»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:01 skids joined
lichtkind S02 say nothing 00:01
00:01 fgomez joined
benabik S09 says .bytes exists on buffer-like things. 00:01
diakopter is there a lazy/pump version of split? match? 00:03
benabik S32/Str says Str.bytes is gone and suggests $str.encode($encoding).bytes insteads, and Str.encode says it returns a Buf.
In short, I see no Buf.bytes directly in the spec, but I see several places that says it should exist. :-D
dalek blets: 644c205 | (Herbert Breunung)++ | docs/appendix-a-index.txt:
.bytes is no longer Str method
lichtkind sound sane thank you very much yes i brought it up becasue i saw in git history of spec that .bytes is gone as str 00:04
diakopter .quads would be nice too.
<bump my last question> 00:05
lichtkind true
what you mena
mean
00:06 fgomez left, fgomez joined
dalek blets: ceabc7f | (Herbert Breunung)++ | docs/appendix-a-index.txt:
.bytes example, benabik++
 00:06
timotimo i'm confused. i'm in the debugger and i step and nothing happens 00:07
the line is return gather { some-function-call($args, $more); }; 00:08
it's run for the second time at that point and the first time it stepped into the gather
rurban sent day 11 advent calendar for preview.
timotimo oh, maybe it didn't match one of the multis? 00:09
lichtkind hai reini
should wrote my calendar leaf too 00:10
00:10 ggoebel_ left
timotimo oh, wat? weird. 00:11
DUH, lazy evaluation
i am silly.
00:21 ggoebel joined 00:23 popl joined
lichtkind everyone is sometimes :) 00:26
no need to prag with that
timotimo so, uh, apparently $ doesn't match end of string? does it only match an actual newline?
oh, it *does* match end of string 00:27
i must be missing something obvious again, like a space at the end.
sorear $ matches only the end of string 00:30
it is different from perl 5 in this regard
perl 5 $ sometimes matches end of lines 00:31
lichtkind yes 00:35
multiline mode is gone
$$ is now end of line
timotimo ah, nice. 00:37
00:39 dayangkun joined
timotimo i'm going insane with this super-meta-programmy test framework 00:45
sorear even in default mode perl 5 $ would match at EOL of the last line
timotimo: you may be making it too metaprogrammy.
timotimo i'm trying to raise a project from the dead 00:46
the testing framework is ... special
i'm struggling with refering to methods from a different module that is apparently not accessible from the test framework
it tries to eval('&' ~ $subname) 00:47
because the list of tests is kind of a tree of strings and the test methods are the same strings but with - instead of " " 00:48
i don't understand what's going on. i have a has Foo $.foo handles <some stuff>; and i try to do $.foo = $foo; 00:52
it says it cannot assign to a readonly variable, but i see no readonly anywhere 00:53
skids try has foo is rw
TimToady that probably won't work 00:54
well, maybe it will
timotimo now i get "cannot look up attributes in a type object" and am confused. does that mean the attribute is uninitialised?
WOW
it starts up
skids might have to play with where "is rw" is 00:55
diakopter ++
timotimo i put the is rw and it works, great!
TimToady if you're inside the class, $!foo = $foo won't need the 'is rw' 00:56
if you're outside the class, why are you setting up a delegation outside the class? 00:57
timotimo i'm doing necromancy on some really old code 00:58
TimToady I was originally thinking your were trying to do $.some = $some or $.stuff = $stuff
timotimo in fact, 4 years old.
TimToady in which case the 'is rw' needs to be in the delegated class
diakopter TimToady: is there a lazy/pump version of split (in case memory is very tight and the string is very large)? 01:00
sorear who says normal split won't be lazy? 01:01
diakopter dunno 01:02
is it? 01:04
01:05 anuby joined 01:06 lue joined
TimToady the spec seems to assume that you'd only return something lazy when splitting or combing something that is already lazy 01:07
but it doesn't prohibit lazy behavior when splitting/combing a string 01:08
diakopter since it's a cpu/memory tradeoff, how about an adverb?
(either way)
TimToady since it's a tradeoff, how 'bout a context?
and because we already have contexts :)
(supposedly) 01:09
you can do the split even faster in sink context :)
timotimo so sink is something like eager? i've been wondering what exactly it was ...
TimToady ponders what a hyper or a raced split would look like... O_O 01:10
sink is a variety of eager, yes
diakopter r: split('','' x 9000000000)
p6eval rakudo a308f3: ( no output )
diakopter n: split('','' x 9000000000) 01:11
p6eval niecza v24-5-g599cbcb: OUTPUT«(timeout)»
TimToady which should, of course, warn you about a Useless use of split in sink context 01:12
timotimo what could be the reason for this not working:
TimToady looks like rakudo is just about the point of being able to provide that very useful Useless warning
timotimo regex one { <[a..z]> }; regex two { \d+ }; regex three { <one><two> }; our $.regex = /^ <three> $/; 01:13
the error i'm getting is "no such method three for invocant of type 'Cursor'"
does it have to be "our regex"?
TimToady generally one puts them into a grammar in that case 01:14
timotimo sounds sensible.
TimToady though you can say 'my regex' and then call it as <&three>
but then it doesn't capture $<three> automatically, since there's a punct
so you have to say <three=&three>, which is a bit suboptimal 01:15
std: regex one { <[a..z]> }; regex two { \d+ }; regex three { <one><two> }; our $.regex = /^ <three> $/;
p6eval std a8bc48f: OUTPUT«Potential difficulties:␤ 'regex' declaration outside of grammar at /tmp/43eKOkyz3J line 1:␤------> regex one ⏏{ <[a..z]> }; regex two { \d+ }; regex t␤ok 00:01 45m␤»
TimToady n: regex one { <[a..z]> }; regex two { \d+ }; regex three { <one><two> }; our $.regex = /^ <three> $/;
p6eval niecza v24-5-g599cbcb: OUTPUT«===SORRY!===␤␤Methods must be used in some kind of package at /tmp/HCXWpLpCqu line 1:␤------> regex one ⏏{ <[a..z]> }; regex two { \d+ }; regex t␤␤Methods must be used in some kind of package at /tmp/HCXWpLpCqu line 1:␤-…
timotimo can i do our $.regex = /^ <mygrammar::three> $/ then? 01:16
it seems i can. next failure to hunt ... 01:18
01:18 anuby left 01:23 anuby joined
TimToady maybe <:&three> should capture to $<three>, much like normal :$foo makes foo => $foo 01:28
timotimo okay, no reason to hold back any more 01:29
i'm trying to resurrect the druid implementation masak++ made and last touched 4 years ago
01:29 Exodist joined
timotimo stumbling over all kinds of things 01:31
i suppose that shows how far perl6 has evolved?
TimToady you should make masak do it :) 01:32
timotimo haha
no, i need practice ... i think!
TimToady you do realize that fixing other people's code is what turns a lot of people off on Perl 5? :) 01:33
timotimo because reading perl5 code can be pretty horrifying? 01:35
01:36 araujo left
timotimo i'm at a point where apparently the game can be played 01:37
the test code pretty much doesn't work at all
and display isn't correct, sadly
01:41 japhb_ joined
japhb_ jnthn++ # Junction autothreader optimizations 01:42
timotimo the perl6 vim syntax highlighting seems to have problems with #=[ multiline comments ] 01:44
japhb_ Can't wait to see what he manages for the X issue
timotimo, perhaps it assumes that # comments are single-line?
mikemol has been spending a _lot_ of time in Python for work.
timotimo it makes one extra line pink, though 01:45
mikemol now wants to try putting together an NQP impl in Python.
japhb_ mikemol, Dare we ask for an NQP in PyPy? 01:46
mikemol japhb_: All that'd require is writing it in pure Python, no?
timotimo that would instantly be the fastest perl6 interpreter :|
japhb_ mikemol, well, RPython (Restricted Python). Actually, I guess RPython : Python :: NQP : Perl 6
TimToady keeps wanting a Go implementation for some reason 01:47
mikemol Just as well to get some practice in that...I want to leverage PyPy for work at some point.
timotimo could it be that "next" jumps out of two for loops at once?
01:47 bowtie left, daniel-s_ joined
timotimo huh! 01:49
mikemol OK, if I want to study NQP by studying its implementations, where's the best place to start? 01:50
01:51 japhb_ left, daniel-s left
TimToady you should ask jnthn that, but he's asleep, like you're supposed to be :) 01:51
mikemol Me? Supposed to be asleep? I'm in GMT-5. :) 01:52
TimToady for some reason I have you in .de mentally 01:53
mikemol Airport code GRR.
timotimo i'm in .de, and i was right next to him in almost every of the last lines
01:54 cognominal_ joined, japhb_ joined
sorear mikemol: that is a great name for an airport 01:54
japhb_ Wheee, managed to overheat laptop until it locked up by compiling Rakudo
timotimo did this ever work? if so, how :| 01:56
01:56 cognominal left
TimToady timotimo: if it ran on ancient rakudo, it probably assumes a lot of stuff is in packages that no longer is 01:56
timotimo do you mean packages as in libraries, or packages as in "what's in a .pm file"? 01:58
japhb_ timotimo, And 4 years ago is an eternity in Rakudo Time
TimToady ancient rakudo tended to put things into packages more like Perl 5, not like Perl 6
01:58 cognominal joined
timotimo this code here seems wrong, but i can't really put my finger on it 01:58
did something about Array.kv drastically change in the last 4 years?
