»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, or /msg camelia p6: ... | irclog: irc.perl6.org or colabti.org/irclogger/irclogger_logs/perl6 | UTF-8 is our friend!
Set by moritz on 22 December 2015.
ugexe maybe method native-descriptor needs to be added to IO::Pipe itself 00:01
skids .native-descriptor is a recent add, and may only have been done for sockets.
commit a819231
00:01 autarch left, BenGoldberg joined
ugexe you could try passing in your own IO::Handle to :in 00:02
rudi_s ugexe: ?
ugexe my $handle = IO::Handle.new(whatever); run @cmd, :in($handle); 00:03
as in, maybe if you create the handle yourself instead of letting Proc do it in .new that the native-descriptor might end up working
just a wild guess though 00:04
00:05 autarch joined
awwaiid What is a control exception? 00:07
skids Things like the "next" in loop { next } use the exeception mechanism to do their work. At least, virtually... the optimizer could remove that. 00:09
Basically anything that does not fall off the bottom of a brace.
Hrmph. nqp::filenofh(nqp::getattr(nqp::decont($p.in), IO::Handle, "\$!PIO")) is also -1. 00:10
rudi_s skids: Hm. Is this a bug in moar/jvm? 00:11
skids I dunno. The nqp::getattr does return an MVMOSHandle. But actually seeing what's in it is difficult. 00:12
00:15 johndau joined
skids Note that "pipes" on windows are strange ISTR. 00:16
00:16 zpmorgan joined
rudi_s Hm. The moarvm implements it for syncfile and syncstream. Not sure if this includes pipes. 00:18
(On the other hand IO::Pipe is a IO::Handle with no special cases.) 00:19
Ah, the pipe-fh comes from syncpipe.c. 00:21
jnthn: ^ - would it be possible to implement the MVMIOIntrospection also for syncpipe.c and if yes would it be much work or can one just add the &introspection to the op_table and it just works? 00:23
skids Likely he
s alseep you may want to .tell 00:24
rudi_s .tell jnthn Would it be possible to implement the MVMIOIntrospection also for syncpipe.c and if yes would it be much work or can one just add the &introspection to the op_table and it just works? - The idea is to support .native-descriptor also for pipes, e.g. when using run().
yoleaux rudi_s: I'll pass your message to jnthn.
rudi_s skids: Thanks.
BenGoldberg If, on windows, you want something *like* a pipe, but which select() will work on, use socketpair, then use shutdown to make one end read-only and the other write-only. If you don't need select() for readability, then please ignore me ;) 00:27
00:29 tmch joined
rudi_s (I don't care about windows, but thanks.) 00:30
00:41 rindolf left 00:49 tmch left 00:54 yeahnoob joined 00:55 labster left 01:02 acrussell joined
dalek kudo-star-daily: 38f4aae | coke++ | log/ (9 files):
today (automated commit)
 01:06
01:24 perlawhirl joined 01:33 grondilu left 01:34 grondilu joined
nadim what's a cleanway to flip flop between two string,preferably held within the same variable/oject 01:34
skids you mean on each read a different string is returned? 01:36
nadim yes
or a circular buffer 01:37
skids remembers an IRC conversation about a true/false flipflop.
nadim I've seen something too but can't remember what 01:38
ugexe ff ?
skids Well, not a "flip flop" exactly 01:39
There was a golfing contest for a variable that returned 0 and 1 on alternate reads.
grondilu m: role flicking { method Str { state $++ %% 2 ?? "flip" !! "flop" }; my $text is flicking; say $text; say $text; 01:40
camelia rakudo-moar 69426b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/wcjJmjqEHO␤is trait on $-sigil variable not yet implemented. Sorry. ␤at /tmp/wcjJmjqEHO:1␤------> 3"flip" !! "flop" }; my $text is flicking7⏏5; say $text; say $text;␤ expecting any of:␤ …»
grondilu m: role flicking { method Str { state $++ %% 2 ?? "flip" !! "flop" }; my $text; $text does flicking; say $text; say $text;
camelia rakudo-moar 69426b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/00ub7NNIgr␤Missing block␤at /tmp/00ub7NNIgr:1␤------> 3ext does flicking; say $text; say $text;7⏏5<EOL>␤»
skids Heheh. Checting though with an implicit .Str 01:41
grondilu m: role flicking { method Str { state $++ %% 2 ?? "flip" !! "flop" }; my $text = Str but flicking; say $text; say $text;
skids *Cheating
camelia rakudo-moar 69426b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/3IEgslkBSg␤Missing block␤at /tmp/3IEgslkBSg:1␤------> 3 Str but flicking; say $text; say $text;7⏏5<EOL>␤»
01:41 gfldex left
grondilu m: role flicking { method Str { state $++ %% 2 ?? "flip" !! "flop" } }; my $text = Str but flicking; say $text; say $text; 01:41
camelia rakudo-moar 69426b: OUTPUT«(Str+{flicking})␤(Str+{flicking})␤»
grondilu m: role flicking { method Str { state $++ %% 2 ?? "flip" !! "flop" } }; my $text = "" but flicking; say $text; say $text; 01:42
camelia rakudo-moar 69426b: OUTPUT«␤␤»
grondilu m: role flicking { method Str { state $++ %% 2 ?? "flip" !! "flop" } }; my $text = "" but flicking; say ~$text; say ~$text;
camelia rakudo-moar 69426b: OUTPUT«␤␤»
grondilu gives up
nadim hehe!
grondilu m: print role { method Str { "duh" } }.new 01:43
camelia rakudo-moar 69426b: OUTPUT«duh»
grondilu m: print role { method Str { state "duh" } }.new
camelia rakudo-moar 69426b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/mu0H6NzrDA␤Malformed state␤at /tmp/mu0H6NzrDA:1␤------> 3print role { method Str { state7⏏5 "duh" } }.new␤»
grondilu m: my $str = role { method Str { state $++ %% 2 ?? "one" !! "two" } }.new; print $str for ^5; 01:44
camelia rakudo-moar 69426b: OUTPUT«onetwoonetwoone»
skids I would suggest a Proxy, but unfortunately, FETCH gets called extra times in the current implementation. 01:47
01:53 kid51 left, labster joined 02:11 Sqirrel left 02:12 kalkin-_ joined 02:16 kalkin- left 02:17 sufrostico left 02:26 johndau left, johndau` joined 02:28 johndau` left 02:29 johndau joined, vendethiel joined 02:31 Actualeyes left 02:44 Actualeyes joined 02:47 jolts left 02:54 vendethiel left 03:16 Sqirrel joined 03:23 Upasaka_ joined, Timbus_ joined, Upasaka_ left 03:24 peter joined, peter is now known as Guest51674 03:26 Upasaka left 03:27 Upasaka joined 03:29 cdg left 03:30 Timbus left, alpha123 left 03:38 Actualeyes left 03:39 Actualeyes joined 03:42 noganex joined 03:45 noganex_ left 03:50 molaf joined 03:58 BenGoldberg left 04:11 cpage_ left 04:23 araujo_ left 04:25 wamba joined 04:41 molaf left 04:47 rickbike joined, cpage_ joined, davido joined 04:57 rickbike left 05:05 John[Lisbeth] left 05:09 jdliver joined 05:10 yqt left 05:19 notbenh left 05:28 wamba left 05:36 perlawhirl left 05:46 johndau` joined 05:49 johndau left, johndau` left 05:50 |Sno| joined, johndau joined 05:51 davido left 05:52 [Sno] left 05:53 Timbus_ left, Timbus joined, Cabanossi left 05:54 khw left 05:55 Cabanossi joined 05:56 zpmorgan left, Spot__ left, pnu left, dalek left, solarbunny left, baest left 05:59 labster left, labster joined 06:01 skids left 06:02 zpmorgan joined, Spot__ joined, pnu joined, dalek joined, solarbunny joined, baest joined, WAAEBUL joined, leguin.freenode.net sets mode: +v dalek 06:15 jdliver left
sortiz m: use NativeCall; sub free(Pointer) is native() { * }; free(Pointer); 06:18
camelia ( no output )
[ptc] timotimo: wrt: the Num/Rat question. I know that 1234/56 is a Rat, however both 1234.56 and 1234/56 are numbers of some sort. Maybe I meant really Numeric... 06:29
timotimo: which then brings me to the real question: is it possible to say in a regex "match a Numeric-looking thing"? 06:30
06:30 cdg joined 06:35 cdg left 06:40 musiKk_ joined
[ptc] timotimo: oops, it seems you answered the question earlier this morning. Sorry! 06:50
07:04 domidumont joined 07:09 domidumont left, pdcawley joined, domidumont joined
nine .tell mst The 13 seconds for loading Inline::Perl6 are mostly spent on compiling Inline::Perl5 which cannot be precompiled currently. So actually my work in that area will directly benefit Inline::Perl5 and ::Perl6 :) 07:09
yoleaux nine: I'll pass your message to mst.
07:10 zpmorgan left, Spot__ left, pnu left, dalek left, solarbunny left, baest left
nine .tell ilmari the REPL locking bug is already fixed on the relocatable-precomp branch 07:18
yoleaux nine: I'll pass your message to ilmari.
07:18 CIAvash joined, darutoko joined 07:19 pdcawley left, pdcawley joined 07:21 zpmorgan joined, Spot__ joined, pnu joined, dalek joined, solarbunny joined, baest joined, WAAEBUL joined, leguin.freenode.net sets mode: +v dalek
a3r0 ~. 07:23
07:29 nakiro joined 07:47 zpmorgan left, Spot__ left, pnu left, dalek left, solarbunny left, baest left 07:49 buharin joined 07:54 zpmorgan joined, Spot__ joined, pnu joined, dalek joined, solarbunny joined, baest joined, WAAEBUL joined, leguin.freenode.net sets mode: +v dalek, yeahnoob left
sortiz .tell lizmat Something is broken in commit 69426bd, with *some* calls, somehow $lib2mangler remains undefined, that results in "This type does not support associative operations … in method setup at … line 276 … in method CALL-ME … line 294"" 08:05
yoleaux sortiz: I'll pass your message to lizmat.
08:23 jsimonet1 left 08:25 jsimonet joined, pdcawley left
dalek rl6-examples/fix-dbi-sql-test: 2f07ded | paultcochrane++ | .travis.yml:
Try installing sqlite

