»ö« | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, alpha:, pugs:, std:, or /msg p6eval perl6: ... | irclog: irc.pugscode.org/ | UTF-8 is our friend!
Set by moritz_ on 25 June 2010.
00:00 Patterner left, Psyche^ is now known as Patterner 00:02 tyru joined, felliott left 00:04 justatheory left, masonkramer joined, hanekomu_9 left 00:07 [bjoern] joined
[bjoern] Shouldn't `say '*'..'~'` print out the characters "*+,-./01234...~"? Rakudo locally consumes lots and lots of memory doing very little, and the bot says ... 00:08
00:08 timbunce left
[bjoern] [02:05] <p6eval> rakudo 0b1212: 00:08
[02:05] <p6eval> ..OUTPUT«************...
jnthn rakudo: say '*'.succ 00:09
p6eval rakudo 0b1212: OUTPUT«*␤»
jnthn I thought the .. bug had been fixed. :-(
00:09 PZt left
[bjoern] It does work with a..z for instance 00:10
dalek kudo: f49e91b | jonathan++ | src/core/Mu.pm:
.can helper is only defined on Any, not Mu. Last bug I fixed masked this issue
kudo: 1231260 | jonathan++ | src/glue/subset.pm:
Need to attach the originally refined type to the subtype so the signature

awesome error and probably much more.
00:11 cdarroch left
tadzik g'night #perl6 00:12
au|irc g'♞ :)
tadzik :)
00:12 tadzik left
[bjoern] www.nntp.perl.org/group/perl.perl6....33775.html seems relevant 00:13
pugssvn r31771 | jnthn++ | [t/spec] Correct a bunch of spectests that relied on the old Int ~~ Num assumption, including changing one test to ensure that this is indeed not the case.
jnthn [bjoern]: Yes, there's been quite a lot of discussion about the .. operator of late. I'm not quite up on the latest. 00:15
[bjoern] s03 even seems to have related tests, so I don't have to bother filing a bug. Good.
jnthn Well, filing may help get it dealt with sooner. On the other hand, this one comes up quite a bit. 00:16
dalek kudo: 3139735 | jonathan++ | src/builtins/Num.pir:
Toss Int ~~ Num hack; from now on, use Int ~~ Numeric. Good to excise this one
kudo: ee1b11b | jonathan++ | src/Perl6/Grammar.pm:
Port over a couple more errors from STD - mostly copy/paste/test.
jnthn std: sub foo {}; foo(:a :b :c) 00:20
p6eval std 31770: OUTPUT«ok 00:01 115m␤»
jnthn Aww
Still don't like that :-)
00:21 Sarten-X left
jnthn Phew, back under 640 tickets :-) 00:24
au|irc yay :)
00:29 lest_away is now known as lestrrat
[Coke] Think we can do <600 for the release this week? 00:30
jnthn [Coke]: That'd be quite a push - especially when there's always more coming in. ;-) 00:32
[Coke]: As well as closing some tonight though, I've given several to moritz_++ for spectesting too.
[Coke]: We're probably sub-630 if you discount those.
00:40 japhb left 00:44 guest__ joined, japhb joined, guest__ left 00:47 whiteknight left 00:58 hudnix left
dalek ecza: d48aa80 | sorear++ | Kernel.cs:
Clean up list assignment, allow for custom FETCH / STORE / INVOKE
 01:05
01:06 takadonet1 joined
takadonet1 hey everyone 01:06
01:07 ashleydev left 01:12 [bjoern] left 01:13 au|irc is now known as au|afk 01:23 PZt joined
sorear hello takadonet1 01:24
takadonet1 sorear: someone is alive!
01:26 payload1 joined 01:27 azert0x left
sorear TimToady: Do you have any thoughts on the matter of ratchet optimization for Perl 6? 01:27
TimToady: just passing and returning 1-element lazy lists as the spec seems to imply turns into a lot of bonus method calls 01:28
01:28 payload left
sorear TimToady: my current plan involves a low-level hack in the method dispatcher 01:28
takadonet1 rakudo: class A { $.var is rw; } A.new(); a.var(3); say $a.var(); 01:40
p6eval rakudo ee1b11: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 22␤»
takadonet1 rakudo: class A { $.var is rw; }; A.new(); a.var(3); say $a.var();
p6eval rakudo ee1b11: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 22␤»
takadonet1 ...
tired 01:41
rakudo: class A { $.var is rw; }; my $ya=A.new(); $ya.var(3); say $ya.var();
p6eval rakudo ee1b11: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 22␤»
01:44 payload1 left
TimToady sorear: not really, though the 'andthen' operator was intended to allow linearization of defined-or-not cascaded hypothesis proving 01:44
(made it to PDX btw) 01:45
01:47 skids joined 01:48 Sarten-X joined 01:53 uvtc joined
dalek kudo: cf3da00 | jonathan++ | src/Perl6/ (2 files):
NYI notes for constant, state and macro. Hopefully we get to at least state by
01:56
takadonet1 jnthn: I just finished doing make spectest;make install ! 01:58
jnthn :P
Well, those are just some better errors rather than anything shiny. :-) 01:59
takadonet1 i know but still :P
hard to keep 3 versions of rakudo in syn
01:59 uvtc left
jnthn Well, I should probably consider some sleep soonish, so I dobut I'll be bothering you with more patches at least until tomorrow. ;-) 02:00
takadonet1 going to bed soon as well. Just one more commit... 02:01
jnthn :-)
takadonet1 jnthn: github.com/cjfields/bioperl6/tree/
jnthn Well, that was my just one more commit. :-) I've been reading while it spectested. :-) 02:02
jnthn looks
Wow, there's quite a lot of work in there. :-) 02:03
takadonet1 it's easy to port something that I used everyday in perl 5 :) 02:04
also without really knowing it, everyone on the channel been helping :) 02:07
jnthn :-)
Next up: make this run fast. :-)
02:08 Sarten-X left
jnthn -> sleep 02:10
02:11 drbean joined
sorear niecza: sub accum() { sub () { state $x = 0; $x++ }; }; my $y = accum; say $y(); say $y() 02:17
p6eval niecza: OUTPUT«===SORRY!===␤Unsupported scope state for simple variable at (eval) line 1:␤------> sub accum() { sub () { state $x ⏏= 0; $x++ }; }; my $y = accum; say $y();␤Attribute (positionals) does not pass the type constraint because: Validation failed for…
sorear oh right 02:19
I *did* ask moritz_ to delay starting the updater... guess I should fix that
02:20 marqos joined
marqos rakudo: const @COLORS = <red green blue>; say @COLORS 02:31
p6eval rakudo ee1b11: OUTPUT«===SORRY!===␤Symbol '@COLORS' not predeclared in <anonymous> (/tmp/l_vmcBpon8:22)␤»
marqos alpha: const @COLORS = <red green blue>; say @COLORS
p6eval alpha 30e0ed: OUTPUT«Symbol '@COLORS' not predeclared in <anonymous> (/tmp/P74wAQT149:10)␤in Main (file <unknown>, line <unknown>)␤»
TimToady std: const @COLORS = <red green blue>; say @COLORS
p6eval std 31771: OUTPUT«===SORRY!===␤Variable @COLORS is not predeclared at /tmp/J0wrrUQS0D line 1:␤------> const @COLORS⏏ = <red green blue>; say @COLORS␤Variable @COLORS is not predeclared at /tmp/J0wrrUQS0D line 1:␤------> @COLORS = <red green blue>; say
..@COLO…
TimToady std: constant @COLORS = <red green blue>; say @COLORS 02:32
p6eval std 31771: OUTPUT«ok 00:01 117m␤»
marqos oh, spelled out.
TimToady looks like that's gonna be one of those common flubs :)
marqos Thanks, Larry. I was just about to dive into the synopsis to see what I was missing.. I remembered that working.
02:33 payload joined
marqos rakudo: constant @COLORS = <red green blue>; say @COLORS 02:33
p6eval rakudo ee1b11: OUTPUT«===SORRY!===␤Symbol '@COLORS' not predeclared in <anonymous> (/tmp/bFdfyFaeTF:22)␤»
TimToady I think alpha does 02:34
marqos alpha: constant @COLORS = <red green blue>; say @COLORS
p6eval alpha 30e0ed: OUTPUT«redgreenblue␤»
TimToady (but it really just means 'my' there)
marqos freshly-built rakudo says it's not implemented yet...
TimToady it's supposed not be a scope declarator anymore either, but more like a subset name 02:36
02:36 masonkramer left
TimToady you might be able to get away with an enum containing one value :) 02:36
sorear TimToady: what do you mean by it being like a subset name?
TimToady std: my subset X of Any where True; 02:37
p6eval std 31771: OUTPUT«ok 00:01 115m␤»
TimToady std: my constant pi = 3;
p6eval std 31771: OUTPUT«ok 00:01 116m␤»
TimToady alpha: my constnat pi = 3;
p6eval alpha 30e0ed: OUTPUT«Malformed declaration at line 10, near "constnat p"␤in Main (file <unknown>, line <unknown>)␤»
TimToady alpha: my constant pi = 3; 02:38
p6eval alpha 30e0ed: OUTPUT«Malformed declaration at line 10, near "constant p"␤in Main (file <unknown>, line <unknown>)␤»
TimToady it thinks constant is like 'my', so doesn't allow a 'my' before it
02:39 takadonet1 left, felliott joined
TimToady dinner & 02:39
sorear std: constant = 5; 02:44
p6eval std 31771: OUTPUT«ok 00:01 116m␤»
sorear Does it really make any sense to allow anonymous constants?
STD.pm6:2938 02:45
02:54 huf joined
dalek ecza: 007fab2 | sorear++ | (2 files):
Implement the constant declator, with and without sigils (they aren't actually
02:56
03:05 brianherman joined
brianherman perl6: say "yay"; 03:05
p6eval pugs, rakudo cf3da0: OUTPUT«yay␤» 03:06
pmichaud good evening 03:07
sorear hello pmichaud 03:08
marqos so is there any way to have an eval affect the surrounding scope? 03:09
03:09 agentzh joined
sorear elaborate on "affect" 03:10
an eval is considered to be an ordinary inner scope
03:10 mrsaturn12 joined 03:11 plobsing joined
marqos so you can use %MY .. or however that's spelled ... 03:12
tylercurtis pmichaud: if, in a HLL::Grammar.O(...), I don't specify an associativity, is the operator taken to be non-associative?
pmichaud tylercurtis: I'd have to look. 03:13
(looking)
marqos rakudo: eval "$OUTER::x=1;"; say $x 03:15
p6eval rakudo cf3da0: OUTPUT«===SORRY!===␤Symbol '$x' not predeclared in <anonymous> (/tmp/5pEe2gpDZT:22)␤»
pmichaud marqos: an eval is not permitted to add to the outer scope at runtime 03:16
(and in this case, $x is not predeclared :-)
marqos rakudo: eval "OUTER::$x=1"; say $x
p6eval rakudo cf3da0: OUTPUT«===SORRY!===␤Symbol '$x' not predeclared in <anonymous> (/tmp/1As4uwbRv_:22)␤»
pmichaud tylercurtis: I'm not sure what associativity is assumed. The algorithm basically follows whatever STD.pm6 did (at the time it was written) 03:17
I think it may be right-associative by default.
tylercurtis pmichaud: is there a way to specify non-associativity?
marqos So how does one create a variable/sub/method with a name that's not known till runtime?
pmichaud :assoc<non>, I think.
looking 03:18
tylercurtis pmichaud++: thanks.
pmichaud :assoc<unary>, :uassoc<non> # looks like
not sure that HLL::Grammar implements true non-associativity yet, though. 03:19
(if non-associativity is missing, it's arguably a bug.)
oh, looks like there's also :assoc<non>
(in STD.pm6)
yes, I'm guessing that HLL::Grammar doesn't implement non-associative yet. We can add it if you need it. 03:20
(I probably don't have it yet because I hadn't really needed it.) 03:21
03:21 skids left
marqos I guess assigning to MY::{$varname} should work... 03:22
pmichaud 03:16 <pmichaud> marqos: an eval is not permitted to add to the outer scope at runtime
marqos Right. Hence my question of how do you add a dynamic name to a scope at runtime?
Since eval ain't it.
mrsaturn12 Why would you want to?
pmichaud From S02: 03:25
You may not use any lexically scoped symbol table, either by name or
by reference, to add symbols to a lexical scope that is done compiling.
(We reserve the right to relax this if it turns out to be useful though.)
03:26 ruoso joined
marqos But that only applies to lexical scopes, right? You can dynamically add to a class or module? 03:26
tylercurtis pmichaud: I'm not sure that I do. Technically, looking at Squaak's grammar in Episode 3 of the tutorial, all the binary ops appear to be associative(although the grammar listed there doesn't describe operator precedence at all, so it's hardly definitive). I'm going to assume left associative for all of them, since it doesn't clearly specify and left associativity is what one would expect for most of its ops.
marqos In this case I was just playing around with porting some code that dynamically creates a bunch of delegation methods that are all very similar... 03:27
pmichaud tylercurtis: best for the tutorial is to explicitly list the associativity, I'd think. 03:28
marqos basically $container.$method($key, *@args) is defined as %.children{$key}.$method(@args) for a bunch of different values of $method
tylercurtis pmichaud: I'm explicitly listing it. I'm just having to guess at what I should explicitly list. Did PGE assume a certain associativity?
pmichaud I think PGE assumed left. 03:29
(it did assume something, yes.)
(I'm pretty sure it was 'left')
sorear Is PGE dead yet?
pmichaud "It's pining for the fjords..." 03:30
(or something like that)
tylercurtis pmichaud: I expect left is the case, too, since geb() or nyr() or fnu() would be very unintuitive if right-associative, for example.
marqos as far as I can tell, "handles" doesn't quite do that. 03:31
sorear marqos: yes, you can add stuff to a class at runtime
try eval "augment Foo { method bar () { } }
brianherman So I am compiling perl6 from source on mac os x and i get this error auto::va_ptr - Test the type of va_ptr...
step auto::va_ptr died during execution: Unknown va_ptr type at config/auto/va_ptr.pm line 42.
can i ignore that?
marqos sorear: right, but in this case the method name is the value of a variable.
sorear brianherman: that doesn't look like a perl 6 build
marqos: eval is a string!
you can interpolate whatever you want
marqos Oh, sorry, missed the eval. :) 03:32
sorear methods and classes don't live in lexical scopes, so you're free to add them in evals
immutability only applies to "my"
pmichaud also, use single quotes.
otherwise the braces are likely to hurt.
(or use some quoting construct other than "...") 03:33
marqos so if I'm inside Foo, should just eval "my $foo='bar'; method $foo \{ ... }" work?
sorear It won't work. 03:34
Perhaps it should.
marqos (in theory, I mean, not necessarily on current rakudo...)
pmichaud well, it's possible to have methods that aren't has-scoped, so I suspect one would need to re-list the class explicitly.
I suppose there might be a way to detect that eval is taking place inside of a class scope. 03:35
brianherman sorear: i am sorry that was parrot 03:36
marqos Wouldn't it have more to do with the fact that the thing being eval'ed is 'method', which defaults to has?
pmichaud brianherman: I'm guessing that you're missing a developers library or module that is needed for va_ptr. (It's just a guess.)
03:36 Sarten-X joined
pmichaud brianherman: you might try asking on irc.perl.org/#parrot 03:37
tylercurtis walk &
pmichaud (and then let us know here so we can document it somewhere.)
03:37 mrsaturn12 left
sorear Methods are very weird, since they can have two scopes at once 03:37
"my method" is really "my has method" 03:38
it goes into both the lexpad and the class
pmichaud ...if it's inside of a class declaration.
sorear my &foo = anon method, by contrast, goes into only the lexpad
pmichaud right.
brianherman only 3 people are in parrot 03:39
pmichaud brianherman: on irc.perl.org, not freenode
brianherman oh
pmichaud (or irc.parrot.org)
brianherman meh i think ill just use linux 03:42
thanks guy 03:43
s
pugssvn r31772 | pmichaud++ | [pm.txt]: Add Pm-21 and Pm-22 to request confirmation of lexical regex semantics. 03:45
03:46 IllvilJa1 joined 03:47 marqos is now known as markjreed
pmichaud phenny: tell jnthn Is there a way from actions to see if an outer scope defines a method (regex, actually) of a given name? See Pm-21 in misc/pm.txt . 03:47
phenny pmichaud: I'll pass that on when jnthn is around.
markjreed sorry, was not myself.. 03:48
pmichaud phenny: it doesn't appear that methods are entered in a PAST::Block's symbol table.
phenny: tell jnthn it doesn't appear that methods are entered in a PAST::Block's symbol table.
phenny pmichaud: I'll pass that on when jnthn is around.
03:48 IllvilJa left, DemoFreak left
markjreed rakudo: class Foo { }; eval 'augment Foo { method foo { say "Hello, world!" } }' 03:48
p6eval rakudo cf3da0: ( no output )
Tene rakudo: class Foo { }; eval 'augment Foo { method foo { say "Hello, world!" } }'; my Foo $foo .= new; $foo.foo; 03:49
p6eval rakudo cf3da0: OUTPUT«Method 'foo' not found for invocant of class 'Foo'␤ in main program body at line 22:/tmp/BLIbIYqzEX␤»
markjreed Locally I get ==SORRY!=== 03:50
In "augment" declaration, typename Foo must be predeclared (or marked as declarative with :: prefix) at line 1, near " { method "
03:50 DemoFreak joined
Tene rakudo: class Foo { }; augment Foo { method foo { say "Hello, world!" } }; my Foo $foo .= new; $foo.foo; 03:51
p6eval rakudo cf3da0: OUTPUT«===SORRY!===␤In "augment" declaration, typename Foo must be predeclared (or marked as declarative with :: prefix) at line 22, near " { method "␤»
Tene rakudo: class Foo { }; eval 'augment ::Foo { method foo { say "Hello, world!" } }'; my Foo $foo .= new; $foo.foo;
pmichaud augment *class* Foo
p6eval rakudo cf3da0: OUTPUT«Method 'foo' not found for invocant of class 'Foo'␤ in main program body at line 22:/tmp/Nn0z2sYf2t␤»
Tene rakudo: class Foo { }; eval 'augment class Foo { method foo { say "Hello, world!" } }'; my Foo $foo .= new; $foo.foo;
p6eval rakudo cf3da0: OUTPUT«Method 'foo' not found for invocant of class 'Foo'␤ in main program body at line 22:/tmp/tZfzclAnQV␤»
pmichaud I'm guessing the eval fails.
markjreed class Foo { }; augment class Foo { method foo { say "Hello, world!" } }
Tene rakudo: class Foo { }; augment class Foo { method foo { say "Hello, world!" } }; my Foo $foo .= new; $foo.foo;
pmichaud probably need "use MONKEY_TYPING"
p6eval rakudo cf3da0: OUTPUT«===SORRY!===␤Can't augment class Foo without 'use MONKEY_TYPING'␤»
Tene yeah 03:52
03:53 ashleydev joined
markjreed rakudo: use MONKEY_TYPING; class Foo {}; eval 'augment class Foo { method foo { say "hello, world" } }'; Foo.new.foo 03:58
p6eval rakudo cf3da0: OUTPUT«Method 'foo' not found for invocant of class 'Foo'␤ in main program body at line 22:/tmp/WnPsmWsN6j␤»
markjreed rakudo: use MONKEY_TYPING; class Foo {}; augment class Foo { method foo { say "hello, world" } }; Foo.new.foo
p6eval rakudo cf3da0: OUTPUT«hello, world␤»
pmichaud might need the MONKEY_TYPING inside the eval.
