pugscode.org/ planetsix.perl.org/ | nopaste: sial.org/pbot/perl6 | evalbot: perl6: say 3; (or rakudo:, pugs:, elf:, etc) | irclog: irc.pugscode.org/
Set by mncharity on 5 January 2009.
00:03 [particle]1 joined, Helios left 00:06 Helios joined 00:18 mdxi left 00:21 [particle] left 00:27 hercynium joined 00:36 alc joined 00:49 gravity joined 00:51 alanhaggai joined 00:54 andrew_ joined 00:55 |jedai| left, |jedai| joined 00:57 wknight8111 left 01:13 Limbic_Region left 01:16 meppl left 01:25 cspencer joined, ft left 01:30 alanhaggai left 01:41 cspencer left 01:42 DemoFreak left 01:43 mncharity2 joined 01:45 cspencer joined 01:46 |jedai| left, agentzh joined, |jedai| joined
mncharity2 elf: say $?VERSION; 01:47
p6eval elf 24900: OUTPUT«Global symbol "$VERSION" requires explicit package name at (eval 117) line 3.␤ at ./elf_f line 3861␤»
mncharity2 elf: $*EXECUTABLE_NAME 01:50
p6eval elf 24900: RESULT«undef␤»
mncharity2 Looks like evalbot is still elf_f. Puzzling.
Btw, <number> on misc/elfish/elfparse/elfgreen is working... it just requires a recent (bugfixed) version of 5.10.0 (at least on 64 bit). 01:52
01:56 mncharity2 left 02:07 andrew_ left 02:09 mj41 left 02:11 cspencer left 02:13 [particle] joined 02:16 Exodist left 02:25 eternaleye joined 02:29 jimmy_ joined 02:30 [particle]1 left 02:31 hercynium left 02:32 a2n joined 02:40 a2n left 02:41 jhorwitz left 02:55 meppl joined 03:08 charsbar_ joined 03:09 |jedai| left, Caelum left, charsbar left 03:15 |jedai| joined 03:19 Caelum joined 03:25 meppuru joined 03:27 gravity left 03:28 broquaint left 03:30 meppl left, meppuru is now known as meppl 03:36 mdxi joined 04:04 |jedai| left 04:05 |jedai| joined 04:08 meppl left 04:10 |jedai| left 04:11 |jedai| joined 04:14 [particle]1 joined 04:15 justatheory left 04:32 hercynium joined, [particle] left 04:40 cspencer joined 04:44 mtnviewmark joined 04:47 xuser joined 04:49 |jedai| left 04:50 |jedai| joined, japhb__ left 04:53 japhb joined
cspencer perl6: my $x = "zz99"; say ++$x 04:55
p6eval elf 24900: OUTPUT«No viable candidate for call to multimethod prefix__43_43($) at (eval 119) line 4␤ at ./elf_f line 3861␤»
..pugs, rakudo 35514: OUTPUT«aaa00␤»
cspencer perl6: my $x = "99aa"; say ++$x 04:57
p6eval elf 24900: OUTPUT«No viable candidate for call to multimethod prefix__43_43($) at (eval 119) line 4␤ at ./elf_f line 3861␤»
..pugs, rakudo 35514: OUTPUT«99ab␤»
cspencer perl6: my $x = "99zz"; say ++$x
p6eval elf 24900: OUTPUT«No viable candidate for call to multimethod prefix__43_43($) at (eval 119) line 4␤ at ./elf_f line 3861␤»
..pugs, rakudo 35514: OUTPUT«100aa␤»
05:00 masak joined 05:02 vanjuggler joined 05:15 cspencer left 05:20 hercynium left 05:23 veritos joined
veritos I have a functionality idea. Currently, we have case insensitive matching. What if we were to add case-fixing matching, so that s:fix/idol/idle/ would convert "idol"->"idle", "Idol"->"Idle", "IDOL"->"IDLE" and so on? 05:24
masak veritos: AFAIK, we have that already. 05:26
veritos: do you want me to find the reference for you, or can you browse S05 yourself?
veritos masak: I can find it; I skimmed too lightly, it seems. 05:27
masak I'm not 100% sure, but it does sound familiar.
veritos :ii or :samecase, apparently. 05:29
masak yes, there you go. 05:31
Perl 6 provides. :)
05:32 alester_ joined
veritos Are we still talking about having unit conversions as inherent capabilities of numbers? 05:34
masak veritos: it's not something I've heard about in a long time. but I don't think the idea was abandoned either.
05:36 |jedai| left, cspencer joined 05:37 |jedai| joined 05:38 veritos left
mtnviewmark oh - masak - you're here! 05:42
masak mtnviewmark: so it seems.
mtnviewmark heh - so "will" -- while it is in S04, it doesn't appear to be an operator 05:43
at least not in the sense of it forming part of the expression grammar
masak that surprises me a lot.
it looks just like an infix operator to me.
mtnviewmark hmmm.... well, it isn't, as you point out, in S03 - and the only use of it in S04 is very tied to variable declarations 05:44
masak so what?
mtnviewmark any clue what precedence it would be?
masak 'but' works about the same.
no, no clues abotut precedence.
mtnviewmark well, 'does' and 'but' seem to have examples applied to values 05:45
masak but TimToady will need to be the arbriter about this anyway. so you might as well ask him.
mtnviewmark (... some big expression...) does (something else)
makes sense
but would "will" or "as" work there?
masak s/arbriter/arbiter/
mtnviewmark okay - what is the incantation to invoke the oracle of TimToady? 05:46
masak mtnviewmark: ah, you mean it only works in var decls? yes, that might be.
mtnviewmark I do worry that there are dozens of language pieces that look somewhat like operators.... but might be "special forms" ... 05:47
and I'm not sure if they should be on the chart or no
masak mtnviewmark: I don't know of any special incantation that works, but this seems to be an issue where he'd reply something, sooner or later.
mtnviewmark: in the meantime, we should check what STD.pm says about 'will'.
ok, it classifies 'will' as a "trait auxiliary", with 'is' and 'does'. 05:48
mtnviewmark ah - and 'but' as well? there seems to be a whole, er, declension, of the verb to be in there 05:49
where can I acquire my own copy of STD.pm?
masak haven't found 'but' yet.
mtnviewmark: got Pugs?
I mean, the SVN flavour. 05:50
mtnviewmark nope
masak mtnviewmark: start by acquiring a copy.
STD.pm is in src/perl6/
mtnviewmark even though it isn't being actively worked on?
masak mtnviewmark: well, there's Pugs itself, and there's community efforts on Perl 6... 05:51
the Pugs repo holds the former as well as the latter.
svn co svn.pugscode.org/pugs
05:52 kisu left
mtnviewmark checkout in progress... 05:52
dukeleto mtnviewmark: the Rakudo test suite languages/perl6/t/spec is an svn external to the pugs repo 05:53
masak traits are allowed in variable declarations, package definitions, routine definitions, method definitions, regex definitions, macro definitions, and paramter lists. 05:54
mtnviewmark okay - well, at the risk of being arbitrary (ha!) I'll claim those aren't full fledged operators
05:54 justatheory joined
mtnviewmark where as 'does' and 'but' are allowed in greater context? 05:54
masak 'doas' and 'but', however, are normal infix ops. yes. 05:55
mtnviewmark oooo the temptaion to type make is high.... but I bet that takes a LONG time.... :-)
masak it seems that 'does' works on both sides in this regard.
mtnviewmark: I haven't done that in a long time. :) 05:56
used to do it all the time.
mtnviewmark gasp! SYN.pm claims <== and family are OPs! 05:59
not terminators!
masak that's surprising. 06:01
mtnviewmark well - thank you for that reference - this is going to take some pouring over....
and just when I was thinking I was close to done!
masak hehe.
I think you and STD.pm will like each other.
I should read it too. 06:02
mtnviewmark on other fronts -- I've been finding good places to have the chart printed in A1 .. or even A0 sized paper
06:02 kisu joined
cspencer perl6: my $x = "\xff"; say ++$x 06:04
p6eval elf 24900: OUTPUT«No viable candidate for call to multimethod prefix__43_43($) at (eval 119) line 4␤ at ./elf_f line 3861␤»
..pugs: OUTPUT«Ā␤»
..rakudo 35515: OUTPUT«ÿ␤»
cspencer perl6: my $x = "Ā"; say --$x 06:05
p6eval elf 24900: OUTPUT«No viable candidate for call to multimethod prefix__45_45($) at (eval 119) line 4␤ at ./elf_f line 3861␤»
..rakudo 35515: OUTPUT«Ā␤»
..pugs: OUTPUT«-1␤»
masak Pugs++ :)
cspencer heh
06:06 cspencer left
masak wonders idly what babble, quibble, tribble and nibble are in STD.pm 06:06
jimmy_ perl6: my $x='b'; say --$x+$x+++++$x 06:07
p6eval pugs: OUTPUT«*** ␤ Unexpected "+++++$"␤ expecting "::", term postfix, operator, ":" or ","␤ at /tmp/zwE0dqmCoH line 1, column 23␤»
..elf 24900: OUTPUT«Can't modify postincrement (++) in postincrement (++) at (eval 119) line 4, near ")++"␤ at ./elf_f line 3861␤»
..rakudo 35515: OUTPUT«0␤»
06:07 justatheory left
jimmy_ perl6: my $x=4; say --$x+$x+++++$x 06:07
p6eval rakudo 35515: OUTPUT«11␤»
..pugs: OUTPUT«*** ␤ Unexpected "+++++$"␤ expecting "::", term postfix, operator, ":" or ","␤ at /tmp/t2fVUZCaOn line 1, column 21␤»
..elf 24900: OUTPUT«Can't modify postincrement (++) in postincrement (++) at (eval 119) line 4, near ")++"␤ at ./elf_f line 3861␤»
masak std: my $a; $a+++++$a
p6eval std 24900: OUTPUT«00:05 86m␤» 06:08
masak I'll be danged.
jimmy_ wonders what is std.
masak jimmy_: it's the official Perl 6 grammar.
mtnviewmark not to mention sibble and quasiquibble
masak mtnviewmark: right. :) 06:09
jimmy_ std: my $x=4; say --$x+$x+++++$x
p6eval std 24900: OUTPUT«00:08 126m␤»
masak I like TimToady names, but sometimes they can be a bit like quasibibble. :)
s/y/y's/
jimmy_ still doesn't know what is std. 06:10
masak jimmy_: got Pugs?
jimmy_ I have no Pugs.
masak ok, hold on.
svn.pugscode.org/pugs/src/perl6/STD.pm 06:11
that's STD.
jimmy_ that's the perl grammar 06:13
masak jimmy_: yes, that's what I said. :) 06:15
jimmy_ std: my $x=10; say --$x-----$x+++++$x 06:17
p6eval std 24900: OUTPUT«00:08 126m␤»
06:22 alc left 06:25 [particle] joined 06:42 mtnviewmark left, vanjuggler left, jhuni joined 06:43 [particle]1 left 06:44 jhuni left
diakopter masak: you around? 06:50
masak diakopter: I'm here.
diakopter t\spec\S04-closure-traits\next. 06:51
doesn't svn co on windows b/c windows doesn't like filenames ending in .
masak seems like a mistake to me.
I'll check the svn logs to see what happened.