It could be that the dbi-sql script needs sqlite to be installed before it can run correctly.		 08:27
sortiz .tell lizmat Trying the binding of nqp::hash at the $lib2mangler declaration, the *same* error is generated early, at rakudo's 'make test', I don't understand why :-( 08:28
yoleaux sortiz: I'll pass your message to lizmat.
08:30 kjs_ joined, geekosaur left 08:31 geekosaur joined
[Tux] test 22.132 08:31
test-t 12.746
csv-parser 52.734
08:32 gfldex joined
nadim Morning all 08:33
my (@sub_elements, %glyphs) = some_sub(...) how do I get @sub_elements to not grab everything? 08:34
08:34 fireartist joined 08:35 cpage left
nadim := hmmm 08:35
It's going to take 100 years to learn 08:36
08:36 cpage joined 08:37 firstdayonthejob left, dakkar joined 08:39 wamba joined 08:41 zakharyas joined 08:43 kjs_ left 08:50 cpage left 08:51 araujo joined, araujo left, araujo joined 08:52 araujo left, cpage joined, araujo joined
dalek rl6-examples/fix-dbi-sql-test: f56a7ae | paultcochrane++ | .travis.yml:
Back out the sqlite installation

  ... it didn't get travis to run the dbi-sql script, so the change can go.		 08:52
rl6-examples/fix-dbi-sql-test: d919e8f | paultcochrane++ | categories/cookbook/14database-access/14-09-dbi-sql.pl:
Use the connect calling style from DBIish docs
08:55 Azry left 08:58 Azry_ joined 09:06 RabidGravy joined 09:08 buharin left
tadzik itz_stmuk++ # nice talk, running the examples now :) 09:09
09:09 tmch joined
masak yes, itz++ 09:10
moritz stmuk++
masak _++
09:11 buharin joined 09:13 brrt joined
[ptc] stmuk++ 09:14
09:14 integral left
Woodi so, any place to gat that (and other) slides ? # nine++ :) 09:16
moritz Woodi: act.yapc.eu/gpw2016/schedule has links to talks (and other days), and some of the talks have links to slides
I've asked the speakers to upload their slides, though stmuk might not have gotten around to it yet 09:17
dalek rl6-examples/fix-dbi-sql-test: 3450c31 | paultcochrane++ | .travis.yml:
Try installing sqlite3

  ... instead of just sqlite		 09:19
09:21 integral joined
tadzik 's lightnink talk slides' interesting part was the one that said "(Live demo here)" :P 09:21
masak tadzik: guess those of us who missed it'll have to wait for the video ;) 09:26
masak 's talks are now online as slide PDFs at act.yapc.eu/gpw2016/talk/6517 and act.yapc.eu/gpw2016/talk/6496
tadzik masak: I can show you on the coffee break :) 09:27
but I'm also planning to write a blog toast on it
moritz masak++
masak tadzik: yes, please! private demo 09:28
I've heard excited whispers about "sixify", so my expectations are high :P
tadzik :D 09:29
so it's settled, there'll be a lightning talk showing
09:41 zhmylove left, zhmylove joined 09:43 [ptc] left 09:44 matt_ joined, matt_ is now known as Guest88249 09:49 rindolf joined
lizmat . 09:50
yoleaux 08:05Z <sortiz> lizmat: Something is broken in commit 69426bd, with *some* calls, somehow $lib2mangler remains undefined, that results in "This type does not support associative operations … in method setup at … line 276 … in method CALL-ME … line 294""
08:28Z <sortiz> lizmat: Trying the binding of nqp::hash at the $lib2mangler declaration, the *same* error is generated early, at rakudo's 'make test', I don't understand why :-(
09:50 nakiro left
lizmat .tell sortiz make test has always been clean for me: do you have an example that consistently fails? 09:51
yoleaux lizmat: I'll pass your message to sortiz.
09:53 Sqirrel left 09:54 Sqirrel joined 10:00 johndau left 10:02 nakiro joined 10:06 pdcawley joined 10:07 FROGGS joined 10:08 pmurias joined
pmurias masak: what's sixify? 10:08
tadzik a lie :) 10:10
I used it on my lightning talk, "I'm now going to port this thing to perl 6: 'sixify'" 10:11
but it was just an alias to 'git stash pop &> /dev/null' :P
arnsholt =D =D =D 10:12
jnthn masak: I'd love to know the story between the Brazil/Argentina correction slide :D
yoleaux 00:24Z <rudi_s> jnthn: Would it be possible to implement the MVMIOIntrospection also for syncpipe.c and if yes would it be much work or can one just add the &introspection to the op_table and it just works? - The idea is to support .native-descriptor also for pipes, e.g. when using run().
10:19 [ptc] joined
jnthn .tell rudi_s It shouldn't be much harder than just filling in that entry in the op_table 10:20
yoleaux jnthn: I'll pass your message to rudi_s.
brrt yeah, i was curious about that too 10:32
lizmat m: use nqp; my @s = "a".comb(1); nqp::join("::",nqp::getattr(@s,List,q/$!reified/)) # what am I missing?? 10:33
camelia rakudo-moar 69426b: OUTPUT«This type cannot unbox to a native string␤ in block <unit> at /tmp/wg0OltfJmO line 1␤␤»
RabidGravy it turns out that it helps to have the microphone volume turned up when testing recording 10:34
jnthn lizmat: Scalars
lizmat so I need to decont ?
jnthn Yeah...see the code in (or just use? :)) the Perl 6 join 10:35
Better just use Perl 6's join and we can put our efforts into optimizing that...
lizmat okidoki :-) 10:36
masak jnthn: guess you had to be there :) 10:38
jnthn heh, then I'll go with assuming you said Argentina and put up a map of Brazil :D
lizmat m: v* # DIHWIDT ?? 10:40
camelia rakudo-moar 69426b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/vWBhqZVKCW␤Missing required term after infix␤at /tmp/vWBhqZVKCW:1␤------> 3v* # DIHWIDT ??7⏏5<EOL>␤ expecting any of:␤ prefix␤ term␤»
lizmat m: v+ # DIHWIDT ?? 10:41
camelia rakudo-moar 69426b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/DOIboYhDvs␤Missing required term after infix␤at /tmp/DOIboYhDvs:1␤------> 3v+ # DIHWIDT ??7⏏5<EOL>␤ expecting any of:␤ prefix␤ term␤»
lizmat m: dd Version.new(*)
camelia rakudo-moar 69426b: OUTPUT«Version.new('*')␤»
lizmat I guess that's why .perl of Version is not outputting v strings
jnthn lizmat: I'm not sure those are common enough to deserve the sugar...
lizmat: It could be smart enough to do v-strings when it can 10:42
lizmat well, the only reason I can think of, is that it would make .perl better readable ?
yeah, will look at that
jnthn And can we not call them v-strings? :P 10:43
lizmat sorry, P5 leftover :-)
Woodi moritz: thanx
jnthn Yeah, just unfortunate if people google :P
RabidGravy but "special Version constructor syntax" is a bit of a mouthful ;-) 10:46
Woodi
.oO( hmm, no one thinked about slides for coffe break so far... )
10:52 Azry_ left
ilmari "version literal"? 10:58
yoleaux 07:18Z <nine> ilmari: the REPL locking bug is already fixed on the relocatable-precomp branch
RabidGravy when did the CArray.new(@a) constructor come in? That has totally revolutionised my life 11:00
11:02 Upasaka left
lizmat RabidGravy: I think that was sortiz 11:04
dalek rl6-examples/fix-dbi-sql-test: 5825851 | paultcochrane++ | .travis.yml:
Try specifying a stable perl6 version
lizmat m: use lib # DIHWIDT :-)
camelia rakudo-moar 69426b: OUTPUT«===SORRY!===␤Cannot iterate object with P6opaque representation␤»
jnthn We could do with a better error for that :) 11:05
lizmat yeah, will look at that
jnthn: feels like it should actually be silent ?
jnthn "What lib, duh!"
lizmat: It's probably a mistake not to specify a lib? 11:06
11:06 Upasaka joined
lizmat did you mean "use lib 'lib'" 11:06
jnthn Something like that
masak .oO( use lib 'lib "lib"' )
RabidGravy sortiz++ # totally revolutionised the thing 11:07
lizmat
.oO( enters doowop choir)
RabidGravy next up let's see if we can record from jack 11:08
11:08 pdcawley left 11:10 TEttinger left
RabidGravy (recording the sound of the computers fan isn't too interesting) 11:11
11:14 pdcawley joined 11:17 pmurias left 11:18 CIAvash left
dalek rl6-examples/fix-dbi-sql-test: 1e343a4 | paultcochrane++ | .travis.yml:
Use stack version 2016.01
 11:22
11:26 sammers left
masak I have a case where I want to turn a `token statement:if` into a `rule statement:if` in my grammar 11:27
it works great, and I can remove all those pesky <.ws> calls when I do
...but...
suddenly this kind of input fails: "if 0 {\n}\nmy x"
I've tracked that down to the fact that my `statement:if` rule gobbles too much whitespace after the "}", and then <.eat_terminator> doesn't work properly 11:28
and I can "fix" the problem by removing absolutely all whitespace at the end of the rule, but that means I cannot indent the closing brace of the *rule* properly
which seems like a high price to pay
when I look at Rakudo's Grammar.npq, it does not seem to have this problem at all 11:29
what am I missing? :)
I tried using unspace, but that only crashed Rakudo with the internal error "Cannot find method 'rxtype'" 11:30
moritz masak: I'd think that <ws> shouldn't eat up a whitespace if it's significant for statement end detection 11:32
<.ws> is explicitly for *in*significant whitespace
masak hm, I *am* overriding the ws rule... 11:33
jnthn masak: blockoid is a token
masak: And it calls <?ENDSTMT> after parsing the } 11:34
So we've already marked statement end
moritz masak: you could also use :!r within the rule to be able to indent the closing curly brace
masak jnthn: I see. 007 doesn't have the <?ENDSTMT> mechanism, though.
jnthn: in fact, this is the first I hear/learn about it :)
moritz: ooh, I'll try that 11:35
jnthn masak: It's the mechanism that makes } always be a statement end
Well, } \n I guess
masak aye
something to look into in the long run
11:36 espadrine_ joined, espadrine_ is now known as espadrine
masak moritz: doesn't :r have to do with ratcheting semantics, though? 11:36
not sigspace?
jnthn Yeah 11:37
:!s
moritz erm, yes
sorry
masak I already tried :!s, and it had no effect
11:37 brrt left
moritz we use :!s in Perl6::Grammar, so it shouldn't be totally broken 11:38
jnthn Where did you try it?
moritz masak: did you put a whitespace before it? :-)
jnthn Yeah, that's what I was wondering ;-)
moritz I hope not
masak :)
I didn't, I put it right where I had to put the "}" of the rule
it *does* work if I put it before the whole optional `else` group 11:39
11:39 sammers joined
masak but I fear that eliminates an implicit <.ws> inside it that I actually wanted 11:39
bah 11:41
I'll just keep it as `token` for now 11:42
I suspect moritz is right, and I brought this on myself by having a too-simple `ws` rule
but that also means that that's where I ought to fix it... some other time
11:48 labster left 11:51 Azry_ joined
dalek pan style="color: #395be5">perl6-examples: 8c950e5 | paultcochrane++ | .travis.yml:
Use perl6 stack version 2016.01

