawns. 21:08
dalek rl6-roast-data: 250b4eb | coke++ | / (3 files):
today (automated commit)
21:08 Mouq joined, tgt joined
lue is :nested(integer) roughly equivalent to =over integer ? 21:08
[Coke] rurban++ 21:10
Mouq lue: I don't think so.. but I'm not sure
also o/
lue \o
Mouq actually 21:11
I think it is
lue Seems like it to me, just wanted to be sure :)
Mouq lue: Looks like it's defined to be a more general nesting than =over, but I think if you implement it the same it should be fine 21:12
lue: Though if this is for the translator, =item already implies nesting in Pod 21:13
*6
lue I just want something to replace =over with, and the only other choice is getting rid of indentation altogether.
Mouq And that's the primary thing I've seen =over used with
21:13 tgt left
lue =over isn't just for item lists though, which is the issue. 21:13
Mouq lue: Yeah, I'd just do it and see how it looks :) 21:14
21:16 flussence left, kaare_ joined, xenoterracide left
lue Right now my converter just rewriting a few directives as seen fit. It'll be a while before it can do fancy stuff like replacing =over/=item/=back with =item1, and nested lists, and... 21:16
*converter is just 21:17
Mouq Well, lue++ anyhow :)
21:17 flussence joined
lue ooh, I just imagined a nasty corner case with POD5 =item * # is for comments 21:17
21:18 cognominal joined
lue *facepalm* guess what I just spotted in S26? "To specify an unnumbered list item that starts with a literal #, either make the octothorpe verbatim:" 21:18
21:19 xenoterracide joined 21:21 airdisa joined 21:23 flussence left 21:25 flussence joined 21:26 thou left 21:27 thou joined 21:33 tgt joined, airdisa left
timotimo Mouq: did you see what happens with allowed formatting codes in code blocks? 21:34
21:36 rurban left
Mouq timotimo: What do you mean? 21:36
Pod::To::HTML works fine with it btw
gimme a sec to finish a commit and I'll push; it's 99% finished
s/sec/few minutes/ :P 21:37
timotimo: I've been testing my changes on S26 actually, and the examples are much nicer when bits of the code are bolded etc.
lue I can't wait, S26 should look much improved :) (it uses :allow I<a lot>)
21:38 beastd joined
Mouq I don't think those tables at the end that have all the C<>s around everything work... I'll have to look into that 21:39
lue timotimo: remember when we were talking about if implicit code blocks should be indented and such? According to S26, code blocks are nested (usually represented as indentation) by default.
Mouq: if they're actually =table tables, then they don't need :allow, each cell should be as a nested =para 21:40
Mouq lue: I know, I don't know what the problem is... 21:43
I'll look at it in a bit
21:43 zby_home left
Mouq Does anyone know if there's a less verbose way in NQP to say "return $bool ?? sub-routine('arg', :named, :sometimes) !! sub-routine('arg', :named)" 21:45
lue Mouq: :sometimes($bool) ? 21:47
timotimo lue: i remember that and i forgot why i didn't go ahead and implement my changes 21:48
Mouq: ah, so you made changes to Pod::To::HTML?
Mouq * "return $bool ?? sub-routine('arg', :named($thing), :sometimes($other_thing)) !! sub-routine('arg', :named($thing))" 21:50
timotimo: Nope
timotimo: I needed to fix Pod::To::Text, but Pod::To::HTML just works :)
lue Run this script on a synopsis (I tried S03), and tell me if you see any problems: github.com/lue/pod_vvi 21:51
21:52 dmol left
lue This is the error I get trying to run that: gist.github.com/lue/9102506 21:56
PerlJam lue: run it through the debugger? 21:57
lue looks like a low-level error, but I can tr.. wait, I don't think perl6-debug-m will work
Nope, stupid IO segfaults 21:58
jnthn IO segfaults? 22:02
lue Yeah, I and someone else (can't remember who) run into issues with perl6-debug-m when IO (like open()) is involved, or so it seems. 22:03
Don't have another backend's debugger to see if it's Moar specific or not.
jnthn Dunno. I fixed a debugger issue today that caused issues with for loops...
btyler lue: yep, that was me. I meant to rakudobug it but didn't get to it yesterday
lue jnthn: irclog.perlgeek.de/perl6/2014-02-17#i_8299165 22:04
btyler I'll rebuild moar & company and see if its still happening 22:05
Mouq :/ I'm having a harder time removing this feature than I would adding it. I'm just going to revert the work on it and push. It's a little off topic to the branch I guess 22:06
jnthn I just tried "my $f = open('README'); for $f.lines -> $l { say $l; }" with perl6-debug-m and it worked out ok 22:07
lemme try the exact one
22:08 tgt left
jnthn btyler: I just tried it now and it worked out for me. 22:08
btyler jnthn: great! I'm rebuilding nqp and will try as soon as that finishes 22:09
jnthn btyler: I did fix a bug with for today. I didn't get a SEGV, but I did not a nasty error on JVM, which is where I discovered it
The bug was independent of any particular backend.