01:59 cognominal_ left
TimToady don't think so 02:00
diakopter TimToady: re fixing other people's code - it's how I earned a Perl 5 living for 4 years. loved it++
TimToady yer just weerd 02:01
diakopter but how much I've forgotten...
mikemol Is niecza really only 28k lines?
timotimo i don't really know how this could have ever been considered correct :|
diakopter mikemol: also see the .cs files
mikemol in /lib, find . -type f -name '*.cs' -print0|xargs -0 wc -l
timotimo or maybe i'm looking at the expected output and am expecting something different 02:02
mikemol I only counted the .cs files.
diakopter oh, there's lots of Perl 6 too
TimToady including the parser :)
timotimo the 3d view doesn't seem to be drawing correctly 02:03
other than that the game seems playable now
i think i'll just throw it up on github and ask masak++ to do the rest?
TimToady that could be changes in print/say semantics (gist and such)
sorear there's also 4k of perl 6 in .setting files
timotimo nah, it's rendering the board fine, what doesn't work is apparently looping over the layers and columns and such 02:06
when i make the "obvious and wrong" fix, it spews lots and lots of errors at me
TimToady is it using unary hypers? those have fishtailed a bit in semantics over time 02:07
also, old rakudo didn't really do Parcels right, or at all, depending
timotimo nah, it's super simple and i think i'm just ten times as tired and twice as stomach-achey as required for understanding perl6 good enough 02:12
02:12 japhb_ left
TimToady relying on order of side effects for a hyper say would also be a problem 02:12
timotimo well, it's simply about the loop body not being executed 02:16
i'll show you the lines if you like
02:17 FROGGS_ joined
timotimo github.com/timo/druid/blob/master/...ext.pm#L92 - this line seems to skip over all of the layers 02:17
02:18 lichtkind_ joined
timotimo @layers, which is assigned in the first line of that function looks like this: 02:18
Array.new([[0, 0, 0, 0, 0, 1, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0], [...], ... [0, 0, 0, 0, 0, 0, 0, 0])) 02:19
02:20 lichtkind left, FROGGS left 02:21 daniel-s__ joined 02:22 Exodist left, exodist_ joined
timotimo so $line would be 7 at the beginning and $row would be [0, 0, 0, 0, 0, 0, 0, 0] 02:22
and some time it would have a 1 in there to signal that there's a block there 02:23
but $line.kv would just be ((0, 7),) :(
changing it to $row.kv.reverse -> $cell, $column was not helpful at all
02:23 araujo joined 02:24 daniel-s_ left
timotimo what a terrible night to have a curse. 02:25
02:30 _daniel-s__ joined 02:31 ggoebel_ joined 02:34 daniel-s__ left, terribleperson joined
timotimo oh 02:35
there already was a "ng-compat" branch that was probably there to make it moderner
02:39 wamba left
timotimo oh, interesting 02:41
there's a SORRY but apparently no filename? 02:42
probably just the filename i passed on the commandline then
no, actually not.
good night happy perl6 people :) 02:44
diakopter o/
02:45 bonsaikitten joined 02:47 terribleperson left
lichtkind_ good night 02:50
02:50 lichtkind_ left 02:52 bonsaikitten is now known as DrEeevil 02:55 hdcool09 joined 02:57 ggoebel_ left 03:21 cognominal left 03:24 cognominal joined 03:35 terribleperson joined 03:50 colomon joined, cognominal left 03:56 cognominal joined 03:59 cognominal_ joined 04:02 cognominal left, noggle left 04:03 noggle joined 04:04 cognominal_ left 04:10 cognominal_ joined, sisar joined 04:17 s1n joined 04:29 cognominal_ left 04:37 cognominal_ joined 04:39 sisar left 04:40 Patterner left 04:41 cognominal_ left 04:43 am0c left 04:45 cognominal joined 04:53 terribleperson left 04:55 terribleperson joined 04:59 dayangkun left 05:05 terribleperson left 05:06 skids left 05:17 sftp left 05:19 telex left 05:21 bluescreen100 left 05:22 bluescreen10 left 05:23 rummik left, rummik joined 05:27 telex joined 05:44 xinming joined 05:48 xinming_ left
dalek kudo/nom: 7797039 | pmichaud++ | docs/parrot-relationship.txt:
Add docs/parrot-relationship.txt, which codifies the relationship manager roles between the Rakudo and Parrot projects.

This was adopted in June 2011, but apparently never made it into either of the repositories. So I'm adding it to the Rakudo repository now.		 06:03
06:08 kaleem joined 06:12 kurahaupo left 06:23 Rix joined 06:24 dayangkun joined 06:28 cognominal left 06:34 cognominal joined 06:43 whitefaced joined 06:44 popl left
dalek ast: 3ac110d | moritz++ | S06-traits/is-readonly.t:
fudge new, failing test for read-only attributes
 06:49
whitefaced codepad.org/SePMyCQR why does the if statement at line 27 cause my program to exit every time, even when i enter input to stdin 06:50
codepad.org/0BvMdcrN 06:51
^
moritz whitefaced: that looks like Perl 5. This channel is about Perl 6. Please try #perl or #perl-help or so 06:52
whitefaced perl wouldnt help me
so i had to resort to here
sorry
moritz so you thought you'd just ignore the 6 in #perl6? :-)
you can also try irc.perl.org
whitefaced ok 06:53
i will
:D
07:00 kaare_ joined 07:22 grondilu joined 07:27 kaare_ left 07:28 whitefaced left 07:32 wamba joined 07:58 samuelvasy joined 08:01 brrt joined 08:04 hdcool09 left 08:05 hdcool09 joined, bphillips left 08:09 bphillips joined 08:24 grondilu left 08:28 marloshouse_ joined, marloshouse_ is now known as marloshouse 08:31 kurahaupo joined, kurahaupo left
jnthn morning, #perl6 08:32
cedrvint o/ jnthn
moritz \o
cedrvint hello #perl6, do you know if there's an upcoming event about Perl 6 in France ? 08:33
moritz jnthn: thanks for the sinking patches, all spectests now pass
brrt hi
08:33 Kharec joined
moritz jnthn: but there's still a known sink bug that I want to track down 08:33
jnthn moritz: yay. And oK :) 08:39
08:41 samuelvasy left 08:43 rummik left 08:49 rummik joined
arnsholt cedrvint: I don't know of any, but cognominal might know 08:50
cedrvint arnsholt: thanks, I'm waiting for his reply then. 08:52
moritz nr: say gather { 42 } 08:53
p6eval rakudo 779703, niecza v24-5-g599cbcb: OUTPUT«␤»
benabik nr: say gather { take 42 } #? 08:56
p6eval rakudo 779703, niecza v24-5-g599cbcb: OUTPUT«42␤»
08:57 brrt left 08:59 kurahaupo joined
cedrvint r: gist.github.com/4257198 09:04
p6eval rakudo 779703: OUTPUT«foo with constraints␤»
cedrvint any idea why infix:<cmp> with constaints isn't call?
*called
moritz cedrvint: because there's an infix:<cmp>(Int, Int) in the setting 09:05
cedrvint: and nominal types bind tighter than contraints
cedrvint moritz: thanks!
moritz write it as 09:06
multi infix:<cmp>(R:D $a, R:D $b) { ... }
felher Good morning, #perl6 09:07
09:08 dayangkun left, wamba left
jnthn o/ felher 09:09
cedrvint moritz: I get this message "Ambiguous call to 'infix:<cmp>'; these signatures all match: \n :(Int:D \a, Int:D \b) \n :(R:D $a, R:D $b)"
moritz grml 09:10
try
R:D $a where { 1 }, R:D $b where { 1 }
cedrvint moritz++ # it works \o/ 09:11
this looks a little bit "artificial" though, is it possible to use "is tighter(...)" here? 09:12
moritz no 09:13
but you can say 'is default' after the signature
cedrvint thanks, it works like a charm too. 09:14
09:14 cognominal left 09:16 hx107 joined 09:17 hx107 left, kst` joined 09:18 LordV joined, Maddingue joined, cognominal joined 09:19 telex left 09:21 telex joined, kst left 09:28 kaleem left 09:29 kaleem joined
cedrvint the post for today isn't published yet. 09:35
irclog.perlgeek.de/perl6/2012-12-11#i_6230127 09:36
moritz publishes it 09:38
cedrvint moritz: maybe the "editor's note" should be removed (at the top)? 09:42
cedrvint isn't sure
09:42 kivutar joined
moritz cedrvint: me neither. If rurban doesn't want it published, he just needs to remove the note from the article 09:43
09:46 dakkar joined 09:47 fhelmberger joined
hoelzro I'm wondering if the "30,000 foot overview of Perl 6" really merits an advent calendar post 09:49
I'm thinking something flashier and more interesting should take its place
cedrvint hoelzro: why?
hoelzro and I should just write up a series of articles about it later
cedrvint: I've been looking into it, and while it's interesting, I don't think it will really "hook" people on to the language
cedrvint hoelzro: it could help to understand what's Perl 6 (langage, implementations, ...) 09:51
hoelzro I just feel that most people know that already; maybe that's presuming too much, though
cedrvint it depends on who these "people" are, I guess most people in #perl6 already have an overview of Perl 6 09:54
but it's not the case in my compagny for instance :) 09:55
(and they are curious)
09:55 cognominal left
hoelzro is it advent calendar material, though? 09:56
I look at some advent calendar posts and think "wow, that's really cool!"