This gives a stable moar/nqp/rakudo version upon which to test the examples, based upon the 6.c perl6 version.		 11:51
[ptc] masak: thanks for the tip about the perl6 version. It turned out that by using a bleading edge version showed up the same error on my machine and not only on travis. 11:54
masak [ptc]: \o/
"environment -- the silent killer" :D
nine tadzik: your segfault is due to "Can't undef active subroutine" which kind of makes sense. We're still inside a Perl 5 subroutine while trying to tear down the interpreter. 11:56
11:57 pdcawley left
tadzik right 11:57
awwaiid nine: what is the use-case for the control { ... } catches in Inline::Perl5 (such as github.com/niner/Inline-Perl5/blob...pm6#L528)? 12:01
nine awwaiid: they're for Inline::Perl6 actually. 12:02
12:04 jack_rabbit left
awwaiid nine: see... I expected you to say something like "well if you call 'next' in a p6 block that p5 is running, this will catch it for ... something" 12:04
12:09 kid51 joined 12:12 ridthyself joined 12:19 pmurias joined
ridthyself Good morning! I'm having trouble loading a module into the REPL. I run: perl6 -I test.pm6, but when i try to call the sub from the REPL, i get the error: Undeclared routine. 12:21
I run the REPL from the directory where the module is located.
llfourn ridthyself: -I is to load a lib path not a module 12:22
there is -M but I don't *think* that works with just any file
ridthyself: you can do require "test.pm6"; as the first line which should work.
ridthyself failed: not a directory? 12:23
12:24 kjs_ joined
llfourn ridthyself: where do you get that error? I think require is the only way to do it. 12:24
perl6 -e 'require "Foo.pm6"' # should work 12:25
ridthyself it appears to work, but doesn't run the REPL, just compiles runs and closes 12:26
llfourn ridthyself: right. So for the REPL you will simply need to type that as the first line.
12:27 pdcawley joined
llfourn ridthyself: alternitvely you can put it into a directory called "lib" and do perl6 -Ilib -MFoo # and then you will have your module already loaded 12:27
of course it doesn't have to be called "lib" it's just convention :) 12:28
jnthn If it's in the current directory then -I. -MFoo would work too
llfourn ^ that's probably easiest then 12:29
ridthyself THAT WORKS! 12:30
llfourn \o/ 12:31
ridthyself the file in the current directory is test.pm6. I ran perl6 -I. -Mtest and it worked. Thank you, ever so much
so perl is an anagram for REPL... coincidence? 12:33
llfourn I'd say so :P
DrForr_ Yep. 12:34
12:34 kid51 left
ridthyself thanks again, 12:35
12:36 ridthyself left 12:38 zpmorgan left 12:41 pmurias left
Azry_ Hi! I have this code : class IO::Pty is IO::Handle { submethod BUILD() { say self.perl; } } why is the result of this a "IO::Handle" and not a "IO::Pty" ? 12:42
12:43 abraxxa left
Azry_ also self.WHAT gives me (Pty) 12:43
12:46 mr-foobar left
llfourn m: class IO::Pty is IO::Handle { submethod BUILD() { say self.perl; } } 12:47
camelia ( no output )
llfourn m: class IO::Pty is IO::Handle { submethod BUILD() { say self.perl; } }; IO::Pty.new
camelia rakudo-moar 69426b: OUTPUT«IO::Pty is disallowed in restricted setting␤ in sub restricted at src/RESTRICTED.setting line 1␤ in method new at src/RESTRICTED.setting line 32␤ in block <unit> at /tmp/gfgsn7Kt7j line 1␤␤»
llfourn m: class IO::Foo is IO::Handle { submethod BUILD() { say self.perl; } }; IO::Foo.new
camelia rakudo-moar 69426b: OUTPUT«IO::Foo is disallowed in restricted setting␤ in sub restricted at src/RESTRICTED.setting line 1␤ in method new at src/RESTRICTED.setting line 32␤ in block <unit> at /tmp/rCP0S9GCZK line 1␤␤»
Azry_ i use it like that IO::Pty.new(path => "/dev/null"); 12:48
llfourn m: class IO::Foo is Str { submethod BUILD() { say self.perl; } }; IO::Foo.new
camelia rakudo-moar 69426b: OUTPUT«""␤»
tadzik download.java.net/jdk9/docs/api/jav...V-K-V-K-V- I found this moderately amusing
llfourn m: class IO::Foo is Perl { submethod BUILD() { say self.perl; } }; IO::Foo.new
camelia rakudo-moar 69426b: OUTPUT«IO::Foo.new(compiler => Compiler.new(id => "D1585FFC2043CFF5D06D7C99A4A788099F9BB225.1457646626.73081", release => "", codename => "", name => "rakudo", auth => "The Perl Foundation", version => Version.new('2016.2.122.g.69426.bd'), signature => Blob, desc…»
llfourn m: class IO::Foo is Perl { submethod BUILD() { say self.^name; } }; IO::Foo.new
camelia rakudo-moar 69426b: OUTPUT«IO::Foo␤»
12:48 mr-foobar joined
lizmat llfourn Azry_ : seems IO::Handle is hard-coded in IO::Handle.perl 12:50
llfourn ah
lizmat PR's welcome
otherwise I'll fix that in the next half hour or so
llfourn Azry_: so it's just cosemetic :)
cosmetic* 12:51
Azry_ uh i don't understand =) PR ?
llfourn Azry_: Pull request on github if you want to help fix it :)
Azry_ ah uh i've never done that, any pointer where to look for it ? otherwise i'd like to see that commit =) 12:52
dalek kudo/nom: 02122e5 | lizmat++ | src/core/Version.pm:
Streamline Version creation / handling and fixes

Makes it 10% (for v1.foo) to 36x faster (for v6)
  - use custom SET-SELF method for faster initialization
  - save parts as native list with deconted values (makes it truly immutable)
  - creates stringification at creation time (faster .perl/.gist/.Str)
  - shares instances for v6, v6.c, v, v*, v+
  - outputs .perl as version string whenever possible
lizmat I'll do it now
llfourn lizmat++
Azry_: it's here: github.com/rakudo/rakudo/blob/nom/...le.pm#L782 # as you can see it's hardcoded 12:53
DrForr_ tadzik: Closed the issue, I was probably out-of-date.
tadzik DrForr_: sounded like that; thanks :) 12:54
12:54 Skarsnik joined
dalek kudo/nom: a2774a7 | lizmat++ | src/core/IO/Handle.pm:
IO::Handle.perl|gist now a bit more subclass aware

Spotted by Azry_++ and llfourn++		 12:57
Azry_ llfourn: thank you very much =) i'll look into rakudo code a bit more now 12:59
*from now on
13:00 geekosaur left
llfourn Azry_: ^ lizmat++ fixed it so if you rebuild you will get a sane .perl :) 13:00
13:00 geekosaur joined
Azry_ yes i saw that, thanks lizmat 13:01
[ptc] m: use DBIish; 13:03
camelia rakudo-moar 69426b: OUTPUT«===SORRY!===␤Could not find DBIish in:␤ /home/camelia/.perl6␤ /home/camelia/rakudo-m-inst-1/share/perl6/site␤ /home/camelia/rakudo-m-inst-1/share/perl6/vendor␤ /home/camelia/rakudo-m-inst-1/share/perl6␤ CompUnit::Repository::Abso…»
13:03 perlawhirl joined
nine Is there a way to overwrite &exit in the dynamic scope? 13:04
Skarsnik star-m: use DBIish;
camelia ( no output )
Skarsnik wow old star x)
llfourn nine: you can't overwrite anything in dynamic scope apart from $* variables?
nine I need to anyway 13:05
llfourn hmmm
well I actually do do this in a project I have
but I overwrite exit to do be sub exit(|args){ $*some-dynamic.exit(|args) } 13:06
(lexically)
13:07 molaf joined
lizmat perhaps we need to make sub exit look at $*EXIT for code to ectually execute ? 13:09
*actually 13:10
but I'm not sure whether that would be 6.c, 6.c.1 or 6.d land
moritz why not simply die, when you want it to be interceptible?
nine Because tadzik decided to use exit in his code and I can't influence that :) 13:11
dalek kudo/nom: c124e57 | lizmat++ | src/core/VM.pm:
Streamline VM.platform-library-name a bit
13:11 musiKk_ left
llfourn can we do signal trapping? 13:12
moritz yes
nine Using exit in a Perl 6 callback called from Perl 5 code causes a segfault during cleanup of the Perl 5 interpreter because that still is inside a running subroutine
lizmat llfourn: exit / die / warn are not signals in the P6 world ?
tadzik :D
moritz die and warn are control exceptions
lizmat moritz: ah, yes... 13:13
duh -)
perhaps exit should also be ?
nine If I could, I would actually replace exit with a sub that dies while we're inside Perl 5 code
lizmat nine: if sub exit would merely execute what is in $*EXIT, you would be in business, right ? 13:15
(and having PROCESS::<$EXIT> set to what sub exit now is) 13:16
13:18 pdcawley left 13:19 buharin left
jnthn Could be &EXIT 13:19
moritz star-m: say 'version?'
camelia star-m 2016.01: OUTPUT«version?␤»
nine lizmat: yes
lizmat jnthn: &*EXIT you mean?
moritz Skarsnik: we now have a contemporary star
Skarsnik nice
tadzik star: use lib: confused; 13:20
camelia star-m 2016.01: OUTPUT«===SORRY!===␤Cannot iterate object with P6opaque representation␤»
tadzik this is even better!
but this is what I meant;
star: use Panda: confuse-me; 13:21
camelia star-m 2016.01: OUTPUT«5===SORRY!5=== Error while compiling /tmp/4K1lGm6Eox␤Confused␤at /tmp/4K1lGm6Eox:1␤------> 3use Panda7⏏5: confuse-me;␤»
jnthn lizmat: I meant in PROCESS
tadzik "Confused" :)
lizmat jnthn nine ok, patch almost ready
tadzik I encountered it mistyping "use HTTP::UserAgent: simple" instead of "use HTTP::UserAgent :simple"
nine I think I found a fix on the Perl 5 side :) 13:23
I copied this code from perl's S_my_exit_jump into p5_destruct_perl: if (PL_e_script) { SvREFCNT_dec(PL_e_script); PL_e_script = NULL; } POPSTACK_TO(PL_mainstack); dounwind(-1); LEAVE_SCOPE(0); 13:24
Seems to clean up perl's callstack enough so the teardown would go through
lizmat nine: nonetheless, being able to capture exit() may be useful for embedding applications 13:25
nine absolutely
mod_perl does it for example 13:26
lizmat hmmmm... problem with creatin restricted setting
chicken/egg problem 13:29
masak hehe -- implementing fizzbuzz in 007 revealed all kinds of parsing and scoping issues! 13:30
and here I thought it would be straightforward 13:31
13:31 sufrostico joined
masak currently, 007 has a fizzbuzz score of 5, because it gets to 5 ("buzz") before it croaks :P 13:31
13:31 kaare_ joined
tadzik hahahh 13:31
dalek line-Perl5: 5160dd1 | (Stefan Seifert)++ | p5helper.c:
Fix segfault on using exit() in a Dancer2 action

