pugscode.org/ | nopaste: sial.org/pbot/perl6 | ?eval [~] <m oo se> | We do Haskell, too | > reverse . show $ foldl1 (*) [1..4] | irclog: irc.pugscode.org/
Set by diakopter on 11 July 2007.
00:00 lichtkind left 00:04 japhb left 00:05 theorb joined
pugs_svnbot r18327 | avar++ | [kp6-lisp] Implemented length, again.. 00:08
diff: dev.pugscode.org/changeset/18327
lambdabot Title: Changeset 18327 - Pugs - Trac
00:17 theorbtwo left
TimToady please call it "chars", not "length" 00:18
00:20 jferrero left
avar it's length in kp6, I suppose it can be changed later:) 00:20
00:21 vbattsAtWork left 00:22 devogon left
TimToady meh, who's the dufus who called it length in the first place? oh wait, nevermind... 00:22
00:22 cognominal__ left
amnesiac heh 00:23
00:38 justatheory left 00:39 kanru joined 00:52 polettix left 00:59 lyokato joined 01:16 nipotaway is now known as nipotan 01:44 weinig left 01:54 kanru left 02:16 daxim_ joined 02:19 mr_ank joined 02:22 zamolxes left, kanru joined 02:29 drbean_ left 02:35 Schwern left 02:44 weinig joined 02:48 vbattsAtWork joined 03:09 drupek12 joined 03:17 REPLeffect_ joined 03:51 jjore-w left 04:03 justatheory joined 04:13 diotalevi joined 04:15 REPLeffect_ left
meppl good morning 04:38
spinclad good morrow, meppl 04:40
meppl good morning spinclad 04:41
allbery_b had to switch back to this tab after doing a quick flash-through and reading "good moring moose"/"good morning squirrel" 05:01
allbery_b thinks it's tme for bed...
meppl good night allbery 05:03
spinclad no, #moose is over there ->
sweet dreams, allbery_b
spinclad changes channel name to #greetings 05:04
05:04 dlocaus joined 05:08 Ashizawa joined 05:15 vbattsAtWork left 05:17 devogon joined 05:41 jisom joined 06:19 thoughtpolice left 06:37 elmex joined 06:49 Aankhen`` joined 06:57 justatheory left 07:02 iblechbot joined, dlocaus left 07:10 shachaf left 07:19 penk joined
pugs_svnbot r18328 | Aankhen++ | * [kp6-lisp] call PERL->DISPLAY from &prefix:<~>. 07:22
diff: dev.pugscode.org/changeset/18328
lambdabot Title: Changeset 18328 - Pugs - Trac
07:38 penk left, penk joined 07:55 franck___ joined 07:59 amnesiac left 08:24 masak joined 08:25 xinming_ joined 08:36 polettix joined 08:37 xinming left 08:50 jisom left, devogon left 08:52 devogon joined 08:54 Kattana left 08:55 iblechbot left 09:03 Kattana joined 09:06 polettix left
pugs_svnbot r18329 | ruoso++ | [yap6] autotools ready to use... just do a make -f Makefile.cvs; configure; make 09:20
diff: dev.pugscode.org/changeset/18329
lambdabot Title: Changeset 18329 - Pugs - Trac
pugs_svnbot r18330 | ruoso++ | [yap6] a roadmap is documented in the README file 09:41
diff: dev.pugscode.org/changeset/18330
lambdabot Title: Changeset 18330 - Pugs - Trac
pugs_svnbot r18331 | ruoso++ | [yap6] clarification of the roadmap in README 09:47
diff: dev.pugscode.org/changeset/18331
lambdabot Title: Changeset 18331 - Pugs - Trac
10:03 Ashizawa left 10:08 fglock joined 10:14 chris2 joined 10:27 snearch joined 10:28 dduncan left 10:44 snearch left
pugs_svnbot r18332 | ruoso++ | [yap6] Namespace seems to be the first step for a MOP (at least this is what MO:: makes me think)... 11:03
diff: dev.pugscode.org/changeset/18332
lambdabot Title: Changeset 18332 - Pugs - Trac
11:06 lyokato left
pugs_svnbot r18333 | fglock++ | [kp6] initial impl of gather/take 11:12
diff: dev.pugscode.org/changeset/18333
lambdabot Title: Changeset 18333 - Pugs - Trac
11:13 drupek12 left 11:20 drupek12 joined
pugs_svnbot r18334 | ruoso++ | [yap6] specifies use of rwlock instead of mutex. define the functions to manage that... 11:27
diff: dev.pugscode.org/changeset/18334
lambdabot Title: Changeset 18334 - Pugs - Trac
11:28 lichtkind joined
pugs_svnbot r18335 | fglock++ | [kp6] added t/kp6/internals/02-gather.t 11:28
diff: dev.pugscode.org/changeset/18335
lambdabot Title: Changeset 18335 - Pugs - Trac
11:30 nipotan is now known as nipotaway 11:37 pina joined
pina ok not trying to troll but where all does python lack in comparison to ruby or perl 11:37
besides syntaxes 11:38
masak pina: what do you mean?