I feel like the advent calendar stuff should get people interested, and then they can read up on it
jnthn hoelzro: Yes, but somebody looking at a bunch of the posts and finding them interested won't automatically know exactly what is what, or how to join in, get hold of stuff, etc. 09:57
cedrvint troo
hoelzro hmm 09:58
I would think that someone interested in Perl 6 after reading a post would just go perl6.org, and start there
10:01 kurahaupo left, cognominal joined
moritz well, if you want to add an overview article to perl6.org, that would be great too 10:03
hoelzro I was thinking of digging into Parrot/NQP/Rakudo and trying to write a basic overview on how things work 10:06
so people interested in contributing find it easier to do so
10:26 SmokeMachine joined 10:29 wamba joined
timotimo huh, there's hyphenation in the middle of the advent calendar post: "since the inter- preter's state is" and "consistency is guarant- teed" 10:38
"taking and releasing locks costs per- formance" 10:39
"a sub does not only con- tain", is this in the source or is my browser acting up? 10:41
10:51 cognominal left 10:52 wamba left, wamba joined 10:53 SamuraiJack joined 11:03 cognominal joined 11:05 Su-Shee_ joined 11:09 Su-Shee left 11:15 hdcool09 left 11:21 anuby left 11:29 birdwindupbird joined 11:30 grondilu joined
grondilu rn: sub f { .(4) given -> $n { $n > 1 ?? &?ROUTINE($n-1)*$n !! 1 } }; say f # expecting 24 11:31
p6eval niecza v24-5-g599cbcb: OUTPUT«(timeout)»
..rakudo 779703: OUTPUT«Too many positional parameters passed; got 1 but expected 0␤ in sub f at /tmp/2AK_oUMygM:1␤ in sub f at /tmp/2AK_oUMygM:1␤ in block at /tmp/2AK_oUMygM:1␤␤»
jnthn grondilu: &?ROUTINE refers to the f 11:32
grondilu indeed, but shouldn't it refer to the anonymous sub?
jnthn No, because it's not a sub, but just a block.
r: say -> { } ~~ Routine
p6eval rakudo 779703: OUTPUT«False␤» 11:33
grondilu rn: sub f { .(4) given -> $n { $n > 1 ?? &?BLOCK($n-1)*$n !! 1 } }; say f # expecting 24
p6eval rakudo 779703: OUTPUT«No such method 'Nil' for invocant of type 'Int'␤ in at src/gen/BOOTSTRAP.pm:839␤ in any at src/gen/BOOTSTRAP.pm:836␤ in sub f at /tmp/SoPBpb2UJY:1␤ in block at /tmp/SoPBpb2UJY:1␤␤»
..niecza v24-5-g599cbcb: OUTPUT«24␤»
jnthn That one should work, apart from I'm pretty sure Rakudo doesn't implement &?BLOCK yet
grondilu ok
11:33 Su-Shee_ is now known as Su-Shee
timotimo trying out bailador and the test suite 05-post-content runs fine in a debugger but not when run by prove or rakudo standalone 11:33
tadzik ack 11:34
jnthn See? it really does de-bugger your programs :P
timotimo it *may* be a version discrepancy between my perl6 binary and the perl6-debug one, though
perl6-debug says "this is perl6 version built on parrot 4.6.0 revision 0", but perl6 is "version 2012.11", so ...
jnthn Hm, that looks wrong 11:35
(the perl6-debug version output)
timotimo hm, it's from the same directory, though.
jnthn yeah, I think it's version reporting is just wrong. 11:36
timotimo tadzik: the error in question is that request.env<psgi.input>.decode claims that decode is called on an Any.
tadzik yeah, I remember this one
timotimo so, just a too old rakudo version? 11:37
tadzik please check
timotimo sure
tadzik I don't think this bug is fixed
cedrvint rn: gist.github.com/4257923 11:38
p6eval rakudo 779703: OUTPUT«debug␤True␤False␤»
..niecza v24-5-g599cbcb: OUTPUT«debug␤Unhandled exception: Cannot call &infix:<cmp>; none of these signatures match:␤ C where { ... }, C where { ... };; |$c where { ... }␤ at /tmp/bjRMGun9mw line 7 (infix:<cmp> @ 8) ␤ at /tmp/bjRMGun9mw line 9 (mainline @ 12) ␤ at /home/p6eval/niecz…
timotimo bleh, i think i'll throw out rakudo star and get panda manually. should be pretty easy, right? 11:39
moritz right
tadzik panda should be able to install rakudo-debugger now too :)
timotimo that's excellent 11:40
cedrvint moritz, jnthn: I thought infix:<..> was generic as it relies on infix:<cmp> (overloaded for my class) 11:42
11:42 rindolf joined
cedrvint does Rakudo implement S03#3389? 11:42
timotimo tadzik: with a very current rakudo it still fails that test 11:44
cedrvint *S03#line_3389 11:47
11:47 noggle left
cedrvint doesn't know how to link to a given spec line. 11:48
11:49 arlinius left
moritz cedrvint: I think it uses cmp by default, but short-cuts for strings and integers 11:49
cedrvint I can't get infix:<cmp> indirectly called by infix:<...> in gist.github.com/4257923 11:51
*infix:<..>
moritz r: gist.github.com/4257923
p6eval rakudo 779703: OUTPUT«debug␤True␤False␤»
cedrvint was not able to find any test-case in roast about that 11:52
timotimo r: gist.github.com/4257997
p6eval rakudo 779703: OUTPUT«debug␤Type check failed for return value; expected 'Bool' but got 'Order'␤ in sub infix:<cmp> at /tmp/jmOdLX_5pn:7␤ in block at /tmp/jmOdLX_5pn:9␤␤»
timotimo er, yeah
r: gist.github.com/4257997 11:53
p6eval rakudo 779703: OUTPUT«debug␤Increase␤False␤»
timotimo indeed
11:55 wamba left, cognominal left
timotimo oh 11:56
11:57 cognominal joined
timotimo i can run the test with perl6, but with panda it breaks? something's terribly off here. 11:57
moritz panda precompiles
grondilu rn: multi talk { callwith "hello" }; multi talk($msg) { say $msg }; talk; 11:58
p6eval rakudo 779703: ( no output )
..niecza v24-5-g599cbcb: OUTPUT«Unhandled exception: Cannot call &talk; none of these signatures match:␤ Any␤ Any␤ at /tmp/Hvxsa4Xiyt line 1 (mainline @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4208 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4209 (module-C…
timotimo moritz: good point. i added lib/ but not blib/ to P6LIB
tadzik timotimo: yeah, that's my experience too. I think it should be skipped, and I'll file a bug and assign it to flussence++ :P
jnthn grondilu: callwith iterates through a *predetermined* candidate list (namely, the list of available candidates which take no arguments, whihc is just one). 11:59
Not re-dispatches each time.
If you want that, just call talk(...) again :)
moritz timotimo: you need to add blib/lib to PERL6LIB
timotimo yes, i did that now
and that's exactly what breaks it, blib vs lib 12:00
arnsholt jnthn: Ooh, that's the confusing change that was made to the {call,next}{same,with} semantics, isn't it?
timotimo but how can my web app be webscale if i don't precompile it with rakudo before deploying it? 12:01
grondilu ok I'll use the name of the subroutine but I would have liked something more generic, and &?ROUTINE would not redispatch :/ 12:02
jnthn arnsholt: No, I don't think it's ever been specified to re-dispatch. 12:03
arnsholt: It's always been iterator-y
tadzik gotta love sloccount and it's COCOMO calculations
arnsholt Right. My bad then
tadzik Total Estimated Cost to Develop = $ 18,216
yeah, sure
felher rurban++ # very long blog post. Hope I will get around to read it after $job
timotimo tadzik: that's for bailador? :) 12:04
tadzik timotimo: nah, just something at $dayjob :)
timotimo heh
12:04 crab2313 joined
tadzik Bailador is just $ 11,328 12:05
12:06 crab2313 left
jnthn wonders what the debugger is :) 12:08
moritz tadzik: and nqp and rakudo?
cedrvint www.ohloh.net/p/rakudo/estimated_cost 12:09
www.ohloh.net/p/nqp/estimated_cost
moritz parrot: 4M 12:11
sorear www.ohloh.net/p/niecza says that it contains 8 man-years of code D:
cedrvint sorear: how many men are you?
