»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.
00:01 rurban1 joined 00:04 rjbs left
dalek kudo-star-daily: a63ba62 | coke++ | log/ (5 files):
today (automated commit)
 00:06
rl6-roast-data: 46d6f65 | coke++ | / (5 files):
today (automated commit)
rl6-roast-data: b71d34a | coke++ | / (5 files):
today (automated commit)
00:12 Psyche^ joined 00:14 btyler left 00:15 stevan__ left 00:16 Psyche^_ left
lue I'll leave it alone for today. I've got a magically called-too-often sub that's also eating my attempts at more granular debug says :( 00:18
00:22 rurban1 left 00:26 markov left
[Coke] woolfy1++ 00:41
00:41 stevan_ joined 00:42 markov joined
timotimo has a prototype of the scan-with-boyer-moore-optimisation moritz made 00:53
for jvm 00:54
dalek ast: 0f86f36 | coke++ | S02-literals/quoting-unicode.t:
Fudge for niecza
 00:55
00:56 dayangkun joined
timotimo ooooh, rakudo.jvm has overtaken rakudo.parrot \o/ 00:56
00:56 dayangkun left
[Coke] 2 days ago, looks like. 01:04
timotimo mhm 01:05
modules/uri doesn't seem to actually do any installing at all
causing some fallout in the following modules
01:07 raiph left 01:08 colomon left
timotimo a beefed up version of the artificial benchmark takes 0.6s with boyer-moore (for 10_000 match attempts) 01:08
and now i'm recompiling the before-version
01:09 colomon joined 01:10 berekuk left
timotimo oh, what, 10x faster? 01:10
apparently. wow.
moritz: ^^
01:11 berekuk joined
[Coke] au: you made some pugs changes. Be happy if you could help fix the install on host06. 01:13
dalek p: d4e274d | (Timo Paulssen)++ | src/vm/jvm/QAST/Compiler.nqp:
teach jvm how to boyer-moore for scan+literal.
[Coke] er, *saw* you made.
01:23 rurban1 joined 01:26 woosley joined, raiph joined 01:29 rurban2 joined 01:31 thou left 01:32 rurban1 left 01:33 rurban2 left 01:35 thou joined
diakopter :-[ 01:38
01:55 mtk left 02:05 PZt left 02:20 jeffreykegler joined 02:25 jnap left 02:26 dmol left 02:30 rurban1 joined 02:36 rurban1 left 02:39 tgt left 02:58 jeffreykegler left
[Coke] both rakudo.parrot and rakudo.jvm have failing tests and could use some fudging. 03:02
(with tickets being opened.)
03:08 ggoebel15 left
lue Hm. Somehow the line "my %meths := $par.method_table($par);" causes the sub it's in to get called when it shouldn't. This is odd... (.method_table simply returns the %!methods attribute) 03:08
03:35 daniel-s left 03:51 labster left 03:52 labster joined 03:56 davecc left, davecc joined 04:06 PZt joined 04:10 SevenWolf joined 04:13 logie joined 04:20 preflex_ joined, ChanServ sets mode: +v preflex_ 04:21 preflex left, preflex_ is now known as preflex 04:22 ssutch left 04:27 wbill left 04:29 wbill joined 04:35 kaleem joined 04:36 davecc left 04:42 logie left 04:43 thou left, thou joined 04:49 nnunley left 04:51 kaleem left 04:55 thou left 05:01 logie joined, rurban1 joined 05:14 SamuraiJack_ joined 05:19 logie left 05:23 nnunley joined
dalek p: 93b9d50 | dwarring++ | docs/ops.markdown:
move nqp::takeclosure doco to misc section
 05:25
05:30 [Sno] left 06:04 cognominal joined 06:14 hoelzro left 06:15 hoelzro joined 06:18 dylanwh joined 06:20 geekosaur joined 06:22 xinming_ left, kaleem joined 06:23 xinming joined, [Sno] joined 06:26 logie joined 06:30 logie left 06:31 rurban1 left 06:32 darutoko joined 06:34 darutoko left 06:35 darutoko joined 06:38 raiph left
Woodi morning everyone 06:56
"Don Knuth"++ , well deserved :) 06:57
moritz P: say 42 07:04
camelia pugs: OUTPUT«42␤»
07:16 dmol joined 07:20 cognominal left 07:30 brrt joined, brrt left 07:34 preflex left 07:36 preflex joined, ChanServ sets mode: +v preflex 07:37 rurban1 joined 07:45 rurban1 left 07:55 atroxaper joined
atroxaper Hello #perl6 07:56
moritz hello atroxaper
atroxaper Hello moritz
07:58 FROGGS joined 07:59 wbill left
atroxaper I have a question about usiong Rakudo grammar again... Is there a way to get a part of parsed data if parsing failed? I mean if i try to call Perl6::Grammar.parse('my $b=4; b=5;', :actions(Perl6::Actions.new())); than i get "Lexical '$*LINEPOSCACHE' not found in dynamic scope" 08:01
Hello FROGGS
FROGGS hi atroxaper
well, you could do before calling .parse: my $*LINEPOSCACHE
but if a match failed, you won't get a match object :/ 08:02
maybe the highwater-mark is sufficient? 08:03
atroxaper What do you mean highwater-mark? 08:04
08:05 darutoko left
moritz it's some way to record how far the match got 08:06
FROGGS atroxaper: the parser remembers how far it could get before failing a rule/token, so you get the right position where something is wrong, for error reporting
moritz (specific to Perl6::Grammar)
08:06 SevenWolf left
FROGGS r: my $x = 42; b # see where the marker is 08:06
camelia rakudo-parrot 2c40f6: OUTPUT«===SORRY!=== Error while compiling /tmp/MmBKTG4TRC␤Undeclared routine:␤ b used at line 1␤␤»
..rakudo-jvm 2c40f6: OUTPUT«===SORRY!=== Error while compiling /tmp/81zdeIVY3r␤Undeclared routine:␤ b used at line 1␤␤»
08:06 darutoko joined
FROGGS umm 08:06
atroxaper I just want to parce Perl6 code and get some structure like which NQPMatch.dump prints... 08:07
FROGGS r: sub a { } { } # see where the marker is
camelia rakudo-jvm 2c40f6: OUTPUT«===SORRY!=== Error while compiling /tmp/_nBCQSRmt7␤Unexpected block in infix position (two terms in a row, or previous statement missing semicolon?)␤at /tmp/_nBCQSRmt7:1␤------> sub a { } ⏏{ } # see where the marker …»
..rakudo-parrot 2c40f6: OUTPUT«===SORRY!=== Error while compiling /tmp/Mq7hX4OxvT␤Unexpected block in infix position (two terms in a row, or previous statement missing semicolon?)␤at /tmp/Mq7hX4OxvT:1␤------> sub a { } ⏏{ } # see where the mark…»
FROGGS there you go
08:07 denisboyun joined
FROGGS atroxaper: then I'd propose you a) write a module that turns an NQPMatch into a Match object, or b) make it an nqp module 08:08
atroxaper FROGGS: and how can i get position where something is wrong? 08:09
08:10 zakharyas joined, iSlug joined
FROGGS atroxaper: you need to check how and where the highwater-mark is stored... and try to get your hands on it 08:11
atroxaper FROGGS: Hmmm. Ok. Thanks for hint) 08:12
FROGGS atroxaper: since I know how wonderful jnthn++ writes code, a grep for 'highwater' in nqp/src should do 08:13
atroxaper FROGGS: you are right! ) 08:14
08:14 nnunley left 08:15 rurban left, rurban joined 08:20 denisboyun left
FROGGS timotimo++ # 10x 08:32
moritz timotimo++ is a 10x programmer! :-) 08:39
08:39 denisboyun joined 08:42 kivutar joined 08:50 woolfy1 left
lizmat good *, #perl6! 08:52
moritz \o lizmat, *
lizmat moritz o/
08:52 woolfy joined 08:53 Maddingu1 joined
hoelzro morning #perl6 09:00
09:08 fhelmberger joined 09:12 Maddingu1 is now known as Maddingue
lizmat morning hoelzro! 09:20
09:21 denisboyun left 09:24 dakkar joined 09:25 denis_boyun joined 09:35 dram left
hoelzro morning lizmat ! 09:41
09:42 Guest2496 is now known as denysonique 09:44 dayangkun joined, dayangkun left 09:46 rurban1 joined 09:58 rurban1 left 10:02 denis_boyun_ joined, denis_boyun left, go|dfish left 10:07 dram joined 10:09 atroxaper left
dram r: my $foo = "FOO"; say qq:ww/$foo "bar baz"/.perl; 10:10
camelia rakudo-parrot 2c40f6, rakudo-jvm 2c40f6: OUTPUT«(("FOO",).list, "bar baz")␤»
dram r: say qq:ww/FOO "bar baz"/.perl; 10:11
camelia rakudo-parrot 2c40f6, rakudo-jvm 2c40f6: OUTPUT«("FOO", "bar baz")␤»
dram Is this a bug?
moritz looks like it 10:12
dram n: my $foo = "FOO"; say qq:ww/$foo "bar baz"/.perl; 10:14
camelia niecza v24-98-g473bd20: OUTPUT«("FOO", "\"bar", "baz\"").list␤»
FROGGS creates rakudobug 10:15
moritz FROGGS++ dram++ 10:16
the fix might be simply calling .list on the result (in the compiler, that is)
dram ok, i'll file a bug. 10:17
FROGGS dram: see rt.perl.org/Ticket/Display.html?id...c0d93413a8
dram thx. 10:18
FROGGS you could try to fix it now that I've filed the bug report for you :P
dram I'm new to perl6. just several days. :) 10:22
10:25 brrt joined 10:26 rurban left 10:27 rurban joined, woosley left 10:35 pernatiy left 10:50 rindolf joined 10:51 pernatiy joined 10:54 rurban1 joined
timotimo so I've been thinking, before i could get any sleep, that to make keeping the NFA's work is not terribly hard 11:05
first, the optimizer needs to know exactly how long the declarative prefix was, and it needs to know that the NFA has, in fact, matched this piece of the QRegex 11:06
then, it needs to analyze the positions of capture groups. if they are constant, it can generate operations that just skip N characters, if not, it has to stop its work
so it would turn [ foo (...) | bar (...) ] into a concat of skip 3, push cstack (or how it does the capturing), skip another 3, push cstack again (for the end of capture, right?) for both branches of the alt. 11:07
the tricky part is: does the NFA hang off of one of these nodes? do i have to be careful not to replace it? 11:10
i know the NFA gets created the moment the QRegex is generated in the Actions, which is long before the optimizer starts running
11:15 jaffa4 joined 11:18 iSlug left
timotimo what does being a 10x programmer mean? :| 11:21
brrt timotimo, i have no idea 11:23
i suspect that 'management is able to get what they want 10x easier'
it certainly doesn't seem to correspond to one particular attribute
timotimo that sounds more like "being a pushover" :P 11:24
brrt true
brrt is thinking
there is, i think, also some part to the 10x engineer meme that says: 'all these other guys are stupid' 11:25
timotimo i guess i'll spend the day trying to write code to analyse the absolute widths of matching thingies
brrt i.e. the implication of being a special human being
or at least, that there are such special humans
:-)
timotimo that doesn't sound much like a compliment to me. at least not one i'd like to get
brrt hmmm 11:26
maybe i'm thinking of it in a different context
11:26 rurban1 left, beee joined, dram_ joined, daxim_ joined
brrt perhaps the simplest way to explain it is: what used to be a 'rockstar developer 11:26
'
timotimo i don't like these people :| 11:27
.o( node.js is badass rockstar tech! )
brrt neither do i 11:28
things can be awesome / exciting in their own right
moritz timotimo: there's a very old meme on the internet that there are some '10x' programmers that are 10 times as efficient as the average programmer, whoever that may be
brrt but to be excited about writing callback based servers is valid, but still a bit geeky :-) 11:29
… what moritz said
moritz and every month there's a blog post denying the fact that such a thing exists
11:29 dram left
brrt the question is what 'efficient' even means 11:29
11:29 dram_ is now known as dram
brrt which is why i offered: get what management wants 11:29
moritz and two others springing up in response citing examples or anecdotes
timotimo i see 11:30
locally, rakudo-jvm is failing isDeprecated and deprecations, is that known?
it fails all of the tests
brrt is bored by the internet
moritz timotimo: I think I've see that too
brrt bye folks :-)
moritz brrt: build a new one :-)
timotimo good
brrt would-like-too
11:31 brrt left, iSlug joined
timotimo moritz: do you see an obvious flaw with my musings above? 11:32
the only thing i can think of is that there are a few dangerous things i'll have to keep in mind, like \n sometimes being 2 characters instead of 1 11:33
moritz well, you have to keep track of match lengths
of all of them
and there are funny things going on with case-insensitive matching 11:34
/ß/i matching 'SS' and all that
FROGGS it matches the uppercase ß unicode thingy too
timotimo i don't have to keep track of all of them, it'll already be a performance benefit if i only do as much as possible 11:35
like, up until i find a match that's possibly mobile
daxim_ I call for abolishment of uc(ß) == SS
FROGGS daxim: me too
timotimo that can't be the only example, though 11:36
but good point. i'll just go with a whitelist-approach instead of a blacklist-approach then :P
moritz iirc it's the only example on the grapheme level
but there are also funny things on the codepoint level
timotimo i'm working on the (wrong) assumption that we are already at grapheme-level :P
moritz like if you have a composed character, and the case-folded equivalent only exists as decomposed (does that case occur in real-life Unicode?) 11:37
timotimo because most of our regex engine does, too, i think
moritz none of it does
11:37 beee left
moritz it works all on the codepoint level :( 11:37
timotimo er
well ... :|
FROGGS but they all *should* work on grapheme level
but only moarvm will get there I fear, right? 11:38
11:39 daxim left, daxim_ is now known as daxim
timotimo hopefully our jvm port will, too 11:39
FROGGS yeah, I just don't know enough of the impl. to muse about NFG 11:41
nwc10 I haven't read scrollback, but the fun one is /ss/i matches "ß". 11:42
OK, so what are the captures for "ß" =~ /(s)(s)/i; # expressed in p5 syntax
daxim burn it! with fire!