(arguably a bug if so.)
markjreed rakudo: class Foo {}; eval 'use MONKEY_TYPING;augment class Foo { method foo { say "hello, world" } }'; Foo.new.foo 03:59
p6eval rakudo cf3da0: OUTPUT«hello, world␤»
markjreed ahh.
That will work. Thanks! 04:00
dalek p-rx: b09bf8d | pmichaud++ | src/Regex/P6Regex/Actions.pm:
Refactor named_assertion method a bit.
 04:04
p-rx: f7e92b4 | pmichaud++ | src/stage0/ (3 files):
Update bootstrap.
markjreed Hm. augment seems to re-trigger role composition... 04:05
04:06 TiMBuS joined
tylercurtis pmichaud: does the bit at trac.parrot.org/parrot/browser/trun...7.pod#L234 beginning with "What happens if you don't specify a pasttype?" still accurate? I'm guessing no, but I wanted to check. 04:09
04:09 snarkyboojum left 04:10 cls_bsd left
markjreed rakudo: role Bar { has @.on_tap }; class Foo does Bar {}; eval 'use MONKEY_TYPING; augment class Foo { method foo { say "hello, world" } }'; say $! 04:11
p6eval rakudo cf3da0: OUTPUT«Attribute '@!on_tap' already exists in the class, but a role also wishes to compose it␤»
sorear is going to need to start thinking about optimization passes soonish 04:12
04:15 markjreed left 04:22 timbunce joined, mrsaturn12 joined 04:24 brianherman left 04:33 justatheory joined
dalek ecza: 663c423 | sorear++ | (3 files):
Implement parsing for simple regexes
 04:40
04:49 PZt left
timbunce jnthn: I hear Zavolaj is broken at the moment. Any ETA for a fix? 04:51
05:04 snarkyboojum joined 05:05 justatheory left 05:07 colomon left 05:08 PZt joined 05:10 plobsing left 05:11 felliott left 05:21 mrsaturn12 left 05:23 payload left 05:24 daxim joined 05:26 timbunce left 05:28 timbunce joined 05:42 cls_bsd joined
sorear std: / :my $x; /; say $x; 05:45
p6eval std 31772: OUTPUT«ok 00:01 118m␤»
sorear I argue this is wrong.
regexes should get a new scope around every quant_atom_list
rakudo: / :my $x; /; say $x; 05:46
p6eval rakudo cf3da0: OUTPUT«Any()␤»
sorear *blink*
05:47 timbunce left 05:52 synth left 05:56 timbunce joined 06:00 ingy left 06:02 timbunce left, ingy joined, cognominal left 06:03 cognomore left 06:04 snarkyboojum left, ashleydev left 06:12 uniejo joined 06:13 PZt left, alfieANDfred joined, PZt joined 06:14 alfieANDfred left, cognominal joined
moritz_ good morning 06:19
mathw Morning
tylercurtis moritz_: good morning. 06:20
moritz_ yawns 06:21
06:21 uniejo left 06:23 uniejo joined 06:25 _mpu joined
sorear good morning moritz_ 06:28
I seem to have forgotten to tell you that I no longer need niecza p6eval updates held
moritz_ crontab installed. 06:30
I think you did tell me, but I forgot to tadd the cron job
sorear: manually triggered niecza rebuild is finished 06:46
sorear: btw if you add a step to make that puts the current revision into a file, I can report that revision from p6eval 06:47
./perl6 -e 'my $x = "foo"; $x ~~ s:g/o/u/; say $x' 06:50
fuu
ingy greetings 06:51
moritz_ oh hai
ingy is sitting alone in the oscon hotel lobby
I just wrote a p6rules-as-yaml reducer in Python 06:52
I have a question
does rakudo use p6rules to parse perl6? 06:53
I'm assuming it does
moritz_ yes
src/Perl6/Grammar.pm
06:53 eternaleye left
ingy so my ambition is to port the p6 grammar to my acmeist p6rules subset called FooParse, that uses a PCRE engine 06:55
and is thus many times faster
moritz_ have fun! :-)
ingy I will!
I feel like I'm really close 06:56
maybe
so is it true that rakudo has no access to a regexp engine? 06:57
(besides p6rules of course...)
tylercurtis ingy: if you're really ambitious, try converting STD.pm6 over to your subset. :)
ingy which is a stretch to call regexp
tylercurtis ingy: not a stretch, a shrink. 06:58
sorear p6rules are far more powerful than regexes and are mostly syntax compatible
ingy I agree
but I'm just asking if rakudo can invoke a "normal" regexp engine 06:59
ie pcre, etc
sorear no
ingy does parrot have one?
tylercurtis ingy: you could use one with NCI. 07:00
sorear parrot has a Regex engine but it's actually a p6rules engine
cxreg tylercurtis: except that without zavolaj, nci is a pain in the ass :) 07:01
ingy so here's an example.... 07:02
I took this 300 line TestML grammar: github.com/ingydotnet/testml-pm6/bl...Grammar.pm
which was ported from this working homebaked-yaml grammar: github.com/ingydotnet/testml-gramma...ammar.yaml 07:03
and turned the original into this: github.com/ingydotnet/testml-gramma...ml.grammar 07:04
and then reduced it into this 90 line, highly optimized: github.com/ingydotnet/testml-gramma...ammar.yaml
which is pcre compatible
sorear moritz_: should I just put the capture-command in all?
all: Setting.dll 07:05
\tgit rev-parse HEAD > VERSION
sjohnson hi
sorear hello
ingy well anyway, I gotta sleep
or drink
night... 07:06
sorear I look forward to seeing how PCRE handles LTM
sjohnson hi sorea
r
oops
ingy sorear: me too :) 07:07
ingy grabs a smop
sorear Are you modifying pcre? 07:08
moritz_ sorear: I'd prefer it if you could truncate the git rev-parse output to 6 characters or so 07:12
nopaste.snit.ch/22136 # hacky s:g/// 07:15
07:15 _mpu left
moritz_ in STD.pm6 line 960, why is there a <circumfix> after a <.sorry> ? 07:18
07:20 [particle]1 joined
moritz_ ah, for error recovery, probably 07:20
07:20 [particle] left 07:23 baest joined 07:24 _mpu joined
dalek ecza: a4a0806 | sorear++ | Niecza/Actions.pm:
Small refactor of declaration handling
 07:31
ecza: 1dad6ad | sorear++ | CgOp.pm:
Rework CgOp::* to preserve much more structure
ecza: cebf46b | sorear++ | (2 files):
Add a version recorder for moritz
07:36 mberends left
moritz_ nqp: $_<v> # just want to see if it parses 07:39
p6eval nqp: OUTPUT«Symbol '$_' not predeclared in <anonymous>␤current instr.: 'parrot;PCT;HLLCompiler;panic' pc 152 (compilers/pct/src/PCT/HLLCompiler.pir:109)␤»
moritz_ nqp: my $_; $_<v> # just want to see if it parses
p6eval nqp: OUTPUT«get_pmc_keyed() not implemented in class 'Undef'␤current instr.: '_block11' pc 0 (EVAL_1:11596908)␤»
moritz_ nqp: my $_; ~$_<v>
p6eval nqp: OUTPUT«get_pmc_keyed() not implemented in class 'Undef'␤current instr.: '_block11' pc 0 (EVAL_1:4)␤»
07:44 [particle]1 left, [particle] joined 07:46 zulon joined, meppl joined 07:54 dolmen_ joined
pmichaud moritz_: ping 08:07
locally:
> my regex abc { ... }; say "abc" ~~ / <abc> /; 08:08
abc
dalek ok: a366001 | pmichaud++ | src/preface.pod:
It's no longer true that variables aren't preserved in the REPL.
 08:10
ok: 6e6ccf8 | pmichaud++ | src/ (3 files):
Various text improvements.
ok: 9a5a82a | pmichaud++ | docs/review-notes.txt:
Add a place for review-notes and discussion about the text.
pmichaud another long day... time for sleep. 08:13
sorear ugh. this big codegen refactor is bigger than I thought
sorear puts most of it on hold
Tene Oh, huh, feeds. Didn't notice when those went in.
Looks like they're eager, though. 08:14
08:16 Mowah joined
moritz_ pmichaud: pong 08:18
probably too late
08:21 thebird joined
pmichaud back 08:25
(not too late :)
feeds are eager? /me wonders why that would be....
08:27 perlygatekeeper1 joined
frettled pmichaud: that was a proper powernap :) 08:27
pmichaud frettled: no, just hadn't made it to sleep yet. 08:28
too busy playing on a few other items.
frettled :)
moritz_ pmichaud: I wonder how <print> should work...
does the lexical lookup check for regex-ness?
08:29 perlygatekeeper left
pmichaud moritz_: at the moment, no. 08:29
08:29 tadzik joined
pmichaud moritz_: it really can't check for regex-ness (and in some sense it should not check for that) 08:30
moritz_ thought so :(
tadzik hello #perl6
moritz_ wouldn't mind renaming <print> to <printable>
pmichaud yes, the reason I haven't committed is because of the strong potential for collisions between lexical symbols and grammar regexes 08:31
moritz_ pmichaud: push it to a branch for now?
pmichaud I have it in a local branch. I can push it, yes. 08:32
branch is 'reglex'
dolmen_ rakudo: say (2i..5i).WHAT
moritz_ oh, the pain...
p6eval rakudo cf3da0: OUTPUT«Range()␤»
dolmen_ rakudo: say (2i..5i)
p6eval rakudo cf3da0: OUTPUT«0 + 2i␤»
dolmen_ rakudo: say (2..5) 08:33
p6eval rakudo cf3da0: OUTPUT«2345␤»
pmichaud rakudo: say (2i).succ
p6eval rakudo cf3da0: OUTPUT«1 + 2i␤» 08:34
pmichaud rakudo: say (2i).succ after 5i
p6eval rakudo cf3da0: OUTPUT«1␤»
dolmen_ what is "after"?
moritz_ you can't have a ponie, and eat it :-)
dolmen_: a comparison operator
pmichaud after returns true if the lhs logically succeeds the rhs
rakudo: say "a" after "z" 08:35
p6eval rakudo cf3da0: OUTPUT«0␤»
moritz_ rakudo: say 3 after 1
dolmen_ rakudo: say (2i.succ) after 2i
pmichaud unlike > or gt, it doesn't do a type coercion on its arguments
p6eval rakudo cf3da0: OUTPUT«1␤»
dolmen_ rakudo: say "3" after 2 08:36
p6eval rakudo cf3da0: OUTPUT«1␤»
dolmen_ should fail, isn't it?
moritz_ nope
it's magic.
pmichaud rakudo: say "b" after 2
p6eval rakudo cf3da0: OUTPUT«1␤»
pmichaud rakudo: say 2 after "b"
p6eval rakudo cf3da0: OUTPUT«0␤» 08:37
dolmen_ "type coercion" is magic. No type coercion means no magic, isn't it?
pmichaud dolmen_: I'd accept that it should fail, yes. Or there's also an interpretation that say that after (and cmp) fall back to stringish semantics
moritz_ I guess in the case of string and number, it coerces the number to Str, and then compares
pmichaud rakudo: say "03" after 2 08:38
p6eval rakudo cf3da0: OUTPUT«0␤»
pmichaud looks like it's falling back to str comparision
*comparison
dolmen_ rakudo: say (2i..0) 08:39
p6eval rakudo cf3da0: OUTPUT«␤»
dolmen_ rakudo: say (1..0)
p6eval rakudo cf3da0: OUTPUT«␤» 08:40
dolmen_ rakudo: say (2i..0).elems
p6eval rakudo cf3da0: OUTPUT«0␤»
pmichaud okay, I'm really off to sleep this time. 08:41
bbl
dolmen_ good night
moritz_ \o night
dolmen_ rakudo: say 2i.WHAT
p6eval rakudo cf3da0: OUTPUT«Complex()␤»
dolmen_ rakudo: say Complex ~~ Num
p6eval rakudo cf3da0: OUTPUT«0␤» 08:42
dolmen_ rakudo: say Complex ~~ Numeric
p6eval rakudo cf3da0: OUTPUT«1␤»
dolmen_ rakudo: say Rat.new(3i, -5i) 08:43
p6eval rakudo cf3da0: OUTPUT«too many positional arguments: 3 passed, 1 expected␤ in main program body at line 22:/tmp/TBqNYMu3aY␤»
dolmen_ rakudo: say 1i*Rat.new(3, -5)
p6eval rakudo cf3da0: OUTPUT«0 + -0.6i␤»
dolmen_ rakudo: say Rat.new(3, -5i) 08:45
p6eval rakudo cf3da0: OUTPUT«too many positional arguments: 3 passed, 1 expected␤ in main program body at line 22:/tmp/0eZVwaQL6q␤»
dolmen_ rakudo: say 1i+Rat.new 08:46
p6eval rakudo cf3da0: OUTPUT«0 + 1i␤»
dolmen_ rakudo: say (3+2i..4).elems 08:47
p6eval rakudo cf3da0: OUTPUT«1␤»
dolmen_ rakudo: say (3+2i..4).perl 08:48
p6eval rakudo cf3da0: OUTPUT«Complex.new(3, 2)..4␤»
dolmen_ rakudo: say (3+2i..4)
p6eval rakudo cf3da0: OUTPUT«3 + 2i␤»
dolmen_ rakudo: say (3+2i..5+5i)
p6eval rakudo cf3da0: OUTPUT«3 + 2i4 + 2i5 + 2i␤»
dolmen_ rakudo: say (.) 08:49
rakudo: say (3.1..5.5)
p6eval rakudo cf3da0: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 22␤»
rakudo cf3da0: OUTPUT«3.14.15.1␤»
dolmen_ rakudo: say (3+2i..5+1i) 08:50
p6eval rakudo cf3da0: OUTPUT«3 + 2i4 + 2i␤» 08:51
dolmen_ rakudo: say (3+2i..^5+2i)
p6eval rakudo cf3da0: OUTPUT«3 + 2i4 + 2i␤»
dolmen_ rakudo: say (3+2i).WHY 08:54
p6eval rakudo cf3da0: OUTPUT«Method 'WHY' not found for invocant of class 'Complex'␤ in main program body at line 22:/tmp/gOA4WP5IPV␤»
dolmen_ rakudo: say (3).WHY
p6eval rakudo cf3da0: OUTPUT«Method 'WHY' not found for invocant of class 'Int'␤ in main program body at line 22:/tmp/JqE9yoU7_4␤»
dolmen_ rakudo: say (3+2i).^methods() 08:56
p6eval rakudo cf3da0:
..OUTPUT«imabsacosechrealsceilingunpolarfloorasechRatroundacosecacoshNumcosechcotanatan2repolartanRealatancosacosnewtanhNumericperlStracotanhfrom-radiansACCEPTSpredatanhcosecexpto-radianssinhacotanrootsComplexlnsecaseccotanhcislogsechlog10randtruncatesinsqrtIntasinBoolcoshsuccsignasinhabsac…
dolmen_ rakudo: say (3+2i).^methods().join(',') 08:57
p6eval rakudo cf3da0: 08:58
..OUTPUT«Str,acotanh,from-radians,ACCEPTS,pred,atanh,exp,cosec,to-radians,sinh,acotan,roots,Complex,ln,sec,asec,cotanh,cis,log,log10,sech,rand,truncate,sqrt,sin,Int,asin,Bool,cosh,succ,sign,asinh,im,acosech,abs,reals,ceiling,unpolar,floor,asech,Rat,round,acosh,acosec,Num,cotan,cosech,atan2,…
dolmen_ rakudo: say (3+2i).im 08:59
p6eval rakudo cf3da0: OUTPUT«2␤»
dolmen_ rakudo: say (3+2i).im.WHAT
p6eval rakudo cf3da0: OUTPUT«Int()␤»
dolmen_ rakudo: say (3+2.0i).im.WHAT
p6eval rakudo cf3da0: OUTPUT«Rat()␤» 09:00
dolmen_ rakudo: say (3+2.3i).im.WHAT
p6eval rakudo cf3da0: OUTPUT«Rat()␤»
dolmen_ rakudo: say (3+2.3i).im
p6eval rakudo cf3da0: OUTPUT«2.3␤» 09:01
dolmen_ rakudo: say (3+2.3i).im.denominator
p6eval rakudo cf3da0: OUTPUT«10␤»
dolmen_ rakudo: say (2.3+2.3i).real.WHAT 09:02
p6eval rakudo cf3da0: OUTPUT«Method 'real' not found for invocant of class 'Complex'␤ in main program body at line 22:/tmp/bJt0QPCU1c␤»
moritz_ .re
09:02 au|afk is now known as au|irc
dolmen_ rakudo: say (2.3+2.3i).re.WHAT 09:02
p6eval rakudo cf3da0: OUTPUT«Rat()␤»
au|irc \o
sorear hmm.
moritz_ o/
sorear Hi au|irc !
au|irc greetings :) 09:03
dolmen_ \o/
sorear it looks like implementing Iterator in Perl 6 is not going to work very well
because $!value doesn't remember the flatness bit
au|irc going into .cs level then?
frettled au|irc: o/~ - nice bugrep 09:04
au|irc thank you :) I tried golfing it...
tylercurtis good night, #perl6.