tadzik moritz: nqp + parrot + rakudo is $ 20,372,602 :) 12:13
150 person-years
I wonder if it says person-years to be politically correct and avoid sexism accusations
cedrvint agrees 12:14
timotimo huh? no, everybody knows that only men are persons!
cognominal the important thing is not the cost but the value, and that, you can't calculate. 12:15
sorear one small step for mankind.
jnthn My magical coding robot is offended at the use of "person-years", and wishes it was "coding-entity-years" :P
tadzik personkind, you sexist pig! :D
sorear apparently I need to be more men, because ohloh is complaining about decreasing year over year cmmits 12:16
oh hey, created by cedrvint 12:17
sorear wondered if ohloh was algorithmically scraping the web for projects
cedrvint :)
sorear sleep& 12:18
12:24 sftp joined
grondilu rn: package Foo { proto talk {*}; multi talk {...}; our talk($msg) { say $msg } }; Foo::talk "hello"; 12:26
p6eval rakudo 779703: OUTPUT«===SORRY!===␤Malformed our␤at /tmp/qgWOh83qB_:1␤------> { proto talk {*}; multi talk {...}; our ⏏talk($msg) { say $msg } }; Foo::talk "he␤ expecting any of:␤ scoped declarator␤»
..niecza v24-5-g599cbcb: OUTPUT«===SORRY!===␤␤Malformed our at /tmp/9pQmveN7ZB line 1:␤------> { proto talk {*}; multi talk {...}; our⏏ talk($msg) { say $msg } }; Foo::talk "h␤␤Parse failed␤␤»
grondilu rn: package Foo { proto talk {*}; multi talk {...}; our multi talk($msg) { say $msg } }; Foo::talk "hello"; 12:27
p6eval rakudo 779703: OUTPUT«Could not find symbol '&talk'␤ in method <anon> at src/gen/CORE.setting:10204␤ in at src/gen/Metamodel.pm:2404␤ in any find_method_fallback at src/gen/Metamodel.pm:2392␤ in any find_method at src/gen/Metamodel.pm:925␤ in at src/gen/BOOTSTRAP.pm:852␤ in any …
..niecza v24-5-g599cbcb: OUTPUT«Potential difficulties:␤ &talk is declared but not used at /tmp/FZF6rCXpEN line 1:␤------> package Foo { proto talk ⏏{*}; multi talk {...}; our multi talk($m␤␤hello␤»
jnthn grondilu: Putting "our" on a single candidate makes no sense (and should probably be an error)
grondilu rn: package Foo { our proto talk {*}; multi talk {...}; multi talk($msg) { say $msg } }; Foo::talk "hello";
p6eval rakudo 779703: OUTPUT«Too many positional parameters passed; got 1 but expected 0␤ in sub talk at /tmp/vRndARrONC:1␤ in block at /tmp/vRndARrONC:1␤␤»
..niecza v24-5-g599cbcb: OUTPUT«Potential difficulties:␤ &talk is declared but not used at /tmp/jDQ4qhG39L line 1:␤------> package Foo { our proto talk ⏏{*}; multi talk {...}; multi talk($msg) ␤␤Unhandled exception: System.NullReferenceException: Object reference not s…
jnthn And now your proto is not liberal enough :)
grondilu oh yeah 12:28
but what if I want a user of a module to access a version of a function, but not an other?
jnthn I think you can selectively export, but if you export the proto that's all or nothing. 12:29
Well, apart from here you're not exporting.
But relying on our semantics.
I'd say it's more Perl 6-ish to export, imho.
grondilu ok
jnthn But there's not a way to have just some candidates visible on the outside and others on the inside. 12:30
(with the our approach)
12:33 dalek left 12:34 dalek joined, ChanServ sets mode: +v dalek
moritz r: gist.github.com/4258258 12:35
p6eval rakudo 779703: OUTPUT«("A", "B", "Z").list␤»
moritz that produces ("Z", "A", "B").list in qast-sink-1 12:36
timotimo just trying out random perl6 modules right now. Inline gives "obsolete use of | or \ with sigil on param $arr" and SDL gives "cannot import symbol &trait_mod:<is> from Inline because it already exists in this lexical scope &trait_mod:<is>" :(
i'm not sure if the last one should be a bug on SDL or on Inline?
FROGGS_ timotimo: it's a bug in rakudo, and it's known
timotimo OK 12:37
jnthn moritz: What if you make it gatehr sink for ... ? 12:40
*gather
moritz jnthn: doesn't make a difference 12:41
jnthn: the problerm is that the inner 'take $_ for @$_;
'
isn't run early enough
if I add 0; after that loop, all is fine 12:42
so I guess the gather block is not sunk
oh, I guess I need to patch up sub coro for that 12:43
in PIR :(
jnthn moritz: See the PIR patching up I did in MapIter the other day for clues
moritz jnthn: that looks roughly like what I would have written myself 12:45
now that nqp::can works with modules and roles, I wonder if I should remove the check for repr_defined 12:46
sergot Could somebody install JSON::Tiny on feather? ;] 12:48
moritz sergot: you should be able to install it in your home dir
clone panda, then perl6 bootstrap.pl
jnthn moritz: It's a fast check.
12:48 cognominal left
jnthn moritz: I re-ordered the checks to make sure we do them in speed order 12:49
12:49 cognominal joined
jnthn repr_defined should be faster than can should be faster than defined 12:49
moritz jnthn: ok
sergot moritz: I'm, but I want to use cronjobs, and then it have to be installed on the whole server I think.
moritz sergot: cron jobs for which purpose? 12:50
sergot moritz: generating feather.perl6.nl/~sergot/modules
moritz sergot: then it should be enough to set a few environment variables for the cron job
like PATH, HOME, maybe some others 12:51
sergot moritz: thanks a lot. :) 12:52
12:53 SamuraiJack left
moritz jnthn: my patch didn't help. And the odd thing is, the gather block seems to be in sink context already 13:02
./perl6 -e 'eager do gather for ^2 { take $_; .say for 42 }'
prints the 42 twice
just as it should 13:03
so it must be in sink context
13:07 wamba joined
moritz hm 13:10
maybe the problem is that some part in there is a Parcel and not a list
and there is no Parcel.sink yet
jnthn moritz: just to check - it's certainly about gather, not about the for as a statement modifier?
moritz jnthn: for as statement modifier generally work 13:11
sergot rakudo is really fast now... ! \o/
moritz though I'll try it with a normal loop, just to be sure
no, doesn't make a difference 13:14
bphillips moritz: I was told you were the smartmatch "resident expert" yesterday. Does this behavior seem right? gist.github.com/4253309 (it goes into the "if" block on each loop iteration) 13:18
r: gist.github.com/4253309
p6eval rakudo 779703: OUTPUT«Matched! c against /a/b/1/d/:e: (match was :c)␤Matched! d against /a/b/1/d/:e: (match was :c)␤Matched! e against /a/b/1/d/3: (match was :e)␤»
moritz bphillips: rakudo currently returns the (possibly modified) string from s///. I believe that's not quite right, but I'm not sure at the moment 13:19
bphillips so is there no way to tell if the smartmatch actually succeeded when using s/// ? 13:20
timotimo r: say $_ for 1, 2, 3;
p6eval rakudo 779703: OUTPUT«1␤2␤3␤»
timotimo hm.
moritz bphillips: no
I should fix that
bphillips also, $/ seems to persist over multiple ~~ s/// operations (specifically, it's not reset when one fails) 13:21
timotimo r: class A { multi method push(@a) { .push $_ for @a; }; multi method push($a) { say $a } }; A.new.push: (1, 2, 3);
p6eval rakudo 779703: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/Qe1k1s0qsD:1␤------> class A { multi method push(@a) { .push ⏏$_ for @a; }; multi method push($a) { sa␤ expecting any of:␤ method arguments␤ postfix␤ infix or meta-…
bphillips that also seems not quite right
moritz timotimo: .push $_ is wrong
timotimo that's what rakudo keeps telling me :D 13:22
moritz must .push($something) or .push: $something
timotimo OK
bphillips moritz: is this something implemented in perl6 (vs. pir) that I could try my hand at fixing? (I've not really delved into perl6 internals but I'd love to give it a try)
moritz bphillips: yes, it is
bphillips: in particular, look at method subst in src/core/Str.pm 13:23
and method quote:sym<s> in src/Perl6/Actions.pm is the codegen for s///
timotimo moritz: i'm getting a "unable to parse expression in block; cound't find final }" self.push <HERE>($_) for $sg.node.v;
bphillips moritz: awesome! I'll take a look 13:24
timotimo but with a : instead of () it works
moritz timotimo: looks like you have a space between 'push' and '('
timotimo so i'll take that
or, derp, yes.