pina i mean where is it inefficient
masak hard to say, I think
pina like python sometimes is inefficient for something perl handles easily
masak regexes are maximally easy in Perl 11:39
in Python you have to use slightly more cumbersome syntax
also, Python is more type strict
Perl is explicitly context-sensitive
11:40 ilogger2 joined
pina ok 11:42
those are pretty basic
just syntaxes
masak well, yes
sufficiently deep down all languages are the same 11:43
though I would say that Python and Perl have different feels to them even mid-way down
many aspects of Python were created with Perl in mind 11:44
Python has "There should be only one obvious way to do it" as one of its slogans
the Python 3000 design process was made to be dissimilar to the Perl 6 design process
I'm not saying one is better than the other, or that people cannot be friends across cultures, or anything like that 11:45
pugs_svnbot r18336 | fglock++ | [kp6] tested gather/take on an infinite loop
diff: dev.pugscode.org/changeset/18336
lambdabot Title: Changeset 18336 - Pugs - Trac
masak it's just that Python in many ways seems to be a reaction to Perl 11:46
it's not only that, of course
it's also a quite clean and easy to read programming languages, with its basis in education
fglock moritz_: would you install the Coro module from CPAN in the evalbot environment? 11:50
11:51 CindyLinz joined 11:53 ilogger2 joined
fglock is the block in 'gather' always supposed to terminate? 11:59
I guess not, since it is evaluated lazily
I suppose the block is killed when the gather return value is no longer needed (gets out of scope)? 12:03
masak fglock: sounds reasonable
in the end, it's just iterators anyway, I suppose 12:04
fglock: is there any case where programs would behave differently depending on whether 'gather' blocks have to terminate or not? 12:08
moritz_ fglock: of course, wait a second... 12:09
12:10 Shillo joined
fglock masak: in the case when you have an infinite loop inside gather; kp6 currently just terminates the main program 12:12
moritz_ kp6: say "hello" 12:13
exp_evalbot r18336: OUTPUT[helloā¤]
masak fglock: that's correct according to me iff the gathered array is not used
moritz_ fglock: does that mean everything is working again with evalbot?
12:13 polettix joined
platypus fglock: How does it detect an infinate loop? 12:14
fglock dev.pugscode.org/browser/v6/v6-Kind...2-gather.t
it is used, but just as needed
lambdabot Title: /v6/v6-KindaPerl6/t/kp6/internals/02-gather.t - Pugs - Trac, tinyurl.com/yodmcp
masak fglock: what do you mean?
fglock platypus: it doesn't detect the loop; it runs until it gets enough data, then pause
platypus nods. 12:15
masak fglock: if I understand you correctly, it sounds excellent
the gather gets stuck in an infinite loop, but the program itself doesn't
fglock masak: see the test
masak oki
fglock yep
12:16 gorax joined
masak re the test: yes, we agree 12:16
excellent use of laziness, by me
why is `gather` spelled `Gather.new`, btw? 12:18
is that a kp6ism?
fglock because it's not in the grammar yet 12:19
this is the internal representation
masak wohoo, sounds like a fun thing to hack into the grammar! :)
fglock masak: because it was just implemented a few minutes ago :P
masak ah
fglock++ 12:20
Shillo fglock: Do you plan to make it truly lazy, eventually? 12:27
I mean, generate nothing when called, and only generate -precisely- as much as needed? 12:28
fglock Shillo: it does already
Shillo Goodie. :)
For a moment I thought you were describing something like unix buffered pipes, so I commented. Sowwy. 12:29
12:30 iblechbot joined
Shillo Come to think of it, making the lazy lists buffered could be very useful. Because then you could autoparalelize to a different thread. 12:30
moritz_ but then you can't guarantee "true" lazyness 12:31
we need a new trait, "is half-lazy" ;)
Shillo Yeah, I know. :)
Well, umm, train creep = evil. ;) Maybe just a proxy-list in standard library
Er, trait creep :) 12:32
fglock Shillo: I don't get it - can you give an example? 12:35
Shillo thinks... 12:36
moritz_ fglock: you can define lazy lists recursively
@a = (0, 1, @[*-1] + @[*-2]) 12:37
lambdabot Maybe you meant: activity activity-full admin all-dicts arr ask . v
moritz_ that should generate the fibonaccy numbers lazyly
Shillo moritz_: You can do that with normal laziness, too
moritz_ now if have a sub that read from that list, you have to switch between the list calculation and the callling sub
pugs_svnbot r18337 | fglock++ | [kp6] added 'gather' to the grammar; needs some AST work
diff: dev.pugscode.org/changeset/18337
lambdabot Title: Changeset 18337 - Pugs - Trac
Shillo Normal gather would create an object that contains list_so_far and a continuation that makes more 12:38
moritz_ but if you did some buffering, you could move the liist calculation into its own thread
Shillo Threaded gather would have list_so_far, a semaphore and a thread
moritz_ and run them in parallel
Shillo list_so_far would now contain more than needed, and the thread would block until more is required
I suspect this is the primitive you need to implement hyperops and pipes with gather 12:39
Not sure if semaphore is the right locking primitive, of course. Possibly not.
moritz_ reboot &
Shillo Could be just a pair of mutexes instead. ACtually, probably is, becuase this is producer-consumer pattern 12:40
But honestly, I don't think this belongs to the core language. Maybe into the runtime library
You actually need it (or something like it) for autothreaded primitives. 12:41
Actually, this is another good problem: Pester @Larry into thinking about interactions about autothreading and laziness 12:42
12:42 pmurias joined
pmurias hi 12:42
Shillo As in, is @a >>+<< @b lazy list? Or does it autothread? Or, if it does both, how? 12:43
Hiya!
masak Shillo: good question
pmurias list context is lazy by default 12:44
(is what i heard)
Shillo That doesn't completely answer my question, though.