09:05 tylercurtis left
sorear Or I'll just try handling flat-bits out of band 09:06
moritz_ has a not-so-hacky-anymore patch for s:g///
au|irc sorear: out of band, as in...? 09:07
sorear au|irc: in a separate array 09:08
but I think I can hack this into working
au|irc looking forward to the commit then :D
au|irc continues translating masak++'s blogpost, part 2 of 5...
sorear well. I already have Iterator, I'm trying to implement a form of map 09:09
getting the sub-parcels to properly flatten is a little tricky
dolmen_ rakudo: say (2.3+2.3i).^methods()[0].do.perl
p6eval rakudo cf3da0: OUTPUT«{ ... }␤»
sorear because of the CPS conversion, the impedence mismatch between C# and Perl 6 is rather large
moritz_ dolmen_: .() for invoking, not do()
sorear my choices are C# (have to do manual CPS, and cannot access CORE:: lexicals without dependency injection) 09:10
Perl 6 (no direct access to CLR facilities like raw arrays, hard to do low-level introspection of values) 09:11
Q:CgOp (easy access to CLR features, with automatic CPS and lexical scoping, but the syntax is awful)
au|irc uh, there wasn't much syntax to start with :) 09:12
dolmen_ realizes that au|irc is Audrey...
au|irc confirms the realization
sorear I'm firmly with TimToady on the "oatmeal with fingernail clippings mixed in" view of sexp languages
they are, however, very expedient 09:13
moritz_ pushed the subst_adverbs branch
dolmen_ moritz_: I wanted the source code of the method. This is what I got, as far as it is possible : { ... }
au|irc it's slightly better if you allow ({[ and ]}) as synonyms, but yeah.
arnsholt pmichaud++ # Fixing classes-with-Q-bug 09:14
moritz_ dolmen_: yes, deparsing (or storing source code) is NYI
arnsholt Now Prolog looks like it parses the really perverted forms of Prolog as well =D
sorear The biggest upshot of switching from RPN to sexps is that I can use % in vim now
09:15 zulon left
arnsholt That sounds like a non-negligeable benefit =) 09:15
moritz_ can I use a PAST tree inside the :value of a PAST::Val? 09:17
std: s:nth[1, 2, 3]/a/b/ 09:18
p6eval std 31772: OUTPUT«===SORRY!===␤Unrecognized regex metacharacter , (must be quoted to match literally) at /tmp/pfKqqYdWHn line 1:␤------> s:nth[1,⏏ 2, 3]/a/b/␤Couldn't find terminator ] at /tmp/pfKqqYdWHn line 1:␤------> s:nth[1,⏏ 2,
..3]/a/b/␤Par…
moritz_ std: s:nth(1, 2, 3)/a/b/ 09:19
p6eval std 31772: OUTPUT«ok 00:03 118m␤»
moritz_ looks LTA to me
09:23 lue left, lue joined
au|irc sorear: R6RS defines [...] to be equiv forms of (...). Would you find a change to cgexp:op that allow (prog [...] [...] [...]) to improve ergonomics, and making writing Iterators in CgOp a tiny bit easier? :) 09:25
sorear Yes 09:26
au|irc on it.
moritz_ jnthn, pmichaud: in the subst_adverbs branch, s:g/o/u/ works. However passing arguments to adverbs (like :g(1)) gives "undefined identifier 'Capture'" - any idea what might be wrong? 09:28
09:29 dakkar joined
frettled Woohoo, parser confusion. 09:30
09:31 snarkyboojum joined
dalek ecza: e538c83 | sorear++ | CgOp.pm:
Start drafting a CgOp-level CPS converter
 09:33
ecza: c9d4a66 | sorear++ | (2 files):
Add Mu.RAWCREATE

anyways for thawing procedutres. Also stub Decl::Regex.
09:42 xabbu42 joined 09:46 kazoo joined
au|irc sorear: alright, it's in. not yet checked for proper circumfix balancing but works :) 09:46
sorear excellent 09:47
09:48 telling2 joined
au|irc finds ":setf lisp" to be great when hacking CgOp 09:48
09:49 telling2 left
dalek ecza: f98b0b1 | au++ | (2 files):
* Grammar: Allow [] in CgOp as synonym forms for ().

readability.		 09:51
ecza: 0fd7044 | au++ | setting:
* Also bracketify the prog args in c9d4a66.
ecza: 0fb98a0 | sorear++ | setting:
Implement List.iterator
 09:57
ecza: 63ae8f5 | sorear++ | rxwork.pl:
First draft of the regex engine
sorear au|irc: so, what are you trying to do with niecza? 09:59
au|irc nothing, really :)
I've never coded C# before
so this is mostly for learning
(I did code some F#, so that helps)
10:00 agentzh left
sorear it's a lot nicer than I expected it would be 10:00
au|irc yeah. erik (meijer) was justifiably proud of it (the post-LINQ incarnation, that is) 10:01
sorear rakudo: "foo".substr(1,1)
rakudo: "foo".substr(1,1).say
p6eval rakudo cf3da0: ( no output )
rakudo cf3da0: OUTPUT«o␤»
10:09 kazoo left
dalek ecza: 69370b6 | sorear++ | (3 files):
Implement Str.chars and Str.substr
 10:16
ecza: 90ad541 | sorear++ | rxwork.pl:
Draft the rest of the regex primitives needed here
au|irc sorear: so I'm filling in the "..." for Op::Yada::code; it reads 10:17
CgOp::subcall(CgOp::fetch(CgOp::scopedlex("&warn")), CgOp::clr_string(">>>Stub code executed<<<"));
but it compiles to
return ((IP6)s0).Invoke(th, new LValue[1] { ((System.String)s1).lv }, null);
and of course strings can't be LVs
maybe I'm missing a RV hint somewhere, or is that a misuse of ::subcall?
ah nvm 10:18
wrong boxing level
::string_var works
sorear yeah. you hit it. 10:19
au|irc warn($), exit() and "..." are in.
dalek ecza: 6fd3e0b | au++ | (3 files):
* Implement warn($), exit(), and the "..." term.
 10:22
ecza: 29b9da5 | au++ | (2 files):
* A bit more bracketification for (prog []) on rxwork.pl and setting.
au|irc sorear: btw, is there an aesthetic reason that the .pm files are using Moose instead of MooseX::Declare? :) 10:26
sorear no 10:27
dalek ecza: e06200c | sorear++ | setting:
Implement Parcel & Nil
 10:28
10:31 azert0x joined
sorear argh. I forgot to compile that last one 10:32
sorear is about to push a fix
this 15,000 line Setting.cs is making me want to increase priority of the code generation refactor 10:33
dalek ecza: b6fa476 | sorear++ | setting:
Implement Parcel.iterator & fix build
 10:34
au|irc alright. MooseX::Declare'ifying Op.pm is pushed too 10:35
sorear how big is the load time penalty?
au|irc 0.7s 10:36
only pays once, though, not x .pm
wolverian au|irc: I like your "irc" suffix. is it to remind you you're on irc? :) 10:37
au|irc well, it's a carryover from irc.socialtext.net
where we all use |afk |irc |brb etc 10:38
it's a makeshift emulation of IM's presence protocol for IRC :)
wolverian "I require conversation"
right.
au|irc nice association!
mathw Conversation you can have
Please select a language 10:39
dalek ecza: 12fb896 | au++ | (2 files):
* Op.pm is now using MooseX::Declare syntax.
 10:40
10:41 zulon joined, colomon joined
rokoteko so how does perl6 fit scientific (where I mean math) computation? 10:42
daxim Rats are built-in, which is *sane* 10:43
rokoteko well irrats builtin would be *sane* but not easily achieved. ;)
anyhow. ++ for rats :) 10:44
how about, let's say power series? that's somehing computers would be useful for.
lazy evalution is something Im waiting for to handle power series. 10:45
daxim pugs: (6/55*17/8).perl
p6eval pugs: ( no output )
daxim rakudo: (6/55*17/8).perl
p6eval rakudo cf3da0: ( no output ) 10:46
daxim pugs: say (6/55*17/8).perl
p6eval pugs: OUTPUT«51/220␤»
daxim aha.
rokoteko are e, pi and i builtins? 10:48
colomon yes
well, maybe not i at the moment.
rakudo: say i
p6eval rakudo cf3da0: OUTPUT«Could not find sub &i␤ in main program body at line 22:/tmp/DJ15Fo0yF0␤»
colomon rakudo: say 1i 10:49
p6eval rakudo cf3da0: OUTPUT«0 + 1i␤»
colomon rakudo: say e, pi
p6eval rakudo cf3da0: OUTPUT«2.718281828459053.14159265358979␤»
daxim rakudo: say 1i.perl 10:50
p6eval rakudo cf3da0: OUTPUT«Complex.new(0, 1)␤»
rokoteko eval: ln(0+1i)
colomon rakudo: say ln(1i)
daxim prefix with one of the p6 implementation names
p6eval rakudo cf3da0: OUTPUT«Could not find sub &ln␤ in main program body at line 22:/tmp/On_AC5ge3Q␤» 10:51
colomon oh, right
rakudo: say log(1i)
p6eval rakudo cf3da0: OUTPUT«0 + 1.5707963267949i␤»
rokoteko eval: say log(e^(i*pi)), log(pi) 10:52
dalek ecza: efce6ed | au++ | CodeGen.pm:
* CodeGen.pm is now in MooseX::Declare syntax.
colomon what do you want to do with lazy eval and power series?
rokoteko colomon: so that I can define a power series and evaluate to as much precision I like without the interpreter choking. why are you asking? 10:53
colomon well, it's an interesting issue to me, and what you want is probably trivially possible with current rakudo. 10:54
daxim you mean by explicit programming, loop until epsilon is small enough?
rokoteko Can you do an eval example for me, let's say with sin() power series that is lazy?
daxim also, when did rakudo get laziness? I didn't notice 10:55
rokoteko daxim: no, Im not meaning a while loop with last inside.
colomon daxim: months ago.
hmmm. 10:56
dalek ecza: 4dc75bb | au++ | (3 files):
* Body.pm, CgOp.pm and Decl.pm are now in MooseX::Declare syntax.
 10:58
rokoteko colomon: something like "take 20 power_series_expansion", which takes the first 20 members of the power series in question 10:59
tadzik rakudo: my $sq = gather for 0..Inf { take $_ * $_ }; $sq[3].say
p6eval rakudo cf3da0: OUTPUT«9␤»
colomon rakudo: sub sine-power($x) { gather for 1, 3 ... * -> $n { $x ** $n / [*] (1 ... $n) }; }; say sine-power(0.1).munch(10).perl 11:00
tadzik now when I leave it without say, REPL hangs. Why?
p6eval rakudo cf3da0: ( no output )
colomon tadzik: because it's trying to evaluate the entire infinite series internally 11:01
tadzik ah, to print the result, right
colomon rakudo: sub sine-power($x) { gather for 1, 3 ... * -> $n { $x ** $n / [*] (1 ... $n) }; }; say sine-power(0.1).munch(2).perl
11:01 ruoso left, tyru left
p6eval rakudo cf3da0: ( no output ) 11:01
colomon hmmm.
oh, duh 11:02
rakudo: sub sine-power($x) { gather for 1, 3 ... * -> $n { take $x ** $n / [*] (1 ... $n) }; }; say sine-power(0.1).munch(2).perl
p6eval rakudo cf3da0: OUTPUT«(0.1, 0.000166666666666667)␤»
rokoteko is munch anti-eager?
colomon: very nice thank you. I suppose that is correct. :) 11:03
colomon rokoteko: munch(N) shifts the first N items off a list
rokoteko: well, it's not completely right, because I left off the sign
11:03 ruoso joined, tyru joined
rokoteko that sounds like that the complete list has to be built first? 11:03
colomon but 11:04
rakudo: say (0.1 * 0.1 * 0.1) / 6
p6eval rakudo cf3da0: OUTPUT«0.000166666666666667␤»
colomon rokoteko: if it's lazy, certainly not
dalek ecza: f4ddde8 | au++ | (3 files):
* Convert RxOp, Sig and Unit to MooseX::Declare. That's all of them. :-)
 11:05
rokoteko Ok, brilliant. Ive been wondering if perl6 would be a good language to learn first. Ive been studying Haskell for a year, be perl is more practical. and I have a relatively strong perl5 background, so it's like ... well, what's the word in English.
tadzik …awesome? :)
au|irc a perfect fit? :) 11:06
colomon rakudo: sub sine-power($x) { my $sign = 1; gather for 1, 3 ... * -> $n { take $sign * $x ** $n / [*] (1 ... $n); $sign *= -1; }; }; say sine-power(0.1).munch(4).perl
rokoteko perfect fit is good. I was thinking in the lines of "obvious" :)
p6eval rakudo cf3da0: OUTPUT«(0.1, -0.000166666666666667, 8.33333333333334e-08, -1.98412698412698e-11)␤»
rokoteko and also Timtoady impresses me. Ever since I read his talk (I think it was the second state of the onion) that mentioned the similarity between triangles and circles. That is a subject that truly fascinates me. 11:08
daxim www.perl.com/pub/a/1998/08/show/onion.html 11:09
rokoteko Using power series I hope that I can get a wider knowledge on the subject. :) As computers have rounding errors with irrationals, then I can decide how long they compute.
daxim: nods, thats the talk. "To wrap up, I'd like to talk about triangles. Here's a sample." 11:10
sorear subtraction appears to be doing something insane. 11:11
mathw sorear: that would be worrying 11:12
colomon sorear: hope that's in niecza and not rakudo... 11:13
au|irc niecza -e 'say 1-2' #worksforme...
11:13 lestrrat is now known as lest_away
rokoteko au|irc: in the power series I think he meant. :) 11:13
au|irc ahh. certainly :)
sorear no
0 - 1 = 0 11:14
colomon rokoteko: in practical terms, however, rakudo doesn't currently have a way of dealing with numbers more precisely than floating point.
sorear but it doesn't happen when I golf it
only in the middle of Parcel.iterator
au|irc is it somehow using uint? #wildguess
sorear no 11:15
I just found it
au|irc whee :)
sorear my $ix = $ll.elems - 1;
this parses as:
(my $ix = $ll.elems) - 1;
au|irc (my $ix = $ll.elems) -1
sorear in the setting
au|irc *nod*
sorear because the earlier mention of sub infix:<=> causes STD to forget its builtin knowledge of = 11:16
and override it with a left-associative additive-prec op
rokoteko colomon: ok, great. Maybe I should concentrate my studied on perl6 then. perl5 is just so great and it has a lot of practical uses also. 11:17
sorear ok I have the build fixed , pushing 11:18
rokoteko merely, about the math, Im interested about why circles have pi as in the ratio of radius the circumference and triangle has the pi as the sum of angles. Ive been studying hyperbolic geometry, complex analysis, topology and what all about this drives me forwards is currently trying to understand Atiyah-Singer theorem. 11:19
it would be nice to have some easy-to-use (ok, perl *is* complex) computing power at use to test thing out. 11:20
au|irc easy-to-use + complex = manipulexity :) #youCanHaveBoth 11:21
rokoteko # CamelCaseBoth or camel_case_both 11:22
colomon rokoteko: I should warn you that Rakudo currently has a bug which will make dealing with more than one power series at once very tricky. basically, you cannot have two infinite generators going from the same code at the same time.
so I should be able to say something like 11:23
11:23 i__ joined
dalek ecza: 50423d2 | sorear++ | test.pl:
Tests for Parcel (todo)
 11:23
ecza: bfb226a | sorear++ | setting:
Two small bugfixes, Parcel is still not quite working
ecza: bd00700 | sorear++ | Kernel.cs:
Disable p6exceptions until they are properly fixed
ecza: 1789938 | sorear++ | Niecza/Actions.pm:
Implement my @x := foo syntax
rokoteko colomon: ok. but it should be fixed relatively soon?
colomon rakudo: sub sine-power($x) { my $sign = 1; gather for 1, 3 ... * -> $n { take $sign * $x ** $n / [*] (1 ... $n); $sign *= -1; }; }; say (sine-power(0.1) Z+ sine-power(0.2)).munch(4).perl
p6eval rakudo cf3da0: OUTPUT«(0.4, 0, 0, 0)␤»
colomon and you can easily see that's not right. 11:24
rokoteko: I've been pressing to get it fixed for several months now, but it seems to be tricky.
au|irc sorear: hrm, I spoke too soon. converting the method arguments into MooseX::Declare imposes a type/signature checking penalty, and on a full "make test" run it slowed things down... (7s vs 18s) 11:26
sorear Aww :( 11:27
au|irc can revert if you'd like; it's a tradeoff between coding time and dev/compile turnaround time I guess :)
11:27 Lorn left
sorear I think I would prefer it; turnaround time is rather a problem 11:27
rokoteko colomon: Z is a zip-operator?
11:28 paul1234 joined
colomon rokoteko: yes 11:28
rokoteko something Ive missed in perl5 core also :)
colomon so Z+ lazily adds two (potentially infinite) lazy lists
au|irc sorear: ok, reverted :)
11:28 snarkyboojum left
dalek ecza: 2a2ae2d | au++ | (3 files):
Revert "* Convert RxOp, Sig and Unit to MooseX::Declare. That's all of them.

This reverts commit f4ddde8e73835aeed9e7f2bd747a70ca6b56f5d5.		 11:29
ecza: 58c7a4a | au++ | (3 files):
Revert "* Body.pm, CgOp.pm and Decl.pm are now in MooseX::Declare syntax."
ecza: 39786c2 | au++ | (2 files):
Revert "* Op.pm is now using MooseX::Declare syntax."
ecza: 65fe5e9 | au++ | CodeGen.pm:
fa2393d | sorear++ | setting:

function. Also fix a thinko with list flattening
11:29 zulon left
rokoteko so you think someone has found the attempts to understand perl5 interpreter code impossible to understand could do something with perl6 to help it get finally released? 11:30
colomon rokoteko: absolutely 11:31
well, maybe not a ton for the R* release next week. :)
moritz_ rokoteko: most of us have no deep knowledge of p5 guts, if at all 11:32
rokoteko: question is, what do you want to do? write modules? compiler? documentation? 11:33
colomon though even then, a good short but useful module could be written in the next week and make it into the release. and that would be entirely working in p6 itself, no compiler guts required
au|irc sorear: alright, I think I grokked niecza :) many thanks for handholding me through! hopefully I didn't abuse my commit bit too much...
rokoteko modules, interpreter, documentatio for the stuff Ive written of course. :) I know some of perl5 guts, as in I have some knowledge of cpansearch.perl.org/src/RURBAN/illg...index.html 11:34
au|irc waves and returns to dinner, then translation, then a repaired MacBookPro and then $dayJobAtApple :)
11:34 au|irc is now known as au|afk
rokoteko moritz_: of course I wish some of the "fun" stuff, as in I enjoy programming perl. I find it fun. :) 11:35
moritz_ rokoteko: I've started with a Perl 6 port of CGI::Application, where you can contribute if you want
tadzik moritz_: are Acme::* modules worth including in proto?
rokoteko au|afk: bye :) Hope you enjoy your forbidden fruit :) 11:36
moritz_ tadzik: yes
tadzik moritz_: I've ported Acme::Meow :) github.com/tadzik/Acme-Meow-perl6
rokoteko moritz_: ah! that's more interesting than catalyst stuff definitely :)
moritz_ tadzik: btw most Perl 6 modules on github put the perl6- in front... but it's not necessary 11:37
rokoteko Ive started to dislike the usual MVC design. That happened while I was programming Java for a while.
tadzik moritz_: well, I wasn't sure how it should be named
11:37 synth joined
moritz_ rokoteko: github.com/moritz/CGI-Application do you have github ID? 11:37
rokoteko moritz_: so most of "you" arent perl5 people?
colomon here's a better version of the sine series: gist.github.com/482850 11:38
moritz_ rokoteko: most of us can use perl 5, but don't hack the guts
colomon rokoteko: I've used perl5 a bunch, but I've never even looked inside the source for it.
rokoteko No, maybe I should. I just set up a git server at work and I kinda liked it.
moritz_: Ill take a look at your project. It well take awhile to take know the syntax but Id be more than happy to help. :) 11:39
or "just set up" is like couple weeks ago, Ive been enjoying my holidays ever since. :)
last nice I dreamt of programming perl, so I guess I kinda fond of it! :) 11:40
s/nice/night/
moritz_ rokoteko: when you've got a github id, just tell me the name, then I can add you as a committer. Most Perl 6 projects are very liberal wrt commit bits :-)
colomon hmmm, I dreamt of playing football (US) again.