nwc10 you can do it with other folding. that's just the most common example. (of "Unicode consortium didn't think it through") 11:43
daxim I'm going to release Lingua::DEU::Casing::Sharp_s in december 11:44
11:45 leont joined, dram left
daxim r: print "\x{1F4A9}" 11:46
camelia rakudo-parrot 2c40f6: OUTPUT«===SORRY!===␤Unrecognized backslash sequence: '\x'␤at /tmp/PhWcX7u3sY:1␤------> print "\⏏x{1F4A9}"␤Unable to parse expression in block; couldn't find final '}' ␤at /tmp/PhWcX7u3sY:1␤------> print "\…»
..rakudo-jvm 2c40f6: OUTPUT«===SORRY!===␤Unrecognized backslash sequence: '\x'␤at /tmp/vFKZ1XXmJV:1␤------> print "\⏏x{1F4A9}"␤Unable to parse expression in block; couldn't find final '}' ␤at /tmp/vFKZ1XXmJV:1␤------> print "\x{1…»
FROGGS r: print "\x1F4A9" 11:47
camelia rakudo-parrot 2c40f6, rakudo-jvm 2c40f6: OUTPUT«💩»
daxim r: print "\x666"
camelia rakudo-parrot 2c40f6, rakudo-jvm 2c40f6: OUTPUT«٦»
FROGGS r: print "\x1E9E" 11:48
camelia rakudo-parrot 2c40f6, rakudo-jvm 2c40f6: OUTPUT«ẞ»
11:48 dmol left 11:50 tgt joined
timotimo made a questhub quest to organize his thoughts 11:57
11:59 go|dfish joined 12:04 iSlug left 12:06 rindolf left
moritz quest: organize my thoughts 12:07
leont Maybe interesting for perl6 people too: www.nntp.perl.org/group/perl.qa/201...13531.html 12:08
Not that I have much of a clue about the state of QA in perl6
moritz sounds like an awesome hackathon 12:09
leont: we're much in need of QA :-)
leont I've heard the Test::Builder 2 is a rather good design (except for it using Mouse), possibly porting it over is useful. 12:12
A better harness may be desirable too (the current one rather reminds me of perl 5.000's, in having the minimum required features) 12:13
But without a good Builder, additional testing modules just wouldn't scale well 12:17
12:25 rurban1 joined 12:28 denis_boyun_ left 12:33 dmol joined 12:38 denis_boyun joined, dmol left
timotimo made the rakudo optimizer unfold junctions insied when, like { when Int | Str { ... } } 12:41
spectests look good, but there's not terribly many spectests for this kind of thing anyway 12:45
moritz add some :-)
timotimo r: my $_ = []; when {$^a.push(100)}|{$^a.push(200)} { say "what am i doing" }
camelia rakudo-jvm 2c40f6: OUTPUT«Potential difficulties:␤ Redeclaration of symbol $_␤ at /tmp/HHsNjTkslR:1␤ ------> my $_ ⏏= []; when {$^a.push(100)}|{$^a.push(200␤what am i doing␤»
..rakudo-parrot 2c40f6: OUTPUT«Potential difficulties:␤ Redeclaration of symbol $_␤ at /tmp/FkGZ_DN6AW:1␤ ------> my $_ ⏏= []; when {$^a.push(100)}|{$^a.push(200␤what am i doing␤succeed without when clause␤ in block at gen/parrot/CORE.settin…»
timotimo r: $_ = []; when {$^a.push(100)}|{$^a.push(200)} { say "what am i doing" } 12:46
camelia rakudo-parrot 2c40f6: OUTPUT«what am i doing␤succeed without when clause␤ in block at gen/parrot/CORE.setting:547␤ in block at gen/parrot/CORE.setting:611␤ in block at /tmp/Q7gX7NTN27:1␤ in any at /tmp/Q7gX7NTN27:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146…»
..rakudo-jvm 2c40f6: OUTPUT«what am i doing␤»
timotimo r: $_ = []; when {$^a.push(100)}|{$^a.push(200)} { say "what am i doing" }; .say;
camelia rakudo-jvm 2c40f6: OUTPUT«what am i doing␤»
..rakudo-parrot 2c40f6: OUTPUT«what am i doing␤succeed without when clause␤ in block at gen/parrot/CORE.setting:547␤ in block at gen/parrot/CORE.setting:611␤ in block at /tmp/3_LyQ_dxru:1␤ in any at /tmp/3_LyQ_dxru:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146…»
timotimo r: $_ = []; when {$^a.push(100)}|{$^a.push(200)} { say "what am i doing" }; $_.say;
camelia rakudo-jvm 2c40f6: OUTPUT«what am i doing␤»
..rakudo-parrot 2c40f6: OUTPUT«what am i doing␤succeed without when clause␤ in block at gen/parrot/CORE.setting:547␤ in block at gen/parrot/CORE.setting:611␤ in block at /tmp/HklmK3NmTz:1␤ in any at /tmp/HklmK3NmTz:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146…»
timotimo well ...
12:47 rurban1 left
timotimo r: $_ = []; given $_ { when {$^a.push(100)}|{$^a.push(200)} { say "what am i doing" }; } $_.say; 12:47
camelia rakudo-jvm 2c40f6: OUTPUT«===SORRY!=== Error while compiling /tmp/KBuM9FVL_y␤Confused␤at /tmp/KBuM9FVL_y:1␤------> push(200)} { say "what am i doing" }; } ⏏$_.say;␤ expecting any of:␤ postfix␤ statement end␤ …»
..rakudo-parrot 2c40f6: OUTPUT«===SORRY!=== Error while compiling /tmp/j21uguGTV6␤Confused␤at /tmp/j21uguGTV6:1␤------> push(200)} { say "what am i doing" }; } ⏏$_.say;␤ expecting any of:␤ postfix␤ statement end␤ …»
timotimo r: $_ = []; given $_ { when {$^a.push(100)}|{$^a.push(200)} { say "what am i doing" }; }; $_.say;
camelia rakudo-parrot 2c40f6, rakudo-jvm 2c40f6: OUTPUT«what am i doing␤100␤»
timotimo seems to still "work"
well, this optimization has less chance to fail, because there's no side-effects that result from evaluating $_ 12:50
... i guess unless you make it a proxy. in which case, what the hell are you doing! :D 12:51
but even that will only evaluate $_ once
so, now that i have a rakudo commitbit, is it "easier to ask for forgiveness than for permission"? 12:53
FROGGS hmm?
moritz +1
12:54 tgt left
timotimo FROGGS: i'm asking if i should push what i think is working code, or if i should ask for reviews before going ahead 12:54
FROGGS timotimo: if you are "sure" that it is all right then go ahead 12:55
12:55 markov left
FROGGS if you feel uncomfortable with it, push to a branch or so :o) 12:55
timotimo i feel comfortable with this, as soon as the spectest run has finished :) 12:56
FROGGS that is a good way to do it :o)
timotimo: what backend are you spectesting?
I can do the other then
timotimo jvm at the moment 12:57
12:57 dmol joined
moritz runs a p-spectest as baseline 12:58
12:58 tgt joined
timotimo notices he doesn't actually have the commitbit 12:59
it's now in timo/rakudo on the "when_junctions_unfold" branch
timotimo publicizes his membership of the rakudo organisation 13:02
github.com/rakudo?tab=members - i'm the only one in it now :P
there's lots of concealed members in the perl6 organisation, too. FROGGS, you for example
FROGGS done 13:03
timotimo: thanks :o)
13:05 lizmat left 13:06 pernatiy left 13:07 tgt left
timotimo so i'm in the organisation, but can't actually commit to the rakudo repository, huh 13:07
FROGGS you are in there due to PRs 13:08
timotimo jvm spectest looks clean
13:09 denis_boyun__ joined 13:10 denis_boyun left
timotimo i think i need zram on my desktop >_< 13:11
yeah, looks clean 13:17
timotimo force-pushed over the branch to make a proper PR 13:18
13:19 pdurbin left
timotimo github.com/rakudo/rakudo/pull/228 - feel free to pull it in 13:19
moritz loves the fact that he can now run a j-spectest and at the same time rebuild rakudo-p
timotimo loves that he can run a j-install or j-spectest without having to rebuild rakudo-p first 13:20
moritz++ # pretty fantastic build system
dalek kudo/nom: b0bf4d5 | (Timo Paulssen)++ | src/Perl6/Optimizer.nqp:
allow junction unfolding for callmethod ACCEPTS
 13:21
kudo/nom: b11094e | (Tobias Leich)++ | src/Perl6/Optimizer.nqp:
Merge pull request #228 from timo/when_junctions_unfold

allow junction unfolding for callmethod ACCEPTS
woolfy timotimo: congrats with your commitbit!
timotimo woolfy: i was mistaken, i don't actually have it yet :( 13:22
woolfy oowowwwww... too bad
13:22 tgt joined
timotimo good things come to those who wait :P 13:22
FROGGS *g* 13:23
woolfy I'll retract the congrats and put them in the fridge for you, I hope they stay good
dalek p: aa14903 | (Timo Paulssen)++ | src/vm/jvm/QAST/Compiler.nqp:
factor out a common instruction
 13:26
p: 5a95294 | (Timo Paulssen)++ | src/vm/ (2 files):
don't try to match \r\n for negated or zerowidth whitespace cclasses
13:27 denis_boyun__ left 13:33 leont left 13:34 ajr joined 13:35 ajr is now known as Guest65187, Guest65187 is now known as ajr_ 13:36 kivutar left 13:40 dakkar left
timotimo is attempting to implement for @arr -> $foo, $bar? in NQP 13:40
13:42 dakkar joined
arnsholt timotimo++ 13:47
timotimo how do i clear a register? 13:48
moritz clear?
assign a defined value to it? 13:49
timotimo pretend i never did, so to speak
my current bug is that for 1, 2, 3, 4, 5 -> $a, $b? will give 1 2, 3 4, 5 4
because i don't clear the $b register if the list is finished
it's a _p register, if that makes a difference 13:50
moritz set it to NQPMu?
timotimo i'll try that
13:51 fhelmberger left
timotimo nqp_get_sc_object $P5006, "7A67D46DDEA3C60BB73DFB1CA4F76715F3D1212C-1384263841.80708", 26 - this apparently gets NQPMu; how do i emit a pirop in Operations.nqp that does the right thing? 13:52
oh, maybe there's an op i can emit that just returns NQPMu? like nqp::null?
nqp: say(nqp::null())
camelia nqp-jvm: OUTPUT«java.lang.NullPointerException␤ in (gen/jvm/stage2/NQPCORE.setting:672)␤ in print (gen/jvm/stage2/NQPCORE.setting:671)␤ in say (gen/jvm/stage2/NQPCORE.setting:678)␤ in (/tmp/ex7apG0lGy:1)␤ in (gen/jvm/stage2/NQPHLL.nqp:1100)␤ in eval (gen/jvm/sta…»
..nqp-parrot: OUTPUT«Null PMC access in get_string()␤current instr.: '' pc 768 (gen/parrot/stage2/NQPCORE.setting.pir:355) (gen/parrot/stage2/NQPCORE.setting:665)␤»
..nqp-moarvm: OUTPUT«␤»
timotimo that's not quite it :|
moritz QAST::WVal.new( :value($*W.find_sym(['NQPMu'])) ); 13:53
arnsholt timotimo: In the code gen, you probably want to set it to something like QAST::WVal.new( :value($*W.find_sym(['NQPMu'])) )
Bah, sniped =)
moritz (copy&paste)++ # faster than arnsholt++ :-)
arnsholt I copy-pasted as well
I just had to open a terminal and get to the directory and then git grep =) 13:54
timotimo can i do something like self.as_post(QAST::WVal.new(...))?
