»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.
masak 'night, #perl6 00:08
TimToady o/ 00:09
00:22 MayDaniel left 00:23 benabik joined 00:31 PacoAir left 00:37 alester left 00:48 fridim_ joined 00:50 leprevost joined 01:02 scott__ joined, leprevost left 01:06 woosley joined 01:19 Patterner left 01:20 Psyche^ joined 01:22 Psyche^ is now known as Patterner 01:26 packetknife left 01:41 lestrrat joined
sorear good * #perl6 01:47
araujo sorear, o/ ! 01:54
02:22 Patterner left 02:30 leprevost joined, leprevost left, lestrrat left 02:31 lestrrat joined 02:35 Psyche^ joined, Psyche^ is now known as Patterner
colomon \o 02:43
03:09 whiteknight left 03:32 orafu joined 03:34 wolfman2000 joined 03:47 lutok left 03:48 pernatiy joined 03:54 jtpalmer joined 04:07 y3llow left, y3llow joined 04:32 mucker joined 04:41 wolfman2000 left 04:57 Trashlord left 04:58 Trashlord joined 05:03 birdwindupbird joined 05:26 sftp left 05:29 birdwindupbird left 05:54 birdwindupbird joined
moritz \o 05:57
TimToady yo
araujo o/ 05:58
moritz nqp: 1.?foo();
p6eval nqp: OUTPUT«Confused at line 1, near "1.?foo();"␤current instr.: 'nqp;HLL;Grammar;panic' pc 23433 (src/stage2/gen/NQPHLL.pir:6734) (src/stage2/gen/NQPHLL.pm:329)␤»
Teratogen nqp: say 1 + 1 06:02
p6eval nqp: OUTPUT«Confused at line 1, near "say 1 + 1"␤current instr.: 'nqp;HLL;Grammar;panic' pc 23433 (src/stage2/gen/NQPHLL.pir:6734) (src/stage2/gen/NQPHLL.pm:329)␤»
Teratogen what the
moritz nqp doesn't parse listops 06:03
needs parens around a sub call
Teratogen nqp: (say 1+1)
p6eval nqp: OUTPUT«Confused at line 1, near "(say 1+1)"␤current instr.: 'nqp;HLL;Grammar;panic' pc 23433 (src/stage2/gen/NQPHLL.pir:6734) (src/stage2/gen/NQPHLL.pm:329)␤»
Teratogen I give up
benabik Teratogen: Around the arguments
nqp: say(1+1)
p6eval nqp: OUTPUT«2␤»
Teratogen oh 06:04
=D
06:07 Tarmvred joined 06:10 kaleem joined 06:12 mikemol left, amirite joined
sorear (say 1+1) 06:13
sorear hugs Teratogen, good Lisp user
Teratogen haha
nqp: (say (1+1) 06:14
p6eval nqp: OUTPUT«Confused at line 1, near "(say (1+1)"␤current instr.: 'nqp;HLL;Grammar;panic' pc 23433 (src/stage2/gen/NQPHLL.pir:6734) (src/stage2/gen/NQPHLL.pm:329)␤»
Teratogen nqp: (say (1+1))
p6eval nqp: OUTPUT«Confused at line 1, near "(say (1+1)"␤current instr.: 'nqp;HLL;Grammar;panic' pc 23433 (src/stage2/gen/NQPHLL.pir:6734) (src/stage2/gen/NQPHLL.pm:329)␤»
Teratogen baw
06:14 amirite left 06:16 Woodi joined
benabik nom: (say ([+] 1, 2, 3)) 06:17
p6eval nom f9fcb2: OUTPUT«6␤»
moritz nom: (say (&infix:<+> 1, 2, 3)) 06:21
p6eval nom f9fcb2: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 2␤»
moritz nom: (say (infix:<+> 1, 2, 3))
p6eval nom f9fcb2: OUTPUT«===SORRY!===␤CHECK FAILED:␤Calling 'infix:<+>' will never work with argument types (int, int, int) (line 1)␤ Expected any of:␤ :(Any $x = { ... })␤ :(Any \$a, Any \$b)␤ :(Real \$a, Real \$b)␤ :(Int:D \$a, Int:D \$b)␤ :(int $a, int $b)␤ :(Num:D \$a…
moritz nom: (say (infix:<+> 1, 2))
p6eval nom f9fcb2: OUTPUT«3␤»
benabik moritz++ 06:22
sorear niecza: (say (infix:<+> 1, 2)) # cuirous 06:24
p6eval niecza v14-42-g1d02da9: OUTPUT«3␤»
Teratogen can you do postfix in perl? 06:27
err, perl 6? 06:28
moritz Teratogen: yes, see S03
sorear "no, you would need async I/O and better socket support to emulate postfix in perl 6"
++ is a postfix operator
moritz I thought ++ was a prefix operator :-) 06:29
sorear niecza: my $x = 3; say postfix:<++> $x; say $x
p6eval niecza v14-42-g1d02da9: OUTPUT«3␤4␤»
sorear it's that too
Teratogen cool
perl 6 does everything
it's like perl + lisp + forth
TimToady
.oO(+ Java + Haskell + BASIC + JS + ...)
 06:30
sorear meh, without homoiconicity perl 6 will never be lisp, and without the ability to run on PIIs (never mind 8085s) Perl 6 will never be Forth
TimToady if homoiconicity were the right thing, we'd work directly with DNA, not proteins :P
sorear oh, TimToady, that reminds me - JavaScript "var foo", Perl5 "my $foo" have strikingly similar semantics. Who stole what from whom? 06:31
TimToady we certainly didn't steal it from them
sorear (apologies if I asked this earlier and forgot)
sorear expects to get an answer like "both got it from $obscure_60s_mainframe_scripting_language" 06:32
TimToady well, Pascal had 'var', if I recall
sorear did pascal allow you to use "var" as an expression? 06:33
Teratogen speaking of obscure mainframe, I used to do multics pl/1 when I was at the Pentagon. 06:34
sorear what I really picked up on was that Perl 5 and JS don't really have initializers. "my $foo" is an lvalue expression
Teratogen: then you'll fit right in here.
Teratogen I had an honest to god VT-100 terminal on my desk.
and I set it to 132 column mode!
I also had a Tektronix 4014 next to my desk, and I used pl/1 to write fractal programs to display on the 4014. 06:35
sorear likewise "var foo = 5" means "(var foo) = 5"
Teratogen Those were the days!
TimToady in Pascal, var just introduce the section where you declare variables 06:36
benabik In my uni machine room, we have a VT-100 that's a serial terminal to all the servers.
TimToady *duced
Teratogen benabik, yes but is it an actual DEC VT-100, or some knockoff clone?
TimToady how fancy; my first CRT was a VT05
72x20, uppercase only 06:37
benabik Teratogen: I think it's a real one... I'll have to check when I work next.
Teratogen Before the VT-100 was the VT-52 I think.
It's bell was a bzzzzzzzzzzzt noise.
TimToady the VT05 was much longer than it was wide 06:38
Teratogen TimToady, I also worked with Decwriters.
They were great serial terminal/printers.
And built like a tank. You could probably drop one off a cliff and it would still work.
sorear vt100 probably wouldn't survive much of a fall, being made of glass and all that... 06:39
TimToady well, the VT05 was much nicer than the ASR33 I used before that... 06:40
the benefits of actually working for the computer center
sorear is, erm, 2e1 06:41
TimToady I thought you were 2.1e1
moritz O(2e1) :-)
sorear closer to 2.13e1, but that's just precision
Teratogen ASR33 was a teletype with a side order of paper tape, right? 06:42
TimToady yup, 10 characters per second
and LOUD
Teratogen We had those in high school, along with 300 baud modems with cups on them. 06:43
When we ran out of paper we'd steal rolls of paper towels from the bathroom.
sorear did people ever use hearing protection in your computer installation(s)?
TimToady eh, what did you say?
sorear you said LOUD not BRIGHT :P
Teratogen We used the 300 baud modems to connect to an HP 2000 minicomputer downtown. 06:44
They would run HP timesharing basic during the day then shut that down so they could run card fortran at night.
HP timesharing basic had the coolest name for the substring function: SST. =D 06:45
TimToady used to be, when the lights flickered, I'd call up the computer's modem bank and whistle at it; if it whisled back, I knew the computer was still up; if not, I knew I'd better head to the computer center
Teratogen The Concorde SST was just making its maiden flight around then, I think, so that was pretty cool.
TimToady our boss in the computer center used to work for HP, but had to switch to DEC when he came :)
sorear TimToady: you could hit a tone the modems understood? 06:46
TimToady I could warble the tone appropriately enough by poking my cheek
sorear is quite impressed 06:47
TimToady a 110 baud modem is not difficult to fool
Teratogen I wrote an hp timesharing basic program that would print large letters on paper tape, some idiot from central high school broke into our account, stole my program, and put his name on it.
plagiarization!
so one of my friends broke into central high school's account, and deleted all their files. 06:48
TimToady you should be flattered
Teratogen BUT HEY IT WAS ALL IN GOOD FUN!
high school rivalry you know
Let's see what else, I was an explorer scout that was hosted by Honeywell so I got to play around with GCOS. 06:49
And in college we had a hulking Univac 1108. 06:50
sorear Teratogen: where in the world was this? 06:51
Teratogen Arizona
TimToady the first time I played adventure was on a DECSYSTEM-20 in the lobby of the DECUS convention in Las Vegas at the MGM Grand Hotel that eventually burned and kill 85 people
sorear I suppose there is probably more than one "Central High" in the US, huh
Teratogen and later on, I was stationed at the Pentagon (those were the Multics years)
06:51 takesako joined
TimToady where I grew up it meant Central Kitsap County 06:51
06:52 replore_ joined
sorear Did it do the burning and the killing during DECUS? 06:52
TimToady I've programmed in Pascal on an Amdahl, and in Algol on a Burroughs mainfraim
Teratogen Although I also got to play around with the first 16 bit computers to arrive at the Pentagon, the Zenith Z-100 and Z-150.
TimToady sorear: nah, that wasn't till 1980 06:53
sorear first computer I used ...
a 486 running Windows 3.11 *lameness*
Teratogen Let's see, I was in high school in the early 1970s, my first computer was the HP-2000.