06:51 ashizawa joined
masak in fact, it ends in '. ' 06:52
moritz_ I'll move it to next.t
pugs_svn r24901 | moritz++ | [t/spec] move 'next. ' to next.t, that was surely a (my?) accident 06:53
diakopter thanks u2
:D
masak moritz_ is faster on the trigger. I analyze too much. 06:54
moritz_ shoot first!
diakopter ponders "shoot first; ask questions later" vs. "'I have a question';'shoot'" 06:55
moritz_ that's the real problem with not asking questions, but asking to ask - you might get shot in between ;-) 06:58
masak ah, so _that's_ why people advise against asking to ask. I've always wondered about that. 06:59
it's for people's own safety.
let us try this, in the interests of public awareness.
diakopter shoot
masak may I ask a question? :)
diakopter glances askance 07:00
masak falls down on the ground, apparently lifeless
moritz_ prepares masak's obituary
masak "he was a curious person, perhaps too curious..." 07:01
diakopter wait, so who shot masak? I merely instructed him to shoot [the question to us]
masak diakopter: I think it must have been a magic bullet.
diakopter [before masak even asked, it should be noted]
moritz_ that masak seems to be a rather unstable person 07:02
diakopter though actually I was saying 'shoot' to moritz_
n-e-waz
masak diakopter: it only goes to show how dangerous that word is.
moritz_: I'm only unstable when being observed. that can make me collapse. 07:03
as with junctions and boxed cats, you should treat me with care. 07:04
moritz_ masak: maybe you'll be interested in a different interpretation of quantum theory that works without the collapse during observations
masak moritz_: I've heard about such things before. and yes, they do intrigue me. 07:05
moritz_ sometimes they are called "many-world theories", but they don't speculate about multiverses and stuff, they have a solid scientific base
diakopter oooooo... testable, even? 07:06
masak that was my line!
diakopter my pre-emption must have collapsed our link. 07:07
moritz_ diakopter: it's a different interpretation, not a differnt theory. All observables are the same, but they are explained in a slightly more intuitive way
diakopter ok
$.ok 07:08
$.ok()
moritz_ in that theory the process of measuring creats an entanglement between the measuring system and the measured object
which allows a knowledge gain, which corresponds to the collopse of the wave function into an eigenstate
diakopter pretends to follow 07:09
moritz_ well, I haven't fully understood it either
but it's the pet topic of one of my mates in the office
so I can't be 100% ignorant ;-) 07:10
diakopter interesting, svn update on that dir did in fact delete the file named "next" (which previously subversion said needed to be added, while also saying "next." was missing
)
moritz_ how surprising 07:11
I somehow expected it to blow all up 07:12
masak svn--
diakopter I don't suppose anyone reading this (this doesn't mean you, bots) knows how to activate Intel VT on a BIOS-crippled Vaio FW170J
perl6: labelit: 07:13
p6eval elf 24901: OUTPUT«Undefined subroutine &GLOBAL::labelit called at (eval 121) line 3.␤ at ./elf_f line 3861␤»
..pugs: OUTPUT«*** ␤ Unexpected end of input␤ expecting Doc block, block declaration, declaration, construct or expression␤ at /tmp/F9irwOKIXN line 2, column 1␤»
..rakudo 35518: OUTPUT«Statement not terminated properly at line 1, near ":"␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
moritz_ perl6: labelit: print 1;
p6eval elf 24901: OUTPUT«Can't locate object method "cb__Label" via package "EmitSimpleP5" at ./elf_f line 1595.␤»
..pugs: OUTPUT«1»
..rakudo 35518: OUTPUT«Statement not terminated properly at line 1, near ": print 1;"␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
diakopter perl6: Superposition.Class.teleport() 07:16
p6eval elf 24901: OUTPUT«Can't locate object method "Class" via package "Superposition" (perhaps you forgot to load "Superposition"?) at (eval 115) line 3.␤ at ./elf_f line 3861␤» 07:17
..pugs: OUTPUT«*** No such subroutine: "&Superposition"␤ at /tmp/dJ8LV1k4Ut line 1, column 1 - line 2, column 1␤»
..rakudo 35518: OUTPUT«Could not find non-existent sub Superposition␤current instr.: '_block14' pc 53 (EVAL_16:38)␤»
diakopter elf thought Class was an object method?
(no parens?)
moritz_ that's the correct behaviour 07:18
well, it woiuld, if Superposition where a known type name
since it isn't, it's correct to parse Superposition as a sub call
diakopter I wasn't referring to rakudo 07:19
was referring to elf
07:19 spx2 joined
moritz_ yes, I know 07:19
07:19 spx2_ left
diakopter which output is correct? 07:20
moritz_ rakudo's and pugs'
they are basically the same, the format of the error message just differs 07:21
masak actually, I don't much like Rakudo's error. 'could not find non-existent sub'. it sort of says the same thing almost twice. 'no such' is nicer.
moritz_ right 07:22
masak can I change it? please? :)
diakopter I can't find non-existent things all the times
masak diakopter: oh, shut up. :)
diakopter in fact, I invariably fail to find non-existent things
masak diakopter: that's not the non-existent point.
moritz_ diakopter: that's because you neglect autovivification ;-) 07:23
diakopter rakudo's message begs the question, "why were you looking for something you knew to be non-existent" or "don't you mean 'found that sub to be non-existent'?" 07:24
07:25 araujo left
diakopter how about "missing sub" or "lost sub" 07:25
or "sub not found" 07:26
moritz_ or "FAIL: attempted to call non-existant sub 'foo'" 07:28
diakopter "voicemail: no answer from call to non-existent sub" 07:29
masak rakudo: my $x = :a<5>; say $x.map({.key => .value + 1}).perl
p6eval rakudo 35518: OUTPUT«Method 'key' not found for invocant of class 'Failure'␤current instr.: '_block14' pc 101 (EVAL_17:49)␤»
masak moritz_: would you expect map to work on a pair?
moritz_ masak: since we support it on Any, yes 07:30
masak so the above is surprising.
moritz_ rakudo: say 1.map({ say 2*$_ });
p6eval rakudo 35518: OUTPUT«2␤1␤»
moritz_ not just that...
but also a...
masak rakudo: my $x = :a<5>; say $x.map({$^k.key => $^k.value + 1}).perl
p6eval rakudo 35518: OUTPUT«["a" => 6]␤»
moritz_ bug
masak submits rakudobug
07:31 namenlos joined
moritz_ so $_ isn't set in the closure in the first example? 07:31
diakopter rakudobugmenot
masak moritz_: seems not. 07:32
moritz_ rakudo: say ({ .key => .value + 1}).arity 07:33
p6eval rakudo 35518: OUTPUT«Method 'key' not found for invocant of class 'Failure'␤current instr.: '_block14' pc 66 (EVAL_16:42)␤»
diakopter maybe it's not parsing it as a closure
moritz_ rakudo: say ({ .key => .value + 1}).WHAT
p6eval rakudo 35518: OUTPUT«Method 'key' not found for invocant of class 'Failure'␤current instr.: '_block14' pc 66 (EVAL_16:42)␤»
moritz_ right
I don't know if that's a bug, or by design 07:34
diakopter rakudo: ({ .key => .value + 1}).WHAT
p6eval rakudo 35518: OUTPUT«Method 'key' not found for invocant of class 'Failure'␤current instr.: '_block14' pc 66 (EVAL_15:42)␤»
diakopter rakudo: ({ .key => .value + 1})
p6eval rakudo 35518: OUTPUT«Method 'key' not found for invocant of class 'Failure'␤current instr.: '_block14' pc 66 (EVAL_15:42)␤»
moritz_ according to the parsing rules I remember (beware...), it should be a hash constructor
oh wait, no
.key isn't and identifier
masak it's a method of Pair 07:35
07:35 maerzhase joined
moritz_ rakudo: given :a<5> { say {.key => .value + 1}.WHAT; 07:36
p6eval rakudo 35518: OUTPUT«Unable to parse block; couldn't find final '}' at line 1, near ""␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
07:36 alester_ left
moritz_ rakudo: given :a<5> { say {.key => .value + 1}.WHAT;}; 07:36
p6eval rakudo 35518: OUTPUT«Hash␤»
diakopter peers to peers 07:37
masak rakudo: my $p = 'mary' => 8; say { .key => .value * 2 }.($p).WHAT 07:38
p6eval rakudo 35518: OUTPUT«Method 'key' not found for invocant of class 'Failure'␤current instr.: '_block14' pc 104 (EVAL_19:50)␤»
07:39 |jedai| left 07:40 |jedai| joined
moritz_ rakudo: say "fo".chars 07:44
p6eval rakudo 35518: OUTPUT«2␤»
diakopter rakudo: say "\"fo\"".chars() 07:48
p6eval rakudo 35518: OUTPUT«4␤»
diakopter rakudo: say "\"fo\"".chars 07:49
p6eval rakudo 35518: OUTPUT«4␤»
07:49 vixey joined
diakopter rakudo: say "\"fo\"".chars () 07:49
p6eval rakudo 35518: OUTPUT«Statement not terminated properly at line 1, near "()"␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
diakopter rakudo: say "\"fo\"".chars\ () 07:50
p6eval rakudo 35518: OUTPUT«4␤»
diakopter rakudo: say "\"fo\"".chars\ .()
p6eval rakudo 35518: OUTPUT«4␤»
masak long dots work?
masak didn't know that
diakopter rakudo: say "\"fo\"".chars.() 07:51
p6eval rakudo 35518: OUTPUT«4␤»
diakopter !
07:52 meppl joined
diakopter rakudo: 4.() 07:52
p6eval rakudo 35518: OUTPUT«invoke() not implemented in class 'Integer'␤current instr.: '_block14' pc 62 (EVAL_17:40)␤»
diakopter rakudo: "fo".chars().()
p6eval rakudo 35518: OUTPUT«invoke() not implemented in class 'Integer'␤current instr.: '_block14' pc 71 (EVAL_17:41)␤»
masak don't provoke the integer! 07:53
diakopter rakudo: say "\"fo\"".chars..()
p6eval rakudo 35518: OUTPUT«␤»
diakopter rakudo: say "\"fo\"".chars...()
p6eval rakudo 35518: OUTPUT«␤»
diakopter rakudo: say "\"fo\"".chars. . . . . .()
p6eval rakudo 35518: OUTPUT«Statement not terminated properly at line 1, near ". . . . . "␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
diakopter rakudo: say "\"fo\"".chars. .() 07:54
p6eval rakudo 35518: OUTPUT«Statement not terminated properly at line 1, near ". .()"␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
diakopter rakudo: say "\"fo\"".chars .()
p6eval rakudo 35518: OUTPUT«Statement not terminated properly at line 1, near ".()"␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
pugs_svn r24902 | masak++ | [u4x] fleshed out README a bit 07:59
masak yesterday, I had the idea that we could name the documentation tool 'psi', for 'Perl Six Whatever-the-"i"-means-in-"ri"' 08:00
('information', it seems)
diakopter masak: wouldn't that be 'PSW'?
masak diakopter: how do I unescape things in explanations of abbreviations? 08:01
diakopter lol
diakopter looks in the synopses
masak :)
we could even alias it to ψ 08:02
or symlink, I dunno.
diakopter 'psi' - the ultimate dwim - because it reads your mind directly.