22:10 Aytan joined
Aytan Hello everyone. Can anyone help me with a "simple" @ARGV question, please? 22:10
masak Aytan: sure. it's not @ARGV, it's @*ARGS 22:12
next!
masak grins 22:13
Aytan perl API-CloudSearch.pl drilling, in my code I need to use the "drilling" as a variable. So, I use @ARGV[0]. But the results are VAR1 = "drilling", how do I get just the value and save it to a string variable?
masak Aytan: you might want to try a Perl 5 channel.
geekosaur or are you asking a perl5 question? that'd be #perl
masak Aytan: this is #perl6.
Aytan Oh ok. Sorry about that.
masak no problem. easy mistake to make.
masak .oO( the languages are, like, right next to each other ) 22:14
22:19 Aytan left
dalek kudo/pod-code-allow: e08bdbb | Mouq++ | src/Perl6/Actions.nqp:
Fix delimited_code for multiple lines + simplify
 22:20
kudo/pod-code-allow: 1217268 | Mouq++ | lib/Pod/To/Text.pm:
Update Pod::To:Text to handle formatted code blocks
kudo/pod-code-allow: 91334ce | Mouq++ | src/Perl6/Grammar.nqp:
Make =code follow spec better
kudo/pod-code-allow: a9cf04d | Mouq++ | src/Perl6/ (2 files):
Add code for "=for code" too
22:22 BenGoldberg joined
Mouq r: =code :this-should-be-literal␤but isn't␤␤say $=pod 22:23
camelia rakudo-parrot 1d2c43, rakudo-jvm 1d2c43, rakudo-moar 1d2c43: OUTPUT«Pod::Block::Code("config" => "\{\"this-should-be-literal\" => \"1\"}").hash␤ but isn't␤␤␤»
22:23 xenoterracide left
Mouq ^^ I was trying to fix that, but it's not really pressing and it causes all kinds of issues 22:24
22:33 rurban joined, airdisa joined
timotimo Mouq: what did i do wrong to get Pod::To::HTML to fail with allowed formatting codes? 22:35
tkhey were formatted properly, but outputted in reverse order
22:35 Rotwang left
Mouq timotimo: Is it like that with the latest? 22:36
timotimo: That was a bug in my own code at a point
22:37 ajr_ left, rurban left 22:39 fridim_ joined
timotimo ah, i should update then 22:40
i'm very glad you're doing this 22:41
it kind of plays right into my secret project 22:44
Mouq timotimo: Glad to hear it :) 22:45
lue timotimo: what kind of secret project? :) 22:46
22:47 johnmilton left 22:49 airdisa left
timotimo lue: the lovely kind 22:50
except i'm not actually putting in lots of work yet
though i really should
22:51 hoverboard is now known as baller 22:53 xinming_ left
Mouq r: my token foo { d+ }; 'abcddcd' ~~ /a b [ c ( <foo> ) ]+/; .say for $/ 22:53
camelia rakudo-parrot 1d2c43, rakudo-jvm 1d2c43, rakudo-moar 1d2c43: OUTPUT«｢abcddcd｣␤ 0 => ｢dd｣␤ foo => ｢dd｣␤ 0 => ｢d｣␤ foo => ｢d｣␤␤»
Mouq r: my token foo { d+ }; 'abcddcd' ~~ /a b [ c ( <foo> ) ]+/; .say for @$/ 22:54
camelia rakudo-parrot 1d2c43, rakudo-jvm 1d2c43, rakudo-moar 1d2c43: OUTPUT«｢dd｣␤ foo => ｢dd｣␤␤｢d｣␤ foo => ｢d｣␤␤»
22:54 xinming joined 22:56 rifkin joined 22:57 FROGGS left
Mouq r: my token foo { d+ }; 'abcddcd' ~~ /a b [ c $<t>=[ <foo> ] ]+/; .<foo>.say for $<t> 22:59
camelia rakudo-parrot 1d2c43, rakudo-jvm 1d2c43, rakudo-moar 1d2c43: OUTPUT«(Any)␤(Any)␤»
Mouq :|
23:01 kurahaupo_mobile joined
lue r: my token foo { d+ }; 'abcddcd' ~~ /a b [ c $<t>=[ <foo> ] ]+/; say $/ 23:01
camelia rakudo-parrot 1d2c43, rakudo-jvm 1d2c43, rakudo-moar 1d2c43: OUTPUT«｢abcddcd｣␤ t => ｢dd｣␤ foo => ｢dd｣␤ t => ｢d｣␤ foo => ｢d｣␤␤»
lue r: my token foo { d+ }; 'abcddcd' ~~ /a b [ c ( <foo> ) ]+/; .<foo>.say for $/ 23:03
camelia rakudo-parrot 1d2c43, rakudo-jvm 1d2c43, rakudo-moar 1d2c43: OUTPUT«(Any)␤»
lue Mouq: What's the problem? Looks to me like you were testing two different things.