Or rather, it doesn't answer 3rd part of it.
Basically, if it splits @a and @b into sublists and assigns them to threads, the question becomes, when do you do the summing? 12:45
... of the middle chunks?
fglock Shillo: you do the sum just when you need the result 12:46
Shillo fglock: This isn't what autothreading implies, though
masak @ask TimToady about the interactions between autothreading and laziness. what happens in the following case? my @a = gather { ... }; my @c = @a >>+<< @b; does the calculation of @c happen lazily, or does it autothread only after the calculation of @a is complete?
lambdabot Consider it noted.
Shillo Because lazy = as needed, as late as possible. Autothreading = random order, hardware/os-dependent time. The two don't seem compatible by default. 12:47
I'm not saying that they can't be made compatible, but it's worthy thinking about. 12:48
masak nod
12:50 cmarcelo joined 12:51 iblechbot left
fglock kp6: "test" 12:56
pugs_svnbot r18338 | fglock++ | [kp6] gather syntax works
diff: dev.pugscode.org/changeset/18338
exp_evalbot r18337: OUTPUT['test'ā¤]
lambdabot Title: Changeset 18338 - Pugs - Trac
12:57 audreyt_ joined
pmurias fglock: why is gather in t/internals? 12:58
fglock kp6: my $a = gather { take 123; take 456; }; say $a[1]
exp_evalbot r18338: OUTPUT[456ā¤]
fglock pmurias: because the Gather class is not part of the spec 12:59
kp6: my $a = gather { while 1 { take 123; take 456; } }; say $a[1]
exp_evalbot r18338: OUTPUT[456ā¤]
fglock yay!
pmurias is Junction.new(...) speced? 13:00
masak fglock++ # cool
fglock kp6: my $a = gather { while 1 { take 123; take 456; } }; say $a[3]
exp_evalbot r18338: OUTPUT[456ā¤]
masak it's almost surreal that that works
how does pugs do in comparison?
fglock pmurias: no - it was created before t/internals :P
pmurias history explain everythin
g 13:01
moritz_ fglock++
moritz_ is truely impressed
Shillo fglock: Nice!
masak apparently pugs hangs on the gather
too eager... 13:02
Shillo masak: Yeah, I tested laziness in pugs a while ago, and that's what happened.
audreyt: Hiya!
masak ownders if there's a test for gather laziness in the pugs test suite
Shillo kp6: my $a = gather { for 1..5 { say $_; take $_ } }; say $a[3]; 13:03
exp_evalbot r18338: OUTPUT[error in Block at lib-kp6-mp6-p5/KindaPerl6/Grammar/Sub.pm line 14, <> line 1.ā¤Illegal declaration of anonymous subroutine at - line 1.ā¤]
Shillo kp6: my $a = gather { for 1..5 -> $x { say $x; take $x; } }; say $a[3];
moritz_ masak: you could add one, just in case
exp_evalbot r18338: OUTPUT[error in Block at lib-kp6-mp6-p5/KindaPerl6/Grammar/Sub.pm line 14, <> line 1.ā¤Illegal declaration of anonymous subroutine at - line 1.ā¤]
masak moritz_: I'll do that
fglock Shillo: Range is unfinished 13:04
pmurias had to dispable valgring support in Coro
masak how best to test against infinite looping? wait a reasonable amount of time for a thread and then kill it?
Shillo kp6: my $a = gather { say 1; take 1; say 2; take 2; say 3; take 3; say 4; take 4; }; say $a[1];
exp_evalbot r18338: OUTPUT[1ā¤2ā¤2ā¤]
Shillo Hmm, how to try this same thing in pugs?
kp6 passed with flying colors :)
pugs: my $a = gather { say 1; take 1; say 2; take 2; say 3; take 3; say 4; take 4; }; say $a[1]; 13:05
masak Shillo: good idea
exp_evalbot OUTPUT[1ā¤2ā¤3ā¤4ā¤2ā¤]
pmurias my $a = gather { while(1) {take 2} };say $a[5];
Shillo Ooooh!