masak diakopter: ooh, didn't think of that.
diakopter: yes, mind reading is one of the features. 08:03
diakopter ... as opposed to reading your mind indirectly.
masak see the latest addition to the README.
diakopter or
psi could also be the name of PerlSix[-In-PerlSix ... [] ] 08:05
though there's probably already a programming language [implementation] named that
hmm, no
08:05 ashizawa left, ashizawa joined
diakopter oh wait, that'd be psip 08:06
pronounced 'ps-ype
or, to follow viv, vivi 08:07
TimToady: see vivi and cetera
08:07 _dh left
diakopter vivivi 08:08
08:08 iblechbot joined
pugs_svn r24903 | masak++ | [u4x] it's official: the tool will be named 'psi' 08:09
08:12 ChanServ sets mode: +o diakopter
pugs_svn r24904 | masak++ | [u4x] added inofficial goals to the README 08:12
08:13 diakopter was kicked by diakopter (no more drunk IRCing tonight)), diakopter joined 08:22 maerzhase left 08:23 broquaint joined 08:25 maerzhase joined 08:30 meppl left 08:32 _jedai_ joined 08:33 |jedai| left 08:35 pdcawley left 08:38 alc joined 08:41 DemoFreak joined 08:45 [particle]1 joined 08:49 [particle] left 09:01 pmurias joined 09:07 mj41 joined 09:11 alc left
masak rakudo: my $x = "foo"; class A { my $.y = $x } # bug? 09:12
p6eval rakudo 35518: OUTPUT«Lexical '$x' not found␤current instr.: 'parrot;A;_block22' pc 113 (EVAL_19:58)␤»
09:14 ludan joined
ludan hi 09:14
masak ludan: greetings. 09:15
09:21 _jedai_ left 09:22 |jedai| joined, maerzhase left
masak rakudo: our $x = "foo"; class A { my $.y = $x } 09:24
p6eval rakudo 35519: OUTPUT«Null PMC access in isa()␤current instr.: '!CALLMETHOD' pc 16769 (src/builtins/guts.pir:74)␤»
masak :(
don't we have tests for this
masak submits rakudobug
moritz_ what is 'my $.x' supposed to do?
masak moritz_: that's a class attribute, sir.
moritz_ masak: and that has a . twigil? 09:25
09:25 pmurias left
masak moritz_: yes, it can. 09:25
then it gets a read accessor.
moritz_ rakudo: our $x = "foo"; class A { has $.y = $x }; say A.new.y
p6eval rakudo 35519: OUTPUT«Use of uninitialized value␤␤»
moritz_ rakudo: our $x = "foo"; class A { has $.y = 3 }; say A.new.y 09:26
p6eval rakudo 35519: OUTPUT«3␤»
moritz_ rakudo: my $x = "foo"; class A { has $.y = ++$x }; say A.new.y
p6eval rakudo 35519: OUTPUT«Lexical '$x' not found␤current instr.: 'parrot;A;_block22' pc 140 (EVAL_20:59)␤»
09:26 elmex joined 09:27 alc joined 09:30 elmex left 09:40 donaldh joined 09:58 maerzhase joined 09:59 mberends joined 10:07 maerzhase1 joined, maerzhase left 10:08 jimmy_ left 10:14 maerzhase joined, maerzhase1 left 10:15 DemoFreak left 10:36 |jedai| left, |jedai| joined 10:49 meppl joined 10:58 aindilis left 10:59 alc left 11:00 |jedai| left, ruoso joined, elmex joined, |jedai| joined
ruoso Hello! 11:01
11:01 barney joined
Matt-W ruoso: hello! 11:04
ruoso in a very hard month... barely working in Perl 6 development :(
11:05 [particle] joined, ejs joined 11:11 ft joined 11:13 |jedai| left, |jedai| joined
Matt-W ruoso: these things happen sometimes unfortunately 11:16
11:18 krunen left 11:20 tux300181 joined, krunen joined 11:22 [particle]1 left 11:39 maerzhase left, maerzhase joined 11:40 araujo joined 11:45 maerzhase1 joined 11:46 maerzhase left 11:49 braceta joined 11:50 |jedai| left 11:51 |jedai| joined 12:11 vixey left 12:13 ashizawa left 12:41 |jedai| left 12:42 |jedai| joined, orafu left
masak on p6u, Richard Hainsworth proposes '.face' as the method for getting the string value of an enum value. while I don't understand the rationale for the method name, I definitely think there should be a method. 12:44
12:44 ruoso left 12:45 orafu joined
pugs_svn r24905 | masak++ | [u4x] added keywords to the README 12:46
mberends hi masak, how's Druid online coming along? I've done my part of the deal :) 12:52
masak mberends: OH HAI
mberends: well, I'm sort of ready to get going, but two components are needed still. 12:53
1: SVG. I'm sort of writing that.
won't be that hard to do, I think.
2: a proof-of-concept page running on feather. 12:54
mberends: do you think you could do that second part? it would make it much easier to know how to hook everything up.
mberends I think SVG may be like Postscript in an XML envelope
masak I understand each individual word in that sentence. but... what? 12:55
SVG is SVG.
mberends I'll add a proof of concept page to autoexec.demon.nl this evening 12:56
12:56 pmurias joined
mberends autoexec.demon.nl:8888 12:56
masak mberends: excellent. I'll try to speed things up on my end.
jnthn
.oO( What is Druid anyway? )
masak jnthn: you haven't played it yet? for shame! 12:57
github.com/masak/druid/
jnthn masak: I'm too busy doing boring stuff, like implementing parametric roles and type registries. ;-)
masak jnthn: I understand. :)
jnthn ...and don't have a git client ;-)
masak you need to relax with this here board game.
jnthn: don't need one.
just click "download".
jnthn masak: FAIL! 12:58
github.com/tarballs/masak-druid-09e...285e3d.zip
That page doesn't exist!
masak :(
it does for me. 12:59
should I email you the .zip file?
jnthn Or the perl6 file ;) 13:01
masak: Oh, tried again and it worked this time.
masak oh, good.
13:04 |jedai| left, pmurias left 13:05 |jedai| joined 13:25 [particle]1 joined 13:29 [particle] left 13:33 DemoFreak joined
Matt-W jnthn: parametric roles are not boring for the people who're going to use them! 13:49
moritz_ and who's going to use them? (except STD.pm)
jnthn Everyone who uses typed arrays? ;-) 13:50
moritz_ erm. 13:51
masak :) 13:52
13:52 ejs1 joined
masak jnthn: did the game work? /me is curious. 13:52
jnthn masak: I'm just doing some @other work at the moment...will get onto Rakudo stuff in 10 mins or so, then I'll try the game. :-)
masak :)
moritz_ should try to get a grant for trying out all those Perl 6 games :-) 13:53
14:00 alexn_org joined
jnthn masak: I'm hoping my type registry addition didn't cause any pain for Druid/November. 14:01
masak jnthn: when did it slide in? 14:02
moritz_ yesterday
14:02 athenot left
masak jnthn: in that case, no. 14:03
14:03 athenot joined
masak at least I think it didn't. 14:03
masak recompiles Rakudo
I just got all the tests in November's master passing after the rvar branch.
14:04 ejs left
jnthn masak++ 14:07
masak :) 14:08
Matt-W November's been really helpful to me - looking at its code to figure out how to do stuff 14:10
moritz_ it's been also really helpful to all developers, it found dozens or even hundrets of bugs 14:11
literal random thought: can macros be used to remove debugging code during compilation so that it won't affect runtime performance?
moritz_ literal: yes
literal ok
masak Matt-W: let's make that step largely unnecessary by making u4x a huge success. 14:12
moritz_: maybe not hundreds, but surely over a hundred by now.
moritz_ masak: t/dispatcher/0{1,2} still fail for me
masak moritz_: nopaste perl6 run of it?
Matt-W masak: yes that would be a good idea :)
jnthn (1) Computer versus human 14:13
(2) Human versus computer
Hmm... :-)
masak jnthn: what? :)
jnthn Vertical: A 14:14
Nasty syntax
Hmm
masak a2
jnthn should read the read-me
masak jnthn: it's under-documented, I know.
moritz_ masak: funny thing, it passes when I run it directly through perl6
masak the move syntax is like a2 or like b3-b5.
the rest is discoverable.
moritz_: :/
moritz_: ok, what warnings/errors do you get in prove?
jnthn masak: It appears to work! :-) 14:15
I don't know how to win though. :-)
moritz_ masak: nopaste.snit.ch/15304
jnthn would hvve to read the docs for that
masak jnthn: if you're vertical, you're supposed to build a chain of pieces from top to bottom.
moritz_: that's from an older version of November. 14:16
14:16 Lorn joined
jnthn masak: Aha! 14:16
masak moritz_: MD5 (Dispatcher/Rule.pm) = 69e8914f10cdc32f1d4d1dd095b629c5
jnthn: and it's 3D.
moritz_ masak: same md5 here 14:17
masak moritz_: ok, strange.
moritz_: old Rakudo? :)
moritz_ masak: Rakudo svn up'ped and built today... 14:18
masak: november git-pulled and
masak moritz_: perhaps even different versions when you run prove and perl6...
moritz_ make clean'ed
and make'd
jnthn moritz_: Does your Rakudo pass all its tests?
masak moritz_: oh. I don't make.
masak tries that.
moritz_ jnthn: I'll try 14:19
jnthn Also, one thing to beware of: I did change the root.in yesterday
So you need to perhaps make Makefile
moritz_ I did a realclean after that
jnthn Also I found some quirkiness with re-generating gen_builtins.pir
Ah, OK.
That should cover it.
masak moritz_: I can't make November. are you saying you can? 14:20
moritz_ jnthn: t/spec/S02-magicals/dollar_bang dies with Null PMC access in isa()
14:21 |jedai| left
moritz_ masak: so it seems :) 14:21
masak oddness.
jnthn moritz_: Odd. I didn't see that yestrday. Doing spectest run now.
masak my make dies on compiling November.pm, though I'm not sure the error is in that module.
14:21 |jedai| joined
moritz_ stupid 14:22
/home/moritz/tmp/rvar2/parrot /home/moritz/tmp/rvar2/languages/perl6/perl6.pbc --target=pir --output=Text/Markup/Wiki/Minimal.pir Text/Markup/Wiki/Minimal.pm
wrong parrot
14:23 pjcj left
moritz_ jnthn: oh, and that test fails because I'm trying Ovid++'s Test.pm 14:23
jnthn moritz_: Ah, OK 14:24
Phew!
moritz_ now I caused very much unneeded confusion due to two stupid mistakes in a row
time to do some physics
masak moritz_: ah. so November passes as well? 14:25
moritz_ masak: November doesn't build with current rakudo 14:26
masak moritz_: that's right.
do don't build it.
just run the tests.
masak will try and figure out why November doesn't build
moritz_: do you also get 'Unable to parse role definition'? don't abort the physics for my sake, just asking for later. 14:29
jnthn masak: Is that something new that you've just got? 14:31
moritz_ masak: aye
jnthn has been doing bits with roles
moritz_ t/01-cgi...........................1/23 too few arguments passed (0) - 2 params expected
masak jnthn: I have not been able to run make for a few days now. I don't remember when it started. I think the lists are in some november-wiki mailing list discussion. 14:32
jnthn masak: OK. 14:33
masak moritz_: oh, I'm getting that too now.
must be the latest Rakudo.
investigating.
moritz_ and many "non-dwimmy hyperops..."
masak moritz_: I don't get those (yet). 14:34
I do, however, get many 'too few arguments passed'
14:34 ruoso joined
masak something must have changed under our feet :) 14:34
moritz_ these seem to be related to grammar invokation 14:35
masak aye.
moritz_ do you use <Grammar::TOP> to invoke it?