13:54 nnunley joined
moritz uhm, not sure if $*W is still around in the optimizer :( 13:54
timotimo this is not in the optimizer 13:55
arnsholt Doing this in the optimizer, though? That sounds weird...
timotimo this is in Operations.nqp
arnsholt Why does it have to be done in Operations.nqp?
moritz no idea if $*W is around then
timotimo arnsholt: because that's where 'for' is implemented. 13:56
arnsholt Oh, right. Of course 13:57
13:57 fhelmberger joined
timotimo :) 13:57
i'll see what happens if i just null the register
14:00 kaleem left
timotimo probably Terrible Things 14:00
14:01 logie joined
timotimo yup, NPMCA, as expected 14:01
er, i don't know how to make a QAST::Op for bind that binds into the register i have allocated for my operation 14:03
as i don't have a var to write to
maybe QAST::VM is needed here 14:04
moritz maybe a QAST::Op.new(:pirop<set>, '$P22', ....)
and yes, QAST::VM then
though I wonder if there shouldn't be a simpler way
timotimo i think there should be, but what do i know :) 14:05
in the worst case, the for opcode will get an argument that passes a NQPMu to it 14:06
diakopter so we were walking to breakfast and were suddenly in the middle of SUSEcon
timotimo i'll have to find out how to do the same for Moar and jvm, too :|
diakopter and tadzik is like, "is Novell still a thing?"
timotimo moritz: feel free to praise me in the blog post about the optimisation for scan + literal :D 14:07
Error while compiling op for (source text: "@a -> $a, $b? { say(\"$a $b\") }"): Method 'find_lexical' not found for invocant of class 'NQPMu'
it seems like i don't have access to $*W at this point 14:08
i may have to generate one call instruction for each number of actually existing arguments :|
arnsholt Yeah, I think $*W is set in the TOP rule, so by the time code-gen rolls around it's not there
timotimo yeah, that makes sense
diakopter so then tadzik is like "we would fit in just fine with our Perl shirts." 14:11
14:11 pernatiy joined
diakopter which isn't true, but mostly because they're all in business formal attire 14:11
14:12 ggoebel15 joined 14:13 woolfy left
diakopter woolfy: what does the oolfy stand for? 14:14
timotimo cleaned his whiteboard 14:17
ready for some sketchion!
14:18 lizmat joined, tgt left
diakopter timotimo: is that a tasty breakfast treat? 14:19
14:20 kivutar joined
timotimo it could be :) 14:21
14:27 lue left 14:28 lowpro30 joined
timotimo made a quest for it: questhub.io/realm/perl/quest/52823b...b3340000c9 14:28
14:31 rurban1 joined
FROGGS I'd guess the output is "1,\n2,\n3,\n" 14:31
like there is missing the comma after the 3 above 14:32
timotimo edited, thanks 14:33
PerlJam greetings!
moritz \o PerlJam 14:34
PerlJam reminds everyone that advent season is nigh.
Add yourselves to github.com/perl6/mu/blob/master/mi...3/schedule
read github.com/perl6/mu/blob/master/mi...instorming for ideas if you don't have any 14:35
(please add to topic-brainstorming too)
moritz has trouble finding interesting topics that haven't been chewed out in the advent before
PerlJam moritz: yeah, I know the feeling
moritz and I must be competent to write about it
PerlJam ditto
moritz so, what have I done this year in the rakudo world? 14:36
build system? not a sexy topic
timotimo 1 2
3 4
5
\o/
FROGGS \o/
moritz a bit constant folding... not too exciting either
timotimo++
14:37 lowpro30_ joined
moritz lots of small cleanups and corrections... not exciting either 14:37
14:37 fhelmberger left
PerlJam moritz: small, unsexy things, in aggregate, can be exciting :) (probably not ever sexy though :) 14:37
dalek p/optional_for_arguments: 38333d3 | (Timo Paulssen)++ | src/ (3 files):
teach nqp-p about optionals in loop blocks
 14:38
timotimo could a post about jnthns workshop be interesting to our audience?
PerlJam timotimo: yes
timotimo feel free to play with my optional_for_arguments branch and try to break it, please
moritz labster: would you blog about dir(), File::Path etc. on the p6 advent calendar? 14:39
14:39 lowpro30 left
moritz I could write about custom EXPORT subs :-) 14:40
14:40 lue joined
moritz oh, I know what I did this year: sink context 14:41
and I wish I hadn't
timotimo :(
14:42 fhelmberger joined
timotimo what did I do all year? :\ 14:42
hm. i think i started my perl6 stuff a month or two before the advent calendar last year 14:43
moritz timotimo: you could blog about your experience contributing to Perl 6, as a relative newcomer 14:44
PerlJam that sounds like an excellent topic
timotimo ... no pressure! :P
14:45 lizmat left, kaare_ joined, raiph joined 14:46 lizmat joined
PerlJam would like to see an advent post (or two) from lizmat about all the clean ups she's done :) 14:46
timotimo i wish i could finish my "is nodal" work before the advent calendar
PerlJam Heck, she could re-package her YAPC::NA lightning talk and that would be awesome 14:47
It would make an excellent Dec 1 post even 14:48
14:50 bluescreen10 joined
timotimo oh! oh crap! 14:50
i don't handle redo at all at the moment :o 14:51
if there's optional parameters, there has to be one redo for each!
(for each amount of optionals that were present)
where can i look to figure out how parrot does label calculation?
moritz label calculation? What's that? 14:52
you mean in the PIR -> PBC step? 14:53
timotimo er, no
14:53 btyler joined
timotimo hopefully i could just replace the redo label just before each call 14:53
moritz doesn't quite understand, but has to leave anyway 14:54
timotimo so, more like "label arithmetics"
a bit like "pointer arithmetics"
or i'd have to build a "dispatcher" for the redo that looks at some integer register and jumps to the right label given its value 14:55
that would suck :(
14:56 rurban1 left
colomon Day 1 should be the state of Perl 6, shouldn't it? talk about JVM and Moar, how to get rakudo installed these days, etc. 14:57
timotimo that's already in the schedule :)
(but not with a name)
PerlJam Well, I make Day 1 be "State of Perl 6" out of habit because I think it's a good way to start off advent. There's nothing that says it *must* be that. 14:58
15:00 rurban1 joined
PerlJam But there should be some post that talks about the Perl 6 ecosystem in any case. 15:01
colomon there's a schedule already?
PerlJam an empty one, but yeah 15:02
github.com/perl6/mu/blob/master/mi...3/schedule
15:02 denis_boyun joined
PerlJam colomon: got any interesting topics you want to do for advent? 15:02
last year the schedule came together kind of at the last minute. That always makes me nervous. 15:03
timotimo how come there's no mention of concurrency/parallelism/asynchronicity?
colomon PerlJam: I've been pondering recoding what I've been working on in C++ in p6 and posting on it. It would be a quick small example of how cool given / when and grammars are. 15:04
PerlJam: I even have some hope I might have enough free time to write it. ;) 15:05
PerlJam colomon: sounds excellent! :)
timotimo: feel free to add it to the topics
(or, better, claim a day for that topic yourself :)
timotimo i don't feel like i'm the right person to write about it 15:06
i don't have much experience in parallelism and stuff
15:07 lowpro30 joined
PerlJam A post on NativeCall for the JVM would be nice too. 15:07
15:08 kaleem joined
timotimo let's hope it's in a releasable state until then 15:09
PerlJam Even if it's just "this is almost done, but here's a tiny bit that works" I think it would use useful/interesting. 15:10
s/use/be/
15:10 lowpro30_ left
colomon PerlJam: another thought: redo the old Mandelbrot code (used twice before for Advent, I think?) using parallel code in JVM. 15:14
might nicely emphasis the performance improvements we've had this year. 15:15
15:16 dmol left, jeffreykegler joined, dmol joined
dalek : 1395dd5 | duff++ | misc/perl6advent-2013/topic-brainstorming:
add some topics
 15:16
15:17 iSlug joined 15:18 ingy2^ joined 15:21 jeffreykegler_ joined, denis_boyun left
PerlJam Now, we just need people to claim days and topics 15:21
15:22 thou joined, ingy2^ left, jeffreykegler left 15:23 tgt joined 15:29 atroxaper joined
timotimo nqp: my $num := 10; while $num > 0 { $num := $num - i; }; 15:30
excellent typo ...
camelia nqp-moarvm, nqp-jvm, nqp-parrot: OUTPUT«(timeout)» 15:31
PerlJam does nqp grok i?
tadzik hahaaha
timotimo i think it is NQPMu instead
nqp: say(i)
camelia nqp-moarvm, nqp-jvm, nqp-parrot: OUTPUT«␤»
timotimo like everything :)
nqp: say(NQPMu)
camelia nqp-moarvm, nqp-jvm, nqp-parrot: OUTPUT«␤» 15:32
tadzik well, at least ita's not i vs $i
15:32 rurban1 left
timotimo :) 15:32
tadzik imagine my $i = 0; i < 5; $i++ 15:33
dat horror 15:34
timotimo :D
15:34 woolfy joined
colomon rn: say i < 5 15:35
camelia rakudo-jvm b11094: OUTPUT«Cannot call 'Real'; none of these signatures match:␤:(Mu:U \v: Mu *%_)␤ in any at gen/jvm/BOOTSTRAP.nqp:1201␤ in sub infix:<<> at gen/jvm/CORE.setting:3971␤ in sub infix:<<> at gen/jvm/CORE.setting:3971␤ in block at /tmp/81k5wPDMMc:1␤ in …»
..rakudo-parrot b11094: OUTPUT«Cannot call 'Real'; none of these signatures match:␤:(Mu:U \v: Mu *%_)␤ in any at gen/parrot/BOOTSTRAP.nqp:1218␤ in any at gen/parrot/BOOTSTRAP.nqp:1209␤ in method Real at gen/parrot/CORE.setting:990␤ in sub infix:<<> at gen/parrot/CORE.s…»
..niecza v24-98-g473bd20: OUTPUT«Unhandled exception: Complex numbers are not arithmetically ordered; use cmp if you want an arbitrary order␤ at /tmp/mME5Yg143X line 1 (mainline @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4583 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/C…»
tadzik niecza wins 15:36
jnthn timotimo: I glanced your patch. It adds 8 bytes to every single QAST node for one feature we only sorta need in NQP, so in its current form it's worse than what it cures... 15:38
timotimo: I'd just assume if we do have arity > 1 set then the others were marked as optional. It may not be true, but if we didn't we're screwed anyway, and also we'll get an error like full Perl 6 gives (not enough args passed) 15:39
timotimo thanks. got a suggestion for how to make it better?
add optarity using a role that only sometimes exists? 15:40
jnthn What I just wrote? :)
No, just keep arity
And assume.
(that all but the first one are potentially optional)
timotimo wait, why?
jnthn I don't think anything else is worth the complexity.
timotimo oh
that's why
yeah, i can do that
do you have a suggestion for the redo problem? 15:41
jnthn r: my @a = 1,2,3; for @a -> $x, $y { }
camelia rakudo-parrot b11094: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in block at /tmp/nHohizSqqN:1␤ in any at /tmp/nHohizSqqN:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146␤ in any eval at gen/parrot/stage2/NQPHLL.nqp:1133␤ in any evalfiles a…»
..rakudo-jvm b11094: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in block at gen/jvm/CORE.setting:7270␤ in block at gen/jvm/CORE.setting:7106␤ in method gimme at gen/jvm/CORE.setting:7524␤ in block at /tmp/bGWE2CZbHX:1␤ in any eval at gen/jv…»
jnthn Note that with what I'm suggesting you'd get a consistnet error with ^^ :)
redo problem?
timotimo yeah. at the moment i have one call for each amount of arguments the block may have populated
a redo would have to find that same call branch again
jnthn But isn't the redo handler around the call? 15:42
timotimo er, i should look closely
jnthn So you could have an approriate one for each on MoarVM as that is really cheap.
On Parrot it's different and would make things more expensive, so I guess "save an integer" or whatever...
This is turning into a lot of complexity for an edge case... 15:43
timotimo i can abort if you'd prefer
15:44 lizmat_ joined
jnthn Well, it will bite people is all. I just hadn't thought about the redo thing when saying "let's fix it"... 15:44
timotimo bite ... in what way?
jnthn Well, the bug god filed because somebody at the internals workshop actually ran into it 15:45
colomon is worried about appeasing the bug god
timotimo :D
jnthn uh
got :D
lol
timotimo if i implement the redo thing, it won't bite people? what am i misunderstanding here? 15:46
geekosaur was wondering...
colomon suspects that masak is beloved by the bug god
jnthn timotimo: I'm just sayin gthat when I thought it should be fixed, I didn't think about the interaction with redo.
timotimo: And now that I do, I realize it's not as easy a fix as I'd first imagined.
timotimo but that doesn't mean i shouldn't attempt the fix, right? 15:47
jnthn I think go for it and just try to make sure the common case of for @a -> $x { ... } does not get more expensive for it.
timotimo sure thing
at the moment, the common case is still a branch that doesn't get any different code at all
i can keep it that way
15:48 lizmat left
jnthn Preferable, provided it doesn't lead to loads of dupe 15:48
OK, I need to go catch boats and trains and stuff
bbl
timotimo have fun!
moritz dupe, the programmer's anti-dope
atroxaper How can i call method of parent class from childe class? I don't find it in Using Perl6... 15:49
moritz self.TheParent::themethod(@args) 15:50
atroxaper moritz: cool. Thanks!
moritz or callsame/callwith, depending on the exact use case
PerlJam moritz: It's odd that you didn't lead off with callsame/callwith :) 15:51
atroxaper Hmm... Very flexible
PerlJam perhaps it's just odd that I think callsame/callwith would be more commoon
er, common even
moritz PerlJam: I cannot remember if call{same,with} work outside the context of multis 15:52
PerlJam: which is why the other alternative seemed safer to mention first
PerlJam ah
perhaps atroxaper really wanted nextsame/nextwith too? 15:53
The question didn't have enough information for me to tell
15:53 jnap joined
atroxaper And one more question: whay i can't type grammar P6Gr is Perl6::Grammar {} ? I mean why P6Gr.parce throwing an exception? 15:54
moritz perhaps because you have misspelled 'parse'?
it would help to know what exception you get
atroxaper No ;)
No such method 'submethod_table' for invocant of type 'NQPClassHOW' 15:55
moritz atroxaper: that's likely related to Perl6::Grammar being an NQP class, not a rakudo class
would be nice if it worked, though
atroxaper use Perl6::Grammar:from<NQP>; use Perl6::Actions:from<NQP>; grammar P6Gr is Perl6::Grammar {}; say P6Gr.parse('my $b=4;', :actions(Perl6::Actions.new())).dump; 15:56
rp: use Perl6::Grammar:from<NQP>; use Perl6::Actions:from<NQP>; grammar P6Gr is Perl6::Grammar {}; say P6Gr.parse('my $b=4;', :actions(Perl6::Actions.new())).dump;
camelia rakudo-parrot b11094: OUTPUT«No such method 'submethod_table' for invocant of type 'NQPClassHOW'␤ in any at gen/parrot/Metamodel.nqp:449␤ in any methods at gen/parrot/Metamodel.nqp:444␤ in any !protoregex_table at gen/parrot/stage2/QRegex.nqp:766␤ in any at gen/parrot/…»
15:57 rurban1 joined, rurban1 left, rurban1 joined
moritz in that case I'd recommend delegation, not inheritance 15:57
that is, keep the NQP Perl6::Grammar as an attribute, don't inherit from it 15:58
atroxaper moritz: I am finding a way to find out a place where grammar failed (when we put some wrong stuff in parse method). And i'm thinking about overriding some method of Perl6::Grammar. 15:59
moritz atroxaper: then it's likely easier to do that in NQP code 16:00
atroxaper moritz: override panic method and get CURSOR object... 16:01
moritz though it might be an easy fix in rakudo, lemmy try something... 16:02
atroxaper How NQP code differ from Perl6? I mean what should i code? Just .nqp file?