exiting Perl 6 while in a callback from Perl 5 leaves the Perl 5 callstack still in place and may confuse Perl 5's cleanup on interpreter shut down. So we try to tear down the Perl 5 callstack first.
  tadzik++ for reporting this issue!		 13:32
rudi_s jnthn: Awesome. - Seems to work fine. Proposed path here (works fine for me): pbot.rmdir.de/YNtqa61WMPIphTA-w5IA5Q - do you want to apply the patch or should I create a pull request on github?
yoleaux 10:20Z <jnthn> rudi_s: It shouldn't be much harder than just filling in that entry in the op_table
rudi_s *patch
moritz masak: regarding your naming of Q trees: you should make a connection to the Q character in the James Bond movies :-)
("awesome tech from the future")
tadzik masak: "is it a man-compiler, or a boy-compiler?" -- "It's a man. A 5 year old man" 13:33
moritz lol
masak moritz: somewhat embarrassingly, I forgot completely about the 007<->Q connection (and it being funny). I'm just so used to it!
jnthn rudi_s: Well, PR is lazy way for me ;)
masak moritz: the causal order is I came up with Q, and shortly after that, 007 was born. so 007 is actually named after Q. 13:34
rudi_s jnthn: As in easier for you? Then I'll create a pull request.
masak heh, turns out it was only a parsing problem. 007 really didn't like that I named my recursive sub `subtract`... :P
moritz masak: that makes sense. Kinda.
masak so now 007 has a fizzbuzz score of 100 \o/
jnthn rudi_s: Yes, easier for me :)
rudi_s jnthn: Ok. Will do. 13:35
jnthn Thanks!
lizmat $ 6 'my &*EXIT = -> $s = 0 { say $s; die }; exit'
0
Died
nine jnthn : something like that? ^^^
nine Looks cool :)
jnthn lizmat: think that's OK
lizmat ok, spectesting now 13:36
rudi_s jnthn: [x] - any idea if this also needs to be implemented for the jvm backend (sadly I can't test that)? 13:37
llfourn will signals call &*EXIT?
jnthn rudi_s: Probably, I don't think .native-descriptor is implemented at on JVM 13:38
llfourn: Not unless you wire up a signal handler
And have that call exit
llfourn jnthn: cool. But would a custom &*EXIT ever be in the signal handler anyway? 13:39
I would guess not...
jnthn Probably not, though you could also set GLOBAL::<&EXIT> instead and that'd work out
llfourn oh GLOBAL::&EXIT is already a thing? 13:40
do you mean GLOBAL::&exit?
jnthn No, I didn't :P 13:41
That compiles into GLOBAL:: & exit, no? :)
llfourn right you need < > 13:42
13:42 avalenn_ left 13:50 musiKk_ joined 13:52 buharin joined
lizmat hmmm.... somehow the &*EXIT is not being registered... 13:55
13:56 pmurias joined
lizmat will look at it when I'm back& 13:56
current state: gist.github.com/lizmat/ad713ff87e8c4c307a81 # for the tinkerers 13:57
afk&
rudi_s jnthn: tests for the pull request complete - btw. I noticed a few other pull requests which look fine, any reason they weren't merged yet? 14:03
14:03 [Tux] left
jnthn rudi_s: Not reviewed, mostly 14:05
rudi_s Ok.
The VS2013 looks correct (have no windows to test though). 14:06
jnthn I think the lint ones I maybe didn't like
rudi_s The lint ones too (if they compile). - Why didn't you like them?
jnthn Will have to look at them more closely 14:07
rudi_s More const sounds always like a good thing, and correct format string specifier too.
jnthn More const is *not* always a good thing, if it leads to new warnings and clutter.
rudi_s Well, assuming they don't introduce more warnings of course. 14:08
Haven't checked the build.
jnthn Yeah, I think some patches that've gone consting in the past did
Also I'm just generally not fond of linting and changing stuff to please some tool.
14:09 virtualsue joined
perlawhirl hi perlers 14:10
if i already have a token, can i "append/concat" to it like i would a string, eg: my token foo { <$bar> }; token foo { <foo> <$baz> };
i can do it with EVAL *spit* but just wondering if there's another way
or alternatvely, can i generate a new token from an array of tokens, eg: my token foo { <{ @list-of-tokens }>; 14:11
masak that latter sounds much better 14:12
perlawhirl yeah it does
can i do it ? :D
14:13 Mouq joined
perlawhirl i am currently doing it that way with EVAL 14:13
ie. my $rx = EVAL '/' ~ @tokens.join('\s') ~ '/'; $sting ~~ $rx
perlpilot Doesn't <@foo> do that? 14:14
Skarsnik m: sub foo($c) { return if $c == 0; { say "B $c"; repeat { say "A $c"; foo($c -1) } while 0 } }; foo(3)
camelia rakudo-moar c124e5: OUTPUT«B 3␤A 3␤B 2␤»
perlawhirl i think i tried, let me check again
14:14 skids joined, fireartist left 14:15 [Tux] joined
perlawhirl doesn't seem to be, no. matches the first token in the array only 14:15
jnthn <@foo> means an alternation of the things in @foo 14:16
llfourn perlawhirl: <{ @list-of-token.join('||') }> # maybe I'm not that good with grammar
masak but <{ }> is basically an EVAL 14:17
jnthn token foo(*@tokens) { [<?{@tokens}> $(@tokens.shift)]* } or soemthing
perlawhirl jnthn: i'll try it out
jnthn I didn't :) 14:18
perlpilot It's funny that @arrays are an order sequence, but in regex we get an alternation which can match in any order.
llfourn grammar Foo { proto token foo {*} }; for @tokens -> $toke { Foo.^add_multi_method('foo',my token { $toke } } # maybe? 14:19
perlawhirl jnthn: it worked! thanks
masak: i know it basically an EVAL, just didn't want no MONKEYS in my code :) 14:20
llfourn jnthn++ solution is fascinating 14:21
perlawhirl it is indeed... but makes perfect sense!
dalek p: cd2c69d | FROGGS++ | tools/build/MOAR_REVISION:
bump moar revision for nativecall feature
llfourn perlawhirl: if you can explain it to me it would be cool :D 14:22
why does it iterate over each token?
perlpilot llfourn: because of the *
perlawhirl because * is 0 or more, so it keeps shifting as it matches
perlpilot though, I would expect that it could also stop early (not sure if that's desirable or not)
perlawhirl maybe *? will stop early... though i am dubious 14:23
llfourn but * checks for 0 or more, once it fails to match the first thing shouldn't it stop rather than try them all?
or is that what you want?
dalek kudo/nom: a8f407c | FROGGS++ | / (4 files):
allow to cast Pointer to "is native" subroutine
 14:24
perlawhirl yes, that's exactly what i wanted, to fail unless all tokens match
llfourn ahh ok I thought you wanted a disjunction
perlawhirl but i wanted to dynamically generate the list of tokens based on some input
llfourn still interesting :) 14:25
perlawhirl actually, it does stop early, but which i mean the $/ has up to the tokens that matched, which is normal behavour of a match object 14:27
but it's probably easy enough to check if all of them matched
14:27 virtualsue left
perlawhirl a basic way would just be check if there's anything left in @tokens 14:27
14:27 Upasaka left
perlawhirl if there is, we didn't shift all of them, so they didn't all match 14:27
14:28 Upasaka joined
perlpilot yep 14:28
And if you wanted to do this in a context where there's backtracking, you may have to worry about the destructive nature of .shift :) 14:30
14:31 musiKk_ left 14:35 FROGGS[mobile] joined 14:36 virtualsue joined 14:41 virtualsue left
[Coke] could use some coffee 14:43
perlawhirl for those still playing at home... this also works a treat: token foo { <{@tokens[$++]}>+ }; 14:49
dalek kudo/nom: 2a87489 | hoelzro++ | lib/NativeCall.pm6:
Fix NativeCall