[Coke] misquotes moritz: "jnthn: thanks for all the stinking patches."
bphillips moritz: Is there a newb guide to checking out the spectests, modifying a test and running one? 13:27
moritz has a stinking feeling in his stomach that this is going to be a hard problem
bphillips: 'make testable'
bphillips moritz: this subst issue is going to be a hard problem, you mean? (or something else) 13:31
moritz bphillips: no, my remaining error in the sink branch
bphillips ah, ok :-)
moritz and I was mostly punning on sink/stink to please [Coke] ;-)
jnthn :) 13:32
moritz: Yeah, tricky...
timotimo r: class A { method new(@items) { .say for @items } }; A.new: [1, 2, 3, 4];
p6eval rakudo 779703: OUTPUT«1␤2␤3␤4␤»
timotimo r: class A { method new(@items) { .say for @items } }; A.new: :items([1, 2, 3, 4]); 13:36
p6eval rakudo 779703: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in method new at /tmp/Swv9Kv9ZfP:1␤ in block at /tmp/Swv9Kv9ZfP:1␤␤»
timotimo it would be cool if it said "but you passed 1 named argument where 0 were expected.", too :P 13:37
moritz method always accept any named parameters
timotimo oh? and if they're not in the signature, they're just silently dropped? 13:38
moritz yes
timotimo is this only for method? what about multi/sub/submethod?
moritz only for methods
jnthn For submethod too, iirc 13:39
timotimo interesting. what's the idea behind it?
jnthn See interface consistency section of S12, but essentially it's to make deferal more useful
timotimo thanks 13:40
13:41 wamba left
moritz thinks this is a good candidate for the FAQs 13:46
13:48 pmurias joined 13:52 mtk left
timotimo r: class A { has %.n; method new(@it) { $.n{$_} = $_ for @it; } }; my $a = A.new([1, 2, 3]); say +$a.n; 13:53
p6eval rakudo 779703: OUTPUT«Cannot look up attributes in a type object␤ in method n at src/gen/CORE.setting:1798␤ in method new at /tmp/8r4QeGkCxk:1␤ in block at /tmp/8r4QeGkCxk:1␤␤»
cedrvint r: role R { }; multi sub infix:<..> (R:D $a, R:D $b) { !!! }; 1 .. 42; 13:55
p6eval rakudo 779703: OUTPUT«===SORRY!===␤CHECK FAILED:␤Calling 'infix:<..>' will never work with argument types (Int, Int) (line 1)␤ Expected any of:␤ :(R:D $a, R:D $b)␤» 13:56
cedrvint why infix:<..>(Int, Int) isn't available anymore?
n: role R { }; multi sub infix:<..> (R:D $a, R:D $b) { !!! }; 1 .. 42; 13:57
p6eval niecza v24-5-g599cbcb: OUTPUT«Potential difficulties:␤ $a is declared but not used at /tmp/_sb7pXOA5Z line 1:␤------> role R { }; multi sub infix:<..> (R:D ⏏$a, R:D $b) { !!! }; 1 .. 42;␤ $b is declared but not used at /tmp/_sb7pXOA5Z line 1:␤------> { }; mult…
cedrvint n: role R { }; multi sub infix:<..> (R:D $a, R:D $b) { +$a .. +$b }; 1 .. 42;
p6eval niecza v24-5-g599cbcb: OUTPUT«Unhandled exception: Cannot call &infix:<..>; none of these signatures match:␤ R where { ... }, R where { ... };; |$c where { ... }␤ at /tmp/V1KSTbtsch line 1 (mainline @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4208 (ANON @ 3) ␤ at /home/p6e…
cedrvint rn: role R { }; 1 .. 42; 13:58
p6eval rakudo 779703, niecza v24-5-g599cbcb: ( no output )
timotimo r: class A { has %.n is rw; method new(@it) { $.n{$_} = $_ for @it; } }; my $a = A.new([1, 2, 3]); say +$a.n; 14:00
p6eval rakudo 779703: OUTPUT«Cannot look up attributes in a type object␤ in method n at src/gen/CORE.setting:1786␤ in method new at /tmp/YkNkK4lh0s:1␤ in block at /tmp/YkNkK4lh0s:1␤␤»
timotimo r: class A { has %.n is rw; method new(@it) { $.n{$_} = $_ for @it; } }; my $a = A.new([1, 2, 3]); say $a;
p6eval rakudo 779703: OUTPUT«Cannot look up attributes in a type object␤ in method n at src/gen/CORE.setting:1786␤ in method new at /tmp/A2uAhYBDBF:1␤ in block at /tmp/A2uAhYBDBF:1␤␤»
moritz timotimo: %.n isn't initialized in .new
timotimo oh, that would explain it
moritz that's done in BUILD
see doc.perl6.org/language/objects#Obje...nstruction or perlgeek.de/blog-en/perl-6/object-c...ation.html 14:01
pmurias what's the difference between nqp::list and nqp::qlist?
timotimo thanks :)
moritz pmurias: nqp::qlist return a QRPA
which is designed to be more performant in some cases, but I don't remember which :(
jnthn shift 14:02
and some others, but shift is the biggie
pmurias what's a QRPA? 14:03
jnthn: it has the same semantics but different performance? 14:04
jnthn pmurias: yeah, the idea is that we use it for nqp::list in the future and then nqp::qlist goes away 14:06
pmurias: It's just a transitional thing.
Basically, Rakudo's iterators need shift to be efficient.
And QRPA optimizes for that. 14:07
14:07 dakkar left
jnthn But I don't think it really makes anything else slower either... 14:07
14:07 dakkar joined
pmurias maps both to js Array 14:07
timotimo i'm not sure how to code this: i would like to have my class Graph take a list of things, like Node and Edge in its .new as a list and i call .push on that list to assign them to the proper attributes %.nodes and @.edges; how do i make the positional argument available to BUILD without having to create a new attribute for my Graph class
14:08 kaleem_ joined
jnthn method new(@things) { self.bless(*, :@things) }; submethod BUILD(:@things) { @!edges = @things.grep(Edge); ... } 14:09
moritz or split up @things into my @nodes, my @edges in method new 14:10
and then call self.bless(*, :@nodes, :@edges);
(though that's not very friendly for subclassing)
timotimo hm, i have a multi method "push" that does the sorting, i'd like to not duplicate that in the constructor if possible :|
moritz timotimo: then put the code somewhere where you can use it both from push and from new 14:11
timotimo i'll probably add more different cases, too. like maybe a bare Str would become a new or existing Node instead
oh, yeah
14:11 kaleem left
moritz sounds like a good task for multi subs 14:11
timotimo well, i'll save it for later and just do named arguments for new until i find an elegant way to structure all of that
cedrvint I need your help for my Advent post: why I can't create a Range of Int anymore once I did: 14:13
r: multi sub infix:<..> (Str $a, Str $b) { !!! }; 1 .. 42
p6eval rakudo 779703: OUTPUT«===SORRY!===␤CHECK FAILED:␤Calling 'infix:<..>' will never work with argument types (Int, Int) (line 1)␤ Expected any of:␤ :(Str $a, Str $b)␤»
jnthn r: say &infix:<..>.is_dispatcher 14:14
p6eval rakudo 779703: OUTPUT«False␤»
moritz because the infix:<..> in the setting isn't a multi sub
jnthn Apparently, 'cus infix:<..> is not declared multi in the setting..
cedrvint is it a bug in the setting?
14:15 PacoAir joined
jnthn I'm a bit surprised it's this way, at least... 14:15
Maybe the spec says otherwise.
moritz r: multi sub infix:<..>($a, $b) { CORE::infix:<..>($a, $b) }; multi sub infix:<..> (Str $a, Str $b) { !!! }; 1 .. 42
p6eval rakudo 779703: OUTPUT«No such method 'Any' for invocant of type 'Parcel'␤ in at src/gen/BOOTSTRAP.pm:845␤ in at src/gen/BOOTSTRAP.pm:839␤ in any at src/gen/BOOTSTRAP.pm:836␤ in sub infix:<..> at /tmp/WTbUa8BQNs:1␤ in block at /tmp/WTbUa8BQNs:1␤␤»
14:16 bluescreen10 joined, fgomez left
moritz r: multi sub infix:<..>($a, $b) { Range.new($a, $b) }; multi sub infix:<..> (Str $a, Str $b) { !!! }; 1 .. 42 14:17
p6eval rakudo 779703: ( no output )
moritz r: multi sub infix:<..>($a, $b) { Range.new($a, $b) }; multi sub infix:<..> (Str $a, Str $b) { !!! }; say 1 .. 42
p6eval rakudo 779703: OUTPUT«1..42␤»
jnthn r: multi sub infix:<..>($a, $b) { &CORE::infix:<..>($a, $b) }; multi sub infix:<..> (Str $a, Str $b) { !!! }; say 1 .. 42
p6eval rakudo 779703: OUTPUT«1..42␤»
jnthn r: multi sub infix:<..>($a, $b) { &CORE::infix:<..>($a, $b) }; multi sub infix:<..> (Str $a, Str $b) { !!! }; say "1" .. "42"
p6eval rakudo 779703: OUTPUT«Stub code executed␤ in sub infix:<..> at /tmp/aPYneBDZlu:1␤ in block at /tmp/aPYneBDZlu:1␤␤»
moritz so it's CORE::... vs. &CORE:: ? 14:19
cedrvint as of my understanding, there's no need for a "multi sub infix:<..>" since it *should* rely on infix<cmp> for user's types. 14:20
this is not the case, sadly.