rokoteko moritz_: ok. Ill contact you then. Ill try to get to know a bit of your project first. Ive mainly been programming in-house projects that also have 5-15 year-old in-house web frameworks. :) 11:42
sorear 1, (2, 3), ((4, 5), (Nil, 7)) --> 132Parcel()<instance>Parcel()<instance> 11:43
parcel flattening is nearly here
rokoteko but it's kind of fascinating also, to see how people have thought is the best way to do the shit.
moritz_ rokoteko: ok. There's no documentation yet, but the methods are all named in the same way as the Perl 5 CGI::Application methods
and there are tests :-)
takadonet morning all 11:46
rokoteko moritz_: personally I find code self-documenting, so it's not a problem. :)
moritz_ :-)
rokoteko moritz_: but as I written documentation ive also found its best to write it as you write code! 11:47
ecza: 7b4cabe | sorear++ | (2 files):
Improve handling of my @x in subs
rokoteko so it's kinda sad its undocumented, but that doesnt scare me. the least so as it is documented in an earlier written code.
moritz_: do you have a todo-list? 11:48
moritz_ rokoteko: it's "port more features from CGI::Application" 11:49
ie from the perl 5 version
rokoteko moritz_: ok.
moritz_ rokoteko: I fear somebody needs to write Cookie module first
rokoteko the code looks rather clear.
moritz_ at least for the cookie handling code in CGI::App
rokoteko some perl6 things I have learn. what on eart is method !foo {} ? 11:50
tadzik private method
rokoteko ahhh!
moritz_ ! means not, so method !foo is a method you don't call from the outside :-) 11:51
rokoteko %ENV{CGI_APP_RETURN_ONLY} .. you have some new CGI headers in ENV?
sorear stefan@stefans:~/niecza$ ./niecza_eval -e 'my $buf = ""; sub cat(*@x) { while @x { $buf = $buf ~ @x.shift; }; $buf; }; say cat(1, (2, 3), ((4, 5), (Nil, 7)))' 11:52
123457
\o/
moritz_ rokoteko: no, that's only internally used for testing
colomon lolibloggedontaylorseries: justrakudoit.wordpress.com/2010/07/...th-series/
moritz_ rokoteko: it's much easier to test if it doesn't print out stuff, but returns it
daxim needs moar loli.
rokoteko moritz_: ok.
dalek ecza: 8e5549c | sorear++ | setting:
Fix lists interpolating backwards
 11:53
ecza: 6d2ff7d | sorear++ | setting:
Fix bad use of signatures - parcel flattening works!
rokoteko so about [<prerun>] what is <prerun> there? a method? also how does the type system work like, every type has a "base-type" as I think Ints are Nums? 11:54
moritz_ rokoteko: <prerun> is the same as qw/prerun/ in Perl 5
and yes, it's a method name here 11:55
rokoteko ah .. so the <> is just a sugar for arrays?
moritz_ [...] is an array, and <foo bar> is a quoted list
rokoteko ok. I think I need to read the synposes also to get a hang of perl6 also. 11:56
moritz_ rakudo: say [<prerun postrun>].perl
p6eval rakudo cf3da0: OUTPUT«["prerun", "postrun"]␤»
moritz_ rokoteko: perlgeek.de/en/article/5-to-6 might be of interest for you
rokoteko moritz_: bookmarked, thanks. :) 11:57
ah wait what. POD syntax is the same? 11:58
moritz_ no
but similar
=begin stuff ... =end stuff
instead of =head1 ... =cut
rokoteko do you have links that absolutely define some functions, so you write like eclipse plugins that browse the docs for you? .. Im a unix+vim user myself but Ive met the just-graduated windows+eclipse user at work. 11:59
so you are able to write ... that is. 12:00
I found that they prefer java also because they can integrate the docs to their editor.
moritz_ not yet
there's been work on a tool for that, named 'grok'
but it's nto really mature 12:01
(afaict)
literal will likely correct me soon enough :-)
literal nah
rokoteko but it doesnt work if the documentation isnt written at the beginning like that? 12:02
or does it?
moritz_ it's meant to work independently of the location of the POD 12:03
literal same as in Perl 5 12:04
rokoteko I just asked him to show how eclipse works with POD and it sucked compare to Java thing. people, at my last workplace where I partly worked progamming java, really tried to force me to use eclipse. but I was just so much faster with vim. 12:07
moritz_ right. vim has the advantage of being a decent editor :-)
rokoteko yes. I took a look, out of interest, at the vim plugin for eclipse. omg. it was nothing like vim :) 12:10
anyhow. afk for now. Im on a holiday, as stated earlier, but just had an urge to do something with perl after the dream last night. :) 12:11
moritz_ btw there's a perl6 vim syntax file at github.com/petdance/vim-perl
it's not perfect, but works more often than not
dalek ecza: d740787 | sorear++ | rxwork.pl:
Bring rxwork into the debugging phase
 12:12
ecza: 9f72975 | sorear++ | rxwork.pl:
rxwork: Fix some iterator issues
takadonet sorear++ :) 12:13
colomon rakudo: say "10" before "100", "10" < "100" 12:15
p6eval rakudo cf3da0: OUTPUT«11␤»
colomon rakudo: say "010" before "100", "010" < "100"
p6eval rakudo cf3da0: OUTPUT«11␤»
colomon rakudo: say "20" before "100", "20" < "100" 12:16
p6eval rakudo cf3da0: OUTPUT«01␤»
hejki rakudo: say 10 before 100 12:17
p6eval rakudo cf3da0: OUTPUT«1␤»
hejki rakudo: say 20 before 100
p6eval rakudo cf3da0: OUTPUT«1␤»
hejki are they Str() for a reason? :)
in your tests i mean
12:18 masak joined
masak hi #perl6! 12:19
moritz_ oh hai masak
looking at the changelog, it's going to be an awesome release.
masak \o/ 12:20
cognominal std: say ,, 12:21
p6eval std 31772: OUTPUT«===SORRY!===␤Preceding context expects a term, but found infix , instead at /tmp/F3hluJKO0T line 1:␤------> say ⏏,,␤Parse failed␤FAILED 00:03 116m␤»
cognominal rakudo: say ,,
moritz_ delegation, binding, Buf and basic binary IO, closures, autoviv, $*ARGFILES, backtracking into captures,
p6eval rakudo cf3da0: OUTPUT«===SORRY!===␤Confused at line 22, near "say ,,"␤»
jnthn o/ folks
phenny jnthn: 03:47Z <pmichaud> tell jnthn Is there a way from actions to see if an outer scope defines a method (regex, actually) of a given name? See Pm-21 in misc/pm.txt .
jnthn: 03:48Z <pmichaud> tell jnthn it doesn't appear that methods are entered in a PAST::Block's symbol table.
masak rakudo: say 3i ~~ (1i .. 5i)
p6eval rakudo cf3da0: OUTPUT«1␤» 12:22
12:22 bluescreen joined
masak lolitsjnthn! 12:22
moritz_ and then small things like Mu.perl
masak rakudo: say (2 + 3i) ~~ (1i .. 5i)
p6eval rakudo cf3da0: OUTPUT«0␤»
masak becomes curious
12:22 bluescreen is now known as Guest62932
moritz_ rakudo: say 1i.succ 12:22
p6eval rakudo cf3da0: OUTPUT«1 + 1i␤»
moritz_ rakudo: say 1i.succ cmp 5i
p6eval rakudo cf3da0: OUTPUT«1␤» 12:23
masak rakudo: say (1 + 1i) ~~ (0 + 0i .. 2 + 2i)
jnthn phenny: tell pmichaud no they shouldn't be unless marked "my regex" or "our regex" in which case they then have an entry. And that's the interesting case for the lexical regex things afaik? Can check...catch me when you're awake, I should be around plenty today anyway :-)
phenny jnthn: I'll pass that on when pmichaud is around.
moritz_ masak: any more questions?
p6eval rakudo cf3da0: OUTPUT«1␤»
masak moritz_: yes. does complex range smartmatching check whether a complex number is on a given line segment? :)
because all the examples I've tried so far suggests that.
moritz_ it does not.
masak s/suggests/suggest/ 12:24
dalek kudo: 3370f07 | moritz++ | docs/ChangeLog:
update ChnageLog
sorear I have the regex engine working 12:26
it takes about .2 seconds to match "xbc" ~~ /ab*c/ though 12:27
jnthn phenny: tell timbunce I'll get to it shortly...just trying to get to a lot of things at the moment
phenny jnthn: I'll pass that on when timbunce is around.
sorear now I just need to add regex syntax and fold rxwork back into setting
sorear -> nap
colomon masak: actually, it might, but only accidentally 12:29
or rather, it will test for an entire region.
masak jnthn: src/glue/enum.pir, line 80. here's where I need to install an instance of the mixed-in value, as we discussed yesterday. unfortunately, that part is PIR. is it still feasible, you think?
colomon: aha. line would be slightly cooler :) 12:30
rakudo: say (1 + 1.5i) ~~ (0 + 0i .. 2 + 2i)
p6eval rakudo cf3da0: OUTPUT«1␤»
dalek ecza: 1dc23fa | sorear++ | setting:
Fix another precendece fail
jnthn masak: You can call '&infix:<does>' from pir
ecza: 534e6b4 | sorear++ | rxwork.pl:
Fix remaining bugs in regex engine
masak yes, the rectangle hypothesis now is the prevailing one.
colomon Range just calls before and after (or maybe cmp) 12:31
jnthn masak: Though the whole thing would be a bunch nicer in Perl 6.
masak jnthn: can I still write the mixin role in Perl 6 and have the PIR see it?
jnthn masak: I guess you it'd need to then live in a namespace somewhere
masak yes. that's the part that I have trouble visualising right now. 12:32
jnthn masak: What's the reason we have any of it in PIR?
masak: Or maybe we could switch to inline PIR?
masak jnthn: IIRC, I tried to do it in Perl 6, failed miserably, and then you did it in PIR.
jnthn And have most of it in Perl 6?
masak jnthn: how it can eventually look might be a post-* concern. 12:33
I just want this particular feature to go into the release :)
jnthn Yeah, I was doing a quick hack to make something work rather than making something it's a good idea to build on...
masak I don't mind building up a slight further tech debt here just to get the feature in.
I'll assume stewardship over the code afterwards if that helps :) 12:34
moritz_ looks like it could be ported to nqp pretty easily
jnthn oh, the way I ended up doing it coulda been written in Perl 6 in the end anyway 12:35
(mostly because this code hasn't got any chance of hanlding lexical enums, for exaple)
12:35 felliott joined
jnthn moritz_: NQP helps little if you want to write $value does role { method key { $name } } and such though. :-) 12:35
colomon rakudo: say ("1", "3", "5" ... 100).perl 12:36
p6eval rakudo cf3da0: OUTPUT«("1", "3", "5")␤»
colomon :\
jnthn masak: Anyway, it probably could be turned into Perl 6 and it'd make your life easier on the roles stuff.
masak jnthn: I could try turning it into Perl 6. it didn't end well last time, though :) 12:37
12:38 skids joined
jnthn I think last time we tried passing a Parrot namespace PMC in 12:39
masak yes, that was it.
maybe that was the big mistake.
by the way, I expect to run out of Internet connection sometime this evening. will be mostly offline tomorrow while travelling home. back Thursday, hopefully with better enumerations :) 12:40
moritz_ is an EnumMap read-only?
jnthn moritz_: Yes. 12:41
masak: \o/
12:42 exodist left
masak rakudo: my $e = EnumMap.new; $e<foo> = "bar"; say $e.perl 12:42
12:42 exodist joined
p6eval rakudo cf3da0: OUTPUT«Cannot modify readonly value␤ in '&infix:<=>' at line 1␤ in main program body at line 22:/tmp/EoUoWpU_sc␤» 12:43
moritz_ jnthn: how do I specify a package in a pasttype('call') ?
like Enum::my_helper
:name('my_helper'), :package('Enum')?
jnthn moritz_: You don't 12:45
moritz_: Don't specify a name on the PAST::Op node
moritz_: just set pasttype to call
moritz_: And then make the first child of it a PAST::Var node that looks up the thing to call 12:46
moritz_ ah 12:47
rakudo: enum A <b c>; say b 12:48
p6eval rakudo cf3da0: OUTPUT«0␤»
tadzik hmm. Looking at rakudo/README on github, Rakudo is also in CRUX ports 12:55
dolmen_ rakudo: enum A <b c>; say b.^methods().join(',') 12:58
p6eval rakudo cf3da0: ( no output )
dolmen_ rakudo: say "5".succ 13:02
rakudo: say ("5".."10")
p6eval rakudo cf3da0: ( no output ) 13:03
[Coke] rakudo: 'whatwhat'.say
p6eval rakudo cf3da0: OUTPUT«whatwhat␤» 13:04
[Coke] juuuust checking.
masak could someone sanity-check my thinking here: gist.github.com/482925 introduces a lot of good things, and I like what I see -- expect .day-name and .month-name, which I think belong in some l10n-y non-core module. 13:06
is this too harsh? should we allow English names in the core module?
it does seem like some formatting letters use those methods, so that's certainly an argument for keeping them... 13:07
mathw if we don't have those, we can't have anything else that renders dates unless it can farm out to the system locale 13:08
I would say
for consistency :)
masak that sounds like a "keep 'em" vote to me. 13:09
moritz_ keep 'em. Worry about l10n later
masak thing is, until now I've been advocating a "numbers yes, names no" line for Temporal.
but this patch actually looks farily solid. Kodi++
I'm also displeased about publishing the name "Calendarical". eww! but everyone knows that by now, and it's still a net win, I think. 13:10
[Coke] masak: if it's any help, I've never heard of it.
dolmen_ Calendarical is a calendar in iCal format? 13:11
masak [Coke]: I think naming something "Calendarical" is a symptom of the general abstract-y-ness that has been haunting S26/Temporal from the start.
[Coke]: there, now you've heard it. :)
dolmen_: if only. it means something like "parent thingy of Date and DateTime". 13:12
moritz_ Dateish!
masak sorry, *Calendrical
jnthn Timey!
masak my point still stands.
jnthn Wheny
dolmen_ was ironic
masak dolmen_: yes, I suspected that. :)
dolmen_: I just tend to answer even jokey questions seriously. 13:13
jnthn It's how Swedes do humor. :P
masak I don't think Calendrical/Wheny/Dateish should be visible. we should be careful about polluting the namespace.
jnthn: yes. it is.
jnthn :-)
dolmen_ masak: Good behavior when there is ambiguity
masak dolmen_: it has been known to lead some conversations into some very interesting tracks. 13:14
dolmen_ doesn't have access to gist.github.com: dns_server_failure
mathw namespace polition is BAD 13:16
pollution
my spelling is also BAD
dolmen_ masak: there is a major problem in the Temporal spec: DateTime is mutable 13:17
moritz_ why is that a major problem? 13:18
jnthn masak: If your internets will be FAIL before #phasers, do preport.
dolmen_ moritz_: this is a problem in the Java API where java.util.Date is mutable 13:19
moritz_ dolmen_: can you try to explain it for somebody who doesn't know the Java API?
dolmen_ moritz_: all APIs use Date objects, but most of the time you just want to expose readonly object
moritz_: for example a database query should return readonly objects 13:20
masak dolmen_: please nopaste a spec patch where you make DateTime immutable.
13:20 macroron joined
masak dolmen_: oh, and a Rakudo patch :) 13:20
moritz_ dolmen_: the problem with immutable DateTime objects is that they make some things more complicated, like setting a timezone 13:21
Leonidas i am thinking about writing something in Perl 6 and for that I'd like to use a template engine. are there any good template engines that work with perl 6?
moritz_ also note that most database queries would probably return an Instant, not a DateTime object
dolmen_ moritz_: I think that a DateTime object must be readonly and a DateTimeCalc object (java.util.Calendar in the Java world) must be used for computations
moritz_ dolmen_: that sounds more like a Java solution than like Perl solution 13:22
masak: your turn (re template) :-)
baest dolmen_: couldn't it just be the reverse? DateTime is readonly and DateTimeImmutable could be returned by databases if needed
isn't readonly 13:23
dolmen_ moritz_: this just a "don't repeat a bad design that will be kept for years"
masak moritz_: sorry, "template"? ENOCONTEXT
Leonidas I found HTML::Template which has been ported to p6 by masak, but actually I think the syntax is ugly ;)
masak ah.
moritz_ what's wrong with returning an Instant (which is a more light-weight anyway, and immutable) and turning it into a DateTime when more sugar is needed?
masak Leonidas: yes, I agree. it's ugly.
Leonidas: don't use that one. 13:24
moritz_ or does that lose timezone information?
Leonidas masak: what other alternatives exist?
moritz_ Leonidas: I don't know of any usable Perl 6 template modules beyond that.
Leonidas would go for something TAL or Template Toolkit-like :)
masak Leonidas: this one is my best bet so far: github.com/masak/web/blob/master/dr...xample.xml
Leonidas: haven't worked on it for over half a year, so I don't remember its status. 13:25
Leonidas: running the tests will tell you.
Leonidas masak: TAL-like. Yeah, looks way better. Thanks.
Pugs had the ability to embed Perl 5, can rakudo do the same?
masak Leonidas: there's also this one: github.com/masak/web/blob/master/t/...1-basics.t
Leonidas: but then we're back on "ugly", I think. :)
Ratel is ugly-but-minimal, Hitomi is pretty-and-flexible. 13:26
baest masak: is this a "you can have working or ugly, pick one" situation?
Leonidas masak: Hitomi looks better, I don't care that much about minimal at the moment %)
masak baest: what kind of choice is that? :P
13:27 tadzik left
baest masak: heh, I meant working or pretty 13:27
masak Leonidas: let me know if there's some feature missing in Hitomi, and I might get the energy to revisit that code.
Leonidas For the record, I came here because masak's 10 years of Perl 6 article :D
masak \o/
au|afk \\o//
13:28 au|afk is now known as au|irc
[Coke] \ö/ 13:28
Leonidas masak: thanks, will do.
masak the article seems to have made Japanese Slashdot: slashdot.jp/submission/39299/Perl-610
PerlJam good morning all
masak by the looks of it, that's like normal Slashdot, but without all the annoying comments :P
PerlJam: bonan matenon!
dolmen_ moritz_: Instant is good for storage. But having the DateTime methods sugar is important 13:30
13:30 rgrau_ joined 13:33 colomon left
masak std: say "OH { my $x } HAI"; say $x 13:35
p6eval std 31772: OUTPUT«===SORRY!===␤Variable $x is not predeclared at /tmp/vd63FOl5fD line 1:␤------> say "OH { my $x } HAI"; say $x⏏<EOL>␤Check failed␤FAILED 00:01 116m␤»
masak std: / :my $x /; say $x
p6eval std 31772: OUTPUT«===SORRY!===␤Bogus term at /tmp/YZFxZG2eeA line 1:␤------> / :my $x /⏏; say $x␤Parse failed␤FAILED 00:01 119m␤»
masak std: m/ :my $x /; say $x
p6eval std 31772: OUTPUT«===SORRY!===␤Bogus term at /tmp/JMDvjwXmnM line 1:␤------> m/ :my $x /⏏; say $x␤Parse failed␤FAILED 00:01 119m␤»
masak oh. 13:36
std: / :my $x; /; say $x
p6eval std 31772: OUTPUT«ok 00:01 118m␤» 13:37
masak fwiw, I agree with sorear that this is wrong.