My first personal computer was an Atari ST. 06:54
TimToady, there was an Algol compiler for the HP-2000, I never fooled around with it, but my friend (the one who deleted central high school's files) wrote a few programs for it. 06:55
06:57 takesako left, wtw joined 06:58 takesako joined
TimToady the first thing I ever programmed had nixie tubes for a display 06:58
Teratogen sorear, Windows 3.1 with Trumpet Winsock and Netscape Navigator on a 28.8k modem was actually pretty peppy back in the day. 06:59
06:59 zby_home left
dalek kudo/sink: 5081a51 | moritz++ | src/ (3 files):
Apply traits in sink context

  ... except in Perl6::World, where
   if pir::can__IPs($ret, 'sink') { $ret.sink }
mysteriously dies because it can't find method sink in Associative. WTF. Also doesn't fix 'use Test;'. Not sure what else is missing.		 07:00
TimToady is looking forward to seeing "Useless use of" warnings from rakudo soon 07:01
Teratogen Or, "Usage is deprecated" ?
TimToady those aren't so useful 07:02
eval: use warnings; 1 + rand; 0
buubot_backup TimToady: Useless use of addition (+) in void context at (eval 20) line 1. 0
TimToady eval; print (1+2),3 07:03
buubot_backup TimToady: 33
TimToady eval: print (1+2),3; 0 07:04
buubot_backup TimToady: 30
TimToady eval: use warnings; print (1+2),3; 0
buubot_backup TimToady: print (...) interpreted as function at (eval 20) line 1. Useless use of a constant (3) in void context at (eval 20) line 1. 30
TimToady eval: use warnings; my $a = 1,2; 0 07:05
buubot_backup TimToady: Useless use of a constant (2) in void context at (eval 20) line 1. 0
TimToady very much more useful than mere deprecations; Useless use warnings often point out a misunderstanding of syntax or precedence 07:06
Teratogen I that's right, it is Java that has all of those deprecated classes. =)
07:08 raoul joined
moritz if we want to be able to generate such useelss use warnings, we need purity annotations or analysis 07:12
TimToady but you need that for constant folding anyway
moritz well 07:13
07:13 Woodi left
moritz it's not that easy 07:13
what would you expect from class A { ... }; A.new() + 1 ? 07:14
07:14 Woodi joined
moritz always warn? only warn if A.Numeric can be determined to be pure? never warn? 07:14
TimToady I would expect + to know it's pure, even if A isn't
moritz then it doesn't use the same kind of purity information as the constant folder 07:15
TimToady the constant folder needs to know more, sure
07:16 mucker left
TimToady but the purity of + is one of things it needs to know 07:16
moritz well, not all + candidates are really pure, which is what I'm trying to say :-) 07:17
TimToady then they probably violate the purity declared on the proto, assuming it's declared there, and assumed part of the API of all + operations
sorear writing impure .Numeric methods is erroneous, so the compiler sould warn 07:18
after all, nowhere do we spec the number of times .Numeric is called by a given compound operation
moritz sorear: that doesn't make writing an impure .Numeric operation errornous
just very stupid
TimToady I'm fine with outlawing impure operations derived from pure protos 07:19
or at least, forcing the declarator of such to work harder
like, by providing a less pure proto
I much prefer people to come up with new operators than to overload existing ones with bogus and unexpected semantics 07:20
Teratogen <@-^-@> 07:21
I have no idea what that does.
But it would be a cool operator.
moritz nom: say <@-^-@>
p6eval nom f9fcb2: OUTPUT«@-^-@␤»
sorear Teratogen: <...> is quote syntax, trampling on that with an operator is considered mildly unkind 07:23
moritz well
there's also < as an infix
and < ... > as a postcircumfix
TimToady perl6: sub infix:['<@-^-@>'] ($a, $b) { @$a (^) @$b }; say <a b c> <@-^-@> <b c d> 07:24
moritz so we already have 3 uses; adding more would be quite confusing
p6eval rakudo f9fcb2: OUTPUT«===SORRY!===␤Preceding context expects a term, but found infix ^ instead at line 1, near ") @$b }; s"␤»
..niecza v14-42-g1d02da9: OUTPUT«set(a, d)␤»
..pugs b927740: OUTPUT«*** ␤ Unexpected ":['<@-^-@>']"␤ expecting "::", "handles", "is", bare trait, subroutine parameters, trait or block␤ at /tmp/hmRx5XM_7K line 1, column 10␤»
TimToady niecza++
Teratogen yikes!
moritz in rakudo you currently need to write infix:«...» for that
TimToady perl6: sub infix:«<@-^-@>» ($a, $b) { @$a (^) @$b }; say <a b c> <@-^-@> <b c d>
p6eval pugs b927740: OUTPUT«decodeUTF8': bad data: '\171'␤decodeUTF8': bad data: '\187'␤*** ␤ Unexpected "(^)"␤ expecting operator or "}"␤ at /tmp/Vznq2HX7wJ line 1, column 36␤»
..niecza v14-42-g1d02da9: OUTPUT«set(a, d)␤»
..rakudo f9fcb2: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 2␤»
TimToady but rakudo doesn't have the Set op yet, alas 07:25
moritz and maybe it also doesn't LTM enough
TimToady I think it probably does...
perl6: sub infix:«<@-^-@>» ($a, $b) { @$a ~ " and " ~ @$b }; say <a b c> <@-^-@> <b c d> 07:26
p6eval pugs b927740: OUTPUT«decodeUTF8': bad data: '\171'␤decodeUTF8': bad data: '\187'␤*** No such subroutine: "&infix:@-^-@"␤ at /tmp/ow_srTtqa4 line 1, column 59 - line 2, column 1␤»
..rakudo f9fcb2: OUTPUT«===SORRY!===␤Non-declarative sigil is missing its name␤at /tmp/fiKr62pYu9:1␤»
..niecza v14-42-g1d02da9: OUTPUT«a b c and b c d␤»
TimToady well, maybe not
Teratogen: but you see why LTM (longest token matching) is so essential to the extensibility of Perl 6 07:27
Teratogen Not really, because I don't know what LTM is. =)
TimToady which is why I defined it in parens
Teratogen is it like greedy regular expression matching?
TimToady yes, like that 07:28
whenever the grammar has a set of alternative tokens, it picks the longest first
so more to do with what it does over | than it does with *
but similar notion
Teratogen Maybe I am old fashioned and showing my age here, but I wish Perl 6 would have included a plain old text preprocessor, like the one C language has. *sigh* 07:29
TimToady when we say "infix" it is really identifying an alternation that happens to be named rather than using |
Teratogen I know Perl 6 has a super duper macro preprocessor of some sort. 07:30
TimToady well, yes, except for the "pre" part, which is evil
Perl 6 is very strict about doing one-pass parsing
sorear Perl 6 is supposed to include everything but the kitchen sink ... that includes a Filter::Util::Call replacement
s/include/make possible/ 07:31
TimToady the problem with multi-pass parsing is that the input language and the output language are two different languages, which confuses people no end 07:32
not to mention theparser 07:33
the only way for a parser to know exactly which language it's parsing at any point is to do it one pass 07:34
anything else ends up being a form of russian roulette
jnthn morning, #perl66 07:35
TimToady you think you're getting away with it, and having fun, till you don't, and aren't
o/
07:35 mj41 joined
jnthn ooh, sticky 6 key... 07:36
TimToady as long as it doesn't stick more than that, nobody will mistake you for the antichrist
hmm 07:38
perl6: sub infix:«<\@-^-\@>» ($a, $b) { @$a ~ " and " ~ @$b }; say <a b c> <@-^-@> <b c d>
p6eval rakudo f9fcb2: OUTPUT«===SORRY!===␤Non-declarative sigil is missing its name␤at /tmp/d2oEDh6Vxa:1␤»
..niecza v14-42-g1d02da9: OUTPUT«a b c and b c d␤»
..pugs b927740: OUTPUT«decodeUTF8': bad data: '\171'␤decodeUTF8': bad data: '\187'␤*** No such subroutine: "&infix:@-^-@"␤ at /tmp/5qdwbLarWE line 1, column 61 - line 2, column 1␤»
TimToady I guess it's not a problem of interpolating insice «», or a different problem 07:39
perl6: sub infix:<\<@-^-@\>> ($a, $b) { @$a ~ " and " ~ @$b }; say <a b c> <@-^-@> <b c d> 07:40
p6eval niecza v14-42-g1d02da9: OUTPUT«===SORRY!===␤␤Non-declarative sigil is missing its name at /tmp/dSv3zN2sS9 line 1:␤------> ) { @$a ~ " and " ~ @$b }; say <a b c> <⏏@-^-@> <b c d>␤␤Use of uninitialized value in string context␤ at /home/p6eval/niecza/b…
..pugs b927740: OUTPUT«*** ␤ Unexpected "infix"␤ expecting "=", subroutine parameters, trait or block␤ at /tmp/txZxhMdPBl line 1, column 5␤»
..rakudo f9fcb2: OUTPUT«===SORRY!===␤Non-declarative sigil is missing its name␤at /tmp/KxFGJ0mcUH:1␤»
jnthn yay, my favorite coffee machine is working again today...
TimToady hmm, that oughta work, methinks
cognominal___ Tdratogen: one of the problem with preprocessors is that you can't stack them because they suppose they parse a instance of the original language.when the very purpose of them is to change it.
TimToady perl6: sub infix:<< <@-^-@> >> ($a, $b) { @$a ~ " and " ~ @$b }; say <a b c> <@-^-@> <b c d> 07:42
cognominal___ the fashion for source filters did not last in Perl 5
p6eval pugs b927740: OUTPUT«*** No such subroutine: "&infix:@-^-@"␤ at /tmp/TkcixSjT5y line 1, column 63 - line 2, column 1␤»
..rakudo f9fcb2: OUTPUT«===SORRY!===␤Non-declarative sigil is missing its name␤at /tmp/6GVRdsDUit:1␤»
..niecza v14-42-g1d02da9: OUTPUT«a b c and b c d␤»
sorear o/ jnthn 07:44
Teratogen perl 5 had -P =P
I wonder who added THAT
sorear most likely TimToady sometime around Perl 2.0 07:45
Teratogen actually
sorear several other languages from Perl's ecological niche support using the C pre-processor even when nothing like C
Teratogen Perl 5 HAS -P.
sorear right, Perl 5 people hate removing features. 07:46
Teratogen Of course it usually doesn't work on Windows, because there is no C Preprocessor on Windows (usually).
sorear there was a riot earlier this week over the fact that 5.16 removes defined(@array)
do you want to guess when it was formally deprecated?
jnthn Last millenium? :) 07:47
doy $ perl -P # Unrecognized switch: -P (-h will show valid options).