irclog.perlgeek.de/perl6/2012-12-11#i_6231704 14:21
jnthn cedrvint: That's odd because I was looking at that code yesterday and it does call cmp
cedrvint: Oh...were your objects really Int or Num + a role? 14:22
pmurias jnthn: do iterators support operations besides nqp::shift?
dalek kudo-js: 274e07a | (Paweł Murias)++ | wrap_nqp:
IO::CaptureOutput optional while building
kudo-js: 4308243 | (Paweł Murias)++ | / (2 files):
[nqp] start of exception handling
kudo-js: 2f16b43 | (Paweł Murias)++ | / (2 files):
Move out p6 ops to the leftover file
kudo-js: 91257ff | (Paweł Murias)++ | / (3 files):
[nqp] implement many nqp:: ops, pass test 59
jnthn pmurias: And boolification
(for if the iterator has more stuff) 14:23
pmurias to check if the list is empty?
jnthn true if there's more, false if not
pmurias ok
nqp-js passes now passes 92 more test assertions ;) 14:24
cedrvint jnthn: (about "were your objects really Int or Num + a role?") I don't understand your question, I tried something like: 14:25
r: gist.github.com/4257923
p6eval rakudo 779703: OUTPUT«debug␤True␤False␤»
pmurias s/passes//
jnthn cedrvint: I think you need to implement a succ in your class too for the range to work out. 14:29
skajp, bbl.
pmurias: nice :)
14:36 am0c joined 14:37 skids joined
cedrvint phenny: tell jnthn that .succ is optional according to S03: "If T also defines .succ, then the range may be iterated. (Otherwise the range may only be used as an interval, and will return failure if asked for a RangeIter.)". Also, this doesn't solve my problem: gist.github.com/4257923 14:38
phenny cedrvint: I'll pass that on when jnthn is around.
pmurias if I want to make nqp-js panda installable what should I check it on? latest rakudo star?
14:39 birdwindupbird left, bphillips left 14:40 bphillips joined
cedrvint phenny: "skajp"? 14:40
phenny cedrvint: "skajp" (sv to en, translate.google.com)
tadzik skype phonetically, I guess
pmurias: we can figure it out
you probably need something like github.com/jnthn/rakudo-debugger/b...r/Build.pm 14:41
note that it currently will also run the original Panda build() method
which may have been a wrong design decision 14:42
anyway, you just write stuff that'll put stuff in bin/ and blib/lib
don't know if that'd be sufficient for a big thing like Rakudo 14:43
pmurias nqp-js is not a big thing yet 14:46
tadzik: I don't currently have rakudo installed so I was wandering what should I test the package against 14:51
(I'm installing curent rakudo star now)
14:54 cognominal left, sftp left 14:55 sftp joined, stopbit joined
pmurias tadzik: if nqp-js is currently run as "nqp bin/nqp-js.nqp" how should that be installed? 14:57
tadzik pmurias: nqp-js thing, executable, having #!/usr/bin/env nqp on top, I think 14:58
pmurias: is that standalone?
15:04 GlitchMr joined 15:08 betterworld joined
[Coke] r: sub asdf (Str $col where $col.chars==2) { say "here" } ; asdf("A"); 15:10
p6eval rakudo 779703: OUTPUT«Constraint type check failed for parameter '$col'␤ in sub asdf at /tmp/CfBCKozMKq:1␤ in block at /tmp/CfBCKozMKq:1␤␤»
[Coke] n: sub asdf (Str $col where $col.chars==2) { say "here" } ; asdf("A"); 15:11
p6eval niecza v24-5-g599cbcb: OUTPUT«Unhandled exception: Constraint type check failed for parameter '$col' in 'asdf'␤ at /tmp/R0ULyRnN4x line 0 (asdf @ 1) ␤ at /tmp/R0ULyRnN4x line 1 (mainline @ 4) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4208 (ANON @ 3) ␤ at /home/p6eval/niecza/lib…
[Coke] Is there a way to write that so the compilers are happy? 15:12
15:14 cooper left
pmurias tadzik: standalone? it require a .pbc library 15:14
15:15 cooper joined
tadzik pmurias: ah. I'm concerned about where will it be looking for that .pbc 15:19
15:24 pmurias left 15:25 BinGOs left, sweet_kid left, zamolxes left 15:26 BinGOs joined, sweet_kid joined, nyuszika7h joined, nebuchadnezzar joined, zamolxes joined, yeltzooo joined, steven___ joined, literal joined, sergot joined, Juerd joined, nyuszika7h left, ruoso left, pmurias joined, ruoso joined 15:27 PacoAir left
pmurias tadzik: I have installed rakudo-star and wrote a Build.pm and META.info 15:28
how do I test if the module builds with panda?
tadzik pmurias: panda install /path/to/directory/
15:30 BinGOs left, sweet_kid left, nebuchadnezzar left, zamolxes left, yeltzooo left, steven___ left, literal left, sergot left, Juerd left 15:32 BinGOs joined, sweet_kid joined, nebuchadnezzar joined, zamolxes joined, yeltzooo joined, steven___ joined, literal joined, sergot joined, Juerd joined, nyuszika7h joined, silug joined 15:34 benabik left 15:35 benabik joined
pmurias tadzik: why does it fetch rakudo-js when I'm installing from hard drive? 15:39
tadzik pmurias: it copies it to a lib where it'll build it 15:40
s/lib/dir
it doesn't build in-place
pmurias shouldn't it be instant? 15:41
pmurias suspects panda prints out the info msgs at the wrong time 15:42
tadzik pmurias: nah, pure perl6 copying is slooow 15:43
don't ask me why :)
it's not "shell cp -r" though 15:44
15:45 hash_table joined
hoelzro tadzik: I'd be troubled if that were only slow under Perl 6 ;) 15:46
pmurias hoelzro: ?
hoelzro shell cp -r 15:47
tadzik :)
15:51 mtk joined
[Coke] was very excited to see a commit from pmichaud. 16:01
jnthn back 16:09
phenny jnthn: 14:38Z <cedrvint> tell jnthn that .succ is optional according to S03: "If T also defines .succ, then the range may be iterated. (Otherwise the range may only be used as an interval, and will return failure if asked for a RangeIter.)". Also, this doesn't solve my problem: gist.github.com/4257923
jnthn cedrvint: OK, will take a look at bit later...gotta worry about $dayjob a bit more yet today 16:10
16:12 kivutar left 16:14 bowtie joined
pmurias tadzik: why don't you use cp? 16:14
16:15 kaleem_ left
jnthn pmurias: Can't rely on having a cp command 16:15
pmurias can\
* can't we test for that? 16:16
16:16 david80s joined
jnthn It's simpler just to do it the same everywhere. 16:16
If it's really *that* slow it's better to look into why and make it faster. 16:17
16:17 cognominal joined
pmurias makes sense 16:18
16:20 david80s left 16:23 david80s joined 16:24 wamba joined
TimToady slurp/spurt are likely not going to be fast unless you do them in binary mode 16:31
and a memory mapping will probably be faster yet 16:32
david80s I'm new in IRC, hope you'll get my question. Why does my 1st test on Perl6 with rakudo-star-2012.11 have a really poor low performance (30sec !! vs 0.04s with Strawberry Perl 5.16) while the source is as simple as it can be: use v6; my Int $max = 1000000; say "max=" , $max; while $max>0 { $max--; } say "loop has ended"; Thanks in advance
TimToady three reasons, or so 16:36
first, parrot is not heavily optimized yet 16:37
second, Perl 6 defaults to arbitrary precision integers, which are a little more work
third and most important, you put a type on the variable, so it thinks it has to check the type on every -- 16:38
[Coke] I would also add "Rakudo is not heavily optimized yet" to that.
TimToady (optimizer not smart enough to realize that -- can only produce another Int)
david80s Ok for two first args, but I've read in perlgeek.de/en/article/5-to-6 that typing variables has the ambition of optimization : "When you have type informations at compile time, you can perform certain optimizations. Perl 6 doesn't have to be slower than C, in principle.". Something went wrong ? 16:39
TimToady [Coke]: I meant that by #1
"ambition"
jnthn Well, the type *should* help in so far as the optimizer could inline the --. But ever since we started doing the :D/:U distinction on the multis it lost its ability to do that.
[Coke] you get a very small gain by elided the Int. 16:40
*eliding
david80s well it's said "motivation" instead ambition...
TimToady oh, sorry, I was running it under niecza, where it did make some gain 16:41
[Coke] and it explodes if you use "int" instead of Int.
TimToady also, niecza runs that particular example about 4 times faster on my machine
[Coke] r: my int $a = 3; $a--; 16:42
p6eval rakudo 779703: OUTPUT«Cannot assign to a non-container␤ in sub postfix:<--> at src/gen/CORE.setting:3085␤ in block at /tmp/d4DFfub3Ao:1␤␤»
[Coke] r: my int $a := 3; $a--;
p6eval rakudo 779703: OUTPUT«===SORRY!===␤Cannot bind to natively typed variable '$a'; use assignment instead␤at /tmp/__Mpnpzewx:1␤------> my int $a := 3⏏; $a--;␤ expecting any of:␤ postfix␤ infix or meta-infix␤ infix stopper␤»…
[Coke] rakudo, go home, you're drunk.