For some reason (role bodies are tricky), we got an error along the lines of "this type does not support associative operations" regarding $lib2mangler. Initializing $lib2mangler at the unit level fixes the issue
14:49 cdg joined
llfourn perlawhirl: nice
perlpilot perlawhirl: how about <{@tokens[$++]}> ** {+@tokens} (to make sure they all match) 14:50
?
perlawhirl my mind was already walking down that path :) let me test
yes =) 14:51
perlpilot cool
perlawhirl fails nicely
llfourn perlpilot++ perlawhirl++
perlawhirl with a big fat Nil
perlpilot perl6++
14:53 tmch left
perlpilot It's still weird that to get the most salient feature of arrays in a regex, you have to be so verbose. It would be nice if there were some built-in way to say "match these things in order" and "match *all* of these things in order" 14:58
jnthn Yeah... We're at least consistent though :) 14:59
< foo bar baz > in a regex is also an alternation
15:00 brrt joined 15:02 brrt left, perlawhirl left
perlpilot It's too bad @array doesn't match all of the items in sequence and you must say |@array or ||@array to get the alternation. (you'd probably need some modifier to make @array stop early though) 15:10
though that this is the first time since the regex reformation that this has come up says something about the demand for such a feature too :)
15:11 sufrostico left
perlpilot anyway ... 15:11
perlpilot stops thinking about it
geekosaur you know, I don;t think I've ever needed that
since, uh, a week after the release of perl 3
15:20 eiro left 15:21 jameslenz left 15:24 sufrostico joined 15:26 Mouq left
dalek c: 1dd6bbf | FROGGS++ | doc/Language/nativecall.pod:
describe how to invoke function pointers
 15:39
Skarsnik duh that really work? 15:42
15:45 MadcapJake left 15:50 FROGGS left 15:52 andreoss joined
andreoss is there way to pass by name in Perl 6? 15:54
15:55 travis-ci joined
travis-ci Doc build failed. Tobias Leich 'describe how to invoke function pointers' 15:55
travis-ci.org/perl6/doc/builds/115346161 github.com/perl6/doc/compare/f6084...d6bbf419bb
15:55 travis-ci left
andreoss or rather call by name 15:56
skids m: 42."say" 15:57
camelia rakudo-moar 2a8748: OUTPUT«5===SORRY!5=== Error while compiling /tmp/NckFTcfdd_␤Quoted method name requires parenthesized arguments. If you meant to concatenate two strings, use '~'.␤at /tmp/NckFTcfdd_:1␤------> 0342."say"7⏏5<EOL>␤»
skids m: 42."say"()
camelia rakudo-moar 2a8748: OUTPUT«42␤»
15:59 xinming left
andreoss something like this docs.scala-lang.org/sips/pending/re...tion-rules 15:59
16:01 eiro joined
andreoss m: sub loop-while(&cond, &body) { &body() if &cond(); &?ROUTINE(&cond, &body) }; my $i = 0; loop-while({ $i < 5 }, { $i++; say $i }); 16:05
camelia rakudo-moar 2a8748: OUTPUT«1␤2␤3␤4␤5␤Memory allocation failed; could not allocate 127424 bytes␤»
jnthn No, you have to explicitly closure them on the callee side 16:06
It's the kind of thing macros might enable
16:06 cognominal joined
jnthn But we don't have those yet. 16:06
andreoss m: sub loop-while(&cond, &body) { &body() and &?ROUTINE(&cond, &body) if &cond() }; my $i = 0; loop-while({ $i < 5 }, { $i++; say $i });
camelia rakudo-moar 2a8748: OUTPUT«1␤2␤3␤4␤5␤»
jnthn fwiw, you can just say body() and cond() for calling them if you bound them with the & sigil 16:07
dalek pan style="color: #395be5">perl6-examples: c7cef31 | raiph++ | categories/tutorial/01-word-wrap.pl:
Rewrite for reasonable naming and bug fix(es)
 16:09
andreoss so closures don't count as `by name` 16:13
llfourn I did some googling on "pass-by-name" ... Don't really get it 16:14
andreoss llfourn: look for scala examples
llfourn andreoss: yeah I have -- they just say it's a syntax trick where instead of doing somefunc(() => math.random) you can just pass somefunc(math.random) # is this right? 16:15
ie it's just shorthand 16:16
but in other places it seems to be some more high level compsci abstraction
andreoss I guess it doesn't use closures at all, but rather operates on AST level somehow 16:17
llfourn so it's just a macro?
16:19 nakiro left
llfourn oh you mean it like compiles a new func each time it sees a call to it and s/arg/math.random/. Kinda interesting. 16:20
but how did this ever get called "call-by-name"
Azry_ is it good practices to use cmake or autoconf to build a module ? I need to check for some libraries/vars/functions. works with cmake but i wonder if it's the way to go 16:25
16:25 ygbn joined
Azry_ Or can Nativecall check for functions/structs/var itself ? 16:26
Skarsnik Nc only use the library file 16:27
source/header are irrelevant, only the symbols in the library files, and proper size for signature of stuff 16:28
Azry_ so i need to check for functions like i would though CHECK_FUNCTION_EXISTS with cmake, is there a perlish way to do that ? 16:32
16:34 Guest51674 is now known as alpha123
Skarsnik I am not sure to get what you want to do. NC work with compiled library, it depend on how the library you want to use had been compiled 16:34
16:34 ygbn left, khw joined 16:36 FROGGS joined
Azry_ i want my code to be portable, and differents OS have differents functions to do the same thing (Pty allocation). I want to know which function to use. I'm currently doing it by initialising variables like "HAVE_PTSNAME" with cmake 16:37
but that implies that the user needs cmake, so i'm not completely satisfied 16:38
16:38 rhmekjfn joined
Skarsnik you can test the os in perl6 and change what you bind if you already know the symbol 16:39
geekosaur might be better to just declare a dependency on one of the C libraries that abstract pty creation
rhmekjfn m: sub loop-while(&cond, &body) { if (&cond()) {&body()} else {return}; &?ROUTINE(&cond, &body) }; my $i = 0; loop-while({ $i < 5 }, { say $i; $i++; }); # andreoss
camelia rakudo-moar 2a8748: OUTPUT«0␤1␤2␤3␤4␤»
geekosaur and bind to that, for a fixed perl6-facing abi
this also protects you from having to deal with e.g. some of the security gotchas around pt_chown 16:41
andreoss m: proto role YYY {*}; 16:42
camelia rakudo-moar 2a8748: OUTPUT«5===SORRY!5=== Error while compiling /tmp/lNjGdeFuWk␤Missing block␤at /tmp/lNjGdeFuWk:1␤------> 3proto role7⏏5 YYY {*};␤ expecting any of:␤ new name to be defined␤»
pmurias what is nqp::closefh_i on a syncpipe supposed to return?
andreoss is there proto roles?
rhmekjfn m: sub loop-while(&cond, &body) { &body() and &?ROUTINE(&cond, &body) if &cond() }; my $i = 0; loop-while({ $i < 5 }, { say $i; $i++; });
camelia rakudo-moar 2a8748: OUTPUT«0␤»
16:44 FROGGS_ joined
FROGGS_ m: use NativeCall; say cglobal Str, 'malloc', Pointer # Azry_ 16:44
camelia rakudo-moar 2a8748: OUTPUT«NativeCall::Types::Pointer<0x7fb9c563ef60>␤»
FROGGS_ m: use NativeCall; say cglobal Str, 'flubber', Pointer # Azry_
camelia rakudo-moar 2a8748: OUTPUT«Cannot locate symbol 'flubber' in native library ''␤ in block at /home/camelia/rakudo-m-inst-2/share/perl6/sources/24DD121B5B4774C04A7084827BFAD92199756E03 line 406␤ in block <unit> at /tmp/lzg2iW_fpm line 1␤␤»
FROGGS_ Azry_: Str is the arg where you'd put the library name
and malloc and flubber are just examples of functions you want to check 16:45
Azry_ FROGGS: wow, so there is a way =) great i'll test this, thanks
Skarsnik is that a typed execption that could be catch?
FROGGS_ if the lib exposes that symbol, you get a valid pointer back, otherwise an exception
pmurias jnthn: is there a reason for nqp::closefh_i to return the exit value shifted by 8 bits? 16:46
16:47 FROGGS left, cdg left 16:49 FROGGS__ joined
FROGGS__ Skarsnik: dont think so 16:49
16:49 FROGGS__ is now known as FROGGS 16:50 FROGGS_ left
andreoss p: use MONKEY-TYPING; use class :open<Foo>; class Foo is open {...}; 16:50
m: use MONKEY-TYPING; use class :open<Foo>; class Foo is open {...};
camelia rakudo-moar 2a8748: OUTPUT«===SORRY!===␤Could not find class in:␤ /home/camelia/.perl6␤ /home/camelia/rakudo-m-inst-2/share/perl6/site␤ /home/camelia/rakudo-m-inst-2/share/perl6/vendor␤ /home/camelia/rakudo-m-inst-2/share/perl6␤ CompUnit::Repository::Absol…»
FROGGS what's that? 16:51
andreoss S12
16:52 rhmekjfn left
FROGGS NYI 16:52
dunno if we ever get that
16:53 FROGGS[mobile] left
pmurias jnthn: it seems we are unpacking it into signal and exit code in the Perl 6 layer so it seems ok 16:54
16:54 zakharyas left 16:55 domidumont left
FROGGS has somebody ever spotted that? travis-ci.org/perl6/doc/builds/115346161#L1674 16:55
llfourn FROGGS: same thing happened to me 16:56
FROGGS ahh
llfourn it's non-deterministic
FROGGS so I am not to blame?
llfourn no your pod is probably not bad enough to cause a segfault
FROGGS I'll let it rebuild
*g*
thanks
nine Is there some way to only enable a pragma if the compiler supports it? 17:02
FROGGS maybe use this? github.com/FROGGS/p6-if 17:03
nine To be precise: to only pass an argument to a pragma if the compiler support it 17:04
llfourn FROGGS++ a World mixin nice idea
in a scary sort of way 17:05
nine I don't even want to think about how that interacts with precompilation :)
skids m: my $c = Channel.new; $c.send($(1,2)); $c.send((1,2)); $c.receive.perl.say; $c.receive.perl.say; # any particular reason why Channel stripping itemization is beneficial behavior?
camelia rakudo-moar 2a8748: OUTPUT«(1, 2)␤(1, 2)␤»
nine my role BetterWorld 17:06
llfourn XD
FROGGS well, can you check the signature of the thing the pragma calls or so? 17:07
then you'd have a somewhat robust test to use with pragma if
nine The pragma is 'precompilation' which is implemented directly in do_pragma 17:08
dalek p: 6d38769 | (Pawel Murias)++ | src/NQP/Compiler.nqp:
[js] Add our custom command line flags to the nqp-js-on-js.
 17:11
p: 3242d3e | (Pawel Murias)++ | src/vm/js/ (3 files):
[js] Implement nqp::closefh_i and nqp::syncpipe.

This allows us to pass test 86. This also makes --beautify on nqp-js-on-js work.
17:13 travis-ci joined
travis-ci NQP build failed. Pawel Murias '[js] Implement nqp::closefh_i and nqp::syncpipe. 17:13
travis-ci.org/perl6/nqp/builds/115369672 github.com/perl6/nqp/compare/cd2c6...42d3ee0d18
17:13 travis-ci left 17:14 gtodd1 joined, gtodd left
pmurias checks why it fails 17:15
weird this seems to be an js only commit that fails 17:16
dalek p: 8b11181 | coke++ | README.pod:
Make MoarVM more prominent than the Parrot VM
 17:19
[Coke] restarts the failed job. 17:20
FROGGS I can reproduce here
17:21 travis-ci joined
travis-ci NQP build failed. Will "Coke" Coleda 'Make MoarVM more prominent than the Parrot VM' 17:21
travis-ci.org/perl6/nqp/builds/115371602 github.com/perl6/nqp/compare/3242d...1118196e03
17:21 travis-ci left
FROGGS pmurias: why does my moar only nqp fail to build with your patch? that does not make any sense 17:22
[Coke] that's the failure in the travis, also: Unhandled exception: While looking for 'nqp.moarvm': no such file or directory
17:22 sufrostico left, travis-ci joined
travis-ci NQP build failed. Pawel Murias '[js] Implement nqp::closefh_i and nqp::syncpipe. 17:22
travis-ci.org/perl6/nqp/builds/115369672 github.com/perl6/nqp/compare/cd2c6...42d3ee0d18
17:22 travis-ci left
FROGGS maybe it does not understand #?if js in NQP::Compiler? 17:23
17:25 sufrostico joined
FROGGS pmurias: found it! 17:26
dalek p: 9b10c93 | FROGGS++ | src/NQP/Compiler.nqp:
fix typo in fudge line
 17:27
17:27 andreoss left
dalek c: 3b5bd22 | paultcochrane++ | doc/Language/testing.pod:
Document the can-ok() test function
 17:29
17:30 travis-ci joined
travis-ci NQP build passed. Tobias Leich 'fix typo in fudge line' 17:30
travis-ci.org/perl6/nqp/builds/115373403 github.com/perl6/nqp/compare/8b111...10c935309f
17:30 travis-ci left, FROGGS left
pmurias sorry for breaking master :/ 17:32
dalek line-Perl5: 827f9bf | (Stefan Seifert)++ | lib/Inline/Perl5.pm6:
Allow for precompilation of Inline::Perl5 itself