13:37 felliott left
PerlJam The regex should delimit a lexical scope for $x ? 13:38
13:38 timbunce joined
masak yes. 13:38
PerlJam yeah, that seems "intuitive" 13:39
masak that's my Least Surprise, at least.
moritz_ if not, you could ask if /:i a / && / b/ matches the b case-insensitive
jnthn rakudo: / :my $x; /; say $x
p6eval rakudo 3370f0: OUTPUT«Any()␤»
moritz_ if it doesn't introduce a scope, why should the :i be confined to just one regex?
jnthn Oh well, at least we're consistent with STD.
masak jnthn: wow, we support that? cool! 13:40
mathw I just had an awesome thought
jnthn masak: Yeah
mathw "That can wait until after R*"
mathw gets very excited
masak mathw: :P
jnthn masak: Thing is though that { ... } defines a scope is a very clear rule.
masak: Maybe we don't want to break that with a special case to remember.
mathw but regexes are a different language
moritz_ jnthn: so you think that :i should leak into the second regex in the same scope? 13:41
what about rx/.../ vs. rx{...} ?
jnthn moritz_: No
moritz_ but why, if it doesn't introduce a new scope?
jnthn moritz_: But variables have meaning in both the regex and in Perl 6.
moritz_ yes, and :i has a meaning both in the first and in the second regex :-) 13:42
mathw :i is more like a function call argument maybe
jnthn rakudo: rx/ :my $x /; say $x
PerlJam why should the lexicalness of :i have any relation to the lexicalness of variables?
p6eval rakudo 3370f0: OUTPUT«===SORRY!===␤Confused at line 22, near "rx/ :my $x"␤»
jnthn rakudo: rx/ :my $x; /; say $x
moritz_ PerlJam: because both are tied to scopes
p6eval rakudo 3370f0: OUTPUT«Any()␤»
jnthn rakudo: rx{ :my $x; }; say $x
p6eval rakudo 3370f0: OUTPUT«Any()␤»
jnthn heh
Well
moritz_ std: rx{ :my $x; }; say $x 13:43
p6eval std 31772: OUTPUT«ok 00:01 118m␤»
13:43 tadzik joined
jnthn That certainly breaks the block rule :-) 13:43
rakudo: regex { :my $x; }; say $x
p6eval rakudo 3370f0: OUTPUT«===SORRY!===␤Symbol '$x' not predeclared in <anonymous> (/tmp/83b6jjp1jm:22)␤»
jnthn Well, I guess in the rx case it's a quoting construct
Not actually a block
masak rakudo: / :my $x = 42; /; say $x
p6eval rakudo 3370f0: OUTPUT«Any()␤»
tadzik what's the difference between rule and token, in grammars? 13:44
masak rakudo: "foo" ~~ / :my $x = 42; /; say $x
p6eval rakudo 3370f0: OUTPUT«42␤»
PerlJam tadzik: rules have :sigspace turned on.
masak \o/
tadzik they're whitespace-sensitive?
moritz_ right 13:45
13:46 Su-Shee joined
cognominal rakudo: eval(("a","b").Seq.perl ).WHAT 13:47
p6eval rakudo 3370f0: ( no output ) 13:48
cognominal rakudo: say eval(("a","b").Seq.perl ).WHAT
dolmen_ tadzik: rules ignore whitespace
p6eval rakudo 3370f0: OUTPUT«Parcel()␤»
moritz_ dolmen_: no. tokens ingore whitespace
PerlJam dolmen_: unless you mean "rules" in a generic sense, but we call those regex. 13:49
cognominal doing .perl on a Seq returns a Parcel :(
mathw shouldn't .perl return a Str? 13:50
although eval() would be the one returning the Parcel there
moritz_ cognominal: it's a bug, but a small one. It's rarely necessary to reproduce the exact list type, as they can all be coerced very easily (and often automatically) 13:51
rakudo: my $x = (1, 2, 3).Seq; say $x.WHAT
p6eval rakudo 3370f0: OUTPUT«Seq()␤»
cognominal mathw: sorry for the imprecision.
tadzik could someone help me with grammars? I'm trying to write one to parse the config files: wklej.org/id/366783/ 13:52
mathw cognominal: don't worry about it, I'm just waiting for the doughnut fairy
13:52 drbean left
cognominal it tripped a test where I checked types and equality 13:52
moritz_ tadzik: you don't match \n anywhere
PerlJam tadzik: you don't have any whitespace eating regex
moritz_ right, change 'parameter' to be a rule, not a token 13:53
tadzik hmm, something like <emptyline>?
cognominal What is the composer for a Seq?
moritz_ @list.Seq 13:54
cognominal ...if any
dolmen_ moritz: S05 says: A "rule" is really short for: regex :ratchet :sigspace { ... }
moritz: S05 says: "rule" is a keyword
moritz_ dolmen_: yes. And it doesn't ingore whitespace. 13:55
13:55 paul1234 left
PerlJam dolmen_: :sigspace means whitespace within the regex implicitly match whitespace in your string. 13:55
dolmen_ moritz_: oh, yes... 13:56
tadzik hmm, so why should my parameter be rule, not token? There are whitespaces inside, whereas the line should look like 'key=value'
dolmen_ doesn't follow enough the sysposis changes... Reading planetsix is not enough 13:57
moritz_ tadzik: there's nothing in your regex that eats newlines
tadzik $$ in paramater is an end of line, no?
PerlJam tadzik: so that you can match the \n at the beginning of your string (for one thing)
13:57 masonkramer joined
timbunce jnthn: what's the status of zavolaj at the moment? 13:57
phenny timbunce: 12:27Z <jnthn> tell timbunce I'll get to it shortly...just trying to get to a lot of things at the moment
PerlJam tadzik: $$ is zero-width
moritz_ tadzik: yes, but only the end-of-line, not the newline
tadzik alright, I added token emptyline { ^^ \s* $$ } 13:58
timbunce jnthn: understood. thanks
13:58 masonkramer left
masak is there a list of *features* that'll go into Rakudo Star? www.reddit.com/r/programming/commen...al/c0unjs4 is asking. 13:58
jnthn timbunce: I suspect it's mostly going to be a case of hunting down how it broke. 13:59
masak: That'd be a long list ;-)
PerlJam tadzik: so ... you want "foo=bar\nads=7" to match <parameter><emptyline><parameter> ?
timbunce bisect?
13:59 ashleydev joined
tadzik PerlJam: no, parameter has to have \ 13:59
… to have \n at the end 14:00
doesn't it? It ends with $$
jnthn timbunce: Once I dig in a can probably debug it pretty quick anyway.
timbunce I'm currently using an older perl6 to I can make progress on DBDI. They'll come a point where that'll become more difficult.
moritz_ tadzik: token key { [^\[] (\w+) 14:01
tadzik: you seem to be using perl 5 character classes
tadzik maybe I'm still making mistakes, I'm not too fluent with Perl6 regexes yet 14:02
that's why I'm trying to write something to learn my way around
moritz_ [...] is just a grouping in Perl 6 regexes
tadzik they're now <[]>, yes?
moritz_ right
tadzik so it should be <[^\[]>?
PerlJam tadzik: $$ is zero-width, it tells you when you're at the end of a line, but it won't consume the newline for matching purposes
moritz_ and <-[...]> for negated char classes
masak I'll do a spectest run, apply Kodi++'s patch, and then do another spectest run.
timbunce what arg values does the perl6 --trace option take? (They're not in the README or the --help)
tadzik oh, ok 14:03
then it should be token parameter { ^^ <key> '=' <value> \\n $$ }, right?
moritz_ right
tadzik token not rule, so I can keep the whitespace inside, yep?
14:03 uniejo left
moritz_ right 14:03
also your string starts with a newline
so you have to think about how to parse that 14:04
tadzik emptyline will match it
wklej.org/id/366787/
moritz_ good
tadzik updated version
still, either it's wrong or I don't know how to extract anything
14:04 ashleydev left
moritz_ if inifile.parse($string) { .say for $<parameter>.flat } 14:05
tadzik: maybe read the chapters on regexes and grammars in the book?
tadzik It look a bit complicated to me at the first glance, but I guess I should, yep :)
timbunce The trace flags also aren't noted in rakudo/docs/running.pod 14:06
moritz_ I guess because they don't produce very interesting output :/
masak anyone see anything not up-to-date on rakudo.org/status ?
I just provided that as a list of things that work for the Reddit person. 14:07
jnthn timbunce: It's near the top of my hit list.
timbunce moritz_: how can I track down Null PMC access in type() and similar errors?
the lack of a trace or stack dump or similar (and the lack of a clue on my part) means I hit a wall whenever I get that kind of error. 14:08
moritz_ timbunce: I rename the backtrace printer, so that I get the parrot-level traces, which then show at least a bit of where the error occured - not a good technique overall
timbunce moritz_: sounds good, but what the heck does "rename the backtrace printer" mean? 14:09
14:09 patcoll joined, gbacon joined
dolmen_ rakudo: qp{/etc/passwd}.say 14:09
p6eval rakudo 3370f0: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<{ }>, couldn't find final '}' at line 22␤» 14:10
moritz_ timbunce: src/Perl6/Compiler.pir, method backtrace
timbunce: rename to something unrelated (like, add a _)
then recompile
dolmen_ rakudo: qp{/etc/passwd}.WHAT
p6eval rakudo 3370f0: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<{ }>, couldn't find final '}' at line 22␤»
14:10 plobsing joined
jnthn qp NYI 14:10
[Coke] (that method overrides the normal parrot backtrace printer.)
14:12 paul1234 joined
moritz_ rakudo: my @a = <a b c>; say @a[1|2] 14:13
p6eval rakudo 3370f0: OUTPUT«any("b", "c")␤»
moritz_ masak: the autothread over array/hash indexes can be removed from the non-working list 14:14
dolmen_ Are qd{} and qt{} used in the spec?
masak sees if he has privs to remove it
PerlJam dolmen_: what are qd and qt ?
14:15 drbean joined
dolmen_ imagines: my $today = qd{2010-07-20}; 14:15
14:15 Mowah left
PerlJam oh. no. 14:15
dolmen_: See perlcabal.org/syn/S32/Temporal.html though 14:16
14:17 daxim left
timbunce I know there's huge effort going into getting language features implemented for R* but as a newbie I'd really appreciate better diagnostic tools. At the moment perl6 seems to slap me in the face regularly with "some error occurred but I'm not going to help you find out what". Saying SORRY! doesn't soften the blows much. If I wasn't so motivated I think I'd be quickly turned off exploring perl6. 14:17
14:18 Mowah joined
PerlJam timbunce: I know the feeling. That's exactly the same feeling I had when I first played with ruby over 10 years ago. I didn't pick it up again until about 4 years ago and the situtation was much improved. 14:18
jnthn timbunce: Many of the patches I've put in of late are aimed at improving errors, fwiw. 14:19
cognominal I guess some sort of pragma would be nice to get the parrot stack. Does not need to be the default though.
14:19 Lorn joined
moritz_ timbunce: if you can reproduce any of these bugs with a little piece of code, please report them, so that we can improve the situation 14:19
14:19 au|irc is now known as au|zzz
timbunce moritz_: disabling backtrace helped. I now see "current instr.: 'perl6;DBDI_pg;!class_init_436' pc 5018 (EVAL_5:1918 (DBDI_pg.pm6:107) called from Sub '' pc 874 (EVAL_5:271) (DBDI_pg.pm6:1)" etc. Very handy. 14:19
takadonet timbunce: the more you describe yours issues, better chance they will get fixed sooner 14:20
PerlJam timbunce: aside from a real debugger, what sort of better diagnostic tools would you like to have? (I fear I've gotten far too used to hacking in says everywhere)
moritz_ timbunce: and btw, it has already improved a lot. Some weeks ago you'd never get proper line numbers in error messages :-) 14:21
cognominal To answer takdonet in the place of timbunce, I think sometimes I could golf the errors faster if I get back the rakudo trace
s/rakudo/parrot/
timbunce cognominal: exactly 14:22
cognominal I mean the stack trace.
moritz_ cognominal: I'd welcome a patch for such a pragma/switch/env variable/whatever
14:22 pmurias joined
moritz_ rakudo: say pir::null 14:23
p6eval rakudo 3370f0: OUTPUT«error:imcc:syntax error, unexpected '\n'␤ in file 'EVAL_8' line 78␤===SORRY!===␤syntax error ... somewhere␤»
moritz_ rakudo: say pir::null__p
cognominal like a 'use trace :stack<parrot>' ?
p6eval rakudo 3370f0: OUTPUT«Null PMC access in type()␤ in main program body at line 22:/tmp/0UAKy4HxeB␤»
moritz_ that even produces a correct line number 14:24
cognominal that would leave other options open?
timbunce It doesn't help that I'm stuck on an older version (currently June 25th) till zavolaj's working again.
moritz_ cognominal: I don't care much for the syntax now
timbunce disabling the backtrace method has helped, thanks moritz_++
dolmen_ masak: my patch for S32 would be to remove the "Set methods" section entirely. A new object should be always created. truncate() would return a new object like succ() does. 14:25
masak: $dt.set(:year(2030)) would be instead written DataTime.new($dt, :year(2030)) 14:30
s/DataTime/DateTime/
tadzik wklej.org/id/366801/ -- straight from the book, but Rakudo complains about it: Useless declaration of has-scoped token in a module; add our or my to install it in the lexpad or namespace. What has changed, how should it look now? 14:31
masak dolmen_: that feels insufficiently flexible to me; always "changing" attributes through the constructor. 14:32
PerlJam tadzik: straight from an old version of the book perhaps 14:33
tadzik probably. 2010.06
nothing newer prebuilt as I see :\ 14:34
cognominal afk&
PerlJam tadzik: stick "my" in front of the token and regex declarations.
tadzik: though I'm not sure rakudo does lexical lookup yet for m/ <dup> / to work 14:35
tadzik now it says Method 'dup' not found for invocant of class 'Cursor'
TiMBuS <&dup> will, work but won't capture
dolmen_ masak: most of the case of use of a DateTime object are readonly: you get the object from a data source (system time, database data, file attribute) and you do not change it. Only transform it (Str)
tadzik yeah, that's probably what You just said
dolmen_ masak: the only case where I see modifying an object would be useful would be as a loop iterator 14:36
masak dolmen_: prepare at least the spec patch, and optionally the Rakudo patch, and I will apply it/them.
TiMBuS (crazy guess) <dup=&dup> might work? 14:37
14:37 macroron left
PerlJam TiMBuS: it does I do believe. 14:37
14:37 Mowah left
dolmen_ masak: what is the repo? 14:38
14:38 plobsing left
masak dolmen_: svn.pugscode.org/pugs/docs/Perl6/Spec/ 14:38
dolmen_: with your change, we can get rid of set-year et al, which are currently a workaround. 14:40
afk # walk
[Coke] tadzik: this works: 14:48
nopaste.snit.ch/22146
the foo=&foo is lame, though. presumably there's a better way.
PerlJam wait for pmichaud's regelx branch to merge 14:49
er, regelx
14:49 timbunce left
PerlJam er, reglex! 14:49
or use a grammar 14:50
tadzik I was about to try some more complex regexes before learning grammars ;) 14:51
14:54 tylercurtis joined
dolmen_ paste? 14:58
14:58 pmurias left
[particle] hrmm, there used to be a paste link in /topic 15:01
dolmen_ masak: nopaste.snit.ch/22147 15:02
15:02 timbunce joined 15:04 baest left 15:09 timbunce left 15:11 snarkyboojum joined 15:13 TiMBuS left
tadzik nopaste.snit.ch/22148 can someone take a look? 15:13
the line before the last prints name=brad as expected 15:14
15:14 patcoll left
[particle] next time, you can paste the output, too. for example, $ cat my-file.p6 ... $ rakudo my-file.p6 ... 15:17
tadzik yeah, my bad
[particle] try $config[0]<key>[0]
tadzik Any() 15:18
[Coke] tadzik: you want: $config[0]<keyval>[0]<key> 15:19
[particle] ok, right, two things.
[Coke] I find it very helpful to dump the match object here. lemme see if I can remembe rhow to do that from perl6. 15:20
[particle] if there's a ? or * or + after a regex, you'll always have an array generated, so you need to index it
PerlJam tadzik: btw, <keyval> is already capturing, you don't need () around it.
tadzik so named regexes are captured automagically? 15:21
PerlJam tadzik: also, <keyval> [ ';' <keyval> ]* can also be written now as <keyval> ** ';'
tadzik that's useful
[Coke] tadzik: "say $config[0].perl" might have been helpful to peek into what the structure of the resulting match was.
tadzik might, but it returns Null PMC access 15:23
Null PMC access in find_method('perl')
PerlJam tadzik: not for me. 15:24
15:24 Eevee joined
tadzik strange 15:24
[particle] token TOP { ^ <keyval> ** ';' $ }
tadzik: are you using rakudo from git-master-head, or a release? 15:25
tadzik git master head
15:25 TiMBuS joined
PerlJam huh, I would have guessed you were using a release 15:25
tadzik This is Rakudo Perl 6, version Kiev-280-g9005478 built on parrot r48110 15:26
PerlJam tadzik: when did you build it?
ah.
tadzik I will paste the whole thing
szabgab rakudo: for <a 1 b 2 c> -> $a, $b { say "$b - $a" } 15:27
p6eval rakudo 3370f0: OUTPUT«1 - a␤2 - b␤ - c␤»
tadzik nopaste.snit.ch/22151
szabgab will this give an exception again any time soon? 15:28
PerlJam szabgab: btw, nice job on the screencasts!
szabgab thanks
which one is better the first or the second?
in terms of view
PerlJam I liked the second better. 15:29
[particle] why not $config.parse?
szabgab I think I'll need to enlarge the fonts a bit
[particle] er, $config.perl 15:30
PerlJam szabgab: have you seen railscasts?
szabgab nope, I should check them out
15:30 mmcleric joined
PerlJam railscasts.com 15:30
That guy uses some good styles, fonts, etc. Copy what you can and what's appropriate and you can't go wrong :) 15:31
oh! He also uses textmate ... it would be really cool to see some screencasts that use padre :)
15:34 salv0 left 15:35 salv0 joined
szabgab but for that I will need to run it from my own server, I guess 15:35
but otoh it clearly looks a lot better than what I have on youtube 15:36
15:43 Mowah joined, envi^home joined, jaldhar left 15:44 timbunce joined 15:45 timbunce_ joined, timbunce left, timbunce_ is now known as timbunce
moritz_ colomon++ # fun-with-series blog post 15:52
ingy morning 15:53
[Coke] ingy: hio 15:54
ingy I was thinking of writing a blog post for oscon/rakudo* about how I ported a ~1000 line p5 framework to p6, and it all worked 15:56
moritz_ do that!
takadonet ingy: can i see it? :)
ingy maybe use.perl.org?
takadonet: I need to write it
:)
I'll do it now 15:57
takadonet hehe
15:57 mmcleric left
PerlJam ingy: but ... perl 6 doesn't even exist yet! ;) 15:57
15:57 mmcleric joined
ingy I think it's an interesting story. A lot of crazy bandaids etc 15:57
dolmen_ ingy: the framework or the blog post? 15:58
ingy but it is workable
dolmen_: rakudo!