14:36 mberends left
moritz_ t/markup/minimal/06-links..........Non-dwimmy hyperoperator cannot be used on arrays of different sizes or dimensions. 14:36
jnthn masak: If you were using things like $foo ~~ Grammar::rule it *will* be broken now.
masak: And the wrong number arguments errors would match up with that. 14:37
14:40 alexn_org left
masak oh, joy. 14:40
what's the correct way to do it?
jnthn masak: pmichaud had been saying it's wrong for a while. :-) 14:41
masak yes, I know, I know.
jnthn masak: $foo ~~ /<Grammar::rule>/ should work.
moritz_ 15:35 <@moritz_> do you use <Grammar::TOP> to invoke it?
masak moritz_: apparently not :/ 14:42
jnthn moritz_: I think I may break parametric roles up into a couple of test files.
moritz_ jnthn: feel free
14:43 alexn_org joined
jnthn moritz_: I want some more basic ones as well as your nice tricky ones. :-) 14:43
moritz_: Plus as mixins vs on a class etc.
moritz_ jnthn: I can very much understand that
14:46 |jedai| left 14:47 |jedai| joined, ft left 15:02 exodist_ joined 15:03 exodist_ left, Exodist joined 15:05 Casan joined 15:06 vixey joined 15:19 Lorn left 15:23 hercynium joined, eric256 joined 15:31 ft joined 15:33 maerzhase joined, maerzhase1 left 15:36 |jedai| left 15:37 |jedai| joined 15:39 mberends joined 15:41 [particle] joined
masak if I use the new /<Foo::Bar::TOP>/ approach instead of the defunct Foo::Bar::TOP one, will $/ be affected in any way? 15:43
moritz_ masak: yes, there will be a $/<Foo::Bar::TOP>, which contains the same thing as the old $/
pmichaud yes, because the match would be $/<Foo::Bar::TOP>
masak ok. 15:44
pmichaud I *think* the new official way to match a grammar is .parse, yes?
at least I saw something like that scroll across my screens a few days ago
masak that explains all the sudden test failures after I fixed the first problems.
pmichaud: sounds familiar, yes.
moritz_ pmichaud: yes, .parse on the grammar
and .parsefile
pmichaud it's not in S05 yet -- is there an IRC discussion I could look at? 15:45
eric256 my $parsed = Foo::Bar.parsefile($filename) ?
masak maybe I should hold out for that change -- putting long grammar names everywhere in working code makes me cringe a bit.
moritz_ eric256: yes 15:46
masak where $parsed would be a Match object, I guess.
Matt-W or whatever Foo::Bar::TOP returns with make... 15:47
If I understand what make does correctly
masak right.
moritz_ Matt-W: it always returns a Match object 15:48
Matt-W: but its scalar value can be an arbitrary object
Matt-W so make sets what that object will be
moritz_ aye
Matt-W such as the AST you've been building in the grammar rules
handy
ruoso moritz_, I think "make" allows you to return something different from a match object 15:49
not only its scalar value
masak :)
ruoso but $/ will still be set, tho
moritz_ ruoso: I asked Larry why there's $/.text when ~$/ would do as well 15:50
ruoso: the answer was "because the scalar value of $/ can be affected by make"
(not word by word, though)
eric256 off topic: anyone have an idea what an average hourly rate for salesforce.com customization might be?
ruoso right... right,..
moritz_, I think that means that when you do "make" it will change the scalar value of $/
but not only that 15:51
it also returns that value from the token/rule instead of the match object
pugs: token a { . { make 123 } }; my $a = 'a' ~~ /<a>/; say $a.WHAT; 15:52
p6eval pugs: OUTPUT«Error eval perl5: "if (!$INC{'Pugs/Runtime/Match/HsBridge.pm'}) {␤ unshift @INC, '/home/evalenv/pugs/perl5/Pugs-Compiler-Rule/lib';␤ unshift @INC, '/home/evalenv/pugs/third-party/Parse-Yapp/lib';␤ eval q[require 'Pugs/Runtime/Match/HsBridge.pm'] or die $@;␤}␤'Pugs::Runti...
ruoso meh
rakudo: token a { . { make 123 } }; my $a = 'a' ~~ /<a>/; say $a.WHAT;
p6eval rakudo 35528: OUTPUT«Statement not terminated properly at line 1, near "make 123 }"␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
moritz_ pugs predates make
ruoso kp6: token a { . { make 123 } }; my $a = 'a' ~~ /<a>/; say $a.WHAT;
p6eval kp6 24905: OUTPUT«syntax error at position 40, line 1 column 40:␤token a { . { make 123 } }; my $a = 'a' ~~ /<a>/; say $a.WHAT␤ ^ HERE␤»
ruoso hmpf;
pugs_svn r24906 | masak++ | [u4x] want to handle special vars and pseudo-packages as well 15:53
pmichaud hmm, that's interesting. I might've guessed that my $a = Grammar.parse() would result in $a getting the Match object itself.
and then one does $( $a ) to get at the result object.
moritz_ rakudo: token a { . {{ .return(23) }} }; my $a = 'a' ~~ /<a>/; say $a.PARROT
p6eval rakudo 35528: OUTPUT«Perl6Scalar->Match␤»
moritz_ pmichaud: that's my understanding as well
masak mine too.
ruoso and what if you "make @@a"
or... make grep { ... }, ...; 15:54
you'll be enforcing one context
where it could be assigned to any context
Matt-W rakudo doesn't support make yet, does it
pmichaud masak: I think I can put .parse in fairly quickly if we can find the original discussion.
masak \o/
pmichaud: I'll go look for it.
pmichaud Note from #parrot, though -- as things currently stand the Parrot repo will be down starting 1900 UTC 15:55
Matt-W Oh dear. Any indication of how long for?
masak pmichaud: yes, I know.
15:56 jgoulah left
ruoso token a { . { make grep { $_ % 2 }, 1..10 } }; my @@a = 'a' ~~ /<a>/ 15:56
moritz_ irclog.perlgeek.de/perl6/2008-12-31#i_797318
pmichaud: that was for you
pmichaud Matt-W: as much as 5 hours
moritz++
masak moritz_++
Matt-W pmichaud: Thanks. 15:57
Matt-W makes a note to svn up rakudo as soon as he gets home
15:57 [particle]1 left
ruoso Matt-W, that's when you want to be using git-svn 15:58
(or svk)
Matt-W Please don't make me overload my brain with more than its limited capacity, it's still trying to learn Perl 6...
Also, if I start using git for anything at home, the svn setup at work is going to get even more intolerable 15:59
masak I just learned to use git. I still think git-svn is a bit tough to use.
ruoso still easier than svk... 16:00
Matt-W home &
mberends in r35522 (and for a few days already) there seems to be a difference between 'parrot perl6.pbc' and 'perl6' when applied to '--target=pir --output=Test.pir Test.pm' in languages/perl6. The former works (as used in make spectest), the latter segfaults here. is that a known problem? 16:05
pugs_svn r24907 | jnthn++ | [t/spec] The beginnings of some simple tests for parameterized roles - more to come. 16:06
masak mberends: no, but thanks for the tip. I've have problems building November, and that might be it.
moritz_ mberends: perl6 is more strict with memory checking than parrot perl6.pbc 16:07
mberends: but if you have a short test case, please submit it as a rakudbug
mberends my suspicion is directed as pbc_to_exe 16:08
the compile if Test.pm is a test case. gotta commute home soon, rakudobug later 16:09
s/if/of/
rakudo_svn r35530 | jonathan++ | [rakudo] class C does R[param1, ...] { ... } now works insofar as it chooses the correct role. 16:10
16:10 |jedai| left, |jedai| joined, ejs1 left 16:11 mberends left 16:12 ejs1 joined 16:14 stephens joined 16:15 maerzhase left
jnthn std: role Foo[] { } 16:32
p6eval std 24907: OUTPUT«00:05 83m␤»
jnthn std: role Foo[] { }; class X does Foo[] { }
p6eval std 24907: OUTPUT«00:05 83m␤»
16:38 ludan left
rakudo_svn r35531 | jonathan++ | [rakudo] Parse does Foo[] (STD.pm does). 16:40
Matt-W masak: reading the updated u4x readme, psi is going to benefit greatly from having STD.pm I think... 16:42
masak Matt-W: perhaps.
I was hoping that wouldn't be necessary, but maybe it is.
maybe we can get away with using a small part of it. 16:43
moritz_ masak: did you follow my earlier "perlhints" attempt?
masak moritz_: oh, it was you! yes, I remember it.
moritz_: I have unconsciously borrowed some ideas from it, methinks. 16:44
moritz_ I noticed some similarities
masak yes, that's because I liked the old idea.
moritz_ the idea was to have a chunk of documentation with much meta information
jnthn std: multi foo(::T) { }
p6eval std 24907: OUTPUT«00:05 83m␤»
jnthn std: multi foo(::T $x) { }
p6eval std 24907: OUTPUT«00:05 83m␤» 16:45
masak moritz_: I've barely started to figure out how to structure the documentation.
moritz_ enough so that a Perl 6 parser could emit HTML where you click on an element, and it gives you an explanation for this thing
masak moritz_: so far, I'm targetting the prompt and some sort of static web site. 16:46
eric256 irclog.perlgeek.de/perl6/2009-01-13#i_829019 moritz_ and i where talking about being able to use the documentation for some thing like "use diagnostics" as wlel
masak (my role models being ri and perldoc, respectively)
eric256: sounds orthogonal to u4x, but a cool idea. 16:47
moritz_ well, if we organize it as some kind of database, it's available to every tool
PerlJam (available to all)++ 16:48
It occurs to me that whatever form that database takes it should probably use the same "database engine" that perl6 will use to manage modules/authors/revisions.
moritz_ for subs and method, they could just be available by name 16:49
masak moritz_: my only thought so far is that I'll have one directory with all the classes, one with all the operators, etc.
moritz_ masak: IMHO there's no reason to separate subs and operators
masak: because ops are just subs with weird name and syntax
masak moritz_: agreed.
16:50 |jedai| left, |jedai| joined
masak as long as things can work as advertised, that's good. 16:50
moritz_ what I worry more about, how do we index information to keywords?
keyword:class perhaps?
masak moritz_: why not just 'class'?
'psi class'
moritz_ masak: because it doesn't work for things like 'is' 16:51
masak moritz_: why not?
moritz_ masak: 'is' can be 1) traits 2) implementation type 3) inheritance 4) sub
masak moritz_: ok, so we need both the fuzzy term and a bunch of exact terms. yes.
the former will either turn up all matches, or a list of them. 16:52
moritz_ masak: that idea is incorporated in the `perlhints' data as well
I use the rule names by which STD.pm matches
as well as the syntax itself
masak moritz_: ooh, good idea!
moritz_ ie infix:<**> and ** 16:53
masak exactly.