16:02 denis_boyun joined
atroxaper moritz: And use MyModule:from<NQP>; Right? 16:03
16:04 telex left
atroxaper What dows 'lemmy' mean? ) 16:07
16:08 telex joined
FROGGS let me 16:08
16:09 denis_boyun left
moritz atroxaper: NQP is more low level than NQP; I recommend looking at the existing example code in src/Perl6/Actions.nqp for example to get an idea how it looks 16:10
atroxaper FROGGS: oh... Ok. I am not an English guy...
FROGGS (we neither)
err, nqp is more low level than perl6
moritz also 6guts.wordpress.com/2013/09/17/mat...ls-course/ has some pointers to documentation for NQP
gfldex does Perl6::Grammar throw exceptions?
FROGGS gfldex: yes
rp: 07; # that is why we have this bug 16:11
camelia rakudo-parrot b11094: OUTPUT«Potential difficulties:␤ Leading 0 does not indicate octal in Perl 6; please use 0o7 if you mean that␤ at /tmp/k3Imm_22mr:1␤ ------> 07⏏; # that is why we have this bug␤ Leading 0 does not indicate octal in Perl 6…»
FROGGS the bug is that it warns twice
16:11 tgt left
atroxaper moritz: ok. thank you. I will try to inherit grammar by NQP. 16:12
16:13 tgt joined 16:14 ingy2^ joined
atroxaper moritz: I know what NQP is. I just dont know how i can write nqp module and import it in Perl6 code ) 16:14
16:14 tgt left 16:15 raiph left
moritz atroxaper: you seem to be doing fine :-) 16:16
atroxaper moritz: glad to hear :-) 16:17
colomon PerlJam: just realized I've already written like a third of the code I was thinking I was going to have to write for the advent post I was thinking of. ;) 16:18
16:18 tgt joined 16:19 dmol left
PerlJam colomon: nice 16:19
16:20 iSlug left
colomon also, I think I found a major shortcoming in my C++ implementation. :\ 16:20
r: say "1f".base(16) 16:21
camelia rakudo-parrot b11094: OUTPUT«No such method 'base' for invocant of type 'Str'␤ in block at /tmp/57ANSoL8Sv:1␤ in any at /tmp/57ANSoL8Sv:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146␤ in any eval at gen/parrot/stage2/NQPHLL.nqp:1133␤ in any evalfiles at gen/parrot/…»
..rakudo-jvm b11094: OUTPUT«No such method 'base' for invocant of type 'Str'␤ in block at /tmp/8zyzVgMb7B:1␤ in any eval at gen/jvm/stage2/NQPHLL.nqp:1086␤ in any evalfiles at gen/jvm/stage2/NQPHLL.nqp:1292␤ in any command_eval at gen/jvm/stage2/NQPHLL.nqp:1196␤ in any…»
colomon r: say "1f".hex
camelia rakudo-parrot b11094: OUTPUT«No such method 'hex' for invocant of type 'Str'␤ in block at /tmp/_K2yVe0vjG:1␤ in any at /tmp/_K2yVe0vjG:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146␤ in any eval at gen/parrot/stage2/NQPHLL.nqp:1133␤ in any evalfiles at gen/parrot/s…»
moritz atroxaper: I have a local patch that gets rid of the error message (inheriting from an NQP class), but a new problem appears then
camelia ..rakudo-jvm b11094: OUTPUT«No such method 'hex' for invocant of type 'Str'␤ in block at /tmp/tG0N9j1Hzk:1␤ in any eval at gen/jvm/stage2/NQPHLL.nqp:1086␤ in any evalfiles at gen/jvm/stage2/NQPHLL.nqp:1292␤ in any command_eval at gen/jvm/stage2/NQPHLL.nqp:1196␤ in any …»
moritz P6opaque only supports type changes where the MRO of the original type is a suffix of the MRO of the new type
colomon r: say :16("1f") 16:22
camelia rakudo-parrot b11094, rakudo-jvm b11094: OUTPUT«31␤»
16:22 [Sno] left
colomon does not know why he always forgets that. 16:22
moritz I don't know why the subcllass'es MRO shouldn't be a suffix of the superclasses' MRO
atroxaper MRO ? 16:23
moritz maybe related to automatically adding Mu?
atroxaper: method resolution order
16:23 zakharyas left
moritz a linearization of the tree of parent classes 16:23
16:23 raiph joined
atroxaper moritz: General i don't want to rid error message. I want to get a CURSON in that moment for getting position of cursor. 16:25
CURSOR i mean 16:26
gfldex atroxaper: did you try to catch the excpetion? 16:30
atroxaper moritz: can i just override Perl6::Grammar.panic method to get my $cursor_class field of HLL::Grammar ? )
FROGGS moritz: I hit the same problem when trying to make v5 a proper module 16:31
colomon r: my $hex = "AFFD" ~~ / [<[0..9]> | <[A..F]>]+/; say $hex
camelia rakudo-parrot b11094, rakudo-jvm b11094: OUTPUT«｢AFFD｣␤␤»
colomon r: my $hex = "AFFD" ~~ / [<[0..9]> | <[A..F]>]+/; say :16($hex)
camelia rakudo-parrot b11094: OUTPUT«Nominal type check failed for parameter '$str'; expected Str but got Match instead␤ in sub unbase at gen/parrot/CORE.setting:6513␤ in block at /tmp/YZPYMJN5ss:1␤ in any at /tmp/YZPYMJN5ss:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146␤ …»
..rakudo-jvm b11094: OUTPUT«Nominal type check failed for parameter '$str'␤ in block at /tmp/zzAFioQMzP:1␤ in any eval at gen/jvm/stage2/NQPHLL.nqp:1086␤ in any evalfiles at gen/jvm/stage2/NQPHLL.nqp:1292␤ in any command_eval at gen/jvm/stage2/NQPHLL.nqp:1196␤ in any c…»
colomon bug or just LTA?
FROGGS moritz: so it would be nice if we could brainstorm a bit about that, maybe with jnthn 16:32
moritz atroxaper: I have no idea
colomon r: my $hex = "AFFD" ~~ / [<[0..9]> | <[A..F]>]+/; say :16(~$hex)
moritz colomon: LTA bug :-)
camelia rakudo-parrot b11094, rakudo-jvm b11094: OUTPUT«45053␤»
atroxaper gfldex: no. but i think that catching dont help me in that case...
moritz colomon: well, kinda, (more)
colomon: I think that :16(...) intentionally doesn't use Cool for its arguments
16:32 tgt_ joined, dbr left
moritz colomon: so that we can catch the case where people pass integers to :16(), because they are confused 16:32
colomon ah 16:33
it's obviously easy to work around...
atroxaper moritz: ok. i will try.
16:34 tgt left, tgt_ is now known as tgt, FROGGS left
dalek kudo/inherit-from-nqp-class: b7de167 | moritz++ | src/Perl6/Metamodel/MethodContainer.nqp:
inch towards making Perl 6 classes able to subclass NQP classes
 16:34
16:35 tgt_ joined, tgt left
tadzik hah, flight from waw to london is like 50€ 16:45
lpw seems more and more tempting
moritz woah 16:46
16:54 ingy3^ joined, ingy3^ left
atroxaper night #perl6 16:58
16:58 atroxaper left 16:59 rindolf joined 17:01 kaleem left 17:09 dmol joined 17:20 stevan_ left
diakopter only $1100 to fly from sfo. more and more attractive. 17:26
quietfanatic agrees with me that Lesser and Greater would be better than Less and More 17:29
since those imply direction slightly more 17:30
jnthn o/ from le train
diakopter jnthn: ahoy buggod
jnthn *groan* 17:31
diakopter jnthn: how about Lesser/Greater 17:32
jnthn context?
tadzik Gesser/Later
jnthn oh, the Increase/Decrease...
timotimo jnthn: if i handle arity > 1 like optional arity, it'll result in more verbose/cautious code in any case :(
but yeah, you're right, we would be getting a nicer error message out of the whole ordeal 17:33
17:33 ajr_ left
tadzik well, I opened the bug because jnthn said that that should work 17:33
jnthn timotimo: Only for arity 2 things, though?
timotimo er, what?
tadzik I'll be fine with agreeing that it shouldn't :)
timotimo tadzik: i'd like it to work, though :) 17:34
jnthn timotimo: 2 or greater
diakopter oh noes
jnthn arity 1 we can assume :)
17:34 ajr joined
tadzik okayo :) 17:34
timotimo yes, 2 or greater.
tadzik jnthn: 2 or more :P
17:34 ajr is now known as Guest71892
jnthn 2 or increase :P 17:34
timotimo btw, i changed the code that increases the arity of the cur_lexpad thingie from "only if it has no default" to "always"
is that dangerous?
17:35 Guest71892 is now known as ajr_
jnthn Well, so long as it doesn'tcount names 17:35
It probably is OK, especially if tests/spectest pass 17:36
timotimo it does not count nameds, no
there is no calculated jump in parrot, is there? 17:37
jnthn yeah, I think there is a local_jump thingy
17:37 [Sno] joined
timotimo do i have to calculate bytecode widths for that? 17:37
jnthn the regex engine uses it 17:38
it's dangerous s hell
no, it's not that bad :)
timotimo heh
jnthn but I *think* you might be able to segfault things with sufficiently good use of it
Unless things got fixed...
timotimo local_jump doesn't seem to exist anywhere in src/ 17:39
jnthn I don't know what it's actually called...
Look somewhere in the regex code postlude
timotimo oh, it's just jump $I123
jnthn oh :) 17:40
timotimo and i can set $I123, label_name[$offset]
17:40 trexy joined
jnthn k 17:41
timotimo huh, is that it? its usage is weird.
jnthn that looks weird to me too but I'm mostly managing to forget PIR :P
timotimo goes with a linear search instead 17:42
jnthn tries to remember if anybody besides timotimo++ now has worked on code-gen on every backend... 17:43
moritz jnthn: I'm pretty sure you did :-)
timotimo "worked on" is a bit much, no? :) 17:44
jnthn I...I did? :P 17:45
17:48 nebuchadnezzar joined 17:50 FROGGS joined 17:51 fhelmberger left, fhelmberger joined 17:52 jnap1 joined
moritz there are really two things you can do: work on the actual code generation framework, or do code generation for specific operations (using the previously mentioned framework) 17:52
I've only ever been involved with the latter
17:52 jnap left
jnthn I guess givne I designed JAST and most of MAST I've done a bit of the former :) 17:53
17:56 fhelmberger left 17:58 lowpro30_ joined
tadzik jost most :) 18:01
18:01 lowpro30 left
timotimo yeah, i also only did the latter 18:06
QAST transformations don't count, IMO
hey jnthn, can you tell me what kinds of nodes are safe to be replaced inside a tree of QRegex during optimisation time, in order not to harm the generated NFA data?