There's your test. Fails without hanging. :)
pmurias kp6: my $a = gather { while(1) {take 2} };say $a[5];
exp_evalbot r18338: OUTPUT[error in Block at lib-kp6-mp6-p5/KindaPerl6/Grammar/Sub.pm line 14, <> line 1.ā¤Illegal declaration of anonymous subroutine at - line 1.ā¤]
moritz_ my $count = 0; my @list = gather { for 1 .. 10 -> $a { take $a; $count++ } }; say @list[2]; say $count 13:06
pugs: my $count = 0; my @list = gather { for 1 .. 10 -> $a { take $a; $count++ } }; say @list[2]; say $count
exp_evalbot OUTPUT[3ā¤10ā¤]
moritz_ you see it's not lazy
Shillo Nice, too :)
masak moritz_: I'll use your method as a test
moritz_ kp6: my $count = 0; my @list = gather { for 1 .. 10 -> $a { take $a; $count++ } }; say @list[2]; say $count
exp_evalbot r18338: OUTPUT[error in Block at lib-kp6-mp6-p5/KindaPerl6/Grammar/Sub.pm line 14, <> line 1.ā¤Illegal declaration of anonymous subroutine at - line 1.ā¤]
pmurias kp6: my $count;while(--$count > 0) {say 2}
exp_evalbot r18338: OUTPUT[syntax error at position 30, line 1 column 30:ā¤my $count;while(--$count > 0) {say 2ā¤ ^ HEREā¤DISPATCH: calling APPLY KindaPerl6::Runtime::Perl5::DispatchSugar::Dispatch=HASH(0x8583850) on invalid object:$VAR1 = {};ā¤ā¤ at
..lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 33ā¤ main::DISPATCH('undef', 'APPLY', 'KindaPerl...]
pmurias kp6: my $count=4;while(--$count > 0) {say 2}
exp_evalbot r18338: OUTPUT[syntax error at position 32, line 1 column 32:ā¤my $count=4;while(--$count > 0) {say 2ā¤ ^ HEREā¤DISPATCH: calling APPLY KindaPerl6::Runtime::Perl5::DispatchSugar::Dispatch=HASH(0x814f504) on invalid object:$VAR1 = {};ā¤ā¤ at
..lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 33ā¤ main::DISPATCH('undef', 'APPLY', 'Kinda...]
moritz_ it doesn't use I/O, and it doesn't loop forever if it's not lazy
pmurias moritz_: it does IO 13:07
I/)
I/O
moritz_ pmurias: yes, but not as a side effect in the gather/take 13:08
13:09 gorax left
fglock kp6: my $a = gather { while (1) {say "take"; take 2} };say $a[5]; 13:09
exp_evalbot r18338: OUTPUT[takeā¤takeā¤takeā¤takeā¤takeā¤takeā¤2ā¤]
pmurias kp6: use IO::File:from<perl5>;my $fh = IO::File.new();$fh.open(">file");
exp_evalbot r18338: OUTPUT[DISPATCH: calling perl on invalid object:$VAR1 = '';ā¤ā¤ at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 33ā¤ main::DISPATCH('', 'perl') called at - line 16ā¤]
pmurias kp6: use IO::File:from<perl5>;my $fh = IO::File.new();
exp_evalbot r18338: OUTPUT[Can't locate object method "perl" via package "IO::File" at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/Wrap.pm line 22.ā¤]
masak there actually is a small laziness test in t/statements/gather.t, but I'll add an explicit one in the end anyway 13:10
the small test is markes :todo<unspecced>
s/markes/maked/ 13:11
pmurias kp6: say "hi"
exp_evalbot r18338: OUTPUT[hiā¤]
pmurias kp6: gather {say "hi";}
exp_evalbot r18338: OUTPUT[no method 'code' in Class 'Gather'ā¤ at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 130ā¤ KindaPerl6::Runtime::Perl5::MOP::__ANON__('HASH(0x85a2494)', 'code') called at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 36ā¤ main::DISPATCH('HASH(0x85a2494)', 'code')
..called at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl6/Gather.pm line 51ā¤ Gathe...]
pmurias kp6: gather {1+2;}
exp_evalbot r18338: OUTPUT[no method 'code' in Class 'Gather'ā¤ at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 130ā¤ KindaPerl6::Runtime::Perl5::MOP::__ANON__('HASH(0x85a84d4)', 'code') called at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 36ā¤ main::DISPATCH('HASH(0x85a84d4)', 'code')
..called at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl6/Gather.pm line 51ā¤ Gathe...]
pmurias kp6: gather {take 1+2;}
exp_evalbot r18338: OUTPUT[no method 'code' in Class 'Gather'ā¤ at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 130ā¤ KindaPerl6::Runtime::Perl5::MOP::__ANON__('HASH(0x85a84f4)', 'code') called at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 36ā¤ main::DISPATCH('HASH(0x85a84f4)', 'code')
..called at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl6/Gather.pm line 51ā¤ Gathe...]
fglock pmurias: one sec, I'll fix it
Shillo Funny.
pmurias moritz_: why dosn't gather use IO? 13:12
moritz_ pmurias: why should it?
pmurias: I think we're not talking about the same thing
pugs_svnbot r18339 | masak++ | [t/statements/gather.t]
r18339 | masak++ | * test for lazy gather
diff: dev.pugscode.org/changeset/18339
lambdabot Title: Changeset 18339 - Pugs - Trac
moritz_ pmurias: I suggested a method to check if gather works lazily, and that didn't use additional I/O 13:13
pmurias moritz_: i see 13:18
had too much school/starcraft 13:19
fglock kp6: gather {take 1+2;} 13:20
pugs_svnbot r18340 | fglock++ | [kp6] fixed Gather.perl (but still have errors)
diff: dev.pugscode.org/changeset/18340
exp_evalbot r18339: OUTPUT[no method 'code' in Class 'Gather'ā¤ at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 130ā¤ KindaPerl6::Runtime::Perl5::MOP::__ANON__('HASH(0x85a84f4)', 'code') called at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 36ā¤ main::DISPATCH('HASH(0x85a84f4)', 'code')
..called at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl6/Gather.pm line 51ā¤ Gathe...]
lambdabot Title: Changeset 18340 - Pugs - Trac
fglock kp6: gather {take 1+2;}
exp_evalbot r18339: OUTPUT[no method 'code' in Class 'Gather'ā¤ at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 130ā¤ KindaPerl6::Runtime::Perl5::MOP::__ANON__('HASH(0x85a84f4)', 'code') called at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 36ā¤ main::DISPATCH('HASH(0x85a84f4)', 'code')
..called at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl6/Gather.pm line 51ā¤ Gathe...]