?
sorear jnthn: very good! 1998, 5.005_something
Teratogen doy, what the 07:48
-P has been removed from Perl!?
sorear my Debian-shipped 5.12.0 also fails -P
sorear half-guesses doy is on 5.16 or bleadperl 07:49
doy looks like it was deprecated in 5.10 and removed in 5.12
(i'm on 5.14.2)
we're really trying to start removing things these days
convincing people is hard, but we're trying 07:50
(:
Teratogen Why not integrate a text preprocessor into Perl 5.10+?
It would make it two-pass, but hey, preprocessors are fun! 07:51
sorear because you could just use Filter::Simple and a pure-Perl cpp
and it would be more portable that way
?
Teratogen is Filter::Simple a core module?
doy yes
Teratogen Nifty.
doy we haven't gotten rid of source filters yet, unfortunately
sorear how *would* you get rid of source filters without breaking D::D? 07:52
doy by getting rid of d::d
sorear you think people can live without mx:d et al?
:)
wait, this isn't #moose, I should probably write the module names in full 07:54
Devel::Declare, MooseX::Declare
doy no, i mean - we have better non-source-filtery options in the works
sorear ah
doy lots of work to make the lexer and parser more pluggable directly 07:55
sorear \o/
doy the p5-mop syntax is actually using one of the new prototypes for that 07:58
moritz can I see the p5-mop syntax somewhere? 08:02
nom: pir::say pir::can__IPs(Associative, 'sink') 08:03
p6eval nom f9fcb2: OUTPUT«1␤»
doy github.com/stevan/p5-mop/blob/mast...-example.t
moritz nom: Assocative.sink
p6eval nom f9fcb2: OUTPUT«===SORRY!===␤CHECK FAILED:␤Undefined routine '&Assocative' called (line 1)␤»
doy it's still very much in progress, but that's what it looks like currently
moritz doy: thanks 08:04
doy the test suite covers a decent range of things 08:05
moritz though it's really class syntax, not MOP syntax 08:06
doy what would 'mop syntax' even mean? 08:07
08:07 replore_ left
doy i was just using 'p5-mop' as the name of the project 08:08
moritz well, Perl 6 has the $class.^meta_method syntax for accessing the MOP
erm, s/class/object/
08:08 replore joined 08:09 replore left 08:15 replore_ joined 08:18 Patterner left 08:19 pomysl_ joined 08:29 fridim_ left 08:34 Psyche^ joined, Psyche^ is now known as Patterner
au whee, got a PD strategy that reliably wins over all other strategies in rakudo (not yet tested niecza). 08:37
moritz PD? 08:38
au iterated prisoner's dilemma
moritz \o/
au: then submit away :-)
au nah :)
I'll just post here instead: gist.github.com/1851858 08:39
you'll see why I won't submit :)
sorear \o/ au \o/
colomon :p
moritz :-)
sorear long time no see
au # 792 au 600 bad good random 502 evil-tit-for-tat tit-for-tat 08:40
sorear: indeed!
moritz wait, WHAT? :-)
ah, now I understand 08:41
au: it does deserve a dishonorable mention :-)
au then mention away :) 08:42
sorear au: What do the numbers mean? 08:43
au sorear: they're respective scores from each strategy
output from moritz++'s gist.github.com/1710688
sorear Why are there numbers before only some strategies? 08:44
I haven't been playing this game.
au it's 600 bad, 600 good, 600 random
I just merged them a bit
moritz that solution does unify the numbers a bit :-) 08:45
au the-monkey-ministry-of-truth.p6
moritz au: I expected a submission somewhat like yours (though I expected fiddling with the harnesses lexicals), but not from you :-) 08:47
au well, the aim of this strategy is precisely to game against harness lexicals
since I expect they'll just tweak truth value but not the .does'ness 08:48
sorear why does gaming .Stringy affect anything? 08:49
au sorear: the lookup table is %scoring{"$current-a $current-b"}.
moritz I thought more along the lines of pir::find_caller_lex__PS('@scores') = (1000, 0);
sorear ah
au moritz: *nod* though that's rakudo specific
sorear au: how've things been for yu?
moritz au: right
sorear @CALLER::scores = (1000, 0); # (: 08:50
moritz sorear: yeah, but according to spec that's only allowed if @scores is declared as contextual, no?
au also that kinda depends on the lexical order of your strategy 08:51
sorear moritz: the specs seem to be very confused about what contextual means
au so you'd need to call your strategy name something like "\x00"
sorear but I concede that
moritz au: well, it'd need to check &a and &b too, to do it the right way 08:52
au sorear: things are good. I've been @Paris this week and staying for two more weeks.
sorear TimToady has made clear that most variables are supposed to be protected from monkeying by default
au moritz: yup. tbh I started with that but couldn't quite get it to work without using rakudo internals
sorear awesome 08:53
sorear wants to be a world traveller someday
moritz @CALLER::scores = &Caller::a === $?ROUTINE ?? (1000, 0) !! (0, 1000) # though NYI in rakudo
au exactly.
moritz oh wait
using === on closures is also a bad idea 08:54
sorear &CALLER::, ne?
moritz yes
and &?ROUTINE
so, clearly untested code :-)
rakudo now has .static-id to compare closures while disregarding their clonedness
nom: sub gen($x) { sub () { $x } }; say gen(3) == gen(5); say gen(3).static_id === gen(5).statid_id 08:55
p6eval nom f9fcb2: OUTPUT«No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤:(Mu:U \$v, Mu %_!)␤␤ in method Numeric at src/gen/CORE.setting:657␤ in sub infix:<==> at src/gen/CORE.setting:2297␤ in block <anon> at /tmp/nyvUb6opy7:1␤␤»
moritz nom: sub gen($x) { sub () { $x } }; say gen(3) === gen(5); say gen(3).static_id === gen(5).statid_id
p6eval nom f9fcb2: OUTPUT«False␤Method 'statid_id' not found for invocant of class 'Sub'␤ in block <anon> at /tmp/c_rpT6JxfG:1␤␤»
moritz nom: sub gen($x) { sub () { $x } }; say gen(3) === gen(5); say gen(3).static_id === gen(5).static_id
p6eval nom f9fcb2: OUTPUT«False␤True␤»
sorear cute, although I still think there should be a way to talk about subs without talking about *any* instantiation
moritz meh. Helps to spell
sorear: probably, yes 08:56
(fwiw I've needed that for the internals of the backtrace printer)
au -> breakfast + more touristy stuff & 08:57
08:57 replore_ left
sorear -> sleep 08:58
09:01 replore_ joined
au moritz: another won't-submit-and-please-don't-mention strategy is this... 09:07
Evil => -> *% { for %strategies.keys.sort -> $k { print 1 if $k eq 'Evil'; say "0\t$k" }; exit }
but I'm sure you've thought of that already :
:)
moritz au: :-) 09:08
I did think of fiddling with %strategies, though not in that way
au I guess this is a bit faster. maybe put it in an INIT block for additional performance gains :) 09:09
au goes afk for real to refrain from turning a simple friendly PD game into CoreWars 09:10
jnthn moritz: Not got a build handy to ry this but 6guts.wordpress.com/2012/01/29/this...omment-148
*try
moritz jnthn: I'll try it 09:11
nom: pir::say pir::can__IPs(Associative, 'sink')
p6eval nom f9fcb2: OUTPUT«1␤»
moritz nom: Associative.sink 09:12
p6eval nom f9fcb2: OUTPUT«Method 'sink' not found for invocant of class 'Associative'␤ in <anon> at src/gen/Metamodel.pm:1163␤ in block <anon> at /tmp/vmOvXm44UM:1␤␤»
09:12 Trashlord left
moritz jnthn: that one currently blocks me from properly sinking traits 09:12
jnthn moritz: er, associate is a role...
*Associative
moritz ... and?
jnthn Just wondering where it's showing up 09:13
In its type object form
moritz jnthn: it's a return value from a trait
09:13 replore_ left
jnthn oh. 09:13
moritz probably from 'does Associative' or so (while compiling the setting) 09:14
jnthn nom: role Foo { }; pir::say pir::can__IPs(Foo, 'bar') 09:15
p6eval nom f9fcb2: OUTPUT«1␤»
jnthn ah
yeah, pir::can calls find_method. On a role this always returns something that when invoked puns the role
It would probably pun and then check. 09:16
And only then return something.
moritz eeks.
jnthn s/would/should/
nom: say so Associative.can('sink') 09:17
p6eval nom f9fcb2: OUTPUT«False␤»
jnthn That is more honest :)
I can probably fix it so the pir::can is honest too :) 09:18
moritz I'd prefer that to emitting .can('sink') calls
jnthn github.com/rakudo/rakudo/blob/sink...Punning.pm 09:20
the find_method in there wants a tweak
instead of just returning the closure, do
pir::can__IPs($!pun, $name) ?? ...the closure... !! nqp::null()
(by "the closure" I mean the pointy at the end)
09:21 kaleem left, kaleem joined
moritz I'll try that 09:23
09:26 ab5tract left 09:27 ab5tract joined
moritz fwiw I can't reproduce the hang in the UP6 example 09:30
with newest rakudo, that is (haven't tried with the release)
and I've checked various typical errors (additinal newline at the end, file not found) and none gave me a hang 09:35
jnthn oddness 09:38
moritz also I don't understand "Program stop on: for $file.lines -> $line" 09:39
where does he get that information?