16:53 maerzhase joined
moritz_ PerlJam: re "Database Engine" maybe that's a good idea, but I'm not fond on waiting for such a thing 16:56
PerlJam moritz_: well, rather than waiting you could prototype something good :-)
masak plans to use the file system as his database engine in the short term 16:57
moritz_ PerlJam: that would require good understanding of what's needed for Modules, and Design skills, and so on
masak: as long as you don't create files named 'infix:<**>' :-)
PerlJam masak: what about dbm files (or bdb)? 16:58
moritz_ PerlJam: are they human readable
masak moritz_: oh. I'll have to think about that.
eric256 is that meta information about modules going to be stored in pod in the module? or externaly?
16:58 iblechbot left
masak PerlJam: do you have a reference? I don't know anything about those formats. 16:58
PerlJam masak: it's basically a glorified hash on disk. keys and values. 16:59
masak eric256: this is first and foremost for core things, not user-written modules. when we get as far as user-written modules, it'll be stored externally.
PerlJam: sounds nice.
PerlJam: why use that and not a Perl hash?
16:59 justatheory joined
moritz_ eric256: well, Perl 6 will have to have some kind of module database, because it must be able to locate modules by names that the file system doesn't support 16:59
PerlJam masak: you could tie the hash-on-disk to a perl hash if you want
masak: perldoc -f dbmopenn # for the old-school method 17:00
masak: perldoc -f tie
eric256 why can't core documentation be in the form of modules? Array.pm, etc. then once prelude gets going it loads all those
PerlJam er, s/dbmopenn/dbmopen/
masak PerlJam: right, right. I was thinking of doing this in Rakudo, however.
eric256 and instantly we can implement core functions in perl6 itself
moritz_ eric256: because we don't have Array.pm right now
masak eric256: also because there is not one Perl 6 implementation.
PerlJam masak: okay ... so you'd have to write the interface glue to attach a dbm file to rakudo. That's a small strike against it I guess. 17:01
eric256 moritz_: doesn't mean that it couldn't be started just to house documentation?
PerlJam :-)
masak PerlJam: but the hash idea is good. that'll get us started.
moritz_ in my vision the documentation can be inside POD, and an external tool will extract that into a doc database 17:02
masak moritz_: yes, I want to use Pod.
admittedly, writing all of the Pod inside hash values sounds a bit cumbersome. 17:06
so maybe make the files pure Pod files instead...
...and treat some level of heading like hash keys when loading them. 17:07
moritz_ that sounds nice
masak I like it, too.
so, starting from the list in the README file, which Pod files will we have? 17:09
subs, metaoperators, macros, keywords, special-variables, pseudo-packages... 17:10
moritz_ I'd rather structure it by topic
masak moritz_: please elaborate.
moritz_ Numbers, Arrays, Hashes, Objects, ...
17:11 maerzhase left, |jedai| left
moritz_ and Numbers would contain subs, methods, ops, macros (if any), ... 17:11
masak moritz_: I'm thinking of it from the viewpoint of fast access. how would pse know where to look?
17:11 maerzhase joined
moritz_ masak: it would read all files, build an index (or a db), ... 17:11
masak hm, maybe that really isn't a problem. or maybe it'll be a problem regardless...
17:11 |jedai| joined
masak moritz_: aye. 17:12
moritz_ masak: and that structure can be optimized for access
what we write should be optimized for flexibility and writeability
masak indeed.
moritz_ that's also the route I took for perlhints 17:13
masak but I guess that was in Perl 5?
moritz_ the programs, yes
masak moritz_: do you see any reasons not to use Rakudo for this? 17:14
moritz_ but I consider them just a prove-of-concept
masak ok.
moritz_ much more effort went into the docs themselves
masak: using Rakudo is good.
masak moritz_: where is perlhint?
moritz_ masak: misc/perlhints in the pugs repo 17:15
masak has a look
moritz_ it has too much duplicate information right now 17:16
masak moritz_: why is that?
moritz_ masak: because I was unsure of how much I needed 17:17
masak: for example token, name and id seem ... it seems overkill to have three of them
s/seem/
17:18 eternaleye left
masak moritz_: ah. 17:18
moritz_ but I like the label approach where you can say "this is an example", "this is a cross reference" etc. 17:19
masak I was thinking I'd use Pod for that. 17:20
haven't thought much about it, though.
17:21 dalek left, jnthn left
moritz_ I found it a bit too clumsy for that 17:21
but that's only personal preference 17:22
17:22 maerzhase1 joined
moritz_ that's why I used ultra-simplistic markup 17:22
17:22 maerzhase left
masak ok. 17:23
will mull over this.
moritz_ that can be parsed with minimal effort, and learned with minimal effort
(but doesn't offer rich possiblities - just key => value pairs)
masak why is it important that the markup be learned with minimal effort? 17:24
I mean, why is that more important than expressivity?
moritz_ it's not really all that important 17:25
more important for me was that I could talk about all kinds of crazy Perl 6 syntax without interfering with markup
masak yes, I can see how that might be an issue. 17:26
moritz_ afk
masak to think that this was only a year and a half ago...
I think Rakudo is to thank for it feeling much longer.
(that's a compliment, by the way. Rakudo++) 17:27
eric256 got pulled away to a metting 17:30
is there stuff in the pod spec for labeling methods, classes, etc, including paramters? or was that just talk on the mailing list 17:32
hard to keep this stuff all straight some times
masak eric256: ISTR there was some talk about JavaDoc-like functionality. I don't know if Damian changed anything because of it. 17:36
to make matters worst, S26 is a draft and out of date.
I plan to read first S26 and then STD.pm to try to get an understanding of Pod. 17:37
17:37 estrabd left 17:41 namenlos left
rakudo_svn r35533 | pmichaud++ | [rakudo]: spectest-progress.csv update: 282 files, 6233 passing, 0 failing 17:50
r35534 | jonathan++ | [rakudo] Ensure parse_name handles names with type-parametric parts.
17:51 [particle]1 joined, pdcawley joined, athenot left, athenot joined
rakudo_svn r35535 | jonathan++ | [rakudo] Handle the case where we have something like multi foo(::T) { ... } or role Foo[::T] { ... }. Added an assertion that isn't in STD.pm, but makes things somewhat easier when building the AST (otherwise we'd get a bogus empty node in the Signature - maybe STD.pm avoids 18:00
..this in some other way, but I can't quite see how).
moritz_ I don't think STD.pm really avoid the creation of empty nodes 18:02
18:03 [particle]2 joined
[particle]2 rakudo: my $s = '+a ++b-b -c d ++b ++/b-b +++c -d -e ++c -f ++/c +++/c -g'; $s ~~ / [ \b ('+''+'+)(<.ident>['-'<.ident>]*) \b ] ~ [ \b $0 '/' $1 \b ] .*? /; 18:04
p6eval rakudo 35532: OUTPUT«Statement not terminated properly at line 1, near ".*? /;"␤␤current instr.: 'parrot;PGE;Util;die' pc 129 (runtime/parrot/library/PGE/Util.pir:83)␤»
[particle]2 feh.
rakudo: my $s = '+a ++b-b -c d ++b ++/b-b +++c -d -e ++c -f ++/c +++/c -g'; $s ~~ / [ \b ('+''+'+)(<.ident>['-'<.ident>]*) \b ] (.*?) [ \b $0 '/' $1 \b ] /; 18:05
p6eval rakudo 35532: RESULT«Method 'perl' not found for invocant of class 'Match'␤current instr.: '_block14' pc 74 (EVAL_17:43)␤»
masak that's a known bug, I think.
[particle]2 rakudo: my $s = '+a ++b-b -c d ++b ++/b-b +++c -d -e ++c -f ++/c +++/c -g'; $s ~~ / [ \b ('+''+'+)(<.ident>['-'<.ident>]*) \b ] (.*?) [ \b $0 '/' $1 \b ] /; say $/[1];
p6eval rakudo 35532: OUTPUT«Use of uninitialized value␤␤»
[particle]2 what's known, the failure of goal matching syntax to work with non-string args?
or .perl?
masak that .perl isn't available on Match objects. 18:07
moritz_ [particle]2: it's known that <atom> ~ <atom> [ ... ] fails
[particle]2 ah, both :)
18:07 [particle] left
moritz_ [particle]2: if you're looking for a small project, a dump of Match objects would be *very* nice 18:08
[particle]2 thank larry there's more than one way to do it. if this were guido's language, we'd be screwed.
18:08 Patterner left
moritz_ even if eval($/.perl) wouldn't recreate the match object exactly, some dump as debuggin info would be great 18:09
[particle]2 moritz: i'm prototyping the api and grammar for a command-line parser atm
agreed that Match.perl is highly desirable 18:10
moritz_ I guess that's also necessary
[particle]2 it's necessary for me to get paid :)
avar somehow I read that as s/p/l/ 18:12
masak Freudian slip of the eye. 18:13
18:13 Psyche^ joined, Psyche^ is now known as Patterner 18:20 dalek joined 18:24 jnthn joined
masak ok, so I'm about to port SVG.pm from Perl 5 to Perl 6. 18:27
I find that the test suite doesn't really cover a lot of the module's functionality. 18:28
from what I can tell, it may even have grown as a sort of guard against regressions.
it definitely does not exercise all possible ways of using SVG.
so, what to do? just start from scratch? any ideas? 18:29
PerlJam I'd start with the tests that are there and incrementally adds ones to cover the functionality as it is ported 18:30
18:31 spx2 left
avar Or, if you just want to port something and don't need SVG in particular port something with proper tests 18:31
99% of cpan probably has "stuff that was screwed at one point" as opposed to anything near 100% coverage though:)
18:32 spx2 joined 18:35 maerzhase1 left
masak thank you both. 18:35
I do need SVG in particular.
18:39 alexn_org left 18:40 donaldh left 18:46 ft left 18:48 iblechbot joined
pugs_svn r24908 | jnthn++ | [t/spec] More tests for parametric roles, including use of parameters in methods. 18:57
jnthn OK, back in a bit...that patch gets parametric roles and methods within them working out about right, I think. 18:59
Some weird issue with attributes within them though.
rakudo_svn r35542 | jonathan++ | [rakudo] Parametric roles now clone methods, meaning that we get them attached to the right parameters. 19:00
19:01 braceta left 19:04 maerzhase joined 19:26 pmurias joined
pmurias hi 19:26
moritz_ hi pmurias 19:30
masak hello pmurias
19:31 [particle]2 left
szabgab rakudo: my $x; $x //= 'DEFAULT'; $x.say; 19:38
p6eval rakudo 35545: OUTPUT«Use of uninitialized value␤␤»
moritz_ szabgab: //= doesn't work atm 19:39
szabgab: there's a ticket for that already
szabgab moritz_: thanks 19:41
19:43 braceta joined
pmurias is code i intend to port to p6 when it's finished off-topic for the pugs repo? 19:47
moritz_ pmurias: the pugs repo has a fairly loose defintion of topic ;-) 19:49
jnthn pmurias: If it's useful to your work on p6 related stuff, then I'd say it's fine. :-)
moritz_ I certainly won't feel offended 19:50
TimToady: we've been discussing built-in testing and adverbs a few months ago - would you mind if I take the disucssion to p6l? 19:51
pmichaud moritz_: I think the answer will be "I don't mind", or "no need to ask". :-) 19:54
I don't speak for TimToady, but that's my guess.
jnthn is back from dinner 19:55
moritz_ pmichaud: probably, but since I have no hurry, it doesn't hurt to ask
pugs_svn r24909 | masak++ | [perlhints] fixed small typo
jnthn pmichaud: I just put a chunk of inline PIR into actions.pm, because we need to spit out the newclosure in the right place, so I couldn't put it in some PIR.