Shillo Didn't update yet. old r number 13:23
kp6: gather {take 1+2; }
exp_evalbot r18340: OUTPUT[gather( Str )ā¤]
Shillo Heh! :)
13:25 mino joined, mino left
pugs_svnbot r18341 | fglock++ | [kp6] fix Gather.perl a bit more 13:26
diff: dev.pugscode.org/changeset/18341
lambdabot Title: Changeset 18341 - Pugs - Trac
fglock note that gather {take 1+2; } never executes the block 13:27
kp6: gather {take 1+2;}
exp_evalbot r18341: OUTPUT[( gather { ... } )ā¤]
13:31 vbattsAtWork joined
moritz_ kp6: my @a = gather { take 1; take 2 }; 13:36
exp_evalbot r18341: OUTPUT[Can't use an undefined value as an ARRAY reference at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/Array.pm line 24.ā¤]
moritz_ kp6: my @a; @a = gather { take 1; take 2 };
exp_evalbot r18341: OUTPUT[Can't use an undefined value as an ARRAY reference at lib-kp6-mp6-p5/KindaPerl6/Runtime/Perl5/Array.pm line 24.ā¤]
moritz_ kp6: gather { take 1; take 2 };
exp_evalbot r18341: OUTPUT[( gather { ... } )ā¤]
13:38 cognominal_ joined 13:40 cognominal_ left 13:41 cognominal_ joined, TJCRI joined
avar is the kp6-kp6 backend completely scewed up atm? 13:44
pmurias fglock: is there much benefit in also testing desugared forms like Gather.new() instead of gather?
13:44 alecc joined
fglock moritz_: Array is not lazy yet, so @a = gather ... doesn't work 13:44
pmurias: not really
just for debugging
pmurias avar: kp6-kp6 never work, it's not realy a backend
avar it doesn't even compile now 13:46
/usr/bin/perl script/kp6 -r kp6 < lib/KindaPerl6/Visitor/EmitLisp.pm | perltidy -pro=util/perltidyrc > lib-kp6-kp6-p5/KindaP\ 13:47
erl6/Visitor/EmitLisp.pm
KindaPerl6/Runtime/Perl6/Pair.pm did not return a true value at lib-kp6-kp6-p5/KindaPerl6/Runtime/Perl5/MOP.pm line 823.
Compilation failed in require at lib-kp6-kp6-p5/KindaPerl6/Runtime/Perl5/Runtime.pm line 5.
pmurias * worked 13:48
13:55 cmarcelo_ joined 13:59 cmarcelo left, cmarcelo_ is now known as cmarcelo 14:05 elmex joined
pmurias ruoso: it might interesting to see home much of the runtime can be implemented using native types 14:06
14:06 iblechbot joined
pmurias some parts of perl6 eq. sub add(int $a,int $b --> int) {$a+$b} can be compiled to c directly 14:07
14:17 ntgrl joined
pugs_svnbot r18342 | fglock++ | [kp6] moved much of the gather impl to Perl 6 land 14:17
diff: dev.pugscode.org/changeset/18342
lambdabot Title: Changeset 18342 - Pugs - Trac
avar pmurias: sial.org/pbot/27962
lambdabot Title: Paste #27962 from Someone at 208.78.101.240
avar pmurias: When kp6-CL supports such type decls we can emit optimized assembly for those cases as that paste shows 14:18
parrot^Hkp6-cl aims to be the fastest perl 6 implementation:)
14:19 polettix left
avar Well, it is at the moment even though the CL we're emitting is completely unoptimized 14:19
pugs_svnbot r18343 | fglock++ | [kp6] implemented .eager on Gather objs 14:23
diff: dev.pugscode.org/changeset/18343
r18344 | fglock++ | [kp6] gather stringification works 14:26
diff: dev.pugscode.org/changeset/18344
14:27 amnesiac joined
pmurias avar: cool 14:34
moritz_ avar: do you have some figures on the speed? 14:35
avar only stuff I posted on pugs.blogs.com 14:36
but the stand-alone exes I've been making feel much faster than pugs, and it's even faster than that inside the lisp image 14:37
(the exes are 30MB)
moritz_ avar: I only saw your "kp6 is now on CPAN" post there
avar pugs.blogs.com/pugs/2007/08/playing...l#comments 14:38
lambdabot Title: Pugs: Playing with the new "Not Quite Perl"
14:38 ajelenak joined
avar sbcl also has built-in support for what parrot was going to do (is doing?), run one parrot image per system and load programs into it 14:39
That's how you're supposed to run it in fact
14:39 ajelenak left
moritz_ avar: ok 14:39
14:44 blindfish joined
[particle] avar: with 30MB exes, let's hope so :/ 14:52
14:54 mncharity joined
pugs_svnbot r18345 | fglock++ | [kp6] Gather.map (doesn't work yet) 15:02
diff: dev.pugscode.org/changeset/18345
Aankhen`` I'm confused. Do packages, modules, classes and roles share the same namespace or do they not? 15:04
Also types and subtypes, I suppose. 15:05
[particle] one part of this that i think i understand clearly is Class has-many Namespace 15:06
in other words, you can alias classnames
Aankhen`` How so? 15:07
[particle] i'm looking for something to support that in the synopses 15:08
however, i know parrot's namespace/class design is that way to allow perl 6 aliasing
pmurias Aankhen``: classes,modules and roles surly share the same namespace 15:09
[particle] and grammars
Aankhen`` Whew.