09:39 cogno joined
jnthn Maybe by putting print statement befoer that line and seeing it outputs? :) 09:40
09:50 woosley left 09:51 cogno left 10:12 scott__ left 10:23 noam left, noam joined 10:30 tarch joined 10:33 noam left, JimmyZ_ joined, noam joined 10:34 noam left 10:35 noam joined
masak good almost-noon, #perl6. 10:47
moritz good lunchtime, masakist 10:48
masak vimeo.com/36579366 has exceeded all my expectations. you must now all watch it.
moritz if that's the one from Mr. Victor, I must disappoint you 10:49
I've already watched it :-)
masak then you don't have to watch it again. 10:50
everyone else does, though.
moritz and yes, it is very inspiring.
it made me want to write Term::ReadKey in Perl 6, so that we can get closer to interactivity
10:51 Tarmvred left
Teratogen noscript is saying no to so many sites on that vimeo url 10:52
I give up. It's just not going to play.
moritz yes, it's quite a hassle. But well worth it
you can just make a temporary excpetion for all sites. 10:53
Teratogen is it on youtube?
masak not that I know of. 10:54
Teratogen ok let me try "allow all scripts" 10:55
although I normally don't do that
masak moritz: when I watch this, I feel like he's successfully written a lot of tools that have only been a glimmer in my eye, or even less than that. 10:58
JimmyZ_ can't viist vimeo.com/36579366 10:59
*visit
masak vimeo is petty-bourgeoise and must be suppressed! 11:00
JimmyZ_ I can't search "学习" in google too 11:02
masak wtf 11:03
JimmyZ_ I don't think study in google should be suppressed :( 11:04
moritz I think China has a political interest to make Google less useful than Baidu
masak to be honest, I don't think anything should be suppressed. :(
JimmyZ_ well, baidu search result is filterd by polity 11:06
*filtered 11:08
11:11 tarch_ joined 11:12 tarch left 11:15 tarch__ joined, tarch_ left
moritz niecza: class A { method x($y) { say $y }; method call-x() { $.x(3) } }; A.new.call-x 11:15
p6eval niecza v14-42-g1d02da9: OUTPUT«Unhandled exception: No value for parameter '$y' in 'A.x'␤ at /tmp/6atFpM7H7T line 0 (A.x @ 1) ␤ at /tmp/6atFpM7H7T line 1 (A.call-x @ 3) ␤ at /tmp/6atFpM7H7T line 1 (mainline @ 4) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3773 (ANON @ 3) ␤ at /h…
moritz that one totally messes up XML::Writer on niecza
moritz submits nieczabug 11:16
masak why is it that all software feels like crap after watching someone like Bret Victor?
moritz because he spends an highly above-average effort on polishing his stuff 11:17
and not just polishing
11:18 daxim joined
masak sure it's polished. but even the underlying ideas should have more recognition and be more commonly known. 11:18
moritz perl6: class A { method x($y) { say $y }; method call-x() { $.x(3) } }; A.new.call-x
p6eval niecza v14-42-g1d02da9: OUTPUT«Unhandled exception: No value for parameter '$y' in 'A.x'␤ at /tmp/G8ufQvzFQQ line 0 (A.x @ 1) ␤ at /tmp/G8ufQvzFQQ line 1 (A.call-x @ 3) ␤ at /tmp/G8ufQvzFQQ line 1 (mainline @ 4) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3773 (ANON @ 3) ␤ at /h…
..rakudo f9fcb2: OUTPUT«3␤»
..pugs b927740: OUTPUT«*** ␤ Unexpected "-"␤ expecting "::", "handles", "is", bare trait, subroutine parameters, trait or block␤ at /tmp/3xe7cdqicc line 1, column 47␤»
11:21 MayDaniel joined 11:27 Tarmvred joined 11:37 birdwindupbird left 11:38 birdwindupbird joined 11:42 am0c joined 11:48 pernatiy left
masak nom: sub step($s) { $s.subst(/^(\(+)(\)+)\(/, -> $/ { [~] "()" x ($1.chars - 1), "(" x ($0.chars - $1.chars + 2), ")" }) }; .say for "()()()", &step ... "((()))" 11:54
p6eval nom f9fcb2: OUTPUT«()()()␤(())()␤()(())␤(()())␤((()))␤»
masak niecza: sub step($s) { $s.subst(/^(\(+)(\)+)\(/, -> $/ { [~] "()" x ($1.chars - 1), "(" x ($0.chars - $1.chars + 2), ")" }) }; .say for "()()()", &step ... "((()))"
p6eval niecza v14-42-g1d02da9: OUTPUT«()()()␤(())()␤()(())␤(()())␤((()))␤»
masak \o/
11:54 ab5tract left
Teratogen masak, can I ask why you used a sigil on step (&step) 11:57
masak because I talk about the function as an object. 11:58
without the sigil, it's a function call.
Teratogen I thought in Perl 5 at least supposed to be moving away from using sigils for subroutines
masak yes, but Perl 5 has a different convention there.
Teratogen ok
Don't get me wrong, I love sigils.
masak basically, in older Perls you had to use the '&' sigil to make a function call.
Teratogen And I think sigil invariance (is that what it is called) is a big mistake. 11:59
masak there were syntaxes such as '&foo()' and 'do foo' to make calls.
sigil invariance is what Perl 6 does. where @array is always @array, even when you @array[$index].
I've met others who think it's a bad idea too. it's a small minority, though. 12:00
Teratogen Yes, I don't like sigil invariance.
masak feel free to load a pragma to make things interesting for you, then :)
Teratogen to me, $array[$index] makes a lot more "sense".
Maybe that's because I have never tried to program in Perl 6.
masak ok.
we won't change it just for you. :) 12:01
Teratogen baw 12:02
12:07 snearch joined 12:08 ab5tract joined, cogno joined
masak eval: my $L = 13; my $s = "[]" x $L; my $t = 1; until ($s eq "[" x $L . "]" x $L) { $s =~ s/ ^ ( \[+ ) ( \]+ ) \[ / "[]" x (length($2) - 1) . "[" x (length($1) - length($2) + 2) . "]" /xe; $t++ }; print $t 12:11
buubot_backup masak: 7429001
masak eval: my $L = 13; my $s = "[]" x $L; my $t = 1; until ($s eq "[" x $L . "]" x $L) { $s =~ s/ ^ ( \[+ ) ( \]+ ) \[ / "[]" x (length($2) - 1) . "[" x (length($1) - length($2) + 2) . "]" /xe; $t++ }; $t 12:12
buubot_backup masak: 742900
masak that's better :)
this takes 3 seconds in the Linux VM on my laptop.
12:12 jferrero left
Teratogen Speaking of VMs, I am thinking about getting a VPS. 12:12
masak the corresponding Perl 6 program (based on those above) is still running, and will probably run for an hour.
so, for this particular use case, with just a lot of strings being generated, and no IO in the middle, there's a factor of 20 difference between Perl 5 and Perl 6. 12:13
12:14 fsergot joined
masak fsergot! \o/ 12:15
fsergot hi o/ :)
jnthn masak: Would be interesting to profile it. 12:16
fsergot how are You masak? :)
masak fsergot: I am good fsergot how are you_ 12:17
?
fsergot masak: I'm good too. :) 12:18
12:20 whiteknight joined 12:21 noam left, noam joined
masak fsergot: written anything cool in Perl 6 yet? 12:24
jnthn nom: say "ice" 12:25
p6eval nom f9fcb2: OUTPUT«ice␤»
jnthn It's easy to write cool things in Perl 6 \o/
masak nom: say Cool
p6eval nom f9fcb2: OUTPUT«Cool()␤»
12:27 pmurias joined 12:31 noam left 12:32 noam joined
moritz works on prisoner's dilemma visualizations 12:35
it seems that generating the visualizations takes about 10x more time than running the strategies :-)
12:37 cogno left
fsergot masak, not yet. I'm still learning p6 :) 12:42
moritz the two don't contradict :-)
fsergot masak: i hope it'll change in the near future. :) 12:43
:)
Right :) 12:44
masak fsergot: it's not possible to learn Perl 6 well without writing it. at least not for me. 12:51
by which I mean "writing Perl 6 code", not necessarily "implementing Perl 6 itself"
fsergot masak: I write a lot of code. :) 12:52
but this code isn't cool :P 12:54
masak hehe 12:57
perl6: say "This code ", <is isn't>.pick, " cool."
p6eval rakudo f9fcb2: OUTPUT«This code isn't cool.␤»
..pugs b927740, niecza v14-42-g1d02da9: OUTPUT«This code is cool.␤»
masak :)
see? whether code is cool or not is a matter of perspective :) 12:58
flussence perl6: say <solid liquid gas plasma handwavey-sciencey-stuff>.pick; # annual vapourware status check :) 12:59
p6eval rakudo f9fcb2: OUTPUT«liquid␤»
..niecza v14-42-g1d02da9: OUTPUT«gas␤»
..pugs b927740: OUTPUT«plasma␤»
moritz :-) 13:00
fsergot :)
masak Niecza is clearly vapourware.
probably because it doesn't have a homepage :P
flussence it's cookin'!
masak .oO( this implementation's a gas! )
Teratogen o/~ hey good lookin' / whatchya got cookin'? o/~
13:02 GlitchMr joined 13:05 noam left 13:07 noam joined 13:11 bluescreen10 joined
masak nom: sub foo { my $value = "HAHAHA DISREGARD THAT"; our sub bar { say $value } }; our &bar; bar() 13:20
p6eval nom f9fcb2: OUTPUT«Any()␤»
moritz perl6: sub foo { my $value = "HAHAHA DISREGARD THAT"; our sub bar { say $value } }; our &bar; bar(); foo; bar 13:21
p6eval niecza v14-42-g1d02da9: OUTPUT«Any()␤Any()␤»
..pugs b927740: OUTPUT«*** Cannot use Undef as a Code object␤ at /tmp/Rcrpv9o_AR line 1, column 88-93␤»
..rakudo f9fcb2: OUTPUT«Any()␤HAHAHA DISREGARD THAT␤»
moritz I'd submit a bug report if I knew which one is wrong :-)
masak rakudo is correct here. 13:22
&bar gets re-bound upon &foo entry.