18:07 zakharyas joined 18:08 cognominal joined
jnthn timotimo: "replaced"? 18:08
jnthn kinda needs more detial to answer :)
detail
timotimo well, can I remove, for example, a QAST::Regex( :rxtype<literal> ) out of the tree? 18:09
what about an :rxtype<alt>?
will NFA data be lost when i pull such a Regex node out of the tree?
i would like to replace nodes like literal or cclass with a new type of node that just skips ahead when i'm certain the NFA has already done the scanning of that particular node before 18:11
questhub.io/realm/perl/quest/528217...1c340000a8 - here's a writeup of my thoughts on this topic
18:12 ajr_ left
jnthn Trouble is that you don't know if you were invoked by the NFA or not... 18:13
18:13 dakkar left
jnthn statically I mean 18:13
timotimo oh? what other cases are there?
jnthn What I mean is given
token foo { bar }
uh, bad exmaple 18:14
timotimo that's right. i can't do this in many cases
but in the cases where i can, i totally could :)
jnthn token foo:sym<bar> { <sym> }
timotimo like the example in the questhub quest: [ "integer: " (\d+) | "identifier: " (\w+) | "based-number: 0" (<[xob]>) (\d+) ]
jnthn Here, you *may* be in it thanks to the NFA
But it's possible somebody explicilty called it too
timotimo yes, i touched on that exact argument in my questhub quest (without a good solution) 18:15
one idea is to introduce QAST::Want into the regex system and have the nfa set a flag that triggers a different branch of the Want to be executed
jnthn I don't think that's it 18:16
Looking more generally, if we have
token foo:sym<bar> { \w+ }
18:17 ggoebel16 joined
jnthn We actually can handle that from the NFA too 18:17
But it means we need to know where the end is
18:17 kivutar left, jeffreykegler_ left
jnthn I'd actually been pondering something in the other direction than you're thinking... 18:17
That is, when we make the NFA, we differentiate automatic "skip to here" point from fate 18:18
timotimo opposite or orthogonal?
jnthn And store the "skip to here" points
timotimo ah, that seems to be opposite
jnthn Well, it's mroe that we make the NFA tell us
It's generally a better design when we let things that know stuff do the telling, than let things that don't know guess ;)
timotimo yes, that's a good point. 18:19
jnthn And if we want to get REALLY cute
timotimo but that's hard, too. the NFA doesn't carry with it from where it has matched, does it?
jnthn We already generate an NFA for everything.
timotimo or where capture groups begin and end etc etc
18:19 ggoebel15 left
jnthn No, capture groups would scupper it 18:19
But that still leaves quite a lot.
timotimo scupper == break? 18:20
jnthn yeah
well, force us not to do that opt
The other part of this though...
If we have a token foo { \d+ }
Nothing to do with a proto-regex
It has an NFA that can do all the matching 18:21
That is, in certain cases we can delegate to that even in non-protoregex/alternation cases.
timotimo so you want to find tokens that are completely declarative
jnthn It's not just about declarative.
timotimo and have no capture groups in them?
jnthn Well, I tink we need to separate out the two notions
Where is the end of the declarative prefix?
And where is the end of the stuff we can just let the NFA do? 18:22
e.g. before we gotta start capturing.
timotimo ah, of course 18:23
18:23 ggoebel16 left
timotimo at the moment it just bails out if something is !.declarative 18:23
it should continue, but no longer count up the length of the declarative prefix
jnthn oh, I wsa more thinking the other way around
the amount we can consider skippable is a subset of what's declarative 18:24
\d+ <exponent>
That whole thing is probably all declarative
But we can only skip the \d+ 'cus the next thing captures
timotimo by skippable you mean "the QRegex implementation can rely on the NFA to have done it"?
yes, that's true
jnthn Right
There's a static and dynamic aspect here
timotimo i was thinking we could tell {} apart from {...} and let the NFA run past empty {} until it finds another stumbling block 18:25
jnthn Statically we know that it's "after the \d+" which is useful for knowing where the jump to in the generated code.
timotimo so it could, in theory, eliminate more branches?
jnthn *to
TimToady and we could, in theory, do a single capture if it starts at an abs position and ends with a known offset to the end of declarative 18:26
timotimo that's true, too
jnthn timotimo: true, but trying to keep this manageable :)
timotimo i was about to say :)
jnthn timotimo: It's not "eliminate more branches"
It's about not re-doing work.
Anyway, the harder part for us to factor is the dynamic aspect 18:27
timotimo if it's going to be me who's to implement this, it should not be a huge monolithic blob of work; instead i'd hope for a bunch of achievable milestones that all have a bit of a reward to them
TimToady well, some of them could be yaks :)
FROGGS timotimo: that was what I was thinking atm, define separate "smaller" tasks
jnthn That is, knowing wheat to set pos to
PerlJam randomly mentions advent posts again :)
18:28 ggoebel16 joined
timotimo PerlJam: we're too busy building castles in the sky 18:28
jnthn I'd love to tell you a lovely robust design for this but I don't have one to hand :)
diakopter TimToady: Mr Toad's Wild Ride is no more...
PerlJam timotimo: but we don't yet have flying cars to get to the sky-castles
timotimo i was thinking the approach i was suggesting is much simpler in the dynamic case
TimToady diakopter: you mean in DW, or in DL?
timotimo PerlJam: i wasn't saying what we were doing was a good idea :) 18:29
jnthn Maybe I'll have some decent thinking time tomorrow...it's mostly travel and stuff. :)
diakopter W
timotimo it would be fantastic if you could come up with a "typical jnthn design" :)
jnthn timotimo: Will ponder it a bit, given you're interested to work on it :) 18:30
TimToady well, it seems obvious to me that whatever calculates the declarative prefix has to also pass along whatever it knows about lengths and offsets that could be useful later 18:31
PerlJam "I'll solve this problem in the style of ... jnthn!"
18:31 ssutch joined, ggoebel16 left
jnthn TimToady: Yes, it's how to factor that passing along :) 18:31
timotimo jnthn: if it's way too hard for me, though, i may not be up to the task just yet; but having a design almost ready to implement it is bound to help :) 18:32
TimToady alternatively, maybe the declarative bits are in their own little structure that can be scanned as a separate pass
(but I'm guessing that would be harder) 18:33
diakopter TimToady: hm, apparently?
it was 1998 18:34
meh
jnthn timotimo: Well, it may be hard in places, but I suspect I can guide you through them if so :) 18:35
18:35 ggoebel16 joined
jnthn I actually will have time to do Perl 6 things besides ranting about stuff on the channel again next week, it seems... :) 18:36
timotimo \o/
FROGGS nice :o) 18:38
dalek d: b469b42 | larry++ | STD.pm6:
\x{} now recommends \x[]
 18:40
diakopter so, the onion logo means the more you dig into it, the more you cry?
TimToady and the less there is left to cry about :)
timotimo the code may be super ugly, but i think my redo handling now works 18:43
first make it work, then make it beautiful
diakopter Melanie just suggested making a camel logo with a camelia drawn on a saddle on the camel
18:43 SamuraiJack_ left
diakopter .. to which I replied, uhm, you're really setting up that critical pun well 18:44
timotimo i kind of wish there was a construct to go through multiple parallel lists in nqp 18:45
that could potentially also give us an optimized for @foo X @bar -> $a, $b { } for rakudo
diakopter nqp::zipzipzip
TimToady you mean Z I presume? 18:47
timotimo indeed, thinko
18:51 darutoko left
timotimo twitter.com/loltimo/status/400336432939544576 - i made a funny 18:57
18:59 rurban1 left 19:02 pernatiy left 19:07 dwarring joined
lue Can anybody tell me offhand why this line my %meths := $par.method_table($par); might call a sub when it shouldn't, or what NQP debugging features might help me figure that out? .method_table consists entirely of "return %!methods;" 19:07
timotimo --target=pir :P 19:08
19:08 tgt_ left
dwarring .u §«¶· 19:09
moritz
dwarring .u § 19:10
moritz ENOYOLEAUX
timotimo nqp has a function that i really think should be exposed to perl6 to get the name of a character 19:11
19:11 tgt joined
moritz +1 19:11
dwarring seems that § and ¶ have been reclassified as punct between unicode 5.2 and 6.0 19:12
so punct tests in github.com/perl6/roast/blob/master...charsets.t are giving variable results depending on version of icu etc 19:13
methinks I should make those punct tests todo 19:14
moritz +1
timotimo +1 19:15
diakopter +1 19:16
dwarring another good reason to bring unicode handling into nqp core 19:18
jnthn Bring them in line with the latest
And then we can todo them by backend, when the backend is behind or inconsistent due to ICU version
I think that on Moar we have >= 6
TimToady has been arguing against icu from the start, having found the other approach worked pretty well for P5 19:19
FROGGS no icu in moarvm \o/ 19:20
timotimo +1 19:21
diakopter I can very easily port/generalize moar's unicu to the other backend
timotimo oooooh!
TimToady it's the very complexity of tracking Unicode that argues against entrusting it to a 3rd party
19:22 ingy2^ left
TimToady and it's one of the spots where P5 is still miles ahead of most other programming languages 19:22
19:23 rindolf left 19:25 tgt left, spider-mario joined
lue
.oO( /usr/local/lib/lib»ö«.so )
 19:25
FROGGS lue: btw, I only see $foo.HOW.method_table($foo) in nqp's src 19:26
lue: did you forgot the .HOW?
lue FROGGS: maaybe, but I don't think that's the issue. /me checks anyway
o.o 19:27
FROGGS O.o
lue FROGGS++
FROGGS \o/
diakopter no one reacted to my "other backend" dig
19:28 ingy2^ joined
lue diakopter: which other backend are we talking about? :P 19:28
moritz diakopter: +2 (to porting it to two other backends)
FROGGS diakopter: would make sense I guess... but I totally wanna see NFG too *g*
Util #ps in 3m
lue FROGGS: but how missing that .HOW caused the sub within which that statement was to get called immediately again is a mystery to me :/
diakopter FROGGS: well that's harder bc it's the whole string impl 19:29
FROGGS diakopter: and in moarvm only?
(I really do like MoarVM)
lue: I guess it wants to call .method_table on the return value of $par or so 19:30
lue FROGGS: thing is, the sub being called again is a sub within a method find_method. Without going through the method again, if the 'die'-induced backtrace is to be believed. 19:31
FROGGS diakopter: I just think that at some point it does not make soo much sense to port stuff over from MoarVM to Parrot, since, it would be a MoarVM-clone with another heartbeat then
lue FROGGS: I don't really care though, because it works now :) 19:32
FROGGS *g*
lue: that's fine :o)
tadzik about 'which other backend': i2.kym-cdn.com/photos/images/origin...94/463.jpg
FROGGS tadzik++
hehe
19:33 ingy2^ left
TimToady timotimo: will your optimization make it harder to optimize when 1|2 {} to a jump table at some point? 19:33
or can the recognizer for that just run in front of this optimization? 19:34
jaffa4 WHat is the state of MoarVM? 19:35
TimToady (Perl 4 would actually make a jump table for the first character of string cases as well as for integer cases!)
diakopter jaffa4: existeth
TimToady nqp bootstrapped on it
just needs rakudo now :) 19:36
jaffa4 what does it mean that nqp bootstrapped on it?
19:36 zakharyas left
TimToady means that nqp can compile nqp running on top of MoarVM, and use that to compile itself on top of MoarVM, and use that... 19:37
jaffa4 ok 19:38
It is a kind of measure of progress I see
diakopter masak: name for p5interop: Voldemoar
jaffa4 What is the advantage of MoarVm?
TimToady it's moar better!
it's designed specifically to support Perl 6 19:39
diakopter jaffa4: it has higher ground
TimToady it natively speaks p6's metamodel
jaffa4 So is it going to be faster than javavm or parrot?
diakopter someday 19:40
tadzik it's what Parrrot wanted to be when it grows u,
* up
lue
.oO(MoarVM is basically Parrot timeshifted about a decade.)
diakopter both directions
FROGGS jaffa4: we are currently working on making rakudo work on it btw
lue: and with the knowledge how a perl6 implementation looks like, to be fair
jaffa4 Is there going to be PIR for MOarvm? 19:41
FROGGS NO
TimToady making rakudo work primarily entails 1) getting the CORE setting to parse, and 2) implementing the run-time support for the semantics of those things that are beyond what npq needs
jnap1 for long lived processes, the javaVM can can really optimized, but startup isn't great. I guess MoarVM will be faster for CLI, from my understanding. At least as a lurker here is what I've heard
FROGGS *NO PIR* *NO PROJECT EULER*
tadzik NO LOL
FROGGS (the latter is a joke)
TimToady jnap1: yes, that seems accurate 19:42
lue FROGGS: yep. The "timeshift" comment was referring to "what if Parrot design started a decade later" than anything else.
*more than
FROGGS jaffa4: it does not spit out yet another readable code, just to parse it again
TimToady lue: no "what if" about it :P
tadzik will moar have pir: i2.kym-cdn.com/photos/images/origin...94/463.jpg
lue :)
FROGGS hihi
diakopter
.oO( sometimes I find myself writing here, imagining chromatic reading the logs, like a convalescent reads the funny papers
)
FROGGS diakopter: I've read more of his stuff yesterday :o) 19:43
tadzik I should make an image-only faq
jaffa4 It means it will convert it into some kind of internal representation I guesrs
diakopter incl questons
FROGGS especially his Parrot-good-bye letter that I didn't know before
dalek ast: 3c83e50 | (David Warring [email@hidden.address] | S05-mass/charsets.t:
mark charset punct tests as todo - results varying depending on unicode version
 19:44
19:44 tgt joined
FROGGS jaffa4: into instructions, as bytecode 19:44
jaffa4 SO it is interpreted
TimToady straight from MAST to bytecode
jaffa4 Like Basic?
tadzik FROGGS: is that very new?