pmurias grammars are classes
Aankhen`` I was worried for a moment.
Types and subtypes as well? 15:10
pmurias they must
Aankhen`` Right.
Thanks, I feel much better now. :-)
moritz_ can't you just use ::= to alias namespaces etc.?
pugs_svnbot r18346 | fglock++ | [kp6] Gather.map runs; possible fencepost error? 15:12
diff: dev.pugscode.org/changeset/18346
fglock kp6: my $a = gather { my $k = 0; while 1 { take $k; $k = $k + 1; } };say ( $a.map: -> $x { $x + 100 } )[10]; 15:13
exp_evalbot r18346: OUTPUT[121ā¤]
15:14 [particle] joined
Aankhen`` Is misc/Perl-Metamodel/ outdated? 15:14
That is, would it be silly to use ideas from there?
fglock Aankhen``: it's a bit outdated
Aankhen``: see Moose and MO instead 15:15
Aankhen`` fglock: What's more up-to-date?
Okay.
fglock looking for link
Aankhen`` Don't worry, I'm familiar with Moose and I came across MO recently. :-)
fglock Aankhen``: nothingmuch.woobling.org/MO/
lambdabot Title: Index of /MO
fglock ok :)
Aankhen`` Oh, perhaps this was the wrong MO I came across. 15:16
Thanks for the link.
15:17 Shillo left
avar [particle]: the 30mb exes are really memory dumps of sbcl 15:25
15:30 IllvilJa joined 15:32 barney joined
pugs_svnbot r18347 | fglock++ | [kp6] moved Gather.INDEX to Perl 6 15:32
diff: dev.pugscode.org/changeset/18347
lambdabot Title: Changeset 18347 - Pugs - Trac
15:43 melissa joined
pmurias 53-attr-bind should use $!member 15:44
i don't know what binding to a method call would/should do
right? 15:47
pugs_svnbot r18348 | pmurias++ | [kp6] private attribute, only binding to private attribiutes is allowed 15:50
diff: dev.pugscode.org/changeset/18348
lambdabot Title: Changeset 18348 - Pugs - Trac
pmurias 2 tests fail on the p5 backend ;) 16:00
pugs_svnbot r18349 | fglock++ | [kp6] gather - more fixes 16:08
diff: dev.pugscode.org/changeset/18349
lambdabot Title: Changeset 18349 - Pugs - Trac
16:10 lisppaste3 joined
TimToady @messages 16:10
lambdabot masak asked 3h 24m 14s ago: about the interactions between autothreading and laziness. what happens in the following case? my @a = gather { ... }; my @c = @a >>+<< @b; does the calculation of @c
happen lazily, or does it autothread only after the calculation of @a is complete?
TimToady the hyperops wants to run to completion, and pretends the array is all there for scheduling threaded execution 16:12
if it isn't all there, then the thread in question will have to wait until its values start arriving from the gather
on the earlier question, I have always felt that laziness should be allowed to happen in batches by default 16:13
you shouldn't be relying on laziness happening in a particular order
that should be considered harmful
so if a thread wants to work ahead a little, I don't see why it shouldn't. 16:14
most of these parallel operators in Perl 6 explicitly promise that you don't care about order of evaluation 16:15
and even laziness doesn't really care about order of evaluation, just order of the resulting values
but we basically give the list optimizer three settings on the dial: lazy, eager, and hyper 16:17
[particle] s/list optimizer/teenager/
TimToady rofl
the basic assumption is that a lazy list may not be interested in all its values; an eager list is interested in all its values and cares about order of evaluation; hyper wants all values but doesn't care about order of evaluation. 16:22
(and junctions don't care about order of evaluation and don't necessarily want all their values)
biab & 16:23
16:23 ruoso joined
Tene Are there currently any implementations of 'lazy' for Perl 5? 16:27
[particle] Moose has lazy 16:28
Tene Oh, Scalar::Defer 16:32
[particle] ah, right!
16:35 jeremyb joined 16:36 alecc left 16:38 arrowax joined 16:39 jjore-w joined 16:48 Daveman joined
TimToady I should say, lazy cares about the order of evaluation, but only with respect to its own list. 16:53
it necessarily follows that it can work ahead if it likes
16:57 justatheory joined
avar Aankhen``: echo 'my $res = do { "1" // 0 }; if $res {}' | perl script/kp6 -lisp | sbcl 16:58
same with 1 (Int)
* Error: #<KP6-STR {AF3AFC9}> fell through ETYPECASE expression.