13:24 shinobicl_ joined
[Coke] (multics years) weird, I know someone whose nickname is multics. 13:30
shinobicl_ perl6: class D is Date { has Bool $.ok is rw; multi method new(Str $datespec, Bool $ok) { $.ok = $ok; nextwith($datespec) } }; my D $var = D.new("2010-01-01", Bool::False); 13:42
p6eval rakudo f9fcb2: OUTPUT«Cannot look up attributes in a type object␤ in method ok at src/gen/CORE.setting:1392␤ in method new at /tmp/eVkxdjuy2k:1␤ in block <anon> at /tmp/eVkxdjuy2k:1␤␤» 13:43
..niecza v14-42-g1d02da9: OUTPUT«===SORRY!===␤␤Unhandled trait Date for this context at /tmp/khN8rum5Ql line 1:␤------> class D is Date ⏏{ has Bool $.ok is rw; multi method new(␤␤Potential difficulties:␤ $var is declared but not used at /tmp/khN8rum5Q…
..pugs b927740: OUTPUT«*** Class prototype occured where its instance object expected: D␤ at /tmp/5KDykIpPlf line 1, column 84-94␤ /tmp/5KDykIpPlf line 1, column 84-94␤»
shinobicl_ i'm not sure how should i use "nextwith"
masak perl6: sub foo { constant value = "OH HAI"; our sub bar { say value } }; our &bar; bar() 13:44
p6eval niecza v14-42-g1d02da9: OUTPUT«Potential difficulties:␤ &foo is declared but not used at /tmp/AsRm9qPC9W line 1:␤------> sub foo ⏏{ constant value = "OH HAI"; our sub bar␤␤OH HAI␤»
..rakudo f9fcb2: OUTPUT«OH HAI␤»
..pugs b927740: OUTPUT«*** ␤ Unexpected "="␤ expecting ":" or "("␤ at /tmp/Lz81jx2WOV line 1, column 26␤»
masak perl6: sub foo { constant $value = "OH HAI"; our sub bar { say $value } }; our &bar; bar()
p6eval niecza v14-42-g1d02da9: OUTPUT«Potential difficulties:␤ &foo is declared but not used at /tmp/3HoPPSo5BX line 1:␤------> sub foo ⏏{ constant $value = "OH HAI"; our sub ba␤␤OH HAI␤»
..rakudo f9fcb2: OUTPUT«OH HAI␤»
..pugs b927740: OUTPUT«*** Cannot use Undef as a Code object␤ at /tmp/iBo9fGxXXM line 1, column 79 - line 2, column 1␤»
moritz shinobicl_: first you can't assign to $.ok in method new, because no object has been created yet 13:45
masak shinobicl_: .new is usually a "static method" (or "class method") in Perl 6.
shinobicl_: generally, a constructor in Perl 6 is just a method that calls .bless on a type object. 13:46
(which is why, as moritz++ said, you can't access the attributes, because 'self' is the type object and not the new instance) 13:47
moritz nom: class D is Date { has $.ok; method new(Str $datespec, $ok) { my $new = self.Date::new($datespec); $new.MYBUILD($ok) }; submethod MYBUILD($ok) { $!ok = $ok } }; say D.new('2012-02-12', True);
p6eval nom f9fcb2: OUTPUT«Nominal type check failed for parameter '$datespec'; expected Str but got Int instead␤ in method new at /tmp/3IQniH5cEy:1␤ in method new at src/gen/CORE.setting:8967␤ in method dispatch:<::> at src/gen/CORE.setting:784␤ in method new at /tmp/3IQniH5cEy:1␤ in bloc…
shinobicl_ if the object does not exists yet, how can i use "self" then? 13:48
moritz shinobicl_: in a class method, 'self' refers to the type object
13:49 Trashlord joined
moritz nom: class D is Date { has $.ok; method new(Str $datespec, $ok) { my $new = callwith($datespec); $new.MYBUILD($ok) }; submethod MYBUILD($ok) { $!ok = $ok } }; say D.new('2012-02-12', True); 13:49
p6eval nom f9fcb2: OUTPUT«Nominal type check failed for parameter '$datespec'; expected Str but got Int instead␤ in method new at /tmp/mvhbMEHl6n:1␤ in method new at src/gen/CORE.setting:8967␤ in <anon> at src/gen/Metamodel.pm:2533␤ in call_with_args at src/gen/Metamodel.pm:2530␤ in block…
moritz why Int?
masak no idea. 13:50
jnthn And, uh, why is your new ABOVE a new method in CORE.setting in the backtrace?
[Coke] moritz: (self refers to class in class method) Ew. 13:51
[Coke] would prefer it were spelled "class" then.
jnthn [Coke]: It refers to the type object.
There IS no class object in Perl 6.
[Coke] moritz: (self refers to type in class method) Ew. 13:52
moritz it is consistent with the rest of the p6 OO model 13:53
jnthn It works out pretty well in general.
[Coke] tries to find the code snippet moritz was commenting on.
jnthn I do miss it being this way in toher languages.
*other
13:54 bluescreen10 left
[Coke] ah, this is related to the distinction between BUILD and new, which is also ew. 13:55
[Coke] looks forward to reading the "perl6 for java programmers" guide that explains all of this to Coke et al. ;)
moritz something is very Ew in the whole example
dalek kudo/sink: f9e18ea | moritz++ | src/Perl6/ (2 files):
fix pir::can on roles. jnthn++
 13:56
moritz but I'm a bit distracted
perl6: my @a = [1, 2], [3, 4]; say @a.map({.reverse.item}).perl 13:58
p6eval rakudo f9fcb2, niecza v14-42-g1d02da9: OUTPUT«([2, 1], [4, 3]).list␤»
..pugs b927740: OUTPUT«(2, 1, 4, 3)␤»
masak [Coke]: +1 on what moritz and jnthn is saying. 'self' referring to the type object is perfectly a-OK with the rest of Perl 6. 14:00
[Coke]: if you want the invocant of a "class method" to be called '$class', consider naming the invocant in the method signature. 14:01
also, cf. Perl 5, where exactly the same rules apply (except that Perl 5 doesn't even have a default term 'self').
shinobicl_ class D { has $.x; has $.y; multi method new(:x!, :y!) { my $ret = self.bless(*, :x, :y) } }; my D $var = D.new(x=>100, y=>200); say $var.perl; 14:05
oops sorry didnt meatn to post that here
masak thanks anyway. looks good. :) 14:06
the only reason you're declaring your own .new there is because your nameds are required, I guess?
shinobicl_ rakudo: class D { has Int $.x; has Int $.y; multi method new(:$x!, :$y!) { my $ret = self.bless(*, :$x, :$y); return $ret } }; my D $var = D.new(x=>100, y=>200); say $var.perl; 14:07
p6eval rakudo f9fcb2: OUTPUT«D.new(x => 100, y => 200)␤»
masak hm, if you make it multi, does that mean that a call without the required named arguments dispatches to the .new in Mu?
moritz yes
masak rakudo: class D { has Int $.x; has Int $.y; multi method new(:$x!, :$y!) { my $ret = self.bless(*, :$x, :$y); return $ret } }; my D $var = D.new(x=>100); say $var.perl
p6eval rakudo f9fcb2: OUTPUT«D.new(x => 100, y => Int)␤»
masak \o/
shinobicl_ i want to have a subclass of Date, and i want to implement as many constructors as Date has
masak ah.
PerlJam multis++ 14:08
14:09 bluescreen10 joined
moritz it might be a bit harder than necessary because the Date constructors aren't writen with inheritance in mind 14:09
it is often not a good idea to inherit from value types; if you do, you have to take extra care 14:10
shinobicl_ so, my problem now is that i want to have a Date plus the x and y members.... so i must call "nextwith" and "glue" it the x and y values
and i'm not sure how to do that
PerlJam shinobicl_: Why do you want a Date to also have and X and Y ? 14:12
14:12 tokuhirom joined
shinobicl_ ok X and Y is just an example, but what i want to do is to add a _complex_ attribute to a date. This attribute is an object itself 14:13
14:13 kaleem left
shinobicl_ i want to represent a date PLUS a "calendar" object 14:13
class Workdate is Date { has WorkdayCalendar $.calendar is rw ........
so i can have a constructor liek this: 14:14
my Workdate $w_date = Workdate.new(year=>2011, month=>12, day=>09, calendar=>WorkdayCalendar.new($calendar));
moritz shinobicl_: also Date objects are immutable. If you make a mutable subclass, make sure to override the WHICH method
otherwise you'll break a whole lot of assumptions
14:14 kaleem joined
shinobicl_ moritz: noted 14:14
so... if i want a constructor like that, i should overload the new methoids in "Date" so they accept the $.calendar attribute, or use a 'default' new calendar if not specified. Something like this: 14:16
(defined inside Workdate class...) multi method new(:$year!, :$month, :$day, :$calendar = WorkdayCalendar.new) { $.calendar = $calendar; return nextwith(year=>$year, month=>$month, day=>$day); } 14:17
well i know that this is not gonna work... is just to express the general idea
moritz nom: class D is Date { method new(|$args) { nextwith(|$args) } }; say D.new('2011-01-14'); 14:21
p6eval nom f9fcb2: OUTPUT«2011-01-14␤»
moritz nom: class D is Date { method new(|$args) { nextwith(|$args) } }; given D.new('2011-01-14') {say .WHAT; .say };
p6eval nom f9fcb2: OUTPUT«D()␤2011-01-14␤»
moritz nom: class D is Date { method new(|$args) { my $obj = callwith(|$args); $obj } }; given D.new('2011-01-14') {say .WHAT; .say };
p6eval nom f9fcb2: OUTPUT«D()␤2011-01-14␤»
moritz nom: class D is Date { has $.x is rw; method new(|$args) { my $obj = callwith(:$x, |$args); $obj.x = $x; $obj } }; given D.new('2011-01-14') {say .WHAT; .say }; 14:23
p6eval nom f9fcb2: OUTPUT«===SORRY!===␤Variable $x is not declared␤at /tmp/3fPKg8BDkK:1␤»
moritz nom: class D is Date { has $.x is rw; method new(:$x, |$args) { my $obj = callwith(|$args); $obj.x = $x; $obj } }; given D.new('2011-01-14') {say .WHAT; .say };
p6eval nom f9fcb2: OUTPUT«D()␤2011-01-14␤»
moritz nom: class D is Date { has $.x is rw; method new(:$x, |$args) { my $obj = callwith(|$args); $obj.x = $x; $obj } }; given D.new('2011-01-14', :x<foo>) {say .WHAT; .say .x.say};
p6eval nom f9fcb2: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 2␤»
moritz nom: class D is Date { has $.x is rw; method new(:$x, |$args) { my $obj = callwith(|$args); $obj.x = $x; $obj } }; given D.new('2011-01-14', :x<foo>) {say .WHAT; .say; .x.say};
p6eval nom f9fcb2: OUTPUT«D()␤2011-01-14␤foo␤»
moritz shinobicl_: is that something you can work on? 14:24
s/on/with/
shinobicl_ yes :D thanks... i just don't understand the |... IIRC that was the "flatten" operator...
moritz correct 14:25
in the signature, it captures all the rest of the argument list
and in the paramter list, it interpolates everthing from $args in the parameter list
shinobicl_ oh, so instead of bless, you create a new object with callwith (that calls the parent 'new'). Now that the object exists you can assign $.x
moritz yes (though inside new it's not called $.x) 14:26
shinobicl_ (brb)
14:28 am0c left
shinobicl_ cool... it now works as i expected... thanks moritz... now i think i can finish my Workdaycalendar library :D 14:30
thank you very much! going back to work... bye!
moritz thinks that mutable subclasses of value types are a kind of a Liskov violation 14:32
shinobicl_ i'm not sure if i understand that whole sentence... hehehe, but i think you are suggesting that i should look for another way to implement that 14:33
moritz shinobicl_: I'd use delegation
ie your object has a date, and not is a date 14:34
shinobicl_ i used to have workdate as a parametric role ($calendar was the parameter). I 14:35
.
masak moritz: odd -- I feel Liskov is so much easier to uphold if the objects are read-only. :)
no way to call .setWidth and .setHeight on the derived Square, etc.