As we can precompile Inline::Perl5 just fine and only modules that use
  :from<Perl5> cannot be precompiled, we "export" the no precompilation
pragma to them.
This takes several seconds off the startup time of any program using Inline::Perl5
  FROGGS++ for contributing the winning idea!		 17:33
tadzik nice! 17:34
now even tests pass faster :) 17:36
and Perldancer.org starts in like 1 second 17:37
17:38 sufrostico left 17:39 sufrostico joined 17:41 FROGGS[mobile] joined
FROGGS[mobile] sergot, tadzik: are you comming? 17:42
tadzik ye 17:43
ilmari nine++ # perl6 -e 'use Inline::Perl5' down from 6.9 to 0.3 seconds on my machine 17:45
FROGGS[mobile] nice!
timotimo oooh
17:47 sufrosti1o joined, sufrostico left, sufrosti1o is now known as Sufrostico, Sufrostico left, Sufrostico joined 17:49 SCHAAP137 joined
hoelzro wow 17:49
Hotkeys Oh god xkcd.com/1654/ 17:53
timotimo <3 17:54
17:59 prammer left 18:00 prammer joined
geekosaur can double that list without even trying, and add even more with some thought... 18:01
18:02 xfix joined 18:03 xfix left 18:05 domidumont joined 18:06 pmurias left 18:08 buharin left, AlexDaniel joined 18:09 kjs_ left
AlexDaniel m: say [1, 2, 3] Z= (4, 5, 6) # heh 18:09
camelia rakudo-moar 2a8748: OUTPUT«(4 5 6)␤»
jnthn nine: Wow, that's evil! :-D 18:11
18:15 dakkar left, Skarsnik left, Skarsnik joined 18:18 maybekoo2 joined 18:19 prammer left
Azry_ is LibraryCheck the best way to test if a library exists ? 18:19
18:20 zpmorgan joined
Azry_ (with nativecall) 18:21
RabidGravy Or do what it does 18:25
it's not very complicated
Azry_ well, i could parse ldconfig -r but i doesn't look clean =) 18:26
RabidGravy all LibraryCheck does is create a native sub on the library and execute it, checking the exception, if it's "library not found" then it returns false 18:28
18:29 jevin left 18:30 atweiden joined
timotimo what if the library actually has the function of that name, and executing that function willy-nilly will give you big trouble? 18:32
i mean, sure, it uses a 15-letters-long random combination of lower and upper case letters 18:33
RabidGravy it tries hard to make an unlikely name ;-)
timotimo but how hard is it really to build a library that has all of these?
18:33 wamba left
Azry_ my problem is : on debian i need to refer to libc.so.6 and on my freebsd i need to refer to libc.so.7 18:34
timotimo did you see the thing froggs showed off with "malloc" and "flubber" using cglobal?
Azry_ yes, but it guesses the name of the library the same way
RabidGravy you shouldn't need to worry about the library for libc, it'll be loaded already 18:35
Azry_ oh, i didn't know that, so i donc need to give an argument to "is native()" then ?
18:36 andreoss joined
RabidGravy no you shouldn''t have to 18:36
Azry_ okay thanks, i look into that =)
andreoss is there a type like Nil on which roles can be applied?
timotimo yesterday i saw a post by an "expert" who said "it's still a board-game, though. you can just calculate every possible move, then every possible counter-move, then every counter-counter move, and so on, until you win. why is everybody so excited about this?"
andreoss m: role Foo { }; my $x = Nil.new ; $x does Foo; 18:37
camelia rakudo-moar 2a8748: OUTPUT«Cannot use 'does' operator with a type object.␤ in block <unit> at /tmp/PtveJTPJbR line 1␤␤»
timotimo er, don't you have to give an argument to "is native" still?
just an undefined one? like Str?
RabidGravy andreoss, "but"
andreoss m: role Foo { }; my $x = Str.new ; $x does Foo;
camelia ( no output )
andreoss m: but.WHAT.say 18:38
camelia rakudo-moar 2a8748: OUTPUT«5===SORRY!5=== Error while compiling /tmp/zkRKjo_rl4␤Undeclared routine:␤ but used at line 1. Did you mean 'put'?␤␤»
timotimo (to be fair, that person didn't claim to be an expert)
RabidGravy m: role F {} ; say (Nil but F}.WHAT
camelia rakudo-moar 2a8748: OUTPUT«5===SORRY!5=== Error while compiling /tmp/foL_YVw9pK␤Unable to parse expression in parenthesized expression; couldn't find final ')' ␤at /tmp/foL_YVw9pK:1␤------> 3role F {} ; say (Nil but F7⏏5}.WHAT␤ expecting any of:␤ stat…»
RabidGravy m: role F {} ; say (Nil but F}).WHAT
camelia rakudo-moar 2a8748: OUTPUT«5===SORRY!5=== Error while compiling /tmp/98CyzKXCdq␤Unable to parse expression in parenthesized expression; couldn't find final ')' ␤at /tmp/98CyzKXCdq:1␤------> 3role F {} ; say (Nil but F7⏏5}).WHAT␤ expecting any of:␤ sta…»
RabidGravy m: role F {} ; say (Nil but F).WHAT
camelia rakudo-moar 2a8748: OUTPUT«Nil␤»
RabidGravy m: role F {} ; say (Str but F).WHAT 18:39
camelia rakudo-moar 2a8748: OUTPUT«(Str+{F})␤»
[Coke] m: role F {} ; my $x = F.new; say $x.WHAT
camelia rakudo-moar 2a8748: OUTPUT«(F)␤»
RabidGravy oh
andreoss RabidGravy: thanks 18:40
lizmat .tell hoelzro your fix makes all native call tests from "make test" fail for me :-( how is that for you ?
yoleaux lizmat: I'll pass your message to hoelzro.
18:40 skids left
andreoss but i still somehow get `slip()` 18:40
RabidGravy from a sub? that's usually when you slipped off the end without a suitable value to return 18:41
18:41 RabidGravy left
ugexe m: my $x = gather for 1..5 { take $_; say $x; }; say $x.perl # is this error message able to be improved? 18:42
camelia rakudo-moar 2a8748: OUTPUT«List_57850416␤List_57850416␤List_57850416␤List_57850416␤List_57850416␤(\List_57850416 = (1 1 2 3 4 5))␤Method 'push' not found for invocant of class 'Mu'␤ in any at /home/camelia/rakudo-m-inst-2/share/perl6/runtime/CORE.setting.moarvm line …»
18:43 maybekoo2 left
andreoss paste.debian.net/414250/ 18:44
anyone did monads in perl6? 18:45
18:45 jevin joined 18:47 maybekoo2 joined 18:48 xinming joined
hoelzro lizmat: oh no =( 18:49
yoleaux 18:40Z <lizmat> hoelzro: your fix makes all native call tests from "make test" fail for me :-( how is that for you ?
hoelzro I didn't spectest, I had to get out the door
timotimo well, those aren't spec tests though :)
those are non-spec tests!
lizmat
.oO( why are people committing changes without testing ?)
there are no nativecall spectests afaik 18:50
hoelzro yeah, that was not cool on my part
Skarsnik well nc is rakudo
lizmat hoelzro: so, I would like to see a piece of code that fails for you
hoelzro my canary was just telling me tab completion was broken, so I wanted to fix it before work
Skarsnik maybe there could be test for the nqp part x)
timotimo i don't really remember what those changes were
hoelzro lizmat: install Linenoise and fire up the REPL
lizmat ok, will do 18:51
hoelzro lizmat: feel free to revert my change for now
I'll do a proper fix after work
lizmat lemme see if I can see what's wrong in the REPL case
hoelzro my commit message describes the situation too 18:52
lizmat yeah, but not *how* you got that message
hoelzro true! 18:53
18:54 skids joined
lizmat it is somehow related to precomp 18:58
if the order is: use lib 't/04-nativecall'; use CompileTestLib; use lib 'lib'; use NativeCall 18:59
it fails, but if the order is: 19:00
if the order is: use lib 't/04-nativecall'; use lib 'lib'; use CompileTestLib; use NativeCall
it works ok
yuck
could it be that we're nuking the mainline of NativeCall somehow? 19:01
19:01 RabidGravy joined
RabidGravy kernel update 19:02
19:07 buharin joined
RabidGravy really struggling with reading from a jack source with portaudio, can't seem to find the right combination of buffer size and latency 19:08
19:09 lichtkind joined 19:16 prammer joined
dalek kudo/nom: 3e2e7f9 | lizmat++ | t/04-nativecall/ (17 files):
Remove superfluous 'use lib "lib"'

This fixes the nativecall test breakage caused by 2a8748979b321635 . As an added bonus, this makes "make test" go 3x as fast.
  hoelzro++ for pointing the way		 19:16
19:16 firstdayonthejob joined
dalek kudo/nom: 5a1e681 | lizmat++ | t/04-nativecall/19-function-pointers.t:
Correct test count
 19:17
RabidGravy is evil going to happen if I close over a channel in a sub passed as a callback that may get executed in another thread? 19:20
moritz RabidGravy: I sure hope not 19:21
it's supposed to be, like, thread safe :-) 19:22
RabidGravy a callback that will get called by native code this is
19:23 buharin left
dalek kudo/nom: b703d67 | lizmat++ | t/0 (41 files):
Make sure all test files start with "use v6"