Wanted one of (KP6-NAMED-PARAMETER KP6-POSITIONAL-PARAMETER).
17:01 theorbtwo joined
pugs_svnbot r18350 | ruoso++ | [yap6] Basic memory management, values are alloced and basic refcount garbage collection is implemented 17:01
diff: dev.pugscode.org/changeset/18350
lambdabot Title: Changeset 18350 - Pugs - Trac
17:03 weinig joined
pugs_svnbot r18351 | avar++ | [kp6-lisp] 17:03
r18351 | avar++ | * Added method support back to class Call, isn't supported by the
r18351 | avar++ | runtime yet but fails more nicely now
r18351 | avar++ | * //, || and && were returning CL (t/nil) object not kp6 objects so they failed on e.g. if (do { 1 || 0 }) {}
r18351 | avar++ | * Re-indented, fix yer settings aankhen:)
r18352 | avar++ | Regened from the last few commits
diff: dev.pugscode.org/changeset/18352
lambdabot Title: Changeset 18352 - Pugs - Trac
17:04 weinig_ joined 17:06 kanru joined 17:07 weinig__ joined, weinig_ left
pugs_svnbot r18353 | fglock++ | [kp6] gather + map works now 17:07
diff: dev.pugscode.org/changeset/18353
lambdabot Title: Changeset 18353 - Pugs - Trac
fglock nested gather has problems 17:10
kp6: my $a = gather { my $k = 0; while 1 { take $k; $k = $k + 1; } };say ($a.map: -> $x { $x + 100 })[10]; 17:12
exp_evalbot r18353: OUTPUT[110ā¤]
17:12 jisom joined
avar oh noes, Aankhen`` broke my precious // :) 17:13
fglock kp6-perl5 feels real slow when working on lists 17:15
avar lie what program for example? 17:17
fglock kp6: my $a = gather { my $k = 0; while 1 { take $k; $k = $k + 1; } };say (($a.map: -> $x { $x + 100 }).map: -> $y { $y - 1 } )[99]; 17:20
exp_evalbot r18353: OUTPUT[198ā¤]
fglock hmm - not that bad
17:21 weinig left 17:24 Psyche^ joined 17:29 pina joined
fglock TimToady: re lazy cares about the order of evaluation, 17:29
@a = gather {...}; @b = @a.map:{ $_+1 }; say @b[10]; # does it need to add 1 to all of @a[0..9] ? 17:30
or maybe you mean "order of results" 17:32
17:33 GeJ joined
pugs_svnbot r18354 | ruoso++ | [yap6] basic dispatch, first tests 17:33
diff: dev.pugscode.org/changeset/18354
lambdabot Title: Changeset 18354 - Pugs - Trac
ruoso fglock, I would say it needs to do all of 0..9 when you try to mess with @b[10]
pmurias me too
it could cheat if there are no side effects, but that's an implementation detail 17:34
ruoso pmurias, it can't, because there's no way to tell "there are no side effects" 17:36
17:37 weinig__ left, weinig joined 17:40 Psyche^ is now known as Patterner
pugs_svnbot r18355 | ruoso++ | [yap6] consts increment the refcount when assigning the dispatcher... 17:42
diff: dev.pugscode.org/changeset/18355
lambdabot Title: Changeset 18355 - Pugs - Trac
avar Aankhen``: echo 'my $x = 5; say $x;' | perl script/kp6 -do Scope,EmitLisp 17:43
17:43 ntgrl is now known as integral
pmurias ruoso: i agree that analysing side effect is difficult but it could be possible if the compiler was smart enough 17:44
although the existence of eval could screw everything up
avar Scope.pm isn't used by the perl emitting stuff.. 17:46
pmurias: yes, eval sucks
It's one of those things along with caller() and other stuff that means you have to carry around stacks, scope info and whatnot 17:47
(eval) in CL for instance does not allow you to access variables in the scope that happens to have the eval call, it just allows you to make a new compilation unit 17:48
17:52 diakopter joined
pugs_svnbot r18356 | fglock++ | [yap6] minor fixes in Makefile, mem alloc, by ruoso++ 17:52
diff: dev.pugscode.org/changeset/18356
lambdabot Title: Changeset 18356 - Pugs - Trac
pmurias avar: it's a conflict between wanting perl6 to be a compiled language a keeping the dynamism 17:53
17:54 ChanServ sets mode: +o diakopter, diakopter sets mode: +o TimToady, diakopter sets mode: -o diakopter
avar pmurias: sure 17:54
pmurias: But as CL shows you can have lots of carefully crafted dymanism without slowing everything else down, if you're ready to drop som features 17:55
pmurias swimming& 17:56
fglock Scope.pm is mostly meant for implementing lexical scopes and closures in backends that don't support it natively 17:58
avar People studying how to design an optimizable dynamic language would do well to study CL, it was basically designed by a bunch of compiler implementors who knew what they were doing and what they needed to fix:) 18:00
*hint* *hint* :)
But it's clear that Perl 6 won't be as fast as CL in the general case since there's a lot of stuff that prevents you from compiling it to optimized assembly, and some of this stuff is hard to figure out statically by a compiler 18:03
integral notes that plenty of people can tell you what makes perl 5 hard to run fast, but little of it has changed for perl 6
18:03 scent joined
avar integral: That's not right, the biggest thing holding perl 5 back I'd say is magic. It's not copied in perl6 unless you bind the variable 18:05
with that and the type system you can optimize stuff like sub add(int $a, int $b --> int) { $a + $b } 18:06
(see sial.org paste earlier)
18:06 kanru left
pugs_svnbot r18357 | fglock++ | [kp6] gather - fixed several workarounds 18:07
diff: dev.pugscode.org/changeset/18357
lambdabot Title: Changeset 18357 - Pugs - Trac
integral avar: But how many people are going to put enough annotations on their code?