(IIUC)
pmichaud: If that's hateful to you, suggestions for how to do better welcome. :-) 19:56
pmichaud okay. But you should also be able to do a :pirop('newclosure') somehow, instead of the inline PIR. Still, I'm not adamantly apposed to the inline PIR. 19:57
I'll review it a bit later to double-check.
TimToady moritz_: well, depends on how much bikeshedding you want :) 19:59
moritz_ TimToady: I know, but I want to hear opinions
TimToady I don't have a problem with it, at least not yet... 20:00
moritz_ and maybe if we give them some food for thought, they spend less energy on .trim ;-)
(ok, that's only wishful thinking)
jnthn pmichaud: Review and let me know. There's a bit more to it than I can see being able to handle with a :pirop. 20:02
20:02 [particle] joined
pmichaud jnthn: that's kinda what I figure. :-) 20:04
20:07 masak left 20:10 dh_ joined
szabgab rakudo: if 1 { say 'ok' } 20:11
p6eval rakudo 35545: OUTPUT«ok␤»
szabgab strange, that does not work for me, checking 20:12
moritz_ szabgab: there are a few known parse failures when an 'if' or 'for' or 'while' is directly after a class (or sub?) definiton
szabgab ah, then that's the reason, thanks 20:13
moritz_ they are always WTF moments when I encounter them
szabgab adding a ; after the sub declaration helps
is there a test with this ? 20:14
moritz_ szabgab: there's a test case in the ticket, but not in the suite 20:15
szabgab I guess it is still in RT?
moritz_ aye 20:16
rt.perl.org/rt3/Ticket/Display.html?id=58544
no, that's the wrong one 20:17
rt.perl.org/rt3/Ticket/Display.html?id=57876
20:18 [particle]1 left
szabgab thanks 20:18
moritz_ np 20:19
20:21 pbuetow joined
pugs_svn r24910 | moritz++ | [t/spec] basic tests that bare print and say die 20:26
szabgab rakudo_svn: @a=(1,2); say @a[-1]; 20:32
rakudo: @a=(1,2); say @a[-1];
p6eval rakudo 35545: OUTPUT«Scope not found for PAST::Var '@a' in ␤current instr.: 'parrot;PCT;HLLCompiler;panic' pc 146 (src/PCT/HLLCompiler.pir:102)␤»
szabgab rakudo: my @a=(1,2); say @a[-1];
p6eval rakudo 35545: OUTPUT«Use of uninitialized value␤␤»
moritz_ rakudo: my @a=(1,2); say @a[*-1];
p6eval rakudo 35545: OUTPUT«2␤»
szabgab ah, I see now, thanks 20:33
but that's ugly :-(
moritz_ I won't argue on this one :/ 20:34
szabgab actually what index -1 means here ?
just another slot before the 0th ?
moritz_ I think so, yes
S09 has the gory details
but mostly it was to avoid accidental indexing from behind 20:35
szabgab yeah, looking at it
20:39 mberends joined 20:40 ovid joined
szabgab I have very strange problem in this script: svn1.hostlocal.com/szabgab/trunk/Pe..._Commas.p6 20:40
it works now, but if I comment out the @_.perl.say; 20:41
part in the first line of the sub, it starts printing foobarbaz
ovid End blocks in Test.pm are not firing. Well, I can put a say() statement in there and it sometimes fires ... at the beginning of the test script.
This means I can't implement no_plan for Test.pm 20:42
szabgab ovid: hi
ovid HI Gabor :)
mberends masak: RAKDO CAN HAZ SVG! autoexec.demon.nl/svg/
masak: RAKDO CAN HAZ SVG! autoexec.demon.nl:8888/svg/
moritz_ szabgab: I remember there was a problem with nested ?? !! - don't know if that still persists 20:43
eric256 mberends++
moritz_ szabgab: but it doesn't really look like that's the problem
szabgab moritz_: the strange thing is that it works if I have the first say in there
moritz_ szabgab: aye
szabgab: seems to be an array passing problem of some sort 20:44
jnthn ovid: Is that a pre-compiled Test.pm?
moritz_ szabgab: if you 'say commify_series(|@names);' it works
jnthn wonders how END blocks and pre-compilation go together...
szabgab what is that | there ? 20:45
ovid jnthn: All I know is that when I alter it, I can type make $test_name and all alterations are picked up, so I think it's being output as Test.pir and then run.
szabgab reading @09 ?
jnthn ovid: Ah.
20:45 donaldh joined
ovid ../../parrot perl6.pbc --target=pir --output=Test.pir Test.pm 20:45
jnthn ovid: I have a suspicion that END blocks and pre-compilation don't go together.
ovid Ouch :( 20:46
eric256 szabgab: i added a sig (@s) and s/_/s/g and it works too
jnthn ovid: Give me a few minutes... :-)
ovid Heh. OK.
moritz_ ovid: plan(*) is probably the more Perl6ish version of 'no_plan'
ovid I can change it once END blocks work. 20:47
moritz_ ovid: oh, and please keep the subs (like ok()) multis, even if you make the description optional
ovid: that way they are easier to extend/override
szabgab but I assume it *should* work the way I wrote too , without the |, or the @s right ?
ovid My friend Gabrielle is flying in and I won't have much time to work on Perl 6 for the next couple of weeks, so I hope to get Test.pm done before then :)
moritz_ szabgab: I think it should, because @_ should be slurpy 20:48
ovid moritz_: why so? We've got the core testing API stabilized for years. How will this help? (Not that I mind. Just curious)
szabgab oh, the data is there and it works iff I have the first say :-)
ovid It just seems like an awful lot of duplicated code.
szabgab ok, I am going in loop
s
20:49 meppl left
moritz_ ovid: when somebody wants to write another 'is' sub that compares things of a specific type, they can add a multi much easier if is() is declared to be a multi 20:49
rakudo_svn r35547 | bernhard++ | [codingstd] trailing spaces 20:50
moritz_ ovid: that doesn't imply that you have to duplicate any code, just add a 'multi ' before the sub, and keep the default value for the description
ovid Ah! I see what you mean. So long as I have $desc="" in the sig, that seems ok.
moritz_ (ideally we'd just declare protos, but I'm not sure how well rakudo does them atm) 20:52
jnthn moritz_: protos only work as tie-breakers at the mo. 20:53
Hope to get that aspect of them in shortlyish though. :-)
20:53 simcop2387 left
jnthn ovid: svn up and give it a try - may have fixed the lexicals in END blocks issue too with that patch. 20:53
ovid ok 20:54
moritz_ jnthn: no hurry
jnthn moritz_: It's on my Hague grant todo list. :-)
moritz_ ah ;)
jnthn moritz_: So if I want to have money for beer a couple of months down the line, hurry. ;-)
ovid Recompiling everything now (and hoping) :) 20:55
rakudo_svn r35548 | jonathan++ | [rakudo] Try to fix END blocks to work with pre-compiled modules and with lexicals. 21:00
r35549 | jonathan++ | [rakudo] Fix make test by tracking roles change in PIR tests for Perl6MultiSub.
r35550 | moritz++ | [rakudo] add test for bare 'say' being an error to t/spectest.data
21:01 vixey left
ovid I assume that with parrot, others are getting regular failures with t/library/protoobject.t? (test 8) 21:03
Wow. languages/perl6 make test just imploded quite spectacularly :)
I did make realclean, but checking out a clean parrot just to be sure. 21:04
jnthn ovid: Really? 21:05
ovid: Where?
ovid Yeah. Tons of failures looking just like this:
'Sub'
current instr.: 'parrot;Any;evalfile' pc 2908381 ((unknown file):-1)
t/spec/S02-builtin_data_types/mixed_multi_dimensional.rakudo ... Dubious, test returned 1 (wstat 256, 0x100)
No subtests run
Like I said, I'm checking out parrot into a new directory and will try again there, just in case I diddled something I should not have. 21:06
jnthn ovid: With no local diffs I get well past that in the spectests, so may be your working copy or something. 21:07
ovid Yeah, I borked Test.pm and forgot to revert it :) 21:08
moritz_ works here too
ovid: I did the same mistake today already ;-)
ovid I need to build better tools to manage this. i already have an alias to handle this, but I need it to run 'svn status' and abort if grep ^M 21:09
Actually, I need to write a bash script to do this. My alias is too long already.
21:12 simcop2387 joined, pmurias left 21:15 ruoso left
ovid Whee! My new 'reparrot' command works beautifully :) 21:16
jnthn Wow! The original Test.pm actually has an END block. Which now gives output...
ovid Sweet! 21:17
jnthn just ran a test invidivually and observed this
Sadly, the test failed. :-| 21:18
jnthn tries to work out the next bug with attributes in roles...
moritz_ can I use a parameterized role everywhere instead of a normal type? 21:19
jnthn Not *yet*. :-) 21:20
moritz_ but in theory I can? 21:21
justatheory hem?
jnthn Depends what you mean by "everywhere" :-)
pmichaud and "theory" :-)
and "can" :-)
moritz_ jnthn: for example in signatures
justatheory Can I what?
jnthn moritz_: Yes.
moritz_: I believe so.
But that's still very much todo. 21:22
moritz_ rakudo: role A[$x] { method f { 2 * $x } }; class B does A[3] { }; B.new.f 21:23
p6eval rakudo 35551: RESULT«6»
moritz_ jnthn++
that's really cool
ovid Sweet :) 21:24
moritz_ rakudo: role A[$x] { method f { 2 * $x } }; my $x = "foo" but A[5]; say $x.f
p6eval rakudo 35551: OUTPUT«elements() not implemented in class 'Perl6Role'␤current instr.: 'postcircumfix:[ ]' pc 2656 (src/classes/Positional.pir:108)␤»
jnthn moritz_: Note done the mix-in form yet.
*not
pasteling "ovid" at 91.85.167.13 pasted "t/library/protoobject.t has been failing for a while now" (14 lines, 250B) at sial.org/pbot/34477 21:26
ovid That output doesn't make a lick of sense to me.
moritz_ Exited with error code: [SIGNAL 10]
it's a bit confusing that the got/expected is shown even though it matches 21:27
jnthn The fact that it...yes, that.
ovid Whitespace? I doubt it, but since there aren't any quotes, tough to say.