And some additional cosmetics		 19:24
19:26 kerframil joined
dalek kudo/nom: 5e59d2d | lizmat++ | src/core/Version.pm:
Oops, we should share v6.c, not v6c
 19:29
19:29 domidumont left 19:33 Sufrostico left
AlexDaniel lizmat: I love your commits :) 19:33
19:33 lichtkind_ joined
AlexDaniel it's always like “oh, things are much better now” 19:33
19:34 Sufrostico joined, ely-se joined
dalek kudo/nom: 41d0007 | lizmat++ | src/core/Version.pm:
Make sure we get stringification correct (*sigh*)
 19:35
lizmat AlexDaniel: well, I'm going for better generally :-) 19:36
19:36 yqt joined 19:37 lichtkind left, spider-mario joined
perlpilot aye, lizmat++ for always improving things 19:37
(even with missteps, you're still headed in the right direction) 19:38
lizmat thank you both for your kind words :-)
Skarsnik 3x faster for NC test
nice
lizmat Skarsnik: I think we need to thank nine++ for that essentially 19:39
with the recent precomp improvements
Skarsnik I though it was the compile time that was slowy then a lot 19:40
*slowing
lizmat I think it was basically precomping all of NativeCall in each test, it doesn't do that anymore 19:41
Skarsnik I mean, the c library compile time
lizmat *test file
hoelzro lizmat: thanks for fixing that!
lizmat++
Skarsnik because 17-18 are quite fast
19:43 cdg joined 19:47 atweiden left 19:49 prammer left, prammer joined
timotimo that's quite good 19:49
19:52 andreoss left 19:54 espadrine left 19:56 pdcawley joined 19:59 prammer left 20:04 prammer joined, musiKk_ joined 20:05 prammer left, rindolf left
Woodi I found in Moust Needed that LDIF files parsing is missing so here is pure Perl6 (grammars) version: github.com/slunski/perl6-text-ldif 20:06
20:08 rindolf joined
RabidGravy there may be a problem "use v6.c;" in "2016.02-128-g3e2e7f9" 20:09
20:09 jjido joined
timotimo RabidGravy: a diferent problem from what liz fixed in 5e59d2d ? 20:10
lizmat and further fixed in 41d0007 ?
RabidGravy I.e. if I load a module which is simply "use v6.c; module F {}" it craps out in "Missing serialize REPR function for REPR NativeRef" 20:11
those just tests though
20:12 prammer joined
lizmat RabidGravy: can't reproduce that here :-( 20:12
RabidGravy maybe it was the latter one that fixed t 20:13
20:13 rindolf left
lizmat I *did* just see that "use v6.c; use Test" runs the CORE setting mainline *twice* 20:13
RabidGravy let's see
timotimo o_O
lizmat m: A.a; class A { method a() { say "a" } } # core setting doesn't complain about this situation 20:14
camelia rakudo-moar 41d000: OUTPUT«5===SORRY!5=== Error while compiling /tmp/B5RcqCwX4M␤Illegally post-declared type:␤ A used at line 1␤␤»
rudi_s Hi. Is there a shortcut for $_ ~~ ...? 20:17
lizmat ...ACCEPTS($_) 20:18
rudi_s I'll stick with $_ ~~, thanks ;-)
20:18 Mouq joined
timotimo given/when will use ~~ for you 20:18
actually, when will use ~~ on $_ for you 20:19
but when also has implications on program flow
20:19 prammer left
rudi_s Yeah. Just an if in this case. 20:19
lizmat m: $_ = 42; when Int { .say }
camelia rakudo-moar 41d000: OUTPUT«42␤»
lizmat m: $_ = "42"; when Int { .say }
camelia ( no output )
20:20 darutoko left
timotimo m: $_ = 64; .say when Int 20:20
camelia rakudo-moar 41d000: OUTPUT«64␤»
RabidGravy lizmat, yes it was the latter commit that fixed it, I must have pulled immediately before you did that one 20:21
lizmat sorry 'bout that one...
RabidGravy Hmm, but I'm still getting "Missing serialize REPR function for REPR NativeRef" in the module 20:23
20:24 labster joined
lizmat RabidGravy: gist ? 20:24
RabidGravy so previously just a "use v6.c" was enough to provoke it, now I don't know what 20:25
the module I'm testing is fine, but a script that uses amother module too gets that 20:26
20:27 prammer joined
dalek kudo/nom: 697fb83 | lizmat++ | src/core/ (2 files):
Implement &*EXIT to catch "exit"

By creating your own &*EXIT, you can catch what "exit" does, e.g.:
   $ 6 'my &*EXIT = -> $a = 0 { $a.say; die }; exit'
   0
   Died		 20:28
lizmat RabidGravy: could you test whether this diff solves it ? 20:29
gist.github.com/lizmat/e683c036e1e5dcc6e385
RabidGravy just rebuilding 20:35
and nope 20:41
lizmat :-(
20:43 Sufrostico left
lizmat brb& 20:43
RabidGravy yeaah, it's Audio::Sndfile 20:44
20:45 lichtkind_ is now known as lichtkind
lichtkind i just rethought roles and i meaby dont get a thing but would it be possible to have just finer grained classes and do multiple inheritance instead of "does" 20:46
yes hter is still ability to mix in at runtime
but for the rest im correct?
20:47 cognominal left, prammer left 20:48 cognominal joined, prammer joined, cognominal left 20:49 cognominal joined, prammer left, rindolf joined 20:50 prammer joined 20:51 prammer left
lizmat RabidGravy: could you try this gist ? gist.github.com/lizmat/1f401586d4098569ebae 20:52
20:53 prammer joined
RabidGravy something strange happened in 2016.02-121-g84a84f2 and 2016.02-133-g697fb83 20:54
between that is
20:55 prammer left, Mouq left
lizmat RabidGravy: inclusive or exclusive ? 20:56
RabidGravy well the earlier one works and the latter one doesn't
20:57 prammer joined, partly joined 20:58 partly left
lizmat RabidGravy: did you try that last gist already ? 20:58
RabidGravy just rebuilding
skids lichtkind: you can do it that way, as long as you can live with the stricture of heirarchy. Most languages felt the need to have "traits" to escape that heirarchy. 20:59
jnthn It's also about conflict detection. 21:00
Roles flatten and methods that are identical in name or, if multis, in signature, will conflict.
21:00 jjido left
jnthn Whereas with inheritance it's just that you get the first one by C3 ordering 21:00
21:01 TreyHarr1 joined 21:02 TreyHarr2 joined
lizmat jnthn: is it correct that a "use v6.c" should cause to run the CORE settings mainline twice ? 21:02
21:02 TreyHarris left
lichtkind skids, thanks for understanding the question :) 21:03
jnthn++ too :) 21:04
21:04 DarthGandalf left
jnthn lizmat: No...that should be nearly impossible... 21:04
lizmat ok, lemme see if I can double verify
jnthn Unless the setting loading got really shagged up somewhere but even then I don't think Moar would actually run load twice on the same bytecode file 21:05
Since it keeps the filenames cached
lizmat ok
jnthn So we'd more likely get a weird error than run twice
lizmat jnthn: BTW, is a REPR NativeRef ever in the works ?
jnthn But yeah, worth confirming
RabidGravy lizmat, no that didn't fix it
jnthn NativeRef? 21:06
21:06 TreyHarr1 left
lizmat my Version changes seems to have caused a regression related to module loading :-( 21:06
RabidGravy I am getting "Missing serialize REPR function for REPR NativeRef" in something that worked an hour ago
lizmat I guess somehow it tries to precomp some native values from a Version object 21:07
jnthn There is a NativeRef REPR; it's the one that's used for passing around references to native attributes etc.
We can probably serialize those in *some* cases...
Maybe all if we hurt our branes enough :)
lizmat jnthn: then why is it complaining about it being missing ? 21:08
jnthn lizmat: Because we didn't implement serializing them yet :)
RabidGravy it is associated with the version thing
skids
.oO(That error message might do with mentioning "constant" somehow since that's how users stub their toe on it.)
 21:11
jnthn skids: Theres's plenty of ways you might 21:12
And we can't easily trace it back to the problem 21:13
lizmat RabidGravy: could you verify that b703d6727714aa24 still works ok ?
jnthn Well, he origin
21:13 prammer left
timotimo i'll be bringing a few snack-ish things to the hackathon, too. among other things, a "fizzbuzz" themed one, and a bag of "crazy python" 21:14
21:14 prammer joined
lizmat wishes she could be there 21:15
instead, I'll be sitting in a car most of the day
21:17 prammer left
timotimo :< 21:17
21:18 SCHAAP137 left
timotimo i got sit on by a cat right now. i think that's better than sitting in a car 21:18
21:18 prammer joined
lizmat :-) 21:18
t.h8.lv/andshare/IMG_20160217_211308~2.jpg.jpg 21:19
timotimo yup 21:22
RabidGravy lizmat, no that doesn't work either
will bisect harder
jnthn hopes to be sleeping lots tomorrow :) 21:24
Then maybe some heap profiler hackery :)
Given I lost a day's worth of working time this week to darn headaches
(They've mostly cleared up again... :))
lizmat :-) 21:25
Hotkeys I'm trying to make an equivalent to /(?<=(.))(?=\1)/ in p6 but I can't do captures in the perl 6 lookarounds
21:25 colomon left
Hotkeys is this intended? 21:25
timotimo i notice a distinct lack of tadzik for the hackathon? :o
21:25 colomon joined
Hotkeys m: m/<?after (.)><?before $0>/ 21:26
camelia rakudo-moar 697fb8: OUTPUT«5===SORRY!5=== Error while compiling /tmp/56pYr0WI__␤Unable to parse expression in metachar:sym<( )>; couldn't find final ')' ␤at /tmp/56pYr0WI__:1␤------> 3m/<?after (.)><?before $0>7⏏5/␤ expecting any of:␤ infix stopper␤»
lizmat RabidGravy: I would try before 02122e5e67d0c9738c and after 21:27
RabidGravy: also, could you post a --ll-exception trace ?
perlpilot m: /<?after (.) ><?before $0>/ 21:28
camelia ( no output )
Hotkeys oh 21:29
perlpilot Hotkeys: )> might be confusing the parser?
Hotkeys I just needed a space?
perlpilot I dunno, it was just a guess
Hotkeys well it didn't get angry
let's see
dalek ast: ebcd568 | lizmat++ | / (76 files):
Make sure that we only have one "use lib"