well I needed to do a few crazy things to make it work
including one pir call from jnthn++
dolmen_ ingy: you're writing rakudo?
ingy gah 15:59
ok, I'll just write the post
PerlJam ingy++
dolmen_ and the framework is done?
ingy dolmen_: the port is 100% complete
but you can read about it in my upcoming blog post 16:00
:P
where did masak post previews of his recent use.perl.org post? 16:01
dolmen_ Is it github.com/ingydotnet/testml-pm6 ?
ingy dolmen_: correct
dolmen_ nopaste.snit.ch/
ingy ah 16:02
jnthn ingy: gist iirc 16:03
16:04 felliott joined 16:05 cdarroch joined, cdarroch left, cdarroch joined
tadzik wasn't it gist? 16:08
ingy use.perl.org/~ingy == Brian Ingerson
it's been a while
I wonder who can change that 16:09
16:10 M_o_C joined
PerlJam ingy: can't you? 16:11
tadzik who is caring about november-wiki.org/? 16:12
ingy PerlJam: doesn't seem so
PerlJam ingy: I would contact Pudge. But I haven't seen him since I stopped hanging out on irc.perl.org:#perl 16:14
dolmen_ ingy: are you aware of blogs.perl.org? 16:15
ingy dolmen_: mst's? 16:16
pmichaud good morning, #perl6
phenny pmichaud: 12:23Z <jnthn> tell pmichaud no they shouldn't be unless marked "my regex" or "our regex" in which case they then have an entry. And that's the interesting case for the lexical regex things afaik? Can check...catch me when you're awake, I should be around plenty today anyway :-)
takadonet pmichaud: morning
ingy that sounds useful
16:17 justatheory joined
jnthn morning, pmichaud 16:17
tadzik hmm, is <-[;\n]> a valid character class, capturing anything but semicolons and newlines?
pmichaud reads backscroll
ingy hi pmichaud
tadzik hello pmichaud
pmichaud tadzik: it's valid, but rakudo doesn't recognize \-escapes in character lists yet.
jnthn tadzik: Believe so.
oh heh 16:18
tadzik oh, ok
so how can I achieve this?
jnthn I think I copy-pasted something like that from STD into Grammar.pm yesterday :-)
pmichaud tadzik: looking
PerlJam tadzik: <subrule> only captures if the character after the < is alphabetic. (just in case "capture" was being conflated :-) 16:20
pmichaud tadzik: sadly, I don't have a quick answer. I might be able to get something in place a bit later. 16:22
tadzik ok, nothing urgent
I'm trying to write config file parser, ini files it is 16:23
pmichaud you could do something like <![;]> \N
which means any non-newline that is not a semi.
tadzik hmm, <![;]> is the same as <-[;]>?
pmichaud no 16:24
<![;]> is zero-width
basically it says "don't match a semicolon at this point"
jnthn It's like <!before ';'> iirc
But more efficient.
tadzik hmm. So [ <![;]> \N ]+ is a positive number of chars not being either semicolon or newline? 16:25
pmichaud correct.
PerlJam pmichaud: you're at oscon now? 16:26
tadzik thanks pmichaud
pmichaud PerlJam: well, I'm in PDX now. :-) 16:27
haven't made it over to the venue yet :)
might just hang out here for a while anyway -- fewer distractions.
PerlJam I bet
16:27 dolmen_ left
pmichaud ...and the wireless may be more reliable anyway. 16:27
last time I was at oscon in PDX my laptop simply wouldn't connect to the OSCON wireless at all
(different laptop then, though)
PerlJam pmichaud: does the reglex branch work correctly? 16:29
pmichaud depends on your definition of "correctly", I suspect.
I suspect I can come up with some cases where it fails.
PerlJam Well, I tried it on one test case and it worked, which is why I asked :)
pmichaud PerlJam: I'm more concerned about interactions between regex names and the setting. 16:30
It should work fine for the simple case of declaring a my-scoped regex.
jnthn pmichaud: Worried about <print> ? 16:31
pmichaud jnthn: well, also worried about things like grammar G { token classify { ... } }
16:31 new_user joined
pmichaud if G then tries to use <classify>, and sees the &classify entry, then it won't use the local one. 16:31
new_user rakudo: for 1..5 { say $_ } 16:32
jnthn Hm
p6eval rakudo 3370f0: OUTPUT«1␤2␤3␤4␤5␤»
16:32 new_user left
PerlJam that's definitely broken then by my definition of "works" :) 16:32
pmichaud any reason why methods can't be in the PAST::Block symtable?
jnthn They're not to do with the block really.
timbunce jnthn: FYI adding "is export" to "sub PQntuples( OpaquePointer $res ) returns Int is native('libpq')" only works if the is export comes before the is native. 16:33
jnthn Better to look at @PACKAGE[0]'s method table.
16:33 phenny left
pmichaud the problem there is figuring out the relationship between @PACKAGE and @BLOCK, though 16:34
jnthn timbunce: It's not surprising taht traits can have ordering dependencies.
timbunce: That particular one surprises me though.
pmichaud: @PACKAGE[0].block should tell you the block associated with the package, if that helps?
pmichaud ...not really. 16:35
jnthn pmichaud: What are you trying to do?
pmichaud I'm having to scan @BLOCK, so I'd need to know when a particular block corresponds to a package
16:35 phenny joined
jnthn pmichaud: Ah...yeah, I don't think we have the relationship marked up in that direction at the moment. 16:35
pmichaud when encountering <foo>, I have to look to see if "foo" refers to a lexical &foo regex or should be left as a grammar method
jnthn Yes 16:36
pmichaud so, we outward-scan the existing symbol tables looking for .symbol("foo")<scope> eq 'lexical'
16:36 alester left
jnthn *nod* 16:36
pmichaud but I should probably stop looking if one of the scopes corresponds to a grammar that has a "method foo" 16:37
jnthn Hmm. That's not going to handle the inheritance case.
pmichaud right.
jnthn But yes, perhaps so.
pmichaud it doesn't handle the inheritance case, which is why I've very much not been a fan of having <foo> automatically find lexical regexes
because I don't see a good way to handle it at compile time. 16:38
jnthn Anyway, my knee-jerk feeling is that method entries in the block symbols table is wrong, for the same kinda reasons that :method being automatically a namespace entry in Parrot was.
So I'd prefer a solution that sets a block's associated package object so that can be queried.
pmichaud jnthn: I can see your point. (Not sure I agree with it, though -- block symbol tables simply say 'these are all of the symbols I'm aware of')) 16:39
anyway -- yeah, it doesn't handle the inheritance case. At the moment I'm still inclined to not merge to master until we get spec clarification.
jnthn I guess if you install it marked as :scope<has> or something it's not so bad.
But I still am not too keen.
pmichaud right, I was thinking it'd be :scope<has>, or even without a :scope at all :-) 16:40
just something to say "this word gets used here."
jnthn I'll think on it some more. I'm not quite sure I've hit upon my real discomfort about it yet. But of coruse, my real discomfort might be bogus too. :-)
One potential issue is that methods are unsigil'd. 16:41
So they may conflict with type names (also unsigil'd) in the .symbol
pmichaud Yes, I was surprised that lexical regexes were going into .symbol without the sigil.
ingy is it more properly Rakudo* or Rakudo *? (in prose)
jnthn pmichaud: We may want to tweak that. 16:42
pmichaud ingy: two words.
PerlJam ingy: I tend to write "Rakudo Star"
pmichaud so "Rakudo *". Or "Rakudo Star" if you want the prose to be clearer.
jnthn pmichaud: Since I'm quite sure we install them into the lexpad as &foo
pmichaud jnthn: yes, we do.
as I said, I was surprised. :)
jnthn Yeah, I think that's a mistake. 16:43
Shoudl be in the block as &foo
But in the meta-class as foo
pmichaud my regex xyz { xyz }; grammar A { regex xyz { XYZ }; }; grammar B is A { regex TOP { <xyz> } }; # which one does TOP see? 16:44
jnthn I guess if we do put methods in as &foo too, it does mean that our compile-time idea of the lexpad and our runtime one afe going to not match up.
*are
pmichaud I wouldn't plan to put methods in as &foo
(unless it's a "my method" of course)
jnthn pmichaud: Yeah, but then we get conflict with a typename foo too
pmichaud right 16:45
jnthn rakudo: role Bar[::foo] { method foo { foo } }; say Bar[Int].foo
p6eval rakudo 3370f0: OUTPUT«Int()␤»
pmichaud which bolsters your argument that it doesn't belong in the block's symbol table :)
16:46 felliott left
jnthn Yeah, I'm a bit worried what'd happen in the above case. 16:47
pmichaud This all reminds me of why we got rid of the "method fallbacks" some time ago. 16:48
jnthn Plus just a general sense of "it's hard enough to keep our runtime and compile time story in sync, let alone deciding they're going to be desync'd in some special cases" :-)
Anyway, that "what does TOP see" example, so far as I grok the spec, would lead to the lexical one being seen.
I can understand why that'd potentially be uncomfortable though. 16:49
PerlJam I would have said it should see B::xyz
pmichaud PerlJam: you mean the one inherited from A? 16:50
PerlJam yes
pmichaud then what about
16:50 supernovus joined
pmichaud grammar A { regex xyz { XYZ }; }; my regex xyz { xyz }; grammar B is A { regex TOP { <xyz> } }; # which one does TOP see? 16:50
still the one inherited from A?
PerlJam yes
pmichaud then what about
PerlJam I know what you're about to show me and I don't know :) 16:51
supernovus phenny: tell Kodi nice work on the Temporal stuff!
phenny supernovus: I'll pass that on when Kodi is around.
PerlJam (my answer is "I don't know")
pmichaud my regex alpha { xyz }; grammar B is Grammar { regex TOP { <alpha> } }; # what does TOP see?
PerlJam I would tend to lean towards the inherited one 16:52
pmichaud maybe we get to be okay if we say that only regexes not part of a grammar declaration participate in the lexical lookups
PerlJam pmichaud: that seems sane
jnthn pmichaud: Could work. 16:53
16:53 felliott joined
pmichaud i.e., a regex that is being installed into a grammar doesn't ever at outer lexicals 16:53
PerlJam if they really wanted the lexical one, they could have said <&alpha>, right?
pmichaud PerlJam: Yes.
and if they really wanted to override alpha, they could do it in the declaration of B
PerlJam right
so, seems really sane.
pmichaud (the difference being that <&alpha> is non-capturing, so it'd need to be <alpha=&alpha> or something like that) 16:54
PerlJam right
seems like a small-price, low-probability sort of thing to me.
moritz_ only 'my' or 'our' regexes in grammars would be picked up as lexicals anways 16:55
PerlJam (i.e. if they *really* wanted the lexical <alpha> and they wanted a named capture, we should encourage the use of grammars anyway)
pmichaud I still feel as though I'd like some definite clarity on this point, though. If at some point we disallow <xyz> from finding a lexical &xyz, we'll end up with a fair bit of code breakage. 16:56
16:56 mmcleric_ joined, mmcleric_ left, mmcleric left 16:57 mmcleric joined
PerlJam pmichaud: so ... we don't do that :) 16:57
pmichaud PerlJam: we may have to
jnthn pmichaud: It worked for the wrong reasons in alpha, and a lot of code depended/depends on it.
pmichaud jnthn: right
jnthn My sense is that people expect that it's going to start to work again. 16:58
To some degree.
pmichaud jnthn: unless we decide that it can't work
PerlJam jnthn: indeed
pmichaud jnthn: because by making it work we're conflating method dispatch with lexical dispatch
PerlJam pmichaud: we'll never know without an implementation that people can hack on :)
jnthn pmichaud: True, which we've gone to lengths to avoid elsewhere. 16:59
PerlJam hmm
pmichaud jnthn: and for good reason.
jnthn pmichaud: These kinds of reasons. :-)
moritz_ brunorc.wordpress.com/2010/07/20/pe...-near-you/
pmichaud the fact that <xyz> finds things lexically is actually left over from the days when we also were saying that xyz $y could fall back to $y.xyz (or perhaps it was the other way around, I forget)
moritz_ still, there are good use cases for it 17:00
for example importing rules from modules
pmichaud either way, we eliminated that particular fallback mechanism, but never updated regexes to match.
tadzik the best way to achieve required parameters in classes is 'has $.a = die "...";'?
moritz_ where you don't want full grammars
TimToady perhaps some way of explicitly importing
moritz_ I'd also be fine with a shorter syntax for explicitly calling lexical rules, and have them capture 17:01
17:01 xabbu42 left
PerlJam in classes, method dispatch always needs a dot, right? perhaps make the same requirement in grammars (and allow <.foo> to capture) 17:01
moritz_ having <&foo> capture, and <.&foo> not capture, for example
pmichaud PerlJam: ouch. that puts huffman all wrong.
17:01 xabbu42 joined
jnthn PerlJam: I almost suggested as logical, but it felt too wrong :-) 17:01
moritz_ but <foo=&foo> feels tedious
TimToady nothing that starts with punctuation captures right now 17:02
don't wanna break that
PerlJam moritz_: or, "if you want capturing, you want a grammar"
pmichaud agree, don't break that.
moritz_ PerlJam: no
jnthn tadzik: heh, that probably works actually... :-)
17:02 plainhao joined
jnthn I hadn't thought of it. :-) 17:02
17:02 dakkar left
tadzik jnthn: is there a better way? 17:02
moritz_ PerlJam: there's nothing wrong with having lexical rules - why force people away from them? 17:03
jnthn tadzik: Well, you could also write a new method with a required param.
pmichaud moritz_: what we're saying is that there may indeed be somethign wrong with lexical rules.
See my examples above.
tadzik and bless myself, like in Perl 5
?
pmichaud until you can resolve those, it's not the case that "there's nothing wrong with lexical rules."
jnthn Go bless yourself!
uh, I mean :-)
tadzik: Yeah
tadzik erm, "call 'bless' myself"
:)
moritz_ or just call self.Mu:::new(|%args) 17:04
jnthn Or that
Note that you can also do like
TimToady maybe the easiest way to fix it is by saying they have to put grammar {...} around everything :)
supernovus Any idea when Kodi's patch will be applied? I will need to modify my recently added %z in the loadable DateTime::strftime library, as it currently depends on the old timezone attribute format in DateTime.
moritz_ pmichaud: if there's a short syntax for calling lexical rules (that captures), there's nothing wrong with lexical regexes per so. Only with our current approach, maybe 17:05
tadzik is there something like sub BUILD in Moose?
jnthn method new(%args (:$required-name!, :$another!, *%)) { self.bless(*, |%args) }
tadzik: Yes, submethod BUILD
tylercurtis Also submethod BUILD.
17:05 stkowski joined
PerlJam pmichaud: how about this ... R* release is too close. Implement something hits the big targets and let the edge cases work themselves out by people using R* :-) 17:05
tadzik sounds like a prettier way for required params
jnthn tadzik: Yes, that'd also work
tadzik good. Will look more sane 17:06
17:06 thebird left
pmichaud PerlJam: there's a problem with "hit the big targets" 17:06
jnthn rakudo: class Foo { has $.x; submethod BUILD($!x) { } }; say Foo.new(x => 42).x;
p6eval rakudo 3370f0: OUTPUT«42␤»
17:06 xabbu42 left
jnthn rakudo: class Foo { has $.x; submethod BUILD($!x) { } }; say Foo.new().x; 17:06
p6eval rakudo 3370f0: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in 'Foo::BUILD' at line 22:/tmp/qAqWBWlP_L␤ in main program body at line 22:/tmp/qAqWBWlP_L␤»
jnthn \o/
pmichaud the way PGE worked was that it fell back to lexical lookup if a method didn't exist at runtime. i.e., it's a runtime check. That's expensive.
tadzik jnthn: that looks even more cryptic to me :P 17:07
pmichaud it may also be wrong, as the lexical lookup should perhaps take precedence over the method lookup.
PerlJam pmichaud: it also sound backwards
right
pmichaud right
TimToady perhaps lexical lookup works only when you *aren't* in a grammar
pmichaud so, if we take the reglex approach, that does a lexical lookup and falls back to method if the lexical doesn't exist
jnthn tadzik: The above works because named parameters get mapped to positional ones of matching names, and $!x in a signature binds to an attribute instead of a lexical :-)
pmichaud TimToady: we speculated something close to that above 17:08
jnthn tadzik: It's pretty logical and declarative when you stop to think about it. :-)
17:08 kalve joined
jnthn Well, somewhat. :-) 17:08
ash_ tadzik: once you know the perl6 syntax for stuff, it makes sense, mostly
tylercurtis rakudo: class Foo { has $.x; submethod BUILD($!x) { } }; say Foo.new(:x(5)).x;
pmichaud PerlJam: but the problem with the reglex approach is that basically every lexical subroutine blocks that name from being used as a regex in a grammar.
p6eval rakudo 3370f0: OUTPUT«5␤»
pmichaud PerlJam: including the ones in the setting.
tylercurtis rakudo: class Foo { has $.x; submethod BUILD($!x) { } }; say Foo.new(5).x; 17:09
p6eval rakudo 3370f0: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in 'Foo::BUILD' at line 22:/tmp/LftQm02JSa␤ in main program body at line 22:/tmp/LftQm02JSa␤»
jnthn tylercurtis: If you want that one to work, you need to write a "new" method.
tylercurtis jnthn: just checking. :)
jnthn :)
TimToady pmichaud: it could ignore anything that isn't declared as method
pmichaud so then it becomes really strange when you define a grammar with regex item { ... } and it doesn't work because <item> keeps finding &item
TimToady though, perhaps item is declared as a method, and exported 17:11
pmichaud TimToady: yes, that also requires a bit more looking-up than I can easily handle in R*
PerlJam Least Surprise is hard sometimes
17:13 pmurias joined
PerlJam I think I've mentioned this before, but how about regex declared outside of a grammar all belong to an implicit grammar? 17:13
17:14 christin1 left
PerlJam (and you always use method dispatch to lookup names) 17:14
TimToady or we require an anonymous grammar {...} around them
timbunce rakudo: say $a; my $a
p6eval rakudo 3370f0: OUTPUT«Any()␤»
timbunce shouldn't that complain that $a is not predeclared? 17:15
TimToady known bug
pmichaud timbunce: yes, it should.
for some reason the check for declared symbols got removed.
tadzik dying inside CATCH will die outside of the try block, right?
ah, whatever
pmichaud tadzik: right now dying inside CATCH will infinite loop in Rakudo.
(still having to work on that one) 17:16
tadzik what I really need is not having try block at all ;)
PerlJam TimToady: so ... is there a singleton anonymous grammar?
pmichaud PerlJam: can't be singleton. needs to be lexically scoped, I think.
17:17 christin1 joined
PerlJam Hmm. 17:17
pmichaud well, I'll phrase it this way
tadzik so, while I'm asking questions anyway: my $asd = INI.parse($string); How can I check if parsing was sucessful?
I know I can do 'if INI.parse(...)', but then I'll have to parse twice to assign it to variable
pmichaud in the normal method space, the closest we have to allowing a bare "foo" to correspond to a method call is via indirect object syntax.