21:28 ovid left
moritz_ ovid: the test script runs individual .pir files, and one of them parrot died, after it produced the desired output 21:28
21:29 ovid joined
ovid Missed me? 21:29
moritz_ ovid: run './parrot t/library/protoobject_8.pir' to get more information
ovid Will do in a moment.
pasteling "ovid" at 91.85.167.13 pasted "Bus error" (5 lines, 70B) at sial.org/pbot/34478 21:30
21:31 wolverian joined
moritz_ ovid: trac.parrot.org/parrot/newticket if you please ;-) 21:31
ovid Will do :)
Ah, sheesh. I don't know if I have an account there. Let me try various pass/email combos. 21:32
moritz_ it's pretty new
21:33 iblechbot left
ovid Yeah, I don't have an account on that (that I can tell) 21:33
21:36 thepler joined
ovid no_plan now works :) 21:37
I'll submit a bug to parrotbug until I get a Trac account. 21:38
jnthn ovid: Great! 21:40
ovid: Lexicals probably should work in end blocks too, if that was getting in your way before.
21:41 |jedai| left
jnthn Turns out attribute initialization in non-parameterized roles didn't work, let alone the parameterized ones. :-) 21:41
(Though there was an extra special bug for paramterized ones too. ;-))
21:43 |jedai| joined
szabgab rakudo: qw(A B C D).pick(5, 1).say; 21:43
p6eval rakudo 35551: OUTPUT«CBDDD␤»
szabgab rakudo: qw(A B C D).pick(5).say;
p6eval rakudo 35551: OUTPUT«DACB␤»
szabgab should not the latter give an error ?
moritz_ dunno 21:44
jnthn didn't know that pick took a second argument...what is that?
21:44 kisu left
szabgab allows repetition 21:44
jnthn Ah, OK. 21:45
szabgab default is no repetition so picking 5 out of 4 is sort of, impossible
moritz_ szabgab: qw() shouldn't work in Perl6, everything of the form identifier() is a sub call 21:47
szabgab: use <a b c d> instead
TimToady note that's supposed to be a named-only parameter
pugs_svn r24911 | jnthn++ | [t/spec] Test for attributes in roles having a default value.
TimToady rakudo: <A B C D>.pick(5, :repl).say
p6eval rakudo 35553: OUTPUT«DACBD␤»
TimToady so the positional call to it should be disallowed
moritz_ rakudo and parrot are too permissive in that respect
pugs_svn r24912 | jnthn++ | [t/spec] Tsts for using role parameters for attribute initialization. 21:48
szabgab moritz_: wel,, now qw() works in rakudo but I stop using it then :-)
21:49 maerzhase1 joined
TimToady std: qw(42) 21:50
p6eval std 24912: OUTPUT«Unknown routines:␤ qw called at 1 ␤00:05 85m␤»
rakudo_svn r35559 | jonathan++ | [rakudo] When composing attributes into a role, we need to carry any properties along with them, otherwise default values and traits and so forth don't work.
r35560 | jonathan++ | [rakudo] Since when we are parsing a signature we may be doing it for a package now (e.g. a parameterized role), we may already have got a block created. So check for this and use the existing one if so (otherwise we lose $block<pkgdecl>).
21:50 maerzhase left 21:51 kisu joined
szabgab TimToady: so you say that 1 as the second param should not be allowed just :repl ? 21:51
jnthn TimToady: Heh, I was about to say, "wouldn't it be better to make it named..." :-)
OK, that's enough parametric roles hacking for one day.
TimToady jnthn: also the parameter to the role needs to be treated as constant 21:52
found this necessary in STD
so that I could say match $stopper and have it treated as part of the token
jnthn TimToady: Clarify what you mean by "constant" 21:53
TimToady: Do you mean that
role Foo[$x] { ... }
class A does Foo['a'] { ... }
class B does Foo['b'] { ... }
Will not trample on each others $x? 21:54
TimToady the optimizer can bank on those not changing for the particular instantiation
in particular, regex semantics need to know what's part of the token, and that implies knowing that the parameter effectively constant
also means that the constant folder can go to work 21:55
jnthn Ah, OK. So you're saying that class A does Foo[$thingy_that_can_change] { ... } should be forbidden?
moritz_ or that a snapshot should be taken from $thingy_that_can_change?
jnthn Or rather, that we take a copy of it as it is at the time we do the role? 21:56
TimToady probably is nonsense to have an "is rw" param there
jnthn It's like an is copy but still readonly kinda thing. :-)
TimToady but since there is just one call for the given instantiation, is effectively constant
jnthn TimToady: Is moritz_ snapshot statement about right? 21:57
pasteling "ovid" at 91.85.167.13 pasted "Known bugs with *?" (25 lines, 1.5K) at sial.org/pbot/34479
ovid Tried to implement plan * as 'no_plan'
TimToady didn't see it
jnthn 22:55 <@moritz_> or that a snapshot should be taken from $thingy_that_can_change?
Basically, is it like we take a snapshot of what the value was?
TimToady yes, snapshot 21:58
jnthn OK.
That makes sense.
pmichaud ovid: those look like rakudobugs to me.
ovid Known? If not, I'll file 'em.
pmichaud I don't think known.
ovid OK.
pmichaud okay to file, even if they're dups :-)
moritz_ rakudo: multi sub plan(Whatever $x) { say "1..0" }; multi sub plan(Int $x) { say "1..$x" }; plan(4); plan(*)
p6eval rakudo 35553: OUTPUT«1..4␤1..0␤»
TimToady otherwise I see little benefit of normal parameterized code
jnthn I don't remember seeing a ticket, but I've seen the first bug.
ovid Ah. I'll stop bugging people about known bugs then :) 21:59
TimToady s/of/over/
jnthn TimToady: Indeed.
[particle] ovid is a known bugger
jnthn Makes sense.
21:59 cspencer joined
TimToady $dayjob & 21:59
ovid particle: shhh! Don't tell anyone!
Actually, that first one I think is a known bug because I seem to recall seeing it in some code where a trailing semi-colon is required. If I supply that, I get the second error. 22:02
22:03 wknight8111 joined
ovid Ah, so as long as I specify the type with a multi-sub , it works. Strange. I'll try that. 22:03
22:05 iblechbot joined
cspencer rakudo: my $x = *; say $x 22:08
p6eval rakudo 35553: OUTPUT«Whatever<0xb5fa2218>␤»
cspencer rakudo: my Int $x = *; say $x
p6eval rakudo 35553: OUTPUT«Type mismatch in assignment.␤current instr.: 'die' pc 15870 (src/builtins/control.pir:204)␤»
22:13 [particle]1 joined 22:16 thepler left
ovid Damn it. How many times am I going to send in patches and forget to attach them? 22:16
PerlJam as many times as it takes ;) 22:17
ovid :) 22:18
PerlJam ovid: since you're here and I'm curious ... Did you convince the BBC or are there other forward thinking people in there that all agreed to help perl 6 ?
ovid This chat is logged. I plead the fifth. 22:19
PerlJam fair enough
ovid++ anyway :)
ovid Let's just say there was a lot of internal support and excitement about it.
Thanks :)
Mmmm ... profanity (apropos of nothing, of course ...) 22:20
22:20 Casan left, donaldh left 22:22 [particle] left 22:25 |jedai| left, maerzhase1 left, |jedai| joined
ovid What I CAN say, and it can be repeatedly publicly so long as it's made extremely clear that this is my opinion and I'm not speaking for the BBC is that we have many excellent developers who are quite excited about Parrot and Perl 6 and they're very excited about its development and how we might use it in the future. 22:32
jnthn :-)
ovid The problem is that the BBC cannot be seen to endorsing "products" as this can distort the market. Just because the developers are excited about Parrot/Rakudo does not mean that it will be adopted. 22:33
dh_ if it ever gets released in the future, and when this happens, hopefully it is as least as fast as perl 5 :)
ovid Again, my opinion and not speaking for the BBC :)
PerlJam dh_: It'll be faster because programmers will be able to do more while writing less. 22:39
moritz_ dh_: I don't think that the first versions will be faster than Perl 5, because that has 20 years of optimizing behind it 22:41
ovid Yes, developer performance != CPU performance. 22:42
jnthn Indeed. It would be a huge diss to the Perl 5 developers to suggest that an unoptimized Perl 6 would beat a Perl 5 that's had so much effort put into it in terms of runtime performance.
moritz_ and I think that Perl 6 compilation will never be as fast as Perl 5 compilation 22:43
simply because in Perl 5 the compilation phase doesn't do all that much
heck, it's even faster to recompile a perl script than to dump the byte code to disk, and re-read it again
jnthn Aye. The parsing will speed up notably over where we're at now in Rakudo. But I agree with moritz_ - we're doing a lot more at compile time. 22:44
*but* running from the bytecode should be good :-) 22:45
22:45 jferrero joined
jnthn (Again, maybe not Perl 5 competitive at first. But I think over time we will have a good chance there.) 22:46
Optimzing before we have a *lot* of regression tests would also be a very bad idea.
moritz_ aye
and writing a type aware optimizer is probably a big task 22:47
araujo ovid, so when you guys in BBC plan to start migrating to Perl 6?
:-D
22:47 barney left
ovid Christmas. 22:47
PerlJam heh
ovid Well, hey ... that was just begging for that response :) 22:49
jnthn moritz_: Aye, I think there will be much fun to be had on type-based optimizations. But...later.
Parametric roles have hurt my brain enough this week.
moritz_ yes, such an optimizer is a well-past-rakudo-1.0-task 22:50
araujo packs the logs and send them to BBC ... (with a typo fix)
jnthn hopes to still be around to participate in writing such an optimizer
ovid hopes to be around after araujo sends those logs :) I was careful about what I said. I doubt they'll be too worried so long as I have enough disclaimers. Of course, that's only my opinion and not the official opinion of the BBC. 22:51
araujo haha
ovid jnthn: Are you worried you'll be gone for a while? 22:52
dh_ are you guys saying it is not possible to design a language that is both equally effecient for the programmer and computer resources?
ovid dh_ Optimization of dynamic languages still lag for behind optimization of static languages. 22:53
araujo ovid, don't worry, you are safe here ...
we even have TimToady
moritz_ dh_: I don't say it's impossible, I just say that it'll be a whole lot of work to optimize Perl 6
jnthn ovid: I don't see anything that'll cause me not to be involved in Rakudo for the foreseeable future, but who knows what life will throw. 22:54
22:54 ejs1 left
jnthn Right now, it's the most interesting project I work on. :-) 22:54
moritz_ dh_: otoh Perl 6 has many more possible hooks for optimization than Perl 5, so after some years or decades I guess that Perl 6 execute speed will be better than Perl 5's, hopefully way beyond
ovid Oh, and if anyone is concerned about my use of 'static' and 'dynamic', I like this description: www.pphsg.org/cdsmith/types.html 22:55
dh_ moritz, that's good to hear, that's what i am hoping for 22:56
PerlJam moritz_: it'll be hard to optimize the parser so much I think.
dh_ i certainly wouldn't expect perl6 to be faster right away
or as fast 22:57
moritz_ PerlJam: don't forget that the parser can, in theory, use pre-compiled DFAs for many regexes
PerlJam moritz_: let's just say I'm hopeful but pessimistic. 22:58
moritz_ ;-)
PerlJam I mean, I *know* the parser might be able to use pre-compiled DFAs ... I just think it's going to be a while before we get there and there will be other factors involved in parsing speed that may negate the advantages. 22:59
ask me about it again after the LTM gets integrated ;) 23:01
ovid I figure some developers might find this useful: use.perl.org/~Ovid/journal/38290 23:02
moritz_ reviews and tests ovid++'s Test.pm 23:03
justatheory werd up, ovid
ovid I'm trying to add the diagnostics, but so many bugs ... (including one that I can't replicate in a smaller test case!)