Mouq lue: I was trying to figure out how to iterate over something in NQP 23:04
23:04 baller is now known as hoverboard 23:08 kbaker left 23:12 tgt joined
jnthn r: my token foo { d+ }; 'abcddcd' ~~ /a b [ c $<t>=[ <foo> ] ]+/; .<foo>.say for %$<t> 23:13
camelia rakudo-parrot 1d2c43, rakudo-jvm 1d2c43, rakudo-moar 1d2c43: OUTPUT«(Mu)␤»
jnthn oh, no. Hmm. 23:14
r: my token foo { d+ }; 'abcddcd' ~~ /a b [ c $<t>=[ <foo> ] ]+/; say $<t>.WHAT
camelia rakudo-parrot 1d2c43, rakudo-jvm 1d2c43, rakudo-moar 1d2c43: OUTPUT«(Parcel)␤»
jnthn r: my token foo { d+ }; 'abcddcd' ~~ /a b [ c $<t>=[ <foo> ] ]+/; say $<t>
camelia rakudo-parrot 1d2c43, rakudo-jvm 1d2c43, rakudo-moar 1d2c43: OUTPUT«｢dd｣␤ ｢d｣␤␤»
jnthn Oh...'cus that doesn't actually make a sub-match I guess. 23:15
23:21 beastd left, skids left 23:23 kaare__ joined
pmichaud good afternoon, #perl6 23:24
timotimo hello pmichaud! :)
i hear the newest parrot already works OOTB with current rakudo (the packfile api change) 23:25
pmichaud sure, rakudo supports the new packfile api
fsvo "supports" (in this case, "doesn't break") 23:26
the next step is to enable --target=pbc, which is on my "todo" list.
timotimo ah, ok. so all i'll have to do is bump parrot to the newest version for this motnh's release
23:27 flussence left, kaare_ left
pmichaud well, we normally only bump if we need a parrot improvement that the bump provides. 23:30
jnthn o/ pmichaud 23:31
pmichaud o/ jnthn
btyler jnthn: sorry to take so long getting back about the debug segfault. after getting the latest moar/nqp/rakudo all installed on jvm/moar, perl6-debug-m gives me "missing or wrong version of src/gen/m-BOOTSTRAP.nqp". perl6-debug-j works though 23:32
so I imagine I messed up a dependency somewhere along the chain
arnsholt "missing or wrong version" usually means one of your precompiled modules has to be recompiled 23:33
btyler arnsholt: ah ha, yep, thanks. needed to make clean/make install on rakudo::debugger 23:34
23:34 rurban joined
btyler and indeed perl6-debug-m works on that gist I posted for me as well. jnthn++ 23:35
jnthn yay
23:36 Ben_Goldberg joined
jnthn OK, I should cut the release... 23:37
23:37 BenGoldberg left
jnthn uh, ww 23:37
23:38 tgt left 23:41 rurban left 23:45 airdisa joined 23:46 hoverboard left 23:47 xenoterracide joined 23:49 airdisa left 23:52 kivutar left 23:57 BenGoldberg joined, arnsholt left, airdisa joined 23:58 Ben_Goldberg left

Powered by Cro and the Raku® Programming Language.

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

Thank you!