but you're right, magic is mostly gone which is nice :-) 18:08
avar integral: few, but you can do it for parts which need optimizing 18:09
pugs_svnbot r18358 | ruoso++ | [yap6] created a doc directory, included some notes about the ideas... 18:13
diff: dev.pugscode.org/changeset/18358
lambdabot Title: Changeset 18358 - Pugs - Trac
avar integral: But I think a lot of people are going to use Num/Int/Str etc. 18:15
18:16 kanru joined 18:18 polettix joined 18:25 snearch joined 18:27 penk joined 18:29 Jedai joined 18:37 ruoso left 18:55 REPLeffect_ joined 19:06 justatheory left, justatheory joined 19:16 _go_ joined 19:18 rindolf joined 19:25 amnesiac left 19:28 blindfish_ joined 19:44 blindfish left 19:48 pina left 19:56 snearch left
pmurias the compiler can create specialize multi variants based on profiling information 19:56
19:57 dduncan joined
pmurias psyco does something similar for python 19:58
20:06 blindfish_ left
pugs_svnbot r18359 | pmurias++ | [yap6] doc fix - I think Rouso meant firstly instead of at first, revert if wrong 20:14
diff: dev.pugscode.org/changeset/18359
lambdabot Title: Changeset 18359 - Pugs - Trac
20:15 vbattsAtWork left 20:28 ofer joined 20:36 shachaf joined 20:37 cognominal_ left 20:43 cognominal_ joined 20:47 drupek12 joined, thoughtpolice joined 20:49 _go_ left 20:54 drupek12 left, rindolf left 21:02 barney left 21:08 jeffreykegler joined 21:14 arrowax left 21:21 TJCRI left
pmurias g'night 21:22
21:22 pmurias left 21:24 iblechbot left 21:30 Daveman left 21:33 Daveman joined 21:56 polettix left 22:20 Limbic_Region joined 22:23 pjcj_ joined
pugs_svnbot r18360 | fglock++ | [kp6] s/length/chars/ 22:27
diff: dev.pugscode.org/changeset/18360
lambdabot Title: Changeset 18360 - Pugs - Trac
22:32 cmarcelo left 22:54 fax joined
avar fax: interested in more hackery?:) 22:58
fax yeah :)
Looks like @source[0] = 2; has broken 22:59
in kp6
hows progress?
avar we're passing a few more tests, but now we need the object model:) 23:00
look at lib/KindaPerl6/Runtime/Perl5/MOP.pm 23:01
The core so to speak being ::DISPATCH 23:02
basically that's the Perl 6 object model in Perl 5. You define an object in a non-std way in Perl 5 that has a name, parents, some functions etc.
Everything works in that model so you can tell a Str instance to tell you its length in chars for instance and it knows to look up its method, its parents method and so on 23:03
I haven't done anything in it but I've been wondering to what amount that can be layered on top of CLOS and some standard slots/methods. So far I haven't seen any reason not to do that 23:04
fax mhm
CLOS is possible you can use the metaobject protocol in CLOS to change most all behavior
avar ...and using CLOS would mean not implementing your own method resolution mechanism from a class higherarchy
fax [if that was needed] 23:05
avar yup
echo 'my %hash = (key => "value"); say %hash.elems' | perl script/kp6 | perltidy 23:08
that's a good example
::DISPATCH( $GLOBAL::Code_say, 'APPLY',
::DISPATCH( $Hash_hash, 'elems', ) )
fax ah ok I get it 23:14
avar basically we've been doing a lot of this stuff wrong:) 23:16
assignment for example in kp6-perl is just: 23:17
'::DISPATCH_VAR( ' ~ $node.emit_perl5 ~ ', \'STORE\', ' ~ $.arguments.emit_perl5 ~ ' )' ~ Main::newline();
23:19 devogon joined
fax hm so are you thinking about implementing the object model and then changing most of the lisp emitter stuff to use it like that? 23:20
avar Yeah
23:21 devogon left 23:23 devogon joined
pugs_svnbot r18361 | avar++ | Docs for Runtime::Perl6::Scope based on small discussion with fglock++ 23:24
diff: dev.pugscode.org/changeset/18361
lambdabot Title: Changeset 18361 - Pugs - Trac
avar I got distracted looking at assignments:)
23:27 melissa left 23:29 amnesiac joined 23:33 FurnaceBoy joined 23:35 jjore-1 joined
pugs_svnbot r18362 | avar++ | nuke some whitespace 23:39
diff: dev.pugscode.org/changeset/18362
lambdabot Title: Changeset 18362 - Pugs - Trac
23:43 drupek12 joined 23:47 SCalimlim joined 23:51 FurnaceBoy left, jjore-w left

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!