  *BEFORE* loading any modules. Looks like this causes less precomp,
resulting in the spectest using ~50 CPU seconds less, and about 10 seconds wallclock less (with TEST_JOBS=8)
jnthn ooh 21:30
lizmat++
lizmat: Did you spot any spectest CPU time reduction from my compiler improvements earlier in the week, btw?
lizmat no, alas, or I would have mentioned it :-)
lichtkind lizmat++
jnthn OK. It shoulda shaved a little of startup time too, but I suspect it's more "measurable with callgrind" :) 21:31
lizmat jnthn: fwiw, the tendency has been upward for bare startup time
115 msecs for me 21:32
21:32 DarthGandalf joined
lizmat pretty far away from <100 msecs of half a year / a year ago :-( 21:32
but at least, loading modules Is now a *lot* faster (once they're precomped) 21:33
21:33 prammer left
jnthn lizmat: Yeah well, the code size of CORE.setting is probably somewhat to thank 21:34
lizmat ah, I realize where the "running mainline twice" came from
from precomping a module (in another process)
jnthn Ah :) 21:35
21:35 prammer joined
lizmat jnthn: had a bit of a puzzle implementing &*EXIT 21:35
m: A.a; class A { method a() { say "a" } } # jnthn: core setting doesn't complain about this situation
camelia rakudo-moar 697fb8: OUTPUT«5===SORRY!5=== Error while compiling /tmp/DwSiMqFL_I␤Illegally post-declared type:␤ A used at line 1␤␤»
jnthn Curious 21:36
21:36 prammer left
perlpilot Hotkeys: BTW, I think /(.) <( )> $0 / may actually do what you want 21:36
jnthn Wonder why
21:37 prammer joined
lizmat m: class A { ... }; A.a; class A { say "running"; method a() { say "a" } } # mainline is run *after* the method call 21:37
camelia rakudo-moar 697fb8: OUTPUT«a␤running␤»
lizmat jnthn: ^^^ that was the actual situation 21:38
jnthn If you have a class A { ... } then it's pre-declared and the method call is OK
lizmat yeah, it was just a gotcha / DIHWIDT
jnthn But yeah, code in a class block just runs as it it was in a bare block
tadzik timotimo: yeah, plane's at 12.ish
Hotkeys perlpilot: you're wonderful 21:39
21:39 SCHAAP137 joined
RabidGravy is the first one where it starts going wrong 69426bd8b3adeffd0a1a23ff240bb1018bfedca4 21:39
lizmat jnthn: not sure whether that could / should warrant a warning
and hoelzro's fix didn't fix ? 21:40
aka 2a8748979b32163563 21:41
RabidGravy every commit after that one doesn't work
21:41 FROGGS[mobile] left
hoelzro I broke everything? =( 21:41
lizmat looks
no, *I* did 21:42
hoelzro I broke it transitively, then =P
timotimo tadzik: sad to hear i'll be missing you by a day :|
RabidGravy just rolling back to the one before that one
jnthn lizmat: No, I don't think it's anything unusual 21:43
lizmat: Anything that actually affects the class should be done at BEGIN time anyway
lizmat in the case of the settings, it's not always clear in which order the files are.. :-( 21:44
but yeah, lesson learned
jnthn We could merge it all into one file to help ;) 21:45
</trololol>
timotimo tadzik: but it's my own darn fault :(
RabidGravy yeah, rolled back to 84a84f218d1af089a311cb52cfbfa5f264a19e74 and it's all good
tadzik timotimo: schade :( 21:47
lizmat RabidGravy: could you try this gist gist.github.com/lizmat/9e0330b039f482c6ae6a after undoing the revert ? 21:48
RabidGravy not quite sure how that plays in but hey 21:50
sortiz \o 21:51
yoleaux 09:51Z <lizmat> sortiz: make test has always been clean for me: do you have an example that consistently fails?
sortiz lizmat, I'll try the gist, seems related to my problems.
lizmat and another of my heroes bites the dust: en.wikipedia.org/wiki/Keith_Emerson :-( 21:53
21:55 perlawhirl joined
sortiz The gist fails the NC tests in rakudo. 21:56
lizmat huh? it doesn't for me :-( 21:57
sortiz: you mean the ones in t/04-nativecall right ?
RabidGravy lizmat, nope that didn't help 21:58
21:58 kaare_ left
sortiz lizmat, wait, my tree wasn't clean... 21:59
lizmat argh, I messed up my repo 22:00
22:00 skids left
sortiz lizmat, pass all tests AND fixes my issues. 22:02
22:03 perlawhirl left
lizmat RabidGravy: ah, I misread, I thought you reverted 84a84f218d1af089a311c 22:04
RabidGravy: so 69426bd8b3adeffd0a1a23f is the first one that fails? 22:05
RabidGravy: or 02122e5e67d0c973 ?
RabidGravy no, I just rolled back to 84a84f218d1af089a311cb52cfbfa5f264a19e74
so the one after that
22:05 prammer left, prammer joined
AlexDaniel github just added “reactions” thingy 22:06
no more lame “+1” comments!
RabidGravy so 69426bd8b3adeffd0a1a23ff240bb1018bfedca4 22:07
and all after that
lizmat adapted gist: gist.github.com/lizmat/d0594826506ed6d0b1aa
it unnativies some attributes 22:08
RabidGravy so this applies to HEAD?
I'm beginning to get confused 22:09
lizmat to HEAD yes
22:09 Mouq joined
lizmat grrrr.... one test fails in "make", works ok with "perl6" 22:11
I hate that
timotimo so, forgot to make install?!
sortiz lizmat, With Int I'm seeing a Segfault in some of my modules, my be a missing unbox
timotimo o_O 22:14
well, with NativeCall, segfaults are provocable, but still ... :| 22:15
lizmat grrr, FROGGS++ commit is interfering with the revert :-(
RabidGravy lizmat, nope that doesn't help
mst damn contributors, going around contributing things
yoleaux 07:09Z <nine> mst: The 13 seconds for loading Inline::Perl6 are mostly spent on compiling Inline::Perl5 which cannot be precompiled currently. So actually my work in that area will directly benefit Inline::Perl5 and ::Perl6 :)
timotimo mst: and Inline::Perl5 is now precompilable, so it's now much faster! 22:16
mst sweeeet
sortiz lizmat, In my case, the change from $lib2mangler to %lib2mangler works great. 22:17
timotimo tadzik: did you ever try the particles or white_noise examples from my SDL2::Raw repo?
22:18 kid51 joined
tadzik timotimo: nope, they don't sound familiar 22:19
timotimo well, maybe you should have a quick look? :) 22:20
tadzik prolly :)
timotimo hrm. i want a script that lets me say "gimme ./foo.txt" on an ssh server i've connected to and let me say "drophere" on my local machine and the file gets transferred immediately 22:22
bonus points if i can say "drophere" on any machine i'm connected to and it'd still work 22:23
22:30 Mouq left 22:31 TEttinger joined 22:34 jameslenz joined
dalek kudo/nom: 5930d11 | lizmat++ | lib/NativeCall.pm6:
Basically revert 69426bd8b3adeffd0a1a

Not sure what is going on here. Waiting for better times, I guess.
Thanks to RabidGravy++ and sortiz++ for support		 22:38
lizmat RabidGravy: ^^^
jnthn 'night, #perl6 22:39
timotimo wow, whoops, seems like i b0rked the profiler's routine tab a tiny bit
lizmat gnight jnthn 22:40
brb
dalek p: 4091c73 | timotimo++ | src/vm/moar/profiler/template.html:
fix icicle graph in profiler
 22:44
Hotkeys m: say "ha" x 10 ** 10
camelia rakudo-moar 697fb8: OUTPUT«repeat count > 1073741824 arbitrarily unsupported...␤ in block <unit> at /tmp/vXqgbNmnLk line 1␤␤»
Hotkeys arbitrary unsupported?
s/ry/rily/
timotimo yeah, sorry about that 22:45
does anybody fancy a bit of work in angularjs on our profiler? i've got a little task 22:48
22:56 Crw joined
timotimo we don't really divide very efficiently when we have Int and Num for example, maybe i'll add candidates for that to the core setting 22:58
22:59 maybekoo2 left, |Tux| joined
timotimo i wonder if that makes any difference at all to my current use case 23:00
RabidGravy lizmat, in summary none of the above worked
all commits after and including 69426bd8b3adeffd0a1a23ff240bb1018bfedca4 give me the "Missing serialize REPR function for REPR NativeRef" 23:02
timotimo oh jesus, that's pretty bad performance 23:03
find_best_dispatchee is back >_>
sortiz RabidGravy, 69426bd is now reverted. Still with the problem? 23:05
23:05 prammer left
RabidGravy yes 23:06
lizmat hmmm.... 23:07
sortiz Oops!
lizmat RabidGravy: apart from completely nuking your install, I've run out of ideas :-( 23:08
timotimo oh come on, why would infix:<+> be calling find_best_dispatchee all the time? >_> 23:09
or is the profile just being messed up? 23:10
sortiz RabidGravy, Are you sure that the error is at the time of a native call?
dalek kudo/nom: 845a95d | lizmat++ | lib/NativeCall.pm6:
Remove a straggling str
lizmat RabidGravy: perhaps it was this
RabidGravy sortiz, it's on module load 23:11
right now Audio::Sndfile won't load 23:12
sortiz So can't be related to the NC setup changes, unless you are doing a call at compile time.
RabidGravy no it doesn't 23:13
lizmat ok, fwiw, I cannot install Audio::Sndfile either 23:14
Missing serialize REPR function for REPR NativeRef
sortiz Let me take a look...
lizmat so, confirmed that the last patch also doesn't fix this
timotimo i changed something in the core setting, find_best_dispatchee appeared in my profile, i changed it back, it didn't disappear
dalek line-Perl5: d231878 | (Stefan Seifert)++ | META.info:
Version 0.6

Changes:
  * Fix segfault on using exit() in a Dancer2 action
  * Allow for precompilation of Inline::Perl5 itself		 23:15
23:15 SCHAAP137 left
lizmat hmmm... the problem is in JSON::Name ??? 23:18
23:18 sufrostico joined
lizmat and there's no NC involved there ??? 23:18
23:18 perlawhirl joined
lizmat nukes install 23:19
RabidGravy that sounds odd 23:20
the module only has about ten lines, it's just to share a role and a trait used by JSON::Marshal and JSON::Unmarshal
23:21 TreyHarr2 left 23:22 WAAEBGW joined
16WAAEBGW Rakudo build failed. Elizabeth Mattijsen 'Basically revert 69426bd8b3adeffd0a1a 23:22
travis-ci.org/rakudo/rakudo/builds/115439103 github.com/rakudo/rakudo/compare/6...30d11e87f2
RabidGravy okay I've got it 23:23
lizmat RabidGravy++
RabidGravy it's a :ver<v0.0.1> 23:24
hahahahaa
23:26 skids joined
sortiz lizmat, so a complete revert wasn't needed, only the change in %lib2mangler. 23:26
lizmat well, I've had it with this particular line of optimization, so I'm not going to touch it 23:27
RabidGravy: so you're saying there is no issue ?
RabidGravy well, I'd actually call it an issue, inasmuch as this has been like this since November 23:29
so even if it was wrong in the first place it should at least make a nice error message 23:32
lizmat agree
lizmat is tired and goes to bed 23:33
good night, #perl6!
sortiz lizmat, o/
23:34 LGD left
timotimo gnite lizmat! 23:34
23:35 musiKk_ left 23:37 Skarsnik left 23:38 geraud left
timotimo oh wow 23:39
what if ... but fast
timotimo doubled the execution speed of the galton box example 23:40
23:40 cpage_ left 23:43 Vlavv_ left 23:44 rindolf left 23:45 kid511 joined, firstdayonthejob left 23:46 kid51 left 23:47 perlawhirl left, sufrostico left 23:48 sufrostico joined 23:55 Vlavv_ joined 23:56 BenGoldberg joined
timotimo a bit better than double, actually 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!