PerlJam pmichaud: so each block would potentially contain a single anonymous grammar that perl knows to look at when someone does / <foo> /
pmichaud or put another way, in normal method space, we don't have a mechanism that doubles as both lexical and method dispatch. 17:18
TimToady alternately, you just have some way of saying 'mix in my current block's regexes'
moritz_ tadzik: if $asd { ... } 17:19
tadzik ah, it will return false in this context? Ok, thanks moritz_
PerlJam TimToady: augment :-)
TimToady not augment, mixin
they are *very* different
pmichaud trying to get <foo> to handle both cases seems to go down the same path we ultimately avoided with methods
TimToady unless the current grammar *is* the current scope 17:20
for instance, simply putting grammar Me is ThatGrammar {...} around the whole thing makes it work 17:21
but maybe there can be something with the same effect that is more lightweight
pmichaud TimToady: does that help with moritz's "import a regex" scenario?
how about this 17:22
perhaps we want a punctuation that means "explicitly capture"
I'll use # here
so:
<&foo> # noncapturing
<#&foo> capturing
TimToady <=&foo> 17:23
pmichaud or yeah, that.
moritz_ +1 to <=&foo>
pmichaud then one could also have <=$foo>
PerlJam wfm
pmichaud and <=@foo>
and the like
masak heh. Unicode *and* smilies. almost too on-topic for this channel: unicode.org/Public/6.0.0/charts/ver...U1F600.pdf 17:25
PerlJam so the first example in the regex chapter of the perl 6 book would become gist.github.com/483251 ?
ash_ .u 1F640 17:26
phenny ash_: Sorry, no results for '1F640 '.
ash_ .u DIZZY FACE
phenny ash_: Sorry, no results for 'DIZZY FACE'.
Su-Shee ash_: without the 0. 17:27
[Coke] .u 1F64
phenny U+1F64 GREEK SMALL LETTER OMEGA WITH PSILI AND OXIA (ὤ)
[Coke] Su-Shee: different character. 17:28
17:28 supernovus left
pmichaud masak: I'm a little surprised they don't have glyphs for the "faces pain scale" in that table. 17:28
masak we should suggest that to them. 17:29
ash_ "FACE WITH NO GOOD GESTURE"
lol
ash_ just realized that is a draft
pmichaud (I've seen that scale a lot in recent months, which is what makes me think of it :-) 17:30
tadzik nopaste.snit.ch/22156 -- could anyone take a look at this? The effect bothering me is at the very bottom
masak second spectest run, now with Kodi++'s patch applied. 17:31
moritz_ pmichaud: btw I've pushed a branch called subst_adverbs which implements s:g/// (but not much more) 17:32
pmichaud moritz_: I'll definitely take a look.
TimToady or <:&foo> to be more like pair syntax
moritz_ it introduces some interesting failure modes, ie ignores s:unknown_pair///
masak "DISAPPOINTED BUT RELIEVED FACE" -- nobody takes emoticons as seriously as the Unicode Consortium.
ash_ tadzik: you found a bug 17:33
pmichaud masak: right, that's why I was surprised faces pain scale wasn't there :)
moritz_ and s:g(1)/// # can't find identifier Capture (or so) - huh?
ash_ rakudo: grammar Foo::Bar { }; class Foo { };
p6eval rakudo 3370f0: ( no output )
TimToady or even, since : is reserved, leave off the <> to call :&foo
ash_ > grammar Foo::Bar { }; class Foo { };
Illegal redeclaration of symbol 'Foo::Bar'
> class Foo::Bar { }; class Foo { };
!class_init_11
tadzik oh, great :)
TimToady std: / :&foo /
p6eval std 31772: OUTPUT«===SORRY!===␤Unrecognized regex metacharacter (must be quoted to match literally) at /tmp/aRSVIiLKw9 line 1:␤------> / :⏏&foo /␤Can't call method "from" on unblessed reference at /home/p6eval/pugs/src/perl6/snap/STD.pm line 53586.␤FAILED 00:01
..1…
moritz_ ash_: please submit rakudobug
ash_ it seems to be an issue with autovivifiying and grammars 17:34
pmichaud bare :&foo looks almost too close to adverbs for me
ash_ moritz_: will do
pmichaud i.e., :my, :ignorespace, etc.
ash_ if you change it from grammar to class it works
moritz_ yes, I'd prefer <:&foo> or <=&foo>
pmichaud i.e., I'd almost want :$ignorecase to be :ignorecase($ignorecase)
moritz_ ash_: as a workaround, you could try 'class Foo::Bar is Grammar'
pmichaud so far I like <=&foo> because we normally use '=' for capture to an alias anyway. 17:35
moritz_ aye
[Coke] phenny: .u ┌
17:36 ashleydev joined
ash_ tadzik: moritz_++ suggestion can give you a fix for now 17:36
tadzik well, I just renamed grammar to INIfile and moved to the file with a class, but thanks 17:37
btw, $.param = die "..." does not work
ash_ tadzik: you could override new if you want to enforce it as an argument for construction of an argument 17:39
tadzik ash_: I rather feel like checking this in BUILD, as I'm using it anyway
rakudo: class foo { has $.bar; submethod BUILD { $.bar.perl.say } }; my $a = foo.new(bar => 'moo') # why? 17:43
p6eval rakudo 3370f0: OUTPUT«Any␤»
jnthn tadzik: You overrode BUILD 17:44
tadzik: BUILD is what sets up the attributes, iirc.
tadzik oh
jnthn tadzik: Oh, maybe we're meant to set them if you didn't touch them in BUIlD, though.
tadzik rakudo: class foo { has $.bar; submethod BUILD { nextsame; $.bar.perl.say } }; my $a = foo.new(bar => 'moo') # this way?
p6eval rakudo 3370f0: ( no output )
tadzik …nope
so, if I want to open a file from a string which is passed to a class constructor, what's the best way to do this? 17:45
PerlJam tadzik: you'll have to be a little more precise, but something like my $handle = open $filename, :r; 17:47
ash_ if you don't need a file handle, just slurp it
my $string = slurp $filename
rakudo: (slurp "/etc/passwd").say # fails for security reasons 17:48
p6eval rakudo 3370f0: OUTPUT«Operation not permitted in safe mode␤ in 'Safe::forbidden' at line 2:/tmp/ddhB2nFXmI␤ in main program body at line 22:/tmp/ddhB2nFXmI␤»
ash_ rakudo: my $string = slurp; $string.say
p6eval rakudo 3370f0: OUTPUT«Operation not permitted in safe mode␤ in 'Safe::forbidden' at line 2:/tmp/2bBI7sCgKJ␤ in main program body at line 22:/tmp/2bBI7sCgKJ␤»
tadzik yeah, I know about slurp, that's what I'm going to use actually. But.. let me write it down
nopaste.snit.ch/22160 17:49
this is what I need
17:50 _macdaddy joined
pmichaud after thinking about it a bit more... 17:52
(1) I think there's a lot of merit to the <=$var> syntax to enable capturing, and we should consider it
(2) but it's just too complex for the common case of "I want a regex here"
PerlJam tadzik: depending on how quickly you pull together your INI grammar, you might want to put it up for inclusion in Rakudo Star 17:53
tadzik PerlJam: what I'm missing is section support (quite an important feature) but I wanted to fire up a working module first 17:54
ash_ rakudo: class Foo { has $.a; has $.b; method new($filename) { self.bless(*, a => $filename, b => $filename.flip); } }; say Foo.new('bar').b
p6eval rakudo 3370f0: OUTPUT«rab␤»
tadzik PerlJam: I have plenty of free time til 29, so it might end up with a success
pmichaud (3) I'm inclined to say that regexes in grammar definitions never do lexical lookup (and use <=&xyz> if that's what you want), while regexes outside of grammar definition can have <xyz> find a lexical &xyz (and fall back to Grammar.xyz if non-existent)
moritz_ tadzik: btw the return value from .parse is usually called a "match object" (or short "match"), not a parser
tadzik oh, ok
moritz_ +1 to all three points 17:55
ash_ rakudo: 'a' ~~ /./; say $/.WHAT
p6eval rakudo 3370f0: OUTPUT«Match()␤»
PerlJam pmichaud: could you elaborate on #2 a bit?
ash_ $/ is the current match object
pmichaud PerlJam: the example from the start of the chapter is the best elaboration 17:56
PerlJam pmichaud: okay :)
jnthn #phsers in ~ 1h
pmichaud having to write =&word to get the thing I just defined seems wrongish.
er, <=&word>
<word> just seems more appropriately dwimmy
TimToady no, the correct approach is to somehow mix the current namespace in as the most derived grammar
moritz_ not sure if I'll make it to #phasers
Tene rakudo: my @fibs := 1,1,{$^a + $^b} ... 50; say @fibs[0..5].perl;
kalve man, trying to read this gives me a headache
p6eval rakudo 3370f0: OUTPUT«(1, undef, undef, undef, undef, undef)␤» 17:57
Tene rakudo: my @fibs = 1,1,{$^a + $^b} ... 50; say @fibs[0..5].perl;
p6eval rakudo 3370f0: OUTPUT«(1, 1, 2, 3, 5, 8)␤»
tadzik so, could someone help me with this one? nopaste.snit.ch/22160
TimToady use regexes; or something
moritz_ where "current namespace" means what?
pmichaud TimToady: I don't intuitively grok the full details of what you're suggesting, but I'd not be opposed to it.
jnthn TimToady: Does "mix in" not imply that things from the current namespace would override those in the grammar we derive from? 17:58
TimToady yes
but as moritz_ points out, the scoping is a bit hard to nail down 17:59
pmichaud (and I'm doubtful it can be readied for R*, fwiw)
jnthn Me too
Well
pmichaud somehow "mix in" doesn't feel rightish to me, though. 18:00
jnthn I'm more worried about the side-effects we may find it has I guess.
pmichaud but I do see how it could potentially clarify a lot of issues, if we can get the scoping right. 18:01
18:01 kensanata joined
ash_ rakudo: class Foo { has $.filename; has $.something_else; method new($filename) { self.bless(*, filename => $filename, something_else => $filename.flip); } }; say Foo.new('bar').something_else; # tadzik slurp in the new maybe? 18:02
p6eval rakudo 3370f0: OUTPUT«rab␤»
TimToady it's really almost a slang issue. when I say /.../, I really mean the language derived thusly
[particle] where is the csv file used to generate the rakudo test passing status?
pmichaud agreed.
and when I say grammar B { ... } in some sense I'm deriving from that current slang
as opposed to deriving from Grammar
[particle] ooh, and now that i think about it, where is the web home for rakudo *, with distro test status, etc?
tadzik ash_: and then I'd be able to use BUILD anyway, yep?
ash_ yea 18:03
pmichaud [particle]: web home for R* will likely be rakudo.org .
ash_ BUILD and new are orthogonal
moritz_ perlgeek.de/blog-en/perl-6/object-c...ation.html 18:05
TimToady well, an explicit grammar takes its derivation from its ISA; the current slang would only influence the language you're writing in, not the language you're writin gabout
pmichaud PerlJam / moritz_ : in the meantime, I suggest we change the book example to use my $word = token { \w+ [ \' \w+ ]? } to avoid &word altogether for now.
18:05 justatheory left
pmichaud TimToady: okay, I can agree with that also. 18:05
TimToady but from a user point of view, it's augment slang Regex { regex ...; token ;;; }
18:05 justatheory joined
moritz_ pmichaud: but then we still have to do <word=$word> to get capturing done, no? 18:06
TimToady and then the rest of that scope uses the Regex slang with those "macros"
pmichaud moritz_: yes. Or maybe we can get the <=$word> syntax approved.
TimToady augment sort of being the wrong word for slang augmentation
moritz_ pmichaud: that's exactly what I want to avoid, because it*s *SO* clumsy
TimToady because it's really lexically scoped, not monkey-typing
moritz_ with <=$word> it would work better
masak let's see... #phasers in 24? 18:07
moritz_ masak: jnthn meant in 54
jnthn I had it as in ...that.
pmichaud moritz_: I think <word=$word> is far easier to explain to a newbie than trying to figure out the relationship between my token word { ... } and <&word> or <=&word>
masak #phasers in 54. goodie.
jnthn masak: ooh you'll has wifis still?
masak: Or internets at least? :-)
moritz_ pmichaud: I disagree. I'd feel obliged to explain why it feels so clumsy, and contains so much repetition 18:08
pmichaud moritz_: in which case that's probably not a good first example anyway :-)
at least, not until we get the handling of anonymous grammars correct per TimToady++
TimToady 'nuther idea. regex declarator inside a grammar adds to meta-object, outside adds to current regex slang
pmichaud TimToady: oh. I thought that's what you meant above,a ctually. 18:09
moritz_ panics
PerlJam we could modify that example and take out the "named regex" section of the regex chapter entirely and leave it to the grammars chapter to explain those. (sort of ignore lexical regex for a while)
[particle] so what happened in the past week that reduced rakudo's spectest count drastically?
moritz_ [particle]: refactored trig tests
[particle] ok, figured that. good. 18:10
pmichaud moritz_: oh, is that the cause?
TimToady <foo=&bar> is fine for R*
18:10 envi^home left
moritz_ PerlJam: I don't want that. Because the grammar chapter is hard to swallow anyway 18:10
pmichaud TimToady: I agree.
moritz_ so is <foo> picking up lexicals now officially dead?
tadzik rakudo: class A { has $.param; method new(Str $a) { self.bless(*, param => $a) }; submethod BUILD { say $.param.perl } }; my $foo = A.new('asd') 18:11
p6eval rakudo 3370f0: OUTPUT«Any␤»
TimToady moritz_: yes
tadzik ash_: ↑, am I doing it wrong?
pmichaud (I did quite a bit of book review on the plane yesterday, but didn't get to the grammars chapter)
moritz_ beats a dead horse
[Coke] trac.parrot.org/parrot/ticket/1707
pmichaud moritz_: we'll likely have something that replaces it, but they won't be lexical.
[Coke] (rakudobug with a parrot core dump special)
TimToady the slang is what's lexical
we can decide later how regex declarations can relate to the current Regex slang 18:12
jnthn [Coke]: Huh, what lame computer doesn't have 2325622477335777280 bytes of RAM available? :-)
[Coke]: Getting a stack trace would be good. 18:13
(C-level)
PerlJam TimToady: what about <=&named_regex> capturing? 18:14
moritz_ that would save a lot of typing to begin with
TimToady PerlJam: that is also probably dead for now
PerlJam :-( 18:15
pmichaud checks the book
[Coke] jnthn: pinged him for one.
jnthn [Coke]++ 18:16
18:19 eternaleye joined
PerlJam I guess if we removed the first example from the regex chapter and explain named regex better in that section of the chapter and added some minor fixups throughout, using <foo=&foo> wouldn't be too bad. 18:20
pmichaud PerlJam: yeah, that's what I'm thinking
moritz_ so, what should be leading example in that section instead?
pmichaud I'm thinking that first example is a little too stark to be a first example.
PerlJam moritz_: dunno.
pmichaud it's likely to frighten newbies off
moritz_ that's why I wanted to a nicer syntactic solution 18:21
pmichaud even with a nicer syntactic solution like <=&xyz> I think it could scare them off. It requires too many leaps in knowledge.
there are quite a few places like that in the book that I've found thus far.
PerlJam pmichaud: in general the chapters are "here's a complete working example, now let me explain it in detail" 18:22
pmichaud PerlJam: that probably needs to be made more explicit in each chapter, then. 18:23
moritz_ I guess that's a trade-off you have to make when you want the intitial examples to be both useful and idiomatic
18:23 macroron joined
moritz_ I know it's problematic, but it's not so easy to find better examples eitehr 18:23
pmichaud moritz_: well, we *do* expect that example to work someday
i.e.,
regex word { ... } 18:24
regex dup { <word> \W+ $<word> }
if / <dup> /
moritz_ yes, but it doesn't
and the examples should be runnable
pmichaud it's just that they won't be "lexicals", and it won't happen in time for R*
moritz_ so it's a moot point wrt the book
pmichaud right
TimToady maybe 'my regex' always adds to the current slang 18:25
pmichaud I'd prefer 'my' continue to create a lexical regex
(which you get to via &regex) 18:26
and bare 'regex' adds to the current slang, in much the same way that defining a new operator modifies a slang
18:27 hercynium left
pmichaud although there's some inconsistency with that approach, I admit 18:27
TimToady possibly confusing inside a grammar, if you're trying to mod the slang instead
pmichaud do we have an explicit "mod the slang" syntax?
PerlJam maybe modding the slang .... blah what pm said
TimToady currently augment slang, but I don't really like using augment that way 18:28
pmichaud I'm afk for a bit to fetch noms for the week. bbi30
PerlJam (I was typing "maybe modding the slang needs its own declarator or something")
TimToady pmichaud: where
pmichaud yeah, "augment slang" seems all wrong.
TimToady: I'm just heading to walgreens up the street from me to get snacks and all-important Dr Pepper to keep in the room :)
(I'm at the LQ again)
TimToady mkay 18:29
pmichaud I don't have explicit lunch plans yet, but will try to be back for #phasers in 31
(lunch after that, likely)
18:29 _macdaddy left
pmichaud afk 18:30
tadzik nopaste.snit.ch/22167 can someone show me the way with this one? 18:31
lue y o hai o/ 18:32
tadzik s/#12/#14/
oh hai
18:33 gbacon left
tylercurtis tadzik: "s/#12/#14/" doesn't parse in Perl 6. ;) 18:34
tadzik oh come on ;)
PerlJam heh
18:34 dolmen joined
tadzik I remember some Perl 6 module using sub new for something 18:35
pugssvn r31773 | tene++ | Fix minor typo
sorear good * #perl6 18:36
18:36 sorear sets mode: +ov tylercurtis pugssvn
Tene Hi! 18:36
[particle] tadzik: you define method new with a signature making $a required (using a trailing !)
masak \o/ sorear
sorear TimToady: Did you see the :my query?