14:36 ab5tract left
moritz masak: yes, that's why it feels weird to introduce mutability in subclasses 14:36
shinobicl_ i'm not familiar with "delegation"... i'll read about that, and maybe return with another few questions :)
moritz nom: class A { has $.str handles <Str split>; } say A.new(str => 'a sentence with spaces).split(' ').perl
p6eval nom f9fcb2: OUTPUT«===SORRY!===␤Confused␤at /tmp/_9p47gi072:1␤»
shinobicl_ the mutability is about the "is rw" part of the $.calendar member? 14:37
moritz nom: class A { has $.str handles <Str split>; }; say A.new(str => 'a sentence with spaces).split(' ').perl
p6eval nom f9fcb2: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 2␤»
moritz shinobicl_: yes
nom: class A { has $.str handles <Str split>; }; say A.new(str => 'a sentence with spaces').split(' ').perl
p6eval nom f9fcb2: OUTPUT«("a", "sentence", "with", "spaces").list␤»
14:37 PacoAir joined
shinobicl_ oh i'll remove that part then. i'm, not expert on p6 oop yet :) i put "is rw" in all my class members hehe 14:38
because until now i didn't understand clearly how to make constructors of that 'bless' part
now i think i got it 14:39
masak I avoid "is rw" on anything whenever I can. 14:40
that's why it isn't the default :)
shinobicl_ nom: class D is Date { has $.x is rw; method new(:$x, |$args) { my $obj = callwith(|$args); $obj.x = $x; $obj } }; given D.new('2011-01-14', x=>"foo") {say .WHAT; .say; .x.say}; 14:41
p6eval nom f9fcb2: OUTPUT«D()␤2011-01-14␤foo␤»
[Coke] moritz++ # handles example
shinobicl_ nom: class D is Date { has $.x; method new(:$x, |$args) { my $obj = callwith(|$args); $obj.x = $x; $obj } }; given D.new('2011-01-14', x=>"foo") {say .WHAT; .say; .x.say};
p6eval nom f9fcb2: OUTPUT«Method 'static_id' not found for invocant of class 'Mu'␤»
moritz oops, that's an error from within the backtrace printer
shinobicl_: wihtout 'is rw' you can't simply assign to $obj.x 14:42
shinobicl_: you could for example add a private method that sets $!x, and call that from within .new
shinobicl_ ok... seems fine to me. 14:43
14:46 jferrero joined
masak for those morbidly interested in macros: my latest musings -- gist.github.com/1853560 14:48
14:53 birdwind1pbird joined 14:54 PacoAir left
dalek kudo/nom: 1e966bb | moritz++ | src/core/Backtrace.pm:
fix thinko in backtrace printer
 14:54
14:55 PacoAir joined
moritz masak: good read. It sounds as if you've understood macros now. 14:59
(well, that part of macros)
moritz managed not to include a "finally" in that sentence o/
masak :) 15:00
jnthn masak has FINALLY understood macros?!
:P
masak the fixup at B was something I hadn't understood until this Sunday.
jnthn All this "Let's talk about X" is making me expect a "WAT" somewhere in here soon :P 15:03
ooh, there'll be a ninaskajp tomorrow 15:07
crap, wrong window!
moritz masak++ # macro post
jnthn masak: nitpick: B need not happen either...
(that is, potentially a macro may be unused) 15:08
moritz and thus boring :-)
jnthn is also pondering how well this will fall out of the closure handling he's doing in the serialization work already 15:09
(for the pre-compiled case)
If it's done by actually have a dummy closure being taken and passed in to the quasi incarnation, then we steal the context from that at fixup time, it probably works out very easily. 15:11
...to the extend that closure serialization is easy. :) 15:12
*extent
moritz well, it's not easy, but jnthn++ does it. So easy for masak++ ;-)
jnthn :P 15:13
Sadly, we can't yet serialize ASTs, since they're PAST nodes. Another reason we need QAST. 15:14
masak jnthn: I pondered mentioning that B needn't happen either. since you saw it, I'll point it out. 15:18
15:20 pomysl_ left, pomysl joined, pomysl left, pomysl joined 15:22 noam left 15:26 noam joined 15:27 pernatiy joined, birdwind1pbird left 15:28 snearch left, mtk joined 15:29 cogno joined 15:32 whiteknight left, whiteknight joined 15:37 wtw left 15:39 kaare_ joined 15:40 colomon joined 15:41 ab5tract joined 15:43 Psyche^ joined 15:45 kst joined 15:46 thou joined
jnthn time for the long journey home & 15:47
15:47 Patterner left, Psyche^ is now known as Patterner 15:50 cogno left 15:51 sivoais joined
gfldex nom: my @a 1,2,3; say [+]@a; 15:54
p6eval nom 1e966b: OUTPUT«===SORRY!===␤Confused␤at /tmp/fAcbghOfZs:1␤»
gfldex nom: my @a 1,2,3; say [+] @a;
p6eval nom 1e966b: OUTPUT«===SORRY!===␤Confused␤at /tmp/pM8afx1sOf:1␤»
gfldex nom: my @a=1,2,3; say [+]@a;
15:54 cosimo joined
p6eval nom 1e966b: OUTPUT«===SORRY!===␤Confused␤at /tmp/N8kuFNoBKZ:1␤» 15:54
gfldex nom: my @a=1,2,3; say [+] @a;
p6eval nom 1e966b: OUTPUT«6␤»
gfldex std: my @a=1,2,3; say [+]@a; 15:55
p6eval std 52f3895: OUTPUT«===SORRY!===␤Two terms in a row (listop with args requires whitespace or parens) at /tmp/df5cE2jCJ6 line 1:␤------> my @a=1,2,3; say [+]⏏@a;␤ expecting any of:␤ POST␤ argument list␤ bracketed infix␤ infix or meta-infix␤
..postfix…
gfldex i like std's error message a lot better :)
15:57 MayDaniel left 16:00 tarch__ left 16:01 am0c joined 16:02 sftp joined 16:05 tarch__ joined, cogno joined
masak we all do :) 16:06
16:06 tokuhirom left 16:07 tokuhirom joined 16:11 tokuhirom left 16:12 tarch__ left 16:13 birdwindupbird left 16:14 cogno left, GlitchMr42 joined 16:18 GlitchMr left 16:21 MayDaniel joined
moritz on my way home I realized why some stuff in the sink branch fails 16:23
traits often return type objects
and I'm putting them in sink context
and List.sink wasn't safe when called on the type object
dalek ast: 2b82f29 | (Solomon Foster)++ | S03-operators/set.t:
Add proper subset tests.
 16:24
ast: 9c421d3 | (Solomon Foster)++ | S03-operators/set.t:
Add superset tests. (with a little help from Rop!)
masak moritz: in other words, the method was written not taking type object invocation into account? 16:26
16:27 tarch__ joined 16:28 kst left, emss joined 16:29 kst joined 16:32 tarch__ is now known as tarch, kaleem_ joined 16:34 kaleem left 16:35 kaleem_ left
masak $ ./perl6 -e 'macro foo { my $a = "OH HAI"; quasi { say $a } }; foo' 16:39
Null PMC access in find_method('gist') in sub say at src/gen/CORE.setting:5932
(this is locally, on the macros2 branch)
I just had a look at the PIR, and to me it *looks* as though the block from the quasi has the correct OUTER. 16:40
16:40 MayDaniel left
masak does anyone have a good way to explain the Null PMC access? 16:40
TimToady does it have it at the right time? 16:41
16:41 emss left 16:42 slavik1 joined
TimToady till bs lands, rakudo still has phase issues 16:42
masak well, the PIR is the product of the compilation, so the macro should've done all it wants by then.
I can nopaste the PIR if that helps. 16:43
TimToady won't help me :)
masak :)
16:43 MayDaniel joined
TimToady has too few braincells to waste any on PIR 16:43
masak hardly helps me either. I find PIR harder to read under nom.
moritz masak: try with --ll-exception 16:44
masak oh! 16:45
moritz jnthn: regarding github.com/rakudo/rakudo/blob/nom/...ld.pm#L764 how can the invocation of $precomp ever work if $precomp is false?
masak I always forget about --ll-exception :) 16:46
moritz jnthn: it seems to be that it should be if $precomp { $precomp(|@pos, |%named) } else { $compiler_thunk() } 16:47
16:48 GlitchMr42 is now known as GlitchMr
moritz I'm getting a failure while trying to invoke Any 16:49
16:50 am0c_ joined, am0c left
masak moritz: stacktrace doesn't tell me much, I think: gist.github.com/1854332 16:51
16:52 aoeuhtns joined
moritz masak: most likely $a ends up being the Null PMC 16:53
lines 4 and 5 are the mainline
masak: am I right in guessing that the macro application just pushes PAST nodes onto the current scope (and doesn't enclose anything in a PAST::Block or so)? 16:54
TimToady masak: re irclog.perlgeek.de/perl6/2012-02-17#i_5162698 maybe we need 'use MONKEY_VALUES;' to turn on mutability :) 16:57
masak moritz: I don't remember whether the PAST::Block gets included or not. I can check. 16:58
yes, clearly the Null PMC is related to $a somehow.
TimToady moritz: re irclog.perlgeek.de/perl6/2012-02-17#i_5162590 I wish we could find some systematic way of detecting when someone has written a method that should be a submethod
masak or rather, to the failed lookup of $a.