FROGGS bytecode just means a pretty packed unreadable mess of binary data here :o) 19:45
tadzik: no, only to me
2011 or so
TimToady jaffa4: it's still just a virtual machine interpreting "bytecode"
FROGGS jaffa4: Basic needs to be parsed, that does not happen with bytecode
jnthn Compiling stuff on MoarVM goes directly from a low-level tree-ish thing to bytecode
The bytecode is the spec'd interface to the VM
TimToady for some definition of "byte" that doesn't include 8-bit opcodes, I suspect
diakopter bytecode is parsed 19:46
tadzik well, what CPU is doing is basically interpreting bytecode, fsvo bytecode
FROGGS like it reads a byte, knows it is the number of an instruction, and it will straight ahead execute it or so
lue
.oO(the new PIR will be MoarVM Objective Abstraction of Runtime Virtual Microcode, clearly)
TimToady is it a bytecode, or is it a wordcode?
diakopter bytecode is a lanfuage too!
jnthn TimToady: Well, yes, but it's stored as a bunch of bytes. :)
lue octetcode!
TimToady jnthn: so is the original Perl 6
jnthn :P
jaffa4 then bycode can be converted into some kind of assembly language
timotimo TimToady: i don't think so. the code that comes out is "if 1.ACCEPTS($_) || 2.ACCEPTS($_)". that should still be optimisable, probably even easier than before. 19:47
jnthn jaffa4: We do have a way to dump the bytecode into something that loosk like assembly.
jaffa4: But that's just a dumping format. Not spec'd in any way, and there's nothing to parse it.
diakopter it's cslled trollcode
lue jnthn: yet
TimToady timotimo: okay, just something to bear in mind that can make a switch statement run much faster if it's jumpable
FROGGS quick! register the MoarVM:: namespace on cpan! 19:48
we will parse its dumpling format!
19:48 raiph left
diakopter if they were still registerable 19:48
lue
.oO(Incoming culinary jokes...)
diakopter jnthn doesn't know it's already parsed
FROGGS so he's the innocent one? 19:49
jnthn lue: Well, we can have something to parse it, but I ain't gonna bless it as "official" in any way :)
TimToady diakopter: shh, that's an NSA secret
diakopter it's in master; he hasn't found it yet
TimToady it's for inserting the NSA's backdoor, so please don't talk aobut it
jnthn Plus it's hardly what you'd want to write by ahnd :)
FROGGS jnthn: in contrast to pir, hahaha 19:50
jnthn I'd rather insert some backdoors for NASA so MoarVM can go into space :P
FROGGS hehe
diakopter it's pretty verbose and repetitititiively redundant
TimToady meant NASA, yeah, that's what I meant...
FROGGS can you tell that again?
jaffa4 person of interest? 19:52
diakopter cbs comedy 19:53
TimToady
.oO(pirson of interest)
diakopter procedural
not very oo
lue
.oO( r: say Q:NSA{███████████████████████████████████} )
TimToady r: say Q:NSA{███████████████████████████████████} 19:54
camelia rakudo-parrot b11094: OUTPUT«===SORRY!=== Error while compiling /tmp/HeDpixyBe4␤Unrecognized adverb: :NSA␤at /tmp/HeDpixyBe4:1␤------> say Q:NSA⏏{████████████████████████████████…»
..rakudo-jvm b11094: OUTPUT«===SORRY!=== Error while compiling /tmp/AWWZL9Drq9␤Unrecognized adverb: :NSA␤at /tmp/AWWZL9Drq9:1␤------> say Q:NSA⏏{█████████████████████████████████…»
TimToady I think that is only recognized when you're compiling yourself
timotimo TimToady: well, when someone comes up with a jump table implementation of given/when, they can ask me to revert that commit if it hinders them. for now, this is faster. 19:55
TimToady funny that whitespace coming out red...
diakopter r: say "y Q:NSA{██████████████████████████████████"
camelia rakudo-jvm b11094: OUTPUT«===SORRY!=== Error while compiling /tmp/XCcotNfc0Y␤Bogus statement␤at /tmp/XCcotNfc0Y:1␤------> say "y Q:NSA{⏏██████████████████████████████████"[0…»
..rakudo-parrot b11094: OUTPUT«===SORRY!=== Error while compiling /tmp/IjVp7h674R␤Bogus statement␤at /tmp/IjVp7h674R:1␤------> say "y Q:NSA{⏏██████████████████████████████████"…»
diakopter r: say "██████████████████████████████████"
camelia rakudo-parrot b11094, rakudo-jvm b11094: OUTPUT«██████████████████████████████████␤»
lue The only other compiler that ████████████ does that ██████ is ████M, if I re██ll.
diakopter r: say "██████████████████████████████████"; nhvh h
camelia rakudo-parrot b11094: OUTPUT«===SORRY!=== Error while compiling /tmp/ICma4THD71␤Undeclared routines:␤ h used at line 1␤ nhvh used at line 1␤␤»
..rakudo-jvm b11094: OUTPUT«===SORRY!=== Error while compiling /tmp/4be5EwdLjs␤Undeclared routines:␤ h used at line 1␤ nhvh used at line 1␤␤»
diakopter r: say "██████████████████████████████████"; . . .
camelia rakudo-parrot b11094: OUTPUT«===SORRY!=== Error while compiling /tmp/Ht5lGp2Qus␤Unsupported use of . to concatenate strings; in Perl 6 please use ~␤at /tmp/Ht5lGp2Qus:1␤------> "█████████████████████████…» 19:56
..rakudo-jvm b11094: OUTPUT«===SORRY!=== Error while compiling /tmp/M32O1OkmYu␤Unsupported use of . to concatenate strings; in Perl 6 please use ~␤at /tmp/M32O1OkmYu:1␤------> "██████████████████████████…»
TimToady ooh, green!
jnthn why did my IRC client just turn into a weird tetris? :P
timotimo %)
lue jnthn: to train our ████ matter in preparation for The ████u██™, ██ course. 19:57
.oO(too bad irclog.perlgeek screws up the camelia output, this doesn't look nearly as cool there :/)
 19:58
FROGGS spectests his fix for #120517
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=120517
19:58 raiph joined
FROGGS lue: it is always better when it is live 19:58
TimToady yes, but as soon as we give instant replay to the refs, it had better be more accurate than their own eyes 19:59
dalek p/optional_for_arguments: ae3226b | (Timo Paulssen)++ | src/ (3 files):
store no optarity, assume arity > 1 means optional
 20:00
timotimo i'm going to clean up that ugly pile of code now :)
TimToady wonders if CRLF should have a pre-assigned NFG codepoint^Wgraphpoint 20:03
20:03 colomon left
jaffa4 if I see black stripes, is that good? 20:05
lue r: say "\r͏\n".ords # :TimToady
camelia rakudo-parrot b11094, rakudo-jvm b11094: OUTPUT«13 847 10␤»
TimToady if you get your eyes checked, you should see alternating black and white 20:06
lue jaffa4: yes, that was intentional on our parts :)
TimToady um, where does that 847 come from? 20:07
lue TimToady: COMBINING GRAPHEME JOINER of course :) 20:08
TimToady o.° 20:09
FROGGS jaffa4: that is how #perl6 works :o)
20:09 Celelibi left
FROGGS jaffa4: welcome in our midst 20:09
lue r: say "█".unicharname; # oh wait, I no can haz
camelia rakudo-parrot b11094: OUTPUT«No such method 'unicharname' for invocant of type 'Str'␤ in block at /tmp/5Ig6Dzklg4:1␤ in any at /tmp/5Ig6Dzklg4:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146␤ in any eval at gen/parrot/stage2/NQPHLL.nqp:1133␤ in any evalfiles at gen/…» 20:10
..rakudo-jvm b11094: OUTPUT«No such method 'unicharname' for invocant of type 'Str'␤ in block at /tmp/zwHl8qo1x7:1␤ in any eval at gen/jvm/stage2/NQPHLL.nqp:1086␤ in any evalfiles at gen/jvm/stage2/NQPHLL.nqp:1292␤ in any command_eval at gen/jvm/stage2/NQPHLL.nqp:1196␤ …»
lue If I vote against ICU (and stand with TimToady), does that mean I could get a .unicharname ?
FROGGS lue: unless there are other charnames the "uni" is redundant
TimToady it might be a function rather than a method, and it might work on integers, not strings 20:11
jnthn detrain, bbiab 20:15
dalek kudo/nom: 9dfecb8 | (Tobias Leich)++ | src/core/Str.pm:
fix #120517, Str:D:.words returns the one elem if it is just one

This way .perl on it looks a bit nicer.
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=120517
20:16 colomon joined
lue TimToady: yeah, I imagine there's a whole class of functions waiting. &charname, &cclass, &ltrdir, etc... 20:17
dalek kudo-star-daily: 41f0a61 | coke++ | log/ (5 files):
today (automated commit)
 20:18
20:19 Celelibi joined
lue would like those functions to work on (integers&(arrays of integers|utf-y bufs/blobs)) 20:21
20:21 stevan_ joined
FROGGS lue: like .chrs? 20:23
lue FROGGS: yeah, like that :)
FROGGS well, it would fit 20:24
timotimo Unable to checkout '240f8d025a72dfd9695290dde45981f238287d79' in submodule path 'modules/uri'
huh.
20:24 pecastro_ joined
dalek ar: e31918b | (Timo Paulssen)++ | modules/uri:
check out an actually existing commit in uri/
 20:25
timotimo [Coke]: that should help with the daily star runs
20:26 pecastro left
dalek ast: 9010356 | (Tobias Leich)++ | S32-str/words.t:
test for RT #120517, Str.words.gist
 20:26
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=120517
FROGGS preflex: tell dram your bug is fixed: rt.perl.org/Ticket/Display.html?id=120517 20:28
preflex Consider it noted.
20:31 rurban1 joined 20:50 kaare_ left
dalek kudo/moar-support: f5a38f2 | nickg++ | tools/build/Makefile- (2 files):
This fixes a build problem when you run Configure with '-backends=moar,jvm'.

We make RUN_PERL6 unique to each build variant.		 20:53
jnthn back 20:56
21:01 stevan_ left
TimToady rather than handling the redo outside the block, it would be a lot simpler conceptually to just turn it into a goto to the first statement 21:02
then it doesn't interact with binding at all 21:03
or am I misunderstanding the difficulty?
timotimo are you talking about my code? 21:04
TimToady I'm talking about the discussion surrounding irclog.perlgeek.de/perl6/2013-11-12#i_7848718
TimToady assumes you're talking about the "redo" op 21:05
slavik will there be a state of the perl6 spec that is like '1.0' type of stable?
or is it already there?
timotimo yes, jnthn said it's not cheap to make a new redo handler for each of the calls
TimToady I'm saying it should not be done that way in the first place 21:06
timotimo oh
well, after the first statement, it will shift items from the iterator
TimToady LABEL: loop { redo LABEL } should turn into loop { LABEL1: goto LABEL1 }
jnthn Beware LEAVE handlers :) 21:07
timotimo well, in that case i'd have to introspect the inside of the loop, which i don't know if i can manipulate freely
TimToady redo should be defined as equivalent to that goto
lue slavik: eventually Perl 6(.0.0) will be 'stable', for some definition :)
slavik yeah, I am waiting for that, moarvm and the concurrency stuffs. :) 21:09
TimToady redo should completely ignore iterators on the outside of the block, as well as phasers that leave/enter the block
it's just supposed to be a shorthand for 'goto'
trying to restart a redo from outside the block is going to be just as painful as trying to goto into a block that has an initializer 21:10
timotimo well, that's how it's been done so far in nqp 21:11
TimToady n: for 1..2 { .say; redo if rand < 0.7; LEAVE { say "oops" } } 21:12
camelia niecza v24-98-g473bd20: OUTPUT«1␤oops␤2␤oops␤2␤oops␤2␤oops␤» 21:13
jnthn Rakudo also, it keeps it in line with the rest of the control exception handlers.
Rather than making it a special case.
TimToady r: for 1..2 { .say; redo if rand < 0.7; LEAVE { say "oops" } }
camelia rakudo-jvm b11094: OUTPUT«1␤oops␤1␤oops␤2␤oops␤2␤oops␤2␤oops␤2␤oops␤2␤oops␤2␤oops␤2␤oops␤2␤oops␤2␤oops␤2␤oops␤2␤oops␤2␤oops␤2␤oops␤2␤oops␤2␤oops␤»
..rakudo-parrot b11094: OUTPUT«1␤oops␤2␤oops␤2␤oops␤»
timotimo well, that seems accurate
TimToady r: for 1..2 { HERE: .say; goto HERE if rand < 0.7; LEAVE { say "oops" } } 21:14
camelia rakudo-parrot b11094: OUTPUT«===SORRY!=== Error while compiling /tmp/0s0xihFsLY␤Undeclared name:␤ HERE used at line 1␤Undeclared routine:␤ goto used at line 1␤␤»
..rakudo-jvm b11094: OUTPUT«===SORRY!=== Error while compiling /tmp/mCxAFpifFu␤Undeclared name:␤ HERE used at line 1␤Undeclared routine:␤ goto used at line 1␤␤»
TimToady n: for 1..2 { HERE: .say; goto HERE if rand < 0.7; LEAVE { say "oops" } }
camelia niecza v24-98-g473bd20: OUTPUT«1␤1␤1␤1␤oops␤2␤oops␤»
timotimo ah, so it shouldn't give a leave for redo. gotcha.