[particle] tadzik: you should look at some examples in t/spec, grep for 'required', 'argument', 'class' etc
tadzik [particle]: will take a look 18:37
ash_ moritz_: i put that bug in, it seems to only happen when you use 'use' 18:38
TimToady sorear: yes, but don't really have much time to think about it till after I'm done writing my talk
ash_ bug 76606 if your curious
tadzik bleh, that's ugly
TimToady but to the first approximation, I'd just as soone keep the my coping rules limitied to curlies
you can declare lexicals from inside strings too 18:39
masak ingy: Just Another Crazy Bandaid Rakudo Hacker :P
rakudo: "foo$(my $x = 42)bar"; say $x 18:40
p6eval rakudo 3370f0: OUTPUT«===SORRY!===␤Redeclaration of symbol $x at line 22, near " = 42)bar\""␤»
masak hm, redeclaration?
std: "foo$(my $x = 42)bar"; say $x 18:41
p6eval std 31772: OUTPUT«ok 00:01 118m␤»
masak rakudo: "foo$(my $x)bar"; say $x
p6eval rakudo 3370f0: OUTPUT«===SORRY!===␤Redeclaration of symbol $x at line 22, near ")bar\"; say"␤»
[particle] what's that dollar for?
masak [particle]: variable interpolation
[particle] oh, i missed the quotes.
ingy masak: :)
masak ingy: I recognize the feeling. it's like, it's really good and really bad, with lots of promises of getting better soon. 18:42
ingy masak: the best part of perl6 is the irc channel 18:43
:)
ash_ rakudo: my $x; "foo$($x = 3)bar"; say $x;
timbunce Is there a perl5-to-perl6 concept map somewhere? ie something that says "if you'd write ... in perl5 then in perl6 you probably want to write ... or ..."?
p6eval rakudo 3370f0: OUTPUT«3␤»
masak ingy: yeah, the language sucks. :P
PerlJam ingy: wait until we have CPAN! :)
masak ash_: o.O
masak submits rakudobug
ash_ timbunce: perlcabal.org/syn/Differences.html
ingy owns bpan.org and is looking for a purpose
masak ash_: oh. you had 'my $x' before the string. nvm 18:44
rakudo: "foo$($x = 3)bar"; say $x;
timbunce ash_++
p6eval rakudo 3370f0: OUTPUT«===SORRY!===␤Symbol '$x' not predeclared in <anonymous> (/tmp/bOCGOeX_Js:22)␤»
ash_ masak: i was just seeing if what you said works if you pre-declare $x
tylercurtis timbunce: perlgeek.de/en/article/5-to-6 might serve that purpose to some extent.
ingy timbunce: wanna pair program with me today on some p6 stuff
moritz_ tadzik: I think you've hit a rakudobug. anyway, has $.b = 2 * $.a; should work just fine 18:45
ingy I know enough to be dangerous :)
masak ash_: right. but the experiment was to declare a variable inside a string which survived to the outside ;)
tadzik moritz_: so that was the right way to do it?
timbunce ingy: :)
ash_ yeah, i know, it was more a sanity check
[particle] a little learning is a dangerous thing.
PerlJam ingy: perhaps together, you and timbunce could be more than dangerous
moritz_ tadzik: yes
ingy PerlJam++
PerlJam ingy: maybe that's "deadly"? :>
tadzik moritz_: I expected BUILD works like in Moose, so the parameters are alredy set and stuff
moritz_: that's how it should work?
ingy is halfway done with the blog post now 18:46
sitting at Powell's cafe. here with @TimToady
tylercurtis tadzik: Perl 6 BUILD lets you take whatever parameter signature you want. However, that requires that you specify what signature you want.
18:46 cxreg left
tadzik tylercurtis: ah, so I don't bless anymore, but I have to set parameters, right 18:47
?
ingy tethered to you via android, cuz teh publik wifi sux
timbunce It takes 1 minute of cpu time to compile DBDI roles and classes, and there's hardly any actual code yet :(
ingy timbunce: seriously though, if I see you later, let's spend an hour... 18:48
PerlJam timbunce: btw, there is also this: perlgeek.de/blog-en/perl-5-to-6
lue quick question: does =:= still exist :)
timbunce PerlJam: yes, tylercurtis++ mentioned it a moment ago. Helpful, thanks.
PerlJam ah, missed it
TimToady lue: it does. don't use it. 18:49
lue rakudo: my $a = 3; my $b := $a; say VAR($a) === VAR($b)
p6eval rakudo 3370f0: OUTPUT«Could not find sub &VAR␤ in main program body at line 22:/tmp/bjDnK0bCYf␤»
tylercurtis rakudo: class Foo { has $.a; has $.b; submethod BUILD (:$.a!, $.b = $.a * 2) { } }; my $a = Foo.new(a => 5); say $a.a; say $a.b;
p6eval rakudo 3370f0: OUTPUT«===SORRY!===␤Can't put required parameter after variadic parameters at line 22, near ") { } }; m"␤»
tylercurtis rakudo: class Foo { has $.a; has $.b; submethod BUILD (:$.a!, :$.b = $.a * 2) { } }; my $a = Foo.new(a => 5); say $a.a; say $a.b; 18:50
p6eval rakudo 3370f0: OUTPUT«Cannot modify readonly value␤ in '&infix:<=>' at line 1␤ in 'Foo::BUILD' at line 6255:CORE.setting␤ in 'Foo::BUILD' at line 22:/tmp/rxwDajrsmU␤ in main program body at line 22:/tmp/rxwDajrsmU␤»
tadzik moritz_: wklej.org/id/366918/ that should be a working Perl 6 code?
tylercurtis rakudo: class Foo { has $.a; has $.b; submethod BUILD (:$!a!, :$!b = $.a * 2) { } }; my $a = Foo.new(a => 5); say $a.a; say $a.b;
p6eval rakudo 3370f0: OUTPUT«5␤10␤»
tylercurtis tadizk: ^
tadzik oh, great
timbunce ingy: it's probably too soon at the moment given the kind of thing I'm doing right now. Should be more useful in a couple of days. Catch me thursday or friday. 18:51
tadzik tylercurtis: what about this one: wklej.org/id/366918/ ?
tylercurtis tadzik: the problem with that is that $.a and $.b are readonly by default. You need to ether declare them "is rw" or use $!a/$!b instead of the accessors. 18:52
masak #phasers in 8.
tadzik tylercurtis: where are they readonly, in BUILD?
ah, oh god. Of course
damn 18:53
ash_ tadzik: they are always read only if you use the accessor without 'is rw';
tadzik yeah, I know, looks like I forgot about it
ash_ if you change Foo.new(5) to Foo.new(a => 5); things should work 18:54
tadzik but I wasn't planning to pass it via name
ash_ then you need to override new
PerlJam tadzik: new is always expecting named things unless you override it. 18:55
tadzik so what's BUILD signature for?
if I need to override new anyway?
TimToady it's for binding the args that BUILDALL will always send as named pairs 18:56
because BUILDALL isn't in the business of trying to coordinate the positional parameters of N different derived classes 18:57
pmichaud back from nom fetch (epic fail btw)
PerlJam pmichaud: they didn't have Dr Pepper?!? :) 18:58
pmichaud only in cans
tadzik Symbol 'self' not predeclared in new (class.pl:8)
pmichaud which is very inconvenient for carrying around the conference venue
tadzik self.bless(*, a => $a) in sub new. What's wrong here?
pmichaud also, they didn't have milk
or about 3 or 4 other things I was planning to buy 18:59
lue pmichaud: D:
TimToady there's a Safeway just north of Lloyd Center
pmichaud (they had spots for them... they were just out of stock)
yeah, Safeway is my next hit
it's a bit of a further walk from here
ash_ tadzik: can you update your nopaste with the current code your working with?
tadzik wklej.org/id/366921/ BUILD might be still broken, but it breaks in new now 19:00
ash_: ↑
ash_ its method, not sub
sub's don't have a self
tadzik oh, sure
TimToady new needs to be amethod
tylercurtis tadzik: I think BUILD still needs a signature, too. 19:01
tadzik the named one?>
jnthn #phasers!
19:01 mberends joined
TimToady if the names correspond to your attributes, you can use the auto-generated BUILD 19:01
BUILD must be a submethod, btw
even more than new must be a method
tylercurtis tadzik: Although in this case, you could just do "method new (Int $a) { self.bless(*, a => $a, b => $a * 2); }" without problem, I think. 19:02
TimToady: he already has BUILD as a submethod, no?
ash_ yeah, then you don't need to do a BUILD
TimToady I'm not actually looking at his code, since I officially don't have time to do that. :)
timbunce If I want to return an undef Str I should just return Str;, right? But then I'm surprised to find that say foo() prints "Str()" rather than complain about undef. Do I just need to get used to this? 19:03
_sri wonders why rakudo is not yet in homebrew (because parrot is)
ash_ TimToady your pretty close to spot on for most of the problems he's had, (but BUILD was already a submethod :P )
TimToady well, I didn't say his wasn't a submethod, now did I? :P 19:04
ash_ timbunce: undef is spelled Mu
TimToady sometimes
std: undef
PerlJam ash_: He wants an undef string though
p6eval std 31772: OUTPUT«===SORRY!===␤Unsupported use of undef as a value; in Perl 6 please use something more specific:␤ Mu (the "most undefined" type object),␤ an undefined type object such as Int,␤ Nil as an empty list,␤ :!defined as a matcher,␤ Any:U as a type
..constraint␤ or fail() as…
TimToady an undef string is spelled Str
ash_ oh 19:05
i see, you mean .defined then
rakudo: say Str.defined, "foo".defined
p6eval rakudo 3370f0: OUTPUT«01␤»
TimToady undefs are typed in Perl 6
ingy is there a canonical purl for p6rules?
*url
jjore_ Is it expected that I get "Malformed UTF-8 string" exceptions when reading text files like perl5's malloc which apparently does something wrong in the string "The Bridge of Khazad-D\xFBm" 19:06
ingy and is that the official term? "Perl 6 Rules"?
hi jjore_
timbunce perhaps there's a need for some way to tell say how undefs should be handled.
PerlJam ingy: Perl 6 does rule! :)
jjore_ Hi ingy
TimToady no, that's the official slogan. they're called Perl 6 Regexes
see S05 19:07
ingy TimToady: is s05 the best reference url?
jjore_: are you in pdx?
TimToady it is if you follow the highlight from the ir clog
jjore_ Nope.
Just Seattle
ingy :'(
jjore_ Yeah, $work.
ingy jjore_: I'll be back on Wednesday (I know you miss me) 19:08
jjore_ Well I do, however I never see you even when you are in town.
I have to go to far-flung places to see you.
TimToady which will send you to perlcabal.org/syn/S05.html
19:08 justatheory left 19:09 lue is now known as mlue
ash_ std: my $a = Str but ''; say $a # should print nothing 19:09
p6eval std 31773: OUTPUT«ok 00:01 118m␤»
ash_ but its not working right now in rakudo
jjore_ you know, just fyi, the unicode on #perl6 really screws with my IRC client. :-( 19:10
TimToady fix your client
it's open source right? :) 19:11
ash_ .u frown
phenny U+2322 FROWN (⌢)
sorear »ö« is our frient
masak jjore_: see topic :)
sorear: isn't it 'prient'? :)
19:12 cxreg joined, kalve left
jjore_ My client is open source. Heck, my client is the editor for itself so I can hot-fix it while it runs. 19:13
but er, lunch apparently &
TimToady ö/
masak yet another vital open source fix foiled by nom. 19:14
rakudo: say "OH HAI".bytes
p6eval rakudo 3370f0: OUTPUT«6␤»
masak Buf should have a .bytes method, I guess. should Str?
and I know we've talked about this, but it's hard and my brain is non-sticky sometimes. 19:15
TimToady perhaps "OH HAI" doesn't feel like committing one way or the other yet
19:15 rurban joined
masak does that mean "no, Str shouldn't have a .bytes method"? 19:15
TimToady no, it means "yes, Str shouldn't have a .bytes method" :) 19:16
19:16 mlue is now known as lue
masak TimToady: so if I go ahead and remove the .bytes method from Str, have I understood you 50% or 100% ? 19:16
tadzik ===SORRY!=== You can not add an attribute to a module; use a class or role 19:17
what does he want? Didn't even show a line
masak tadzik: you tried to do 'has' in a non-class/role.
tadzik …no I didn't :) will show the code
masak nopaste!
tadzik wklej.org/id/366935/
whoops, too late
pmichaud it should be possible to implement =:= without Var. 19:18
tadzik on 'use Config::INI' rakudo cries as I showed
pmichaud er, without VAR()
ash_ line 2
tylercurtis tadizk: "has $.file;"
ash_ has $.file
tadzik oh
wtf is it doing there
ash_ i think you mean our
or maybe it shouldn't be there...
tadzik no, I didn't mean it at all. Suprise keystrokes in vim I guess
[particle] vim curses in perl6?
tadzik why curses? 19:19
must have 'p' accidentally
ash_ tadzik: did you know you can specify :actions for a grammar?
tadzik ash_: nope, what does it do? 19:20
ash_ GrammarName.parse($string, :actions(Actions.new));
whenever it parses a token/rule/regex it invokes the action associated with the token/rule/action
takadonet tadzik: you will love them :)
ash_ regex*
tadzik oh 19:21
triggers it, yeah?
takadonet yes
tadzik wklej.org/id/366937/ another problem
ash_ allowing you to do any magic you need mixed right into the parsing
tadzik so in this example, where does Actions.new object go?
tylercurtis tadzik: And you can use the actions to build up a AST or similar from the parse, if that's necessary.
Tene tadzik: perl6advent.wordpress.com/2009/12/2...d-actions/
ash_ Tene++ for giving a real example
Tene tadzik: that's a quoting problem in your shell. 19:22
tadzik /o\ I'm failing today
Tene tadzik: look, you used the ' quote for the outer, and then ' to try to quote inifile.
tadzik yeah, I see :(
Tene but it just ended the shell quote.
tadzik what a day
Tene :)
tadzik but at least this grammar subset works :)
19:23 timbunce left
TimToady masak: yes, you may remove it; my point was just that there could be an ascii buf type for literals that also knows how to promote to Str 19:23
Tene ash_: That post was inspired by stuff I was really doing at work at the time.
masak TimToady: oki. removing. 19:24
TimToady just as a utf8 buffer knows how to promote
ash_ Tene: neat, even better, an actual example based off real work
sorear rakudo: say (1.0 .. 10.0).flat.Str
Tene the real source I had was XML, and the real program I wrote was in Perl 5 and used XML::Parser, iirc. 19:25
p6eval rakudo 3370f0: OUTPUT«1 2 3 4 5 6 7 8 9 10␤»
sorear rakudo: say (1.0 .. 10.5).flat.Str
p6eval rakudo 3370f0: OUTPUT«1 2 3 4 5 6 7 8 9 10␤»
sorear rakudo: say (10.5 .. 1).flat.Str
p6eval rakudo 3370f0: OUTPUT«␤»
sorear rakudo: say (1.5 .. 10.0).flat.Str
p6eval rakudo 3370f0: OUTPUT«1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5␤»
TimToady arguably, defaulting to the indicated precision would be more useful, I suppose 19:26
sorear well, I actually care about Num 19:27
TimToady say (10.0).perl
rakudo: say (10.0).perl
p6eval rakudo 3370f0: OUTPUT«10/1␤»
TimToady I guess it always reduces, though, so you lose the info
sorear but really what I'm checking is how infix:<..> deals with 1. reversed arguments 2. dense types
(in list context)
19:27 rotterdam joined 19:28 rotterdam left
TimToady reversed will always produce Nil, and "add 1" is the simpleminded rule, on the assumption anything more complicated should be done with ... anyway 19:28
ash_ rakudo: say 1.0 ... 10.5 19:30
p6eval rakudo 3370f0: OUTPUT«12345678910␤»
ash_ rakudo: say 1.5, 2.0 ... 10
p6eval rakudo 3370f0: OUTPUT«1.522.533.544.555.566.577.588.599.510␤»
ash_ i have no idea what i just did... 19:31
oh
its just missing spaces
nevermind
sorear I think you just asked p6eval's phone number.
ash_ lol
mathw Evening 19:34
masak mathw! \o/ 19:35
Tene rakudo: say 8675309
p6eval rakudo 3370f0: OUTPUT«8675309␤» 19:36
dalek ecza: cb5b9dd | sorear++ | Kernel.cs:
Allow for subclassing Sub
ecza: 5b329a4 | sorear++ | rxwork.pl:
Implement class Regex & smart-matching
19:36 rurban left 19:37 gbacon joined
mathw hey masak 19:37
19:38 plainhao left
mathw time for form-fixing again 19:38
pugssvn r31774 | moritz++ | [evalbot] niecza VERSION
tadzik method postcircumfix:<< >> should work for an object like $object<asd>? 19:40
mathw hmm
how to do this 19:41
tadzik it acts funny
Tene <> is the autoquoting form of postcircumfix:<{ }>
you want to override the latter, not the former.
tadzik wklej.org/id/366945/
well, same was happening for the former
19:41 i__ left
Tene Hmm. Dunno. 19:44
tadzik bug?
Tene Probably, yes.
It certainly should not fail like like that.
tadzik it's alredy two bugs today, let's wait til moritz_ gets back from #phasers
moritz_ upgrades one of his computers to a newer debian version in the hope of then being able to build the book 19:45
pmichaud actually, you probably want to override "at_key" instead of postcircumfix:<{ }>
moritz_ what exactly is at_key?
pmichaud return a value at a given key location
it's what postcircumfix:<{ }> calls when it has a single key to resolve 19:46
moritz_ ah
pmichaud if you override postcircumfix:<{ }>, you're basically taking over all slicing forms
moritz_ evalbot control restart
19:46 p6eval left
ash_ so... method at_key($key) { } ? 19:46
19:46 p6eval joined
pmichaud ash_: yes. 19:46
ash_ rakudo: class F { method at_key($b) { 'got it' } }; my F $f .= new; say $f<a>; 19:47
p6eval rakudo 3370f0: OUTPUT«got it␤»
tadzik same effect
ash_ rakudo: class F { method at_key($b) { 'got it ' ~ $b } }; my F $f .= new; say $f<a>;
p6eval rakudo 3370f0: OUTPUT«got it a␤»
tadzik and I'm not evev calling
dalek ok: aa1199a | pmichaud++ | docs/release-guide.pod:
Update with pmichaud release date.
ash_ your error was happening in BUILD though...
tadzik hmm 19:48
what can cause this?
ash_ might be related to your for loop 19:49
take that out and see if it dies still
tadzik no, it lives
funny, all the code works if it's not inside the class 19:50
ash_ hmm
might be another bug :P
19:51 mmcleric_ joined, mmcleric left, mmcleric_ is now known as mmcleric
tadzik saying [0] works 19:51
and elems prints 4, so it's finite 19:52
jnthn rakudo: (1,2,3,4,5).grep(* > 2).grep(* < 5).perl.say
p6eval rakudo 3370f0: OUTPUT«Cannot resume dead coroutine.␤ in 'Any::join' at line 1␤ in 'List::perl' at line 2528:CORE.setting␤ in main program body at line 22:/tmp/tP945QDNyS␤»
tadzik %!contents<toplevel>{$param<key>} = $param<value>; this fails 19:53
assigning anything fails 19:54
pmichaud inside of BUILD, is %!contents already initialized at this point?
tadzik say $param<key>.WHAT; return Match()
19:54 pyrimidine joined
tadzik pmichaud: I don't think so 19:54
it isn't being touched before
pmichaud tadzik: I mean, is it already a Hash?
or at least an Any? 19:55
tadzik well, it has a sigil
if Perl 6 hasn't made it a hash the no
19:55 takadonet left, whiteknight joined
tadzik now %!contents<toplevel>{$param<key>.Str} = $param<value>.Str; works 19:55
so just %hash{Match()} is not a good idea 19:56
19:56 eternaleye left
tadzik it didn't stringify by itself 19:56
19:57 eternaleye joined
pmichaud afaik, it's not supposed to stringify by itself 19:58
(but perhaps there was a spec change there)
tadzik it doesn't, there have to be .Str in both calls
but I have yet another something, will paste 19:59
wklej.org/id/366953/ 20:00
Before returning .perl shows the hash, outside the class it Any
perhaps some issue with returning a private value? 20:02
[Coke] (Perl 6 R* Us)
20:03 mberends left, mberends joined 20:06 patspam joined
moritz_ any objections