TimToady which is, of course, pretty silly, since C has no problem with those semantics :)
jnthn [Coke]: You can't use -- with native ints yet since the native/rw interaction is a hard problem.
[Coke] figured. 16:43
jnthn (Which hasn't been solved yet... :/)
TimToady yes, we basically need native pointers to do that
on some level or other
jnthn The :D and :U stuff I know how to fix, I just need a chunk of undisturbed time to concentrate on it.
[Coke] TL;DR - we know, we're working on it.
[Coke] pokes jnthn. "Are you free now?" 16:44
[Coke] had a plan to keep poking you for a while, but coke is too lazy and has already given up. :)
jnthn [Coke]: No, right now I'm trying to get the material done for a class I'm meant to teach on Thu/Fri :P 16:45
TimToady david80s: anyway, we're still "motivated" :) 16:47
[Coke] why is it hard? overflow? 16:48
TimToady probably recognizing that a set of static candidates has a collective property that does not arise directly from the proto 16:49
jnthn Yeah, a lot of it is knowing we need to pass a reference rather than a value. But it's also the case that we really want it to inline well too. 16:50
TimToady -- on Int cannot produce an undef, but that's hard to infer currently
jnthn (for the native case) 16:51
Oh, the :U/:D thing at least somewhat needs the whole concept of those as types to be a bit more deeply embedded.
TimToady coercive types like Int() you mean?
jnthn No, but those are in the same problem space. 16:52
TimToady oh, okay, you mean like reference types, a la int** and such in C
jnthn At the moment, "as Int" (which Int() style things replace) is special cased.
TimToady and how those relate to boxed types
jnthn As is the handling of :U/:D
david80s TimToady: I hope so. I assume not being the only one looking forward to code seriously with Perl6. I've tried my test avoiding swaping '$max - -' by '$max=$max+1', and it spit the result in 9s. If this can help jnthn too... 16:53
jnthn So only a handful of places are clueful about them.
TimToady we probably need more return type annotation of the "Just" sort, like Haskell
jnthn But yes, there's certainly a reference issue for the ++
Well, returns Int:D would probably be enough.
(for the definedness case) 16:54
We're talking about 3 different problems at once here :)
TimToady 4, if you count the fact that you're not preparing your slides :)
jnthn D'oh :)
TimToady david80s: yes, nobody wants a fast Perl 6 more than we do :)
but also want it to be Perl 6, and not a fast some other language... 16:56
which is hard
16:58 cognominal left 16:59 cognominal joined 17:02 leont joined
[Coke] is reminded that he needs to move some [spec] tickets from rakudo's queue to spec's. 17:03
david80s TimToady: Ok, try to keep in mind a particularity, good or bad, is strongly highlighted if it's facing a huge gap comparing to others products. I'm confident on the future and your hard work. It was a pleasure. 17:05
[Coke] question on rt.perl.org/rt3/Ticket/Display.html?id=116014 - 17:06
given:
rn: my Str $a = "Z"; $a++ 17:07
p6eval rakudo 779703, niecza v24-5-g599cbcb: ( no output )
17:07 david80s left
[Coke] rn: my Str $a = "Z"; $a++; say $a; 17:07
p6eval rakudo 779703, niecza v24-5-g599cbcb: OUTPUT«AA␤»
[Coke] how is 'A' ... 'a' supposed to ever hit 'a' ?
(argh. the RC link in the ticket has been smushed and is unusable) 17:08
cedrvint: do you have the original URL for that ticket?
n: 'A'...'a'.say; 17:09
p6eval niecza v24-5-g599cbcb: OUTPUT«a␤»
[Coke] n: ('A'...'a').say;
p6eval niecza v24-5-g599cbcb: OUTPUT«A B C D E F G H I J K L M N O P Q R S T U V W X Y Z [ \ ] ^ _ ` a␤»
[Coke] n: ('A'..'a').say;
p6eval niecza v24-5-g599cbcb: OUTPUT«"A".."a"␤»
17:12 Kharec left 17:18 pmurias left 17:20 cognominal left 17:21 kurahaupo joined
FROGGS_ does somebody know if there is a spec about the module search paths? 17:21
17:23 cognominal joined 17:26 kurahaupo left 17:27 Vlavv` joined 17:28 cognominal left 17:38 benabik left, benabik joined, wamba left 17:42 Chillance joined
TimToady FROGGS_: S11 tries to discourage the concept, reserving it only for private module development 17:42
errands &
FROGGS_ TimToady: k, thanks 17:52
17:55 PacoAir joined, wamba joined 17:57 lorn left 18:01 am0c left, hash_table left 18:03 lorn joined 18:09 SmokeMachine left 18:13 MayDaniel joined
GlitchMr Oh, I've found actual Perl 6 module in CPAN 18:17
metacpan.org/module/Perl6::Conf
This doesn't seem to work well
18:22 MayDaniel left
[Coke] ughs at metacpan's web page. 18:33
ahhh. most of that pain is probably the podsix. 18:34
timotimo probably :|
18:41 fhelmberger left 18:42 arlinius joined
arlinius huh? they serve totally different purposes 18:43
sorry, wrong window
skids r: multi sub a ($a is rw) { "rw".say; $a; }; multi sub a ($a) { "ro".say; $a }; my $a = 1; a($a); a(1); # should mutability factor into multidispatch? 18:47
p6eval rakudo 779703: OUTPUT«Ambiguous call to 'a'; these signatures all match:␤:($a is rw)␤:($a)␤␤ in block at /tmp/46e0fTEDL2:1␤␤» 18:48
jnthn skids: yes, just nyi 18:50
skids Wow I was expecting differently :-) 18:51
18:51 kaare_ joined 19:00 dakkar left 19:16 fgomez joined
pmichaud good afternoon, #perl6 19:17
jnthn o/ pmichaud
FROGGS_ hi pmichaud
[Coke] pmichaud: hio! 19:18
PerlJam pmichaud: greets! LTNS 19:19
moritz \o
pmichaud I'm about to convert the rakudo github repo to an organization repo... any objections or comments before I do that? 19:26
PerlJam: yes, ltns. Hope things are going well there. 19:27
jnthn pmichaud: If you're comfortable doing so, I'm fine with it. 19:28
[Coke] pmichaud: wfm. 19:36
rjbs seen mauke 19:39
aloha mauke was last seen in #perl6 145 days 19 hours ago joining the channel.
[Coke] r: say 145 / (365.24/12) 19:40
p6eval rakudo 779703: OUTPUT«4.763991␤»
19:41 jaldhar_ left 19:50 spider-mario joined
sorear good * #perl6 19:54
GlitchMr Perl 6 makes impossible without modifying implementation things possible - nikic.github.com/2012/07/27/How-to-...o-PHP.html - 'in' operator in PHP 19:56
dalek rl6-roast-data: 7b9a396 | coke++ | / (4 files):
today (automated commit)
GlitchMr perlcabal.org/~glitchmr/examples/in - in operator in Perl 6 :-)
I prefer Perl 6 version
niecza> my \in = '2' 19:57
2
niecza> in in in
True
But perhaps I shouldn't abuse Perl 6 like that. 19:58
Woodi hallo :) 20:06
rjbs buffalo operator
Woodi GlitchMr: want to explain me something ?
just --> in signature is strange - it declares type of return value ? 20:07
GlitchMr --> is return value 20:08
It can be even constant value instead of type, but why?
Woodi just learning Perl 6 :)
two more strange/unknown things for my eye in that example... 'so' (creating booolean answer?) and why :exists and not .exists ? 20:10
GlitchMr :exists because I ask %b, not %b{$a} 20:11
so converts value to boolean
Just like !! 20:12
Or ?
grondilu r: say !! True
p6eval rakudo 779703: OUTPUT«True␤»
masak evenin', #perl6
grondilu did not know this one
Woodi o/
20:12 rindolf left
Woodi but 'so %b{$a}' could work too ? 20:13
GlitchMr No 20:14
%b{$a} could exist but be falsy value
I wanted it to look like in Python
>>> "a" in {"a": False}
True
This is Python
20:15 leont left
Woodi >>> is prompt or operator ? :) 20:15
GlitchMr Woodi: It's Python prompt
Woodi oki :)
20:24 fgomez left 20:31 wamba left 20:34 bruges left 20:35 noggle joined 20:36 bruges joined
skids r: my @g = 1,2,3; my ($a, $b, $c) := @g[0,1,2]; $a.say; $c.say; my $d := @g[1]; $d = 5; $b.say; @g.say; $b = 6; # newly implemented my (...) and slice binding not happy together 20:42
p6eval rakudo 779703: OUTPUT«1␤3␤2␤1 5 3␤Cannot assign to a readonly variable or a value␤ in block at /tmp/qjlELtsLRF:1␤␤»
20:43 fgomez joined
masak skids: if you bind directly to a value (like the 2 in @g[1]), you can't assign later. 20:45
20:47 cognominal joined 20:48 popl joined 20:49 fgomez left 20:59 fgomez joined 21:03 GlitchMr left 21:04 kaare_ left
masak timotimo: ooh, you're resurrecting Druid? nice! 21:08
timotimo: I'll have more time next week. I could help you. 21:09
probably needs lots more tests.