TimToady: heh, I don't find 'is rw' as horrible as monkey patching :P 16:59
TimToady the average Haskell programmer would classify them in the same bucket :)
maybe we should change 'is rw' to 'is mutable-under-monad(IO)' :) 17:00
masak locks TimToady inside a monad until he lets go of his axes to grind :P 17:01
TimToady you're think of it inside out; it's not a monadstery, but a monadic tribal existence 17:02
*ing
au "wow, all these tourists live behind iron bars!" 17:04
-- monadic monkey 17:05
pmurias -/exit 17:06
17:06 pmurias left, mikemol joined 17:07 pmurias joined
masak .oO( the OOK monad ) 17:07
moritz is that related to OOKness?
TimToady noodnood
17:09 raoul left 17:12 mj41 left 17:15 pernatiy left 17:16 kst left 17:19 kst joined 17:23 lateau_ joined 17:26 daxim left 17:27 JimmyZ_ left 17:30 lateau_ left, pyrimidine left 17:32 whiteknight left 17:36 pyrimidine joined 17:38 lateau_ joined 17:39 pmurias left 17:41 Chillance joined 17:43 mucker joined 17:44 mucker left 17:48 whiteknight joined 17:50 pyrimidine left 17:54 kst left 17:55 simcop2387 joined
[Coke] # 02/17/2012 - niecza at 99.88% 17:57
"niecza", 20055, 8, 751, 1573, 22387, 23257
"rakudo", 20078, 28, 610, 1863, 22579, 23267
(no change)
17:58 kst joined
masak interesting how, as Niecza and Rakudo started comparing test pass numbers, they "got stuck" only a few dozen tests apart. :) 17:59
[Coke] observational bias?
masak not if you count actual spectests passed. Niecza hasn't been passing this many for more than a month or so. 18:00
whereas Rakudo's level has been much more stable in the past months.
18:00 pyrimidine joined 18:03 kst left, lateau_ left
[Coke] yes. we started observing when niecza's growth was stable, I thought. then observation, then rapid growth to match rakudo, now stability. 18:06
colomon [Coke]: ... hmm, I added over 100 new tests at about 11am this morning... 18:08
18:09 kst joined
masak hm, I didn't notice before about www.lispcast.com/modern-language-wishlist -- that's a really weak definition of "homoiconic". 18:10
my definition is much stronger: *all* code is also available as data. 18:11
18:14 whiteknight left
PerlJam why do all venn diagrams seem to relate 3 things? 18:16
masak because that's the maximum venn diagram that doesn't hurt to look at? 18:17
PerlJam perhaps 18:18
masak you quickly run up against planarity.
[Coke] colomon: that was probably after my run. 18:19
PerlJam ah, that makes more sense
and I guess it'a another variation on "doens't hurt"
colomon my all-time favorite venn: flowingdata.com/2011/10/28/venn-dia...-a-keytar/
18:24 sivoais left 18:27 sivoais joined 18:29 fridim_ joined
masak colomon: :D 18:33
18:35 whiteknight joined
TimToady masak: the schedule of school terms has much more to do with niecza's progress curve than anything else 18:40
sorear++ for not blowing off school *all* the time :)
masak oh, indeed. 18:41
18:41 revdiablo joined, revdiablo left, revdiablo joined
masak I wasn't implying being anything less than utterly impressed by the progress in Niecza of late. 18:41
18:42 kst left, lateau_ joined
masak just remarking on my expectations that once Niecza and Rakudo got very similar test pass numbers, either implementation would quickly overtake the other. 18:42
18:43 y3llow left
moritz is stuck with sink 18:43
18:43 y3llow joined, lateau_ left 18:44 birdwindupbird joined
TimToady and pmichaud++ for keeping his family first, and au++ for learning to pace herself 18:44
flussence
.oO( Perl 6: everything, but the kitchen sink's clogged )
18:45 kst joined
TimToady "Sink me!" -- the Scarlet Pimpernel 18:45
18:46 preflex left
masak "Frankly, my dear, I don't give a sink." 18:46
18:47 revdiablo left, preflex_ joined, revdiablo joined 18:48 preflex_ is now known as preflex
PerlJam TimToady++ for being our fearless language designer 18:49
:)
[Coke] Iunno, a little more fear might be a good thing. :) 18:50
18:51 lateau_ joined
masak have the appropriate amound to fear. 18:51
amount*
TimToady maybe a mound is the appropriate amount 18:52
but don't be afraid of me, just be afraid of what I might do :)
moritz now that it turns out that we rarely write Mu as a smart-matcher, can we rename it to Object again?
18:53 am0c_ left
masak oh yes please yes 18:54
TimToady well, it's really Any that fills the role of Object; I think of objects as something other than concepts like junctions or failure
18:54 am0c joined
PerlJam Calling something "Object" doesn't seem very perlish ;) 18:54
moritz it was perl6ish about 8 years long
PerlJam yeah, but that was just a first approximation and everyone knows those are always wrong 18:55
TimToady Illuminati...Nomenclatura...the Cosmic All...non-OO OO...
but it's still "most undefined" 18:56
commuting &
19:01 GlitchMr left 19:02 GlitchMr joined, fridim_ left 19:04 pernatiy joined 19:09 fridim_ joined 19:26 pyrimidine left
TimToady & computing 19:27
19:27 fridim_ left 19:29 kst left 19:42 lateau_1 joined 19:45 lateau_ left 19:46 lateau_ joined 19:47 alester joined 19:48 lateau_1 left 19:51 kst joined 19:56 zby_home_ joined
PerlJam watches DHH's "Why Ruby?" talk and it sounds like he's talking about Perl 19:57
20:01 lateau_ left
masak url? 20:06
PerlJam vimeo.com/17420638
rubyconf 2010
masak dz
20:07 lateau_ joined, lateau_ left 20:11 raoul joined
PerlJam argh ... he says monkey patching is the one thing that makes him happy about ruby 20:14
20:14 Patterner left
masak I've heard people wax lyrical about '5.times { ... }' too. :) 20:16
20:17 mj41 joined
moritz that was stolen from smalltalk, no? 20:18
20:19 ksi_ joined
moritz whenever I see that, I wonder "is control flow really the task of an integer"? 20:19
20:19 Psyche^ joined, Psyche^ is now known as Patterner
doy i feel that way about python's "'+'.join(['a', 'b'])" too 20:21
20:22 snearch joined
moritz looks like somebody did a pass over the perl 5 builtins, and then used the first argument as invocant :-) 20:24
Teratogen cutting and pasting code is never a good idea 20:25
but it comes in handy sometimes
except in python
if someone uses a different indentation scheme than you do 20:26
good luck cutting and pasting their code
moritz as a vim user I can just say gg=G 20:27
doy that doesn't work if the changing indentation is meaningful 20:28
Teratogen oh, I use vim, but I don't know what that does
what does gg=G do?
20:28 alester left
moritz gg goes to the start of the file 20:29
= re-indents code
and G stands for the end of the file
TimToady sometime soon we need to start working on py2p6 and rb2p6 translators :)
moritz so it re-indents the whole file
Teratogen how is p52p6 going? 20:32
20:33 localhost left 20:34 localhost joined, GlitchMr42 joined
Teratogen not going? 20:37
going once...
going twice...
20:37 kst left
masak no-one is working on that right now. 20:37
Teratogen I thought Perl 6 was supposed to be able to run Perl 5 code 20:38
masak it is.
20:38 GlitchMr left
[Coke] Teratogen: If that's an itch you want to scratch, awesome! 20:38
masak has been thinking about it recently 20:39
[Coke] wonders if there is currently a better plan than "embed perl5"
masak Perl 5 on Parrot would be nice. 20:40
PerlJam "embed perl 5" seems like the best plan to me.
[Coke] masak: No comment!
masak please elaborate :) 20:41
moritz wants a Ponie!
masak too!
au uh, there is a working p5top6 compiler if I'm not mistaken...
...that doubles as p5tojs: perlcabal.org/~fglock/perlito5.html 20:42
masak right, perlito.
moritz au: but that only does a tiny subset of p5
awwaiid yummy perlito
masak I sjould look into perlito, to get a handle on what it can and can't do. 20:43
should*
awwaiid recent commits, github.com/fglock/Perlito
au moritz: concurred. it's somewhat useful though, if one'd like to write javascript in perl syntax (instead of, say, in coffeescript syntax)
moritz au: aye
masak I'm coming up on such a project.
20:44 ksi_ left
PerlJam likes coffeescript 20:44
well ... except for making the ternary op more wordy 20:45
[Coke] might be odd, but I'd rather write JS in JS. ;)
au yeah. that ternary thing (among a lot other things) was apparently lifted straight from haskell.
20:46 kst joined
[Coke] would like to get pugs running on his feather login so he can add it to the daily spectest runs! 20:47
(speaking of sends from au ~~ /haskell/) 20:48
au set PATH to a dist/pugs built on some other home directory? :)
[Coke] to get at /a/ running copy, or to somehow bootstrap my own? 20:49
au to get at a running copy
PerlJam au: btw ... are you "back" or just visiting? :)
[Coke] au: any idea why this failure? gist.github.com/1827728
au PerlJam: yes :) 20:50
PerlJam good answer :)
au [Coke]: I had no idea; haskeline should've defined that instance.