TimToady that's how redo is supposed to work
21:14 lowpro30 joined
timotimo that's just a small bit outside of my comfort zone to fix right now 21:15
21:16 lowpro30__ joined
TimToady jnthn: seems to me the whole point of defining loop control constructs in the language is because each of them is a special case 21:17
21:18 lowpro30_ left 21:20 lowpro30 left
timotimo i think Proc::Status should be mentioned in this month's changelog 21:20
FROGGS++
TimToady but in this case, it should just be handled like a goto to an implicit label, not by escaping the block in question
to the extent that goto is also a control exception at need 21:21
I think we can restrict 'redo' to work only lexotically, so we don't have to install a conjectural label on every loop, which could impact optimization 21:23
otherwise we can't unroll any loops 21:24
21:25 Celelibi left
timotimo that sounds wise indeed. 21:29
dalek ecs: 20876cc | larry++ | S04-control.pod:
notes on redo semantics, restrict to lexotic
 21:30
21:34 bluescreen10 left 21:35 bluescreen10 joined, bluescreen10 left 21:36 bluescreen10 joined
TimToady inlining is just one of many possible optimizations that we want to be careful not to make impossible... 21:36
21:37 raiph left
dalek p: dba785d | (Timo Paulssen)++ | src/vm/parrot/QAST/Operations.nqp:
presize list{,_*} only for 2 or more elements
 21:37
timotimo inlining and unrolling, yeah
TimToady though I suppose one could view loop unrolling as a form of inlining some of a tail recursion on the loop block
timotimo mhm
TimToady dividing a list into work pieces also plays into possibly hypering 21:38
though in that case you're trading for cores rather than memory
timotimo i think we can safely assume that many cores will be available in the future 21:39
at least 4 on normal machines. even phones have quad-cores now.
but rakudo should learn to treat memory more sparingly, i think 21:40
that doesn't mean we shouldn't do inlining. we just have to be more intelligent in other places that we have yet to find >_<
TimToady well, loop unrolling using instruction memory, not data memory
it's the data multipliers that kill you 21:41
timotimo you mean something like "8 extra bytes per QAST node" like i almost introduced into nqp? ;)
TimToady one generally doesn't store 1000000 versions of the same code, but data can easily do that
yes, well, that's data :P
timotimo that sounds entirely logical 21:42
i see you're no longer jetlagged in your brane :D
TimToady well, now it's just the innate branedamamge...
jnthn I dunno how many QAST nodes CORE.setting is, but it'll be easily in the tens of thousands, maybe rather more. 21:43
TimToady hence "multiplier"
jnthn So I tend to be kinda watchful on that stuff :)
FROGGS maybe we should run-length encode our memory :o)
timotimo say, why is arity in QAST::Node rather than QAST::Block? 21:44
FROGGS or even LZW or so
TimToady code like an unrolled loop doesn't tend to multiply unless you inline it, or unroll everything :)
is there any effort to store natives in decreasing size order? P5 takes pains to put 32 bits before 16 bits before 8 bits... 21:45
otherwise you take alignment hits
jnthn timotimo: Good question...
timotimo jnthn: i'll try to move it into Block and see if it breaks
jnthn TimToady: Certainly various structs in Moar have had that thought out carefully 21:46
TimToady good
jnthn timotimo: Please do; it'd be good to know what breaks.
[Coke] timotimo (deprecations) known, but probably not ticketed yet.
it's been failing on jvm since it was added.
timotimo [Coke]: very well
[Coke]: i'm looking forward to the next star daily run, since i fixed the checkout of uri
or at least i think i fixed it. i may have made it a tiny bit less good :P 21:47
21:47 raiph joined
timotimo jnthn: Error while compiling op unless: Method 'arity' not found for invocant of class 'QAST::Op' - so, should i duplicate arity in QAST::Op and have it in QAST::Block, too? 21:47
21:48 denis_boyun joined
jnthn timotimo: Um...first, let's understand what arity's an op... 21:48
timotimo i'm already looking :) 21:49
i think it's being done to calculate $*HAVE_IMM_ARG 21:50
21:50 Rotwang joined 21:51 spider-mario left
timotimo and here it tries to find the arity of a QAST::Var, that's *certainly* not right :) 21:51
maybe it should only look at it if it nqp::istype QAST::Block
jnthn Where, ooc?
I thought the immediate stuff looked at the arity of the block 21:52
timotimo src/vm/parrot/QAST/Operations.nqp 441
jnthn Oh....is it simply that it just blindly checks arity of the children of if/unless/while/blah that can take a block with a param?
timotimo while does the same thing 21:53
jnthn And so far we always give back 0?
timotimo yes, i think so.
jnthn Yeah
Well, nqp::istype check like you check will cut it.
timotimo \o/
jnthn I'm pretty sure it will anyway
That'd be a rather nice saving.
21:53 denis_boyun left
timotimo that's 8 byte saved for a lot of QAST nodes :D 21:53
jnthn :D
21:54 logie left
timotimo how do you feel about knowing the allocation numbers for the different QAST nodes for... the setting for example? 21:54
jnthn Can't hurt. 21:55
21:55 denis_boyun joined
jnthn We may learn something from it :) 21:55
timotimo i'll whip up something
[Coke] looks like modules-test worked. 21:57
github.com/coke/rakudo-star-daily/...odules.log
Testing modules/Math-Model...
Saw 1 call to deprecated code during execution.
jaffa4 what do you use guys to make slides?
dalek ast: 3531185 | (Tobias Leich)++ | S32-str/encode.t:
unfudge now-passing test
timotimo Perl6/Actions also seems to blindly look at a Stmts's arity and crashes, i can fix that, too, though 21:59
21:59 jaffa4 left
timotimo hah 22:00
if nqp::defined($past.arity)
FROGGS ummm 22:01
a nice no-op
timotimo no, a nice crash :)
FROGGS at least, in the $past
lue jaffa4: depends (I'm guessing LaTeX beamer most often). The one constant is that we tend to publish them as pdfs.
FROGGS yeah :o)
japhb___ jnthn: From the point of view of the task doing the publishing, is a publish action blocking? In other words, can the publisher expect all current subscribers to have had a chance to look at the published item before the publish action completes? (Apologies for likely out of date terminology, I haven't had a chance to soak up the name-mangled version of the concurrency spec.) 22:02
timotimo i don't think it should be blocking
jnthn japhb___: That makes two of us :P
japhb___ heh 22:03
jnthn japhb___: A publisher doesn't introduce any asynchrony of its own
timotimo oh
jnthn japhb___: When there was a Publisher class, which I can't remember what is called now, there was gonna be an AsyncPublisher too.
timotimo so it's blocking in the normal case
Supply, no?
jnthn timotimo: Right
japhb___ Yeah, I definitely want both use cases. 22:04
jnthn japhb___: Yeah, we should have it. It's just part of my general "don't introduce async where it ain't asked for" design sense.
timotimo .o(the duplication of QAST nodes through the autosink thingie is probably a big factor in memory usage actually)
japhb___ Sure, I'm fine with that default. 22:05
jnthn timotimo: Note that it's not duplicating the two trees in the want... :)
timotimo: As in, just referencing the same nodes, even if they dump twice.
timotimo ah, that's good
japhb___ And in fact, for the use case that spawned the question, I do want blocking behavior. I just also have a need for async in other cases.
jnthn japhb___: Yup, both are needed in different places.
japhb___: Rx also provides both. 22:06
japhb___: The other thing we need is replay semantics.
timotimo nqp: say(105 % 10)
camelia nqp-moarvm, nqp-jvm, nqp-parrot: OUTPUT«5␤»
jnthn japhb___: uh, control over...
japhb___ jnthn: By which you mean being able to subscribe "late" and get the history, right?
jnthn japhb___: Right.
japhb___ Yep, very needed.
TimToady
.oO(cockpit recorders)
timotimo like zero-mq :P
jnthn japhb___: Probably you get replay of everything as a * where you can also put a buffer size. 22:07
timotimo: Yes, this is very much in messaging pattern space.
TimToady the application probably needs a way to limit replay to avoid sending all its memory down that primrose path 22:08
japhb___ In fact, we need that semantic general enough that I'd almost say we need replay to be a method that people override to e.g. do "Remove events that logically replace old events, so as to shrink the replay to the minimum necessary to create a given state"
jnthn TimToady: Yes, "replay everything from this supply" is often overkill.
TimToady key frames, basically
japhb___ client/server game engines use state sync whenever a new client connects, and then minimal diffs when a client loses sync. 22:09
jnthn That's a nice analaygy
uh, spelt right...
:)
japhb___
.oO( method sync(:$from) { ... } )
 22:11
sync there, because it's different from replay in that the latter feels like it imposes the concept of replaying events in the same order they were initially issues, rather than getting fastest-possible resync to a published stream in progress 22:12
*issued
jnthn Yeah, that's more window-y 22:15
timotimo jnthn: i'll have your numbers in about 4 minutes
TimToady the question is, do you want to see the whole movie again, or just get to the particular point in the movie?
FROGGS well, both 22:17
timotimo hey everybody. want to bet on how many of each QAST node type are allocated in the rakudo setting? (plus optimizer work, which adds a few nodes here and there as well)
FROGGS 4_231_456_200
timotimo what type of node would that be? 22:18
jnthn FROGGS: ffs, for a moment I thought that was timotimo posting the total :P
FROGGS ahh, that was a guess for all in total
22:18 rurban1 left
timotimo i'll calculate that number, too 22:18
FROGGS I'd say 35% blocks
the rest is hard to guess 22:19
timotimo java.lang.StackOverflowError
FROGGS I parse that as NaN
timotimo :) 22:20
jnthn I don't think 35% will be blocks
I think QAST::Op will be the most numerous
timotimo sounds likely
jnthn Perhaps followed by QAST::Want
timotimo followed by Var?
FROGGS well, 55% QAST::Op then :o)
timotimo ah, Want, yeah
FROGGS < 5% QAST::Regex
jnthn And lots of QAST::Var too 22:21
QAST::Stmts and QAST::Stmt will be above QAST::Var
FROGGS jnthn: you have to provide exact numbers :P 22:22
timotimo finally made the counter work %) 22:23
~589350 nodes in total 22:24
FROGGS O.o 22:25
timotimo Var is in the lead with 123350, followed by Op with 121550
WVal, SVal and IVal: 98900, 92850, 74950
Stmts, Stmt, Want: 29350 18100 16650
Block, BVal, InlinePlaceholder: 5950 5050 1350 22:26
finally: 1050 QRegex, 150 CompUnit, 50 Node, 50 NVal
22:26 denis_boyun left
jnthn 150 CompUnit is a useful number. That tells us we run 149 BEGIN time things :) 22:26
timotimo r: say "{ (589350 - 5950) * 8 } bytes saved by timotimo++" 22:27
camelia rakudo-jvm 9dfecb: OUTPUT«(timeout)»
..rakudo-parrot 9dfecb: OUTPUT«4667200 bytes saved by timotimo++␤»
TimToady if you guys were better functional programmers there'd be fewer Vars :P
jnthn 4 MB. I'll take that.
FROGGS 4.6MB, nice timotimo++ :o)
timotimo r: say "that's {4667200 / 1024} kbytes"
camelia rakudo-parrot 9dfecb, rakudo-jvm 9dfecb: OUTPUT«that's 4557.8125 kbytes␤»
timotimo jnthn: and compile_in_context from the optimizer, i bet
FROGGS aww, way too much for a floppy :( 22:28
jnthn timotimo: Yeah, that counts as BEGIN time
timotimo 4.6 MB is far from a big amount of ram usage compared to the 2gb or what it takes most of the time :(
TimToady well, let's go for a live CD
jnthn Well, kinda... :)
FROGGS hehe, a live CD of rakudo's saved bytes :o) 22:29
timotimo r: say "timotimos computer generates about {589350 / 110} QAST Nodes per second during parse time."
camelia rakudo-parrot 9dfecb, rakudo-jvm 9dfecb: OUTPUT«timotimos computer generates about 5357.727273 QAST Nodes per second during parse time.␤»
timotimo i like { } interpolation a lot 22:30
TimToady it used to be controversial :) 22:31
FROGGS especially when you think of it as an embedded foreign language
(about liking it)
timotimo it's powerful and if you don't like it, just use qq:!c
(is that the right spelling?)
r: say qq:!c(heyo {1 + 1}!) 22:32
camelia rakudo-jvm 9dfecb: OUTPUT«===SORRY!=== Error while compiling /tmp/kzu0ognnWE␤Argument not allowed on negated pair␤at /tmp/kzu0ognnWE:1␤------> say qq:!c⏏(heyo {1 + 1}!)␤»
..rakudo-parrot 9dfecb: OUTPUT«===SORRY!=== Error while compiling /tmp/PqCPuzeMeM␤Argument not allowed on negated pair␤at /tmp/PqCPuzeMeM:1␤------> say qq:!c⏏(heyo {1 + 1}!)␤»
timotimo r: say qq:!c/heyo {1 + 1}!/
camelia rakudo-parrot 9dfecb, rakudo-jvm 9dfecb: OUTPUT«heyo {1 + 1}!␤»
FROGGS r: say :!c
camelia rakudo-parrot 9dfecb, rakudo-jvm 9dfecb: OUTPUT«␤»
FROGGS r: say (:!c).perl
camelia rakudo-parrot 9dfecb, rakudo-jvm 9dfecb: OUTPUT«"c" => Bool::False␤»
FROGGS I didn't know that 22:33
timotimo you didn't?!