timotimo well, yeah 21:10
i was just trying to get it running. then after a few hours of digging and trying to understand what has changed in the last 4 years i see your branch that tries to do resurection, too (but 2 years ago rather than 4) 21:11
on my branch you can apparently already play, but the 3d display doesn't work and i didn't understand why the last time i looked
21:12 Chillance left 21:14 wamba joined 21:19 spider-mario_ joined, spider-mario left
timotimo masak: it should only take you five minutes :D 21:20
21:23 fgomez left
masak I'm spent tonight, unfortuntely. need to use the remaining mana to locate the bed. 21:23
maybe I'll have a look tomorrow. but definitely next week at the latest.
diakopter
.oO( under piles of laundry? )
timotimo no problem :)
it's not a high priority for me
i was just boredly clicking through the perl6 module index, looking at all the failing and/or abandoned modules 21:24
diakopter timotimo: do you have a list of which still work?
jnthn diakopter: Ementaller (how they heck do you spell it...) produces such a list automatically 21:26
timotimo no, i also only clicked on a few of them
diakopter it'd be nice to have that info on the module list 21:29
(but perhaps embarrassing)
timotimo well, can't an automated test run do that?
ideally, the tests in those programs would be exhaustive
jnthn diakopter, timotimo: I think sergot++ is working on building something like that, and already has quite a lot of it done. 21:30
timotimo that sounds great 21:32
21:32 fgomez joined
tadzik yeah, it's pretty working 21:35
feather.perl6.nl/~sergot/modules/
skids masak: you'll notice I can and did assign to g[1] through $d. But $b can't.
tadzik some trouble with cronjobs
there are those Street Lights on the right 21:36
masak TimToady: S02 contains *one* example of qq:to, and that example doesn't address the subtle question of what happens first: .indent(*) or variable interpolation. what's your take on which one happens first? 21:39
21:40 cedrvint_ joined
masak 'night, #perl6 21:40
cedrvint_ o/ masak 21:41
jnthn Well, .indent before would be weird, given we don't have a string to call indent on at that point...
cedrvint_ phenny: tell [Coke] the URL is rosettacode.org/wiki/Calendar_-_for...rogrammers 21:42
phenny cedrvint_: I'll pass that on when [Coke] is around.
cedrvint_ phenny: tell [Coke] the regression appears when Rakudo (star) has switched to NQP 21:44
phenny cedrvint_: I'll pass that on when [Coke] is around.
[Coke] phenny: I'm not convinced it's a regression. 21:48
phenny [Coke]: 21:42Z <cedrvint_> tell [Coke] the URL is rosettacode.org/wiki/Calendar_-_for...rogrammers
[Coke]: 21:44Z <cedrvint_> tell [Coke] the regression appears when Rakudo (star) has switched to NQP
[Coke] er, cedrvint
cedrvint_ o/ [Coke]
[Coke] that snippet doesn't look like your ticket. 21:49
cedrvint_ I reduced the problem as much as possible, indeed
r: "\0"..."~" 21:50
p6eval rakudo 779703: ( no output )
[Coke] cedrvint_: so, my question is: why should 'A'...'a' work, when if you get to 'Z', 'Z'++ == 'AA' ?
cedrvint_ ?
r: ("\0"..."~").say
p6eval rakudo 779703: OUTPUT«(timeout)» 21:51
[Coke] or, when doing using alphas with ..., what function generates the next value?
r: ("A"..."AA").say
p6eval rakudo 779703: OUTPUT«A B C D E F G H I J K L M N O P Q R S T U V W X Y Z AA␤»
cedrvint_ [Coke]: I don't know
[Coke] r: ("A"..."a").say
p6eval rakudo 779703: OUTPUT«(timeout)» 21:52
cedrvint_ I was just trying to execute this piece of RC code, then noticed a regression
[Coke] so, given a transition from Z to AA, 'a' will never show up in the list. so it's infinite, and then you try to print it out, it times out.
ok, but the ticket doesn't show that code, does it?
jnthn It's possible (sorry, don't have time to check right now) that the current behavior is actually the spec one.
cedrvint_ maybe, then it's a bug in Niecza :) 21:53
and in old Rakudo versions
[Coke] ... which was my other point, there's no doubt been some spec churn since the RC entries started.
21:54 cognominal left
[Coke] r: my $a = "\0"; say $a++; 21:54
p6eval rakudo 779703: OUTPUT«␀␤»
[Coke] r: my $a = "\0"; my $i = 1000; while $i { $a++; $i--}; say $a; 21:55
p6eval rakudo 779703: OUTPUT«␀␤»
[Coke] r: my $a = "\0"; my $b = $a; my $i = 1000; while $i { $a++; $i--}; say so $b == $a;
p6eval rakudo 779703: OUTPUT«Cannot convert string to number: base-10 number must begin with valid digits or '.' in '⏏␀' (indicated by ⏏)␤ in method Numeric at src/gen/CORE.setting:10197␤ in sub infix:<==> at src/gen/CORE.setting:2809␤ in sub infix:<==> at src/gen/CORE.setting:2809␤ in blo…
[Coke] r: my $a = "\0"; my $b = $a; my $i = 1000; while $i { $a++; $i--}; say ($b == $a); 21:56
p6eval rakudo 779703: OUTPUT«Cannot convert string to number: base-10 number must begin with valid digits or '.' in '⏏␀' (indicated by ⏏)␤ in method Numeric at src/gen/CORE.setting:10197␤ in sub infix:<==> at src/gen/CORE.setting:2809␤ in sub infix:<==> at src/gen/CORE.setting:2809␤ in blo…
[Coke] r: my $a = "\0"; my $b = $a; my $i = 1000; while $i { $a++; $i--}; say ($b eq $a);
p6eval rakudo 779703: OUTPUT«True␤»
21:56 skids left
[Coke] r: ("\0"..."~")[1000000].say 21:56
p6eval rakudo 779703: OUTPUT«(timeout)» 21:57
[Coke] (so, string increment of \0 is \0 in current rakudo.)
21:58 rickza joined
cedrvint_ r: say "\0".ord; say ("\0"++).ord 21:58
p6eval rakudo 779703: OUTPUT«0␤Cannot assign to a non-container␤ in sub postfix:<++> at src/gen/CORE.setting:1611␤ in block at /tmp/jeiUgeLf_5:1␤␤»
[Coke] you have to assign it to a var first. 21:59
(before ++'ing it)
cedrvint_ r: my $c = "\0"; say $c.ord; $c++; say $c.ord 22:00
p6eval rakudo 779703: OUTPUT«0␤0␤»
22:00 rickza left
cedrvint_ [Coke]: do you mean: 1. the bug I reported is not a bug; and 2. there's a bug elsewhere in this RC code ? 22:01
[Coke] I'm saying I don't see much similarity between the 2 snippets except that they use the "..." operator. 22:03
22:03 cognominal joined
[Coke] and yes, that the bug you reported with A...a doesn't look like a bug to me. I don't see in the spec where strings get anything other than the default ++ behavior to get the next value. 22:03
cedrvint_ [Coke]: my mistake, sorry. 22:04
[Coke] no worries!
I'm not positive, which is why I was trying to ping the channel about it earlier. no one but you responded. :)
cedrvint_ :) 22:05
[Coke] ah! and I'm an idiot, because I see the ... range from the RC at the /end/ of the ticket.
but I think that's the same bug. the RC code and your 2 samples expects that you're going to iterate over the ords, but I don't think that's the case.
r: (0...1000).map(ord(*))[100]; 22:06
p6eval rakudo 779703: OUTPUT«===SORRY!===␤CHECK FAILED:␤Calling 'ord' will never work with argument types (Whatever) (line 1)␤ Expected: :(Cool $s)␤»
[Coke] ... "close enough!" 22:07
22:11 spider-mario_ left 22:19 cedrvint_ left 22:21 sahadev1 joined
sahadev1 perl6: [1, 2] ~~ [1, 2] 22:21
p6eval rakudo 779703, niecza v24-5-g599cbcb: ( no output )
sahadev1 perl6: say [1, 2] ~~ [1, 2]
p6eval rakudo 779703, niecza v24-5-g599cbcb: OUTPUT«True␤»
sahadev1 perl6: say [1, 2] ~~ [1, *] 22:22
p6eval rakudo 779703, niecza v24-5-g599cbcb: OUTPUT«False␤»
22:26 colomon left 22:28 armitage81 joined, armitage81 left, armitage81 joined 23:00 bluescreen10 left 23:05 whiteknight joined 23:09 exodist_ left 23:12 stopbit left 23:14 cognominal left 23:21 PacoAir left
jnthn No Perl 6 tuits today, it seems :( 23:21
'night o/
23:22 cognominal joined 23:23 fgomez left 23:24 wamba left
timotimo how long until the next perl6advent post? :) 23:47
23:50 Vlavv` left 23:51 Vlavv joined 23:58 colomon joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!