[Coke] has, btw, checked out Pugs.hs, cd'd to Pugs, and run "make". (each time running cabal install as it complained). wondering if there's a step that isn't documented. 20:53
20:53 birdwindupbird left
au [Coke]: have you tried just "cabal install Pugs"? 20:54
(affter a "cabal update", of course)
[Coke] I /think/ that's what make does. I just reset the git repo, trying that. 20:56
(does .../Pugs require a pre-built pugs for bootstrapping?) 20:57
au no. it was designed after Hugs, a stand-alone interpreter.
masak oh, good. leveraging au to get a working Pugs.hs -- [Coke]++ 20:58
masak does a 'cabal update', Justin Case 20:59
[Coke] masak: it feels like cheating, but I'm not above that. ;)
au: once I get it running, I'm going to spend some time refudging the spectest suite, i think. You see any issues there? 21:00
au marvels at how a complete Pugs compilation with recursive dependencies took just 10min here
(it used to take ~3hrs.) Moore++
[Coke]: sounds good. 21:01
21:01 GlitchMr42 left
[Coke] lining! 21:02
*linking!
flussence
.oO( silver linking! )
au . o O ( slivers linking! )
[Coke] linking slooww 21:03
colomon . o O ("Lowered him down on a silver chain")
[Coke] $ /home/coke/.cabal/bin/pugs -e "'woot'.say" 21:04
woot
so, I'm guessing the Makefile is broken.
make is running "cabal build", but au said "cabal install Pugs"
au it's like "cpanm" vs "cpanm Pugs". 21:05
one is building local checkout, one is fetching from hackage.
geekosaur yeh, that just means "do the build phase in the current directory"
au (fwiw, both worked for me on feather with fresh github audreyt/Pugs.hs checkout.) 21:08
[Coke]: mind if I run "make" in your ~coke/sandbox? 21:10
[Coke] NOOOOOO, go right ahead. ;) 21:11
(but now that I did the build, it might ack differently. Iunno)
au: how would you normally run the spectests? 21:12
sorear good * #perl6
au uh, the last time I hacked perl6, it was before spectests :)
PerlJam Is someone going to update perl6.org/compilers/features.html with pugs info? :)
au [Coke]: /home/coke/sandbox/Pugs.hs/Pugs/dist/build/pugs/pugs built just fine. 21:13
(under "sudo su - coke")
so maybe the cabal update picked up a good haskeline release version, or some such.
[Coke] au - might be because I had a successful install already. *shrug*
right.
no matter, it works now. 21:14
au \o/
masak sorear! \o/
PerlJam idly wonders if all of the Perl 6 implementors have ever been in #perl6 simultaneously
masak [Coke]: whatdidyoudo whatdidyoudo!?
au PerlJam: it'll be yellow ± on pretty much all rows :) 21:15
so, not exactly informative
masak we'll change that :)
PerlJam au: sure it will ... as people update it, those rows will change color
au *nod* 21:16
[Coke] masak: try "cabal install Pugs" instead of "make" 21:17
masak ooh
masak tries
[Coke] github under attack again? :(
21:18 ksi_ joined
sorear o/ masak 21:18
o/ o/ au
au hey sorear :)
flussence [Coke]: cat thing says yes. status.github.com/ 21:19
well yes as in "our five nines are under attack!"
au wow, it's past 10pm already... have a good localtime hackers. -> {sleep} 21:20
flussence eek, page just went red 21:21
21:23 kst left 21:25 bluescreen10 left
masak \o/ # Pugs built! 21:26
au++
[Coke]: we should do something about the instructions in INSTALL, though. 21:27
[Coke] masak: I'll let you. I'm glaring at spectests.
masak oh wait, INSTALL is full of 'cabal install Pugs'
huh!
had I only read this properly, instead of trying to run 'make', it'd probably have worked the first time... 21:28
PerlJam reading++
[Coke] masak: one of the things does say to run make, though. 21:29
gist.github.com/1855576
masak running 'make' still doesn't work.
same error.
21:30 kst joined
masak yes, .gist is very new. 21:30
au masak: try "rm -rf Pugs ; git reset --hard"?
masak does so 21:31
[Coke] the spectest cleanup is going to be monumental. ;) 21:34
[Coke] will get something committed this weekend that has at least enough so folks can say "make spectest" and run /something/
s/cleanup/refudge 21:35
sorear au: do you know about git clean -dfx ?
masak au: same error after "rm -rf Pugs ; git reset --hard"
the one saying 'No instance for (MonadIO (InputT IO))'
[Coke]: aren't you getting this error any more?
au sorear: I didn't. thanks! 21:36
masak: I'm reinstalling haskeline for you, a sec. 21:39
masak thanks. I had an error message about haskeline, so I did 'cabal install haskeline' two 'make's ago. 21:40
maybe that wasn't enough, though.
21:44 pomysl_ joined, pomysl left, Tarmvred left 21:46 mj41 left
au yeah. I ended up doing 21:46
"ghc-pkg unregister haskeline; cabal install --reinstall haskeline"
and that worked.
masak: you have an entirely unoptimized build of pugs in ~/Pugs.hs/Pugs/dist/build/pugs/pugs now -- share & enjoy :) 21:47
21:47 thou left
au (might want to git clean -dfx and rebuild again to get optimized version instead of the fast-building-slow-running "cabal install -O0" one.) 21:47
g'localtime for real :) &
jnthn home :) 21:53
masak au++! 21:55
au: thanks!
[Coke] moritz++ - very easy to copy over the t/* utilities I need from niecza/rakudo 21:59
jnthn masak: #110918 is a dupe
[Coke] (and quite a few spectests still pass in pugs with no additional fudging!)
au++ 22:00
22:01 pmurias joined
masak jnthn: hm -- it felt a little like one when I submat it. 22:01
22:02 shinobicl_ left
[Coke] if I want a t, t/spec hierarchy in pugs, where do I put it? top level? .../Pugs subdir? 22:04
[Coke] thinks toplevel.
masak +1 22:05
[Coke] "pugs", 637, 0, 1, 0, 638, 22744 22:07
... that's with about 10m of effort.
PerlJam Coke++ 22:08
masak \o/
this'll be fun!
Coke++
[Coke]: don't forget to push whatever you create in t/spec :) 22:09
(the rest of us want to play, too!)
[Coke] masak: commuting shortly. You'll have something to play with in the morning, OK? 22:10
22:10 kst left
masak deal. 22:13
jnthn: tickets merged, thanks.
jnthn asbra :)
masak phenny: sv en "asbra"? 22:14
phenny masak: "asbra" (sv to en, translate.google.com)
masak translator I am disappoint.
phenny: sv en "bra"?
phenny masak: "good" (sv to en, translate.google.com)
masak phenny: sv en "as-bra"?
phenny masak: "as- well" (sv to en, translate.google.com)
jnthn fail :) 22:15
masak phenny: sv en "så jävla bra"?
phenny masak: "so damn good" (sv to en, translate.google.com)
masak :)
jnthn Ain't jävla a place just a bit north of Stockholm? :) 22:16
22:16 MayDaniel left
[Coke] masak: hurm. I changed my mind. Here you go: 22:16
dalek gs.hs: 466f840 | coke++ | / (5 files):
Run (some) spectests.

This commit adds in a bunch of tools from niecza/rakudo to run the spectests, as well as a small list of known-to-pass spectests in t/spectest.data
[Coke] I added some directions/caveats in HACKING. 22:17
Please do not eat ALL the low hanging fruit. ;)
22:18 kst joined
[Coke] Should be possible to bump that 637 number up to several thousand this evening. 22:18
(with only minimal fudging) 22:19
22:19 lestrrat left
[Coke] ping me with any problems on the HACKING notes. back online in a few hours. 22:19
22:20 lestrrat joined
masak [Coke]++ 22:20
jnthn: no, that's "Gävle" :) 22:21
but I can see the possibility of confusion there.
pmurias sorear: what's the slow part of compiling NieczaActions.pm? 22:22
sorear: or generally what slows down the compilation the most? 22:23
22:33 aoeuhtns left 22:34 rgc joined
sorear pmurias: I don't really have a clue. 22:35
I think the memory usage could probably be cut some by hacking in a method to throw away branches of the parse tree that we're done with 22:36
sorear checks calendar ... good, a whole week until release day
pmurias sorear: should 2GB of ram be enough? 22:38
sorear: for the memory usage not to be a problem? 22:39
22:39 arlinius joined
sorear absolutely 22:40
remember, my dev box has 0.375GB :D
jnthn You can even compile *RAKUDO* with 2GB and it's currently a memory pig.
pmurias sorear: is the command flag to show time use per compilation phase still present? 22:42
sorear: do you recompile niecza with make after making a change or just the file? 22:43
sorear I use make 22:45
if you have a sufficiently new mono (2.8.1+), use MONO_ENV_OPTIONS=--gc=sgen 22:46
that activates the Mono native GC, which saves a bit of memory per object
pmurias does that speed up stuff? 22:50
sorear usually
it makes allocations a lot faster, but slows down writes (write barrier)
niecza makes a lot of garbage... 22:51
diakopter gigo :P 22:52
masak .oO( Affirming the Consequent ) 22:58
23:00 ilogger2 joined, ChanServ sets mode: +v ilogger2
diakopter wb ilogger2 23:00
masak ilogger2 is the silent, stoic kind of bot.
ilogger2: aren't you? aren't you the silent, stoic kind? yes, you are! yes, you are! 23:01
ilogger2 masak: I am a logger bot. Lines starting with [off] won't be logged. Extra help available at colabti.org/irclogger/irclogger_logs.
masak ruffles ilogger2's fur
whoa :)
not so silent, I guess. but polite :) 23:02
pmurias sorear: niecza doesn't seem to print the time usage per compiler part anymore
with the --verbose flag
sorear pmurias: yeah, there are no stages anymore 23:04
pmurias: stuff had to be interleaved to make BEGIN work
23:05 kst joined
pmurias the stack traces from exceptions are getting annoying :( 23:10
diakopter lol [off] the magic dragon 23:13
masak I thought you might like that one :)
but you just blew our cover :P
diakopter oops
maybe it means off not [off] 23:14
no
pmurias sorear: why do we need Run.*.dll? 23:22
benabik Otherwise it won't know how to run?
pmurias benabik-- 23:23
benabik awh 23:24
pmurias benabik: there's both Run.Kernel.dll and Kernel.dll
sorear pmurias: Run.Kernel.dll is compiled from the current version of lib/*.cs
Kernel.dll is copied from niecza-14.zip
by giving them different names, they don't interfere with each other 23:25
pmurias sorear: eval(:from<perl5>,...) throws a casting C# exception when run from NieczaActions.pm6
sorear: could it have anything to do with it?
sorear yes 23:26
I do not endorse calling eval from within the compiler
the idea is that there are two sandboxed spaces - you can have NieczaActions in user code, and it won't interfere with the compiler 23:27
(this is necessary for bootstrapping)
eval(:from<perl5>,...) doesn't make the switch from compiler space to user space, so even if it worked, it would be wrong 23:28
pmurias i understand 23:32
so how can i call the user space eval?
sorear I'd probably do it by adding a new method to $*unit 23:38
$*unit.foo calls the "unit_foo" method in lib/CodeGen.cs in user space
that is fundamentally the only way the transition happes
23:44 kst left 23:57 kst 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!