FROGGS no 22:34
timotimo but it's so nice! :)
FROGGS looks a bit sad though
timotimo i'm pining for a commitbit :P
FROGGS don't ask me :o)
timotimo i wonder if TimToady would be able to verify that my CLA has arrived? 22:35
and give a commitbit
TimToady has nothing to do with that end of the business
timotimo OK
FROGGS he is just our brane :o)
TimToady isn't even sure he has a commitbit himownself
though he did turn in a CLA for Parrot lo these many years ago
FROGGS dunno if somebody else than pmichaud++ is able to 22:36
TimToady: really?
that is kinda weird
jnthn timotimo: I *think* I can do the commit bit part, but not the CLA check part :(
timotimo: Where'd you send it to? 22:37
FROGGS walnut I suppose
TimToady FROGGS: the early parrotfolk chided me for all the macros I put into Perl 5, and then proceeded to put a nearly identical set of macros into parrot :) 22:38
jnthn doesn't know whos' on the receiving end of that...
timotimo FROGGS: t.h8.lv/0001-.arity-will-only-be-on...soon.patch would you be so kind real quick?
22:38 djanatyn left
timotimo Karen Pauley got it 22:38
22:38 hummeleB1 joined
jnthn timotimo: Oh...she has it? 22:38
timotimo yes, she passed it on for me, she said
22:39 djanatyn joined
dalek kudo/nom: 1371ccc | (Timo Paulssen)++ | src/Perl6/Actions.nqp:
.arity will only be on QAST::Block soon.
 22:39
22:39 colomon left
FROGGS let me be your commitbot :o) 22:40
timotimo :3
many thankulations
.o( gewürz thankulatius )
FROGGS no problemotics
hehe
22:40 colomon joined
FROGGS I don't like them :o( 22:40
dalek p: eada3ad | (Timo Paulssen)++ | src/ (4 files):
move $!arity from Node to Block

Saves about 4.6 MB of RAM when compiling the rakudo core setting		 22:41
FROGGS TimToady: I felt a bit home in Parrot land, it is so XS-ish :o) 22:42
jnthn Somebody at TPF acknowledging timotimo++'s CLA is enough for me, so I've granted the commit bit. :)
FROGGS \o/
TimToady welcome to the exclusive club...oh wait... 22:45
FROGGS hehe
*g*
22:46 iSlug joined
Ulti was there a R* .msi that needed testing? 22:46
22:46 woolfy left
FROGGS Ulti: sort of 22:47
Ulti I guess I should try building my own at some point
jnthn TimToady: Moar has its share of macros too. Not all of which I'm glad of... :)
FROGGS Ulti: irclog.perlgeek.de/perl6/2013-11-02#i_7804199
Ulti thanks FROGGS I'll give it a go 22:48
FROGGS Ulti: panda has some problem installing some modules
Ulti I'm on 64bit Windows 8 so might have a go sorting out building stuff 22:49
FROGGS but that is mostly due to outdated included modules
Ulti all my machines that run Rakudo are either OSX or Ubuntu... havent touched setting up a dev environment on Windows for more than a decade
FROGGS: ahh ok
22:49 woolfy joined 22:50 lowpro30__ left
Ulti one nasty thing on windows 8 you get a great big warning banner when installing 22:50
FROGGS yeah, we need to sign it somehow :/
Ulti well it said finished...
FROGGS I think `panda install Math::RungeKutta` failed to delete a tempdir and aborted installation due to that 22:51
TimToady you can't really write portable C without a lot of macros
timotimo hm, there's no spectests for List.combinations 22:52
22:53 berekuk left
jnthn TimToady: Oh, I agree. Turning smallish static functions into macros rather than leaving the inliner to get them is what I'm less fond of. 22:53
22:53 berekuk joined
FROGGS and we're fudging the rakudo source code, that is the same 22:53
jnthn: *nod* 22:54
TimToady and to be fair, there was no such inliner when Perl was first written
FROGGS AND THEN YOU EVEN HAVE TO SHOUT THEM
timotimo If the range option is taken, returns all subset combinations in order of increasing size, up to the specified max, or the size of the list, whichever is smaller. - i don't understand. what is the range for, now?
jnthn TimToady: Yes, but I'm talking about Moar, being written in 2013 :)
TimToady: I've read to little of the Perl 5 source to have anything agaisnt it ;) 22:55
timotimo and what if i start the range not at 0 but at, say, 5?
jnthn *too
TimToady then it throws away any combinations that are 0..4 at a time
Ulti FROGGS: I get a problem just running panda
FROGGS Ulti: what exactly? 22:56
TimToady r: say <a b c>.combinations.perl
camelia rakudo-jvm 9dfecb: OUTPUT«No such method 'combinations' for invocant of type 'Parcel'␤ in block at /tmp/aUrIbjwi3d:1␤ in any eval at gen/jvm/stage2/NQPHLL.nqp:1086␤ in any evalfiles at gen/jvm/stage2/NQPHLL.nqp:1292␤ in any command_eval at gen/jvm/stage2/NQPHLL.nqp:1196…»
..rakudo-parrot 9dfecb: OUTPUT«No such method 'combinations' for invocant of type 'Parcel'␤ in block at /tmp/BF2MRLDgs5:1␤ in any at /tmp/BF2MRLDgs5:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146␤ in any eval at gen/parrot/stage2/NQPHLL.nqp:1133␤ in any evalfiles at …»
timotimo either my non-native englishness is kicking into gear here or that sentence could be reworded a bit
TimToady r: say [<a b c>].combinations.perl
camelia rakudo-jvm 9dfecb: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in block at /tmp/wgFc3WP3vS:1␤ in any eval at gen/jvm/stage2/NQPHLL.nqp:1086␤ in any evalfiles at gen/jvm/stage2/NQPHLL.nqp:1292␤ in any command_eval at gen/jvm/stage2/NQPHLL.nqp:11…»
..rakudo-parrot 9dfecb: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in method combinations at gen/parrot/CORE.setting:7979␤ in block at /tmp/_ATFXc2dnM:1␤ in any at /tmp/_ATFXc2dnM:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146␤ in any eval …»
Ulti FROGGS: Could not open C:\rakudo\bin\C:\rakudo\bin\panda.bat. Unable to open filehandle from path 'C:\rakudo\bin\C:\rakudo\bin\panda.bat' 22:57
TimToady r: say [<a b c>].combinations($_) for 0..3
camelia rakudo-parrot 9dfecb, rakudo-jvm 9dfecb: OUTPUT«␤a b c␤a b a c b c␤a b c␤»
Ulti perl6 runs fine though
FROGGS hmmm
is "C:\rakudo\bin\" set in environment somewhere? 22:58
(except path)
or.. how did you invoke it? just by "panda"? 22:59
Ulti FROGGS: yeah
I invoked it from the directory as 'panda'
FROGGS hmmm, that worked on my box 23:00
23:00 BenGoldberg joined
FROGGS why should it even want to open the bat file? 23:00
Ulti im doing it from a powershell terminal though.. one sec I'll try cmd
FROGGS it should execute it to run `perl6-p panda`
yeah, maybe... I have no powershell installed 23:01
23:01 PacoAir left
Ulti yeah PowerShell doesn't like it but under cmd.exe everything works fine including installing Math::RungeKutta ^___^ 23:01
FROGGS nice 23:02
23:02 bluescreen10 left
Ulti I'll try and work out whats up with pash 23:02
FROGGS I'll install powershell these days then :o)
cool!
Ulti++
Ulti Windows 8.1 has it from the 'start menu' as a link
FROGGS only have XP and 7
Ulti nice to know you are forwards compatible :)
FROGGS yes :o) 23:03
Ulti compile now run an OS version later
FROGGS Windows++ here
that is something where Windows Just Works, and linuxes are pain
dalek ecs: 570c49c | larry++ | S32-setting-library/Containers.pod:
clarify wording of .combinations(Range)
 23:06
FROGGS s/Dec/Nov/ 23:07
TimToady oops
dalek ecs: 7e0441c | larry++ | S32-setting-library/Containers.pod:
s/Dec/Nov/ FROGGS++ timotimo++
 23:08
TimToady larry-- 23:09
FROGGS :o)
LEAVE; /* gnight all */
TimToady o/ 23:10
23:10 FROGGS left, rurban1 joined
timotimo r: [1, 2, 3].combinations(-1) # what should this do? 23:21
camelia rakudo-parrot 9dfecb: OUTPUT«(signal KILL)»
..rakudo-jvm 9dfecb: OUTPUT«Unhandled exception: java.lang.StackOverflowError␤ in count (gen/jvm/CORE.setting:7107)␤ in count (gen/jvm/CORE.setting)␤ in (gen/jvm/CORE.setting:7525)␤ in reify (gen/jvm/CORE.setting)␤ in (gen/jvm/CORE.setting:7949)␤ in reify (gen/jvm/…»
timotimo r: [1, 2, 3].combinations(0) 23:22
camelia ( no output )
jnthn std: LEAVE; 23:23
camelia std b469b42: OUTPUT«===SORRY!===␤Whitespace required after keyword at /tmp/4M843Awg8G line 1:␤------> LEAVE⏏;␤Parse failed␤FAILED 00:01 120m␤»
jnthn r: LEAVE;
camelia rakudo-parrot 9dfecb: OUTPUT«===SORRY!=== Error while compiling /tmp/6pcNl5uiAQ␤Undeclared name:␤ LEAVE used at line 1␤␤»
..rakudo-jvm 9dfecb: OUTPUT«===SORRY!=== Error while compiling /tmp/1yabGdR6Qh␤Undeclared name:␤ LEAVE used at line 1␤␤»
jnthn Discrepancy!
23:28 logie joined
timotimo S32-list/combinations.t sounds like a good place to put combination tests? 23:30
n: say [1, 2, 3].combinations(2)
camelia niecza v24-98-g473bd20: OUTPUT«Unhandled exception: Unable to resolve method combinations in type Array␤ at /tmp/4TSkLQkbfF line 1 (mainline @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4583 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4584 (module-CORE …»
jnthn timotimo: wfm
23:30 Rotwang left
timotimo for tests, am i allowed to rely on the ordering of combinations? 23:30
it's not specced, right? 23:31
jnthn If it's not spec'd the no...
r: say [1, 2, 3].combinations(2) 23:32
camelia rakudo-jvm 1371cc: OUTPUT«(timeout)»
..rakudo-parrot 1371cc: OUTPUT«1 2 1 3 2 3␤»
timotimo aaw, jvm, come on ;_;
jnthn eval server gone wild I guess
timotimo r: say all([1, 2, 3]) == one([2, 1, 3]); 23:33
camelia rakudo-parrot 1371cc, rakudo-jvm 1371cc: OUTPUT«all(one(True))␤»
23:33 logie left
timotimo r: say [1, 2, 3].combinations(3); 23:34
camelia rakudo-parrot 1371cc, rakudo-jvm 1371cc: OUTPUT«1 2 3␤»
timotimo er, wha?
r: say [1, 2, 3].combinations(2);
camelia rakudo-parrot 1371cc, rakudo-jvm 1371cc: OUTPUT«1 2 1 3 2 3␤»
timotimo ah, they are supposed to be sets 23:35
i was thinking permutations
r: say [1, 2, 3].permutations(3);
camelia rakudo-jvm 1371cc: OUTPUT«Too many positional parameters passed; got 2 but expected 1␤ in method permutations at gen/jvm/CORE.setting:7970␤ in block at /tmp/9f3KlR8RB7:1␤ in any eval at gen/jvm/stage2/NQPHLL.nqp:1086␤ in any evalfiles at gen/jvm/stage2/NQPHLL.nqp:1292…»
..rakudo-parrot 1371cc: OUTPUT«Too many positional parameters passed; got 2 but expected 1␤ in method permutations at gen/parrot/CORE.setting:7994␤ in block at /tmp/RuDIFxQ4sW:1␤ in any at /tmp/RuDIFxQ4sW:1␤ in any at gen/parrot/stage2/NQPHLL.nqp:1146␤ in any eval at…»
timotimo r: say [1, 2, 3].permutations();
camelia rakudo-parrot 1371cc, rakudo-jvm 1371cc: OUTPUT«1 2 3 1 3 2 2 1 3 2 3 1 3 1 2 3 2 1␤»
23:35 rurban1 left 23:42 iSlug left, REPLeffect left 23:43 amkrankruleuen left
dalek ast: 25ccede | (Timo Paulssen)++ | S32-list/combinations.t:
test List.combinations a bit.
 23:57
kudo/nom: 5e56c98 | (L. Grondin)++ | src/core/List.pm:
implement List.combinations(Range $)
kudo/nom: 7942362 | (L. Grondin)++ | src/core/List.pm:
fixing indent
kudo/nom: 4e29ebf | (Timo Paulssen)++ | src/core/List.pm:
don't stack-overflow on negative arguments
kudo/nom: e71f994 | (Timo Paulssen)++ | t/spectest.data:
run S32-list/combinations.t
 23:58
timotimo when i've already merged in commits from a pull request and push "merge pull request" on github, what happens?
i'll just try it, worst case i'll force-push immediately 23:59
jnthn timotimo: I guess it'll tell you they're already there...

Powered by Cro and the Raku® Programming Language.

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

Thank you!