Thanks David :)
justatheory fer what? 23:04
moritz_ ovid: is there a good reason for plan(Whatever) not to set $testing_started? 23:05
ovid moritz_ Hmm, didn't think about that. Didn't impact the test suite passing, though.
I'll look again. 23:06
moritz_ ovid: afaict the only difference is that it will never output the "Looks like you failed $num_of_tests_failed tests of $num_of_tests_run" message, I think
pmichaud when looking at Test.pm, keep in mind two items: (1) there's a good case to be made that it should remain simple, so that other budding Perl 6 implementations can use it; (2) we expect/hope many of these functions to become part of the Perl 6 specification.
ovid justatheory: I thought "werd up" was a compliment. My bad.
:)
pmichaud (in which case Test.pm becomes obsolete, or part of the prelude) 23:07
ovid pmichaud: Interesting point. I thought about incorporating some of the new TAP diagnostics, but I thought a) they're not finalized and b) Test.pm needs to go away because it's too primitive.
moritz_ ovid: t/spec/S03-junctions/boolean-context.t fails with your Test.pm, but passes with the old one
ovid Hmm, not getting that failure. I'll look again. 23:08
pmichaud there's a purpose for having a primitive Test.pm (the idea that these may become part of the spec)
moritz_ pmichaud: from my POV it's ok to sacrifice some simplicity to get good diagnostics
pmichaud this isn't to say that there can't be a more sophisticated testing harness -- we just don't want to make that a requirement for using the spectest suite
justatheory ovid: Just saying hello. 23:09
pmichaud I'd like to avoid (or help others avoid) the problem I run into with Python-on-Parrot. Python has a test suite, but you can only use it if you have a fully implemented version of Python. 23:10
ovid pmichaud: the test harness consumes TAP. Test.pm produces it. Different things. That being said, I completely understand your point. Maybe I should write a different Makefile target which uses a more fully featured Test.pm (renamed, of course) to avoid said issues? 23:11
moritz_ pmichaud: sure, but that shouldn't keep us from improving Test.pm... as long as it's not hard to write a very primitive Test.pm when we start a new implementation
pmichaud moritz_: we're agreeing. :-)
moritz_ good ;-)
pmichaud thus I said "keep in mind" as opposed to "don't do it because..." :-) 23:12
I'm very receptive to improvements to Test.pm 23:13
ovid pmichaud: I agree, too. That's why I'm thinking there should be a separate Makefile target for the spectests and a better "Test.pm". I welcome other strategies, though. (And the fact that I'm soon on vacation and may not have much time is a frustration. I want to keep working on this) :)
pmichaud ovid: the issue will likely still be around when you return from vacation :-) 23:14
jnthn pmichaud: Fixing END blocks in pre-compiled modules gets tests spitting out a little more diagnostics at the end when run individually. :-)
justatheory ovid: Work on it during your vacation.
ovid Stop. Yer killin' me :)
pmichaud I'm find with separate makefile targets for a while.
*fine
moritz_ ovid: the tests that cause problems for me are of the form 23:15
(1|$undef && pass '1|undef in boolean context') || flunk '1|undef in boolean context';
ovid OK. I'll have to reconfigure vim to not kill all tabs. I'll take a s(tab) at it.
pmichaud make sure that the test functions accept Object and not Any
moritz_ ovid: in this case, poss pass() and flunk() are run
ovid moritz_: Ah, I just fixed that. The 'sub proclaim' needs to return '$cond' or "$passed" or whatever it's called.
moritz_ ovid: I thought along the same lines
ones that's fixed, I'm positive to check it in (awaiting pmichaud's veto) 23:16
but now: bed
jnthn pmichaud: ping 23:18
pmichaud pong
jnthn hi
Does PGE yet handle interpolating variables into tokens/rules/regexes?
As in token foo { $x }
pmichaud no. 23:21
that's coming relatively soon.
jnthn OK.
pmichaud (because when PGE was written, lexicals didn't work. Like, at all.)
jnthn *nod*
Will it be smart about things marked constant?
23:21 iblechbot left
pmichaud you mean for assignments? 23:21
jnthn Once we define a way of marking thingies constant... :-)
pmichaud I'm confused. 23:22
jnthn In my example above, if you know the variable $x cannot change, it'd be able to take its value at compile time and inline it?
As in, match its content as a literal.
ovid Since the test explicitly say 'use Test', is there yet a way to get Perl 6 to pick up a different path (e.g. perl6 -l lib/Test.pm)? I can't think of any other way for the tests to pick up a better Test.pm.
moritz_ ovid: set PERL6LIB env variable 23:23
pmichaud ovid: change PERL6LIB env var
ovid Ah. Thanks!
moritz_ ovid: or BEGIN { @*INC.push('/path/') }
pmichaud jnthn: I'm not sure that we can assume that $x doesn't change.
anyway, inlining it won't really be that big an improvement, I don't think.
jnthn pmichaud: In the case when we _know_ it can't?
pmichaud jnthn: I don't think there's a significant speed difference between looking up the PMC and simply having a string constant. 23:24
jnthn Maybe I should just point you upwards to earlier discussion with TimToady on this. :-)
ovid Gotta hit the hey. Thanks for everyone's help!
s/hey/hay/
jnthn But basically, when you pass parameters to a parametric role they become "constant".
As in, snapshot-ish.
23:24 ovid left
pmichaud sure, I followed that part. 23:24
oh, you want the _grammar_ to do the substitution for you? 23:25
jnthn No, no.
Well, I don't *think* so. :-)
pmichaud let me phrase it this way
jnthn I don't quite follow what needs to happen for the parsing stuff to work out.
pmichaud PGE will take "token foo { $x }" and compile it to a sub that does a lexical lookup of $x and attempts to match that value. 23:26
jnthn I just want to make sure however I make this happen, it's going to tie up with what you need.
pmichaud because PGE is really a regular expression compiler.
jnthn OK 23:27
Then I'll just make sure, you can do that. :-)
pmichaud it's possible that at the point of compilation we might know that $x is constant. But in the general case we don't know that, and in order to figure it out PGE would have to be smart enough to look at the outer PAST::* scopes to figure out that $x is really a constant lexical.
jnthn I was thinking more than if know we can label it with a properly.
*property
23:27 Limbic_Region joined
pmichaud (and yes, it does mean we have to figure out how to handle the case where $x is a package scoped variable) 23:27
by the time we get around to executing the regex, we could check if the variable has a constant property, but the regex itself is already compiled. 23:28
jnthn This is the difficultly I was seeing.
TimToady point is that the regex compiler has to know it
jnthn This is what I was picking up. 23:29
pmichaud okay.
TimToady or, actually, just the code that harvests the longest token
pmichaud that's a point I hadn't foreseen and we'll fix it.
TimToady (is how Cursor does it)
pmichaud oh, harvesting the longest token I had thought about handling, yes.
I knew I would have to handle that.
TimToady in particular, STD uses it for tokens embedded in parametric roles
expecting starters and stoppers to become part of LTM 23:30
pmichaud no problem.
PGE doesn't come anywhere close to handling that now.
It will.
TimToady line 2298 in STD.pm for instance
pmichaud (the "It will" part might be a problem for _me_, but I'm willing to tackle it when I get to that point.) 23:31
23:31 |jedai| left
moritz_ TimToady: speaking of which, should the ~ combinator backtrack to find the terminator? so will 'a' ~ 'a' 'a'* ever match? 23:31
(I don't care either way, just want to know if my test cases are correct)
23:32 |jedai| joined
TimToady I suppose it depends on :ratchet, but probably it would be dwimmier if it didn't autoratchet :) 23:32
pmichaud TimToady: S05 currently has 23:33
$<OPEN> = '(' <SETGOAL: ')'> <expression> [ $GOAL || <FAILGOAL> ]
which means after parsing an expression (greedy 'a'* ), we either match the goal or fail.
I'm thinking it should be
$<OPEN> = '(' <SETGOAL: ')'> [ <expression> $GOAL || <FAILGOAL> ]
(perhaps)
TimToady or something. STD doesn't even handle non-string goals yet... 23:34
pmichaud so that <expression> can backtrack if we don't immediately find the goal (and if expression is backtracking, which depends on :ratchet)
PGE is currently implementing the first version, which is what moritz++ is seeing.
TimToady I suspect we should do it right rather than wrong :) 23:35
pmichaud we're curious what "right" is :-)
I lean towards the second, but didn't know if there was a side effect I'm not seeing.
TimToady me too
pmichaud we can make it the second for now and see what side effects pop up :-) 23:36
TimToady TIAS
pmichaud okay, we'll do that. moritz: I'll make the change later tonight or early tomorrow.
TimToady also have an issue with the notion that make replaces what is returned by a rule, because under STD it must always be a Cursor that is returned
moritz_ pmichaud: ok 23:37
TimToady so currently it only replaces the scalar element
and then maybe we pull that out at the last moment somehow
otherwise we have a type issue
pmichaud "make replaces what is returned" ... you mean something other than the result object? 23:38
TimToady since we always have to return the new match state in order to continue anyway
the result object is in $($/), doesn't replace $/
pmichaud right.
TimToady is what I'm saying
pmichaud that's the way I've always understood it to be.
TimToady I think S05 isn't always clear on the subject 23:39
pmichaud okay.
certainly rakudo relies on that behavior.
but rules return Cursor objects, or Match objects, or there's a relationship between Match and Cursor I haven't detected? (I haven't looked at STD Cursor objects yet.) 23:40
because PGE/Rakudo has rules always returning Match objects.
TimToady maybe Cursor does Match, or something 23:41
pmichaud anyway, I've always understood "make" to set an element of a Match object, and not set the return of the match itself. 23:43
(where that element is the Match object's scalar value)
(or more precisely, how a Match evaluates if explicitly given an item context)
TimToady which is, unfortunately, somewhat ill defined 23:44
pmichaud yes.
TimToady a strong argument for an object always behaving as *itself* in item context 23:45
pmichaud anyway, over the next couple of weeks I expect to review STD Cursor (in order to do something similar in PGE), so I can hopefully speak on this more usefully then.
23:45 spx2 left
TimToady I'll be glad to answer Qs 23:45
pmichaud I'm certain I'll have them :-) 23:46
23:46 eric256 left
pmichaud afk for me for a bit 23:46
23:46 cspencer left
Limbic_Region pmichaud - when you come back, do any of the rakudo spec tests open a network socket? 23:48
the test suite is hanging for me, and the only thing I can come up with is my firewall is interfering 23:49
23:49 spx2 joined
jnthn Limbic_Region: AFAIK we don't implement socket IO in Rakudo yet. 23:50
Limbic_Region jnthn - I didn't think so either 23:51
23:51 wknight8111 left 23:54 wknight8111 joined 23:56 |jedai| left 23:57 |jedai| joined
pasteling "Limbic_Region" at 68.48.165.70 pasted "rakudo spectest output on Win32/MinGW with latest rev" (69 lines, 4.4K) at sial.org/pbot/34484 23:57
Limbic_Region jnthn - does that look right to you?
Limbic_Region hasn't been running spectest for quite a while 23:58

Powered by Cro and the Raku® Programming Language.

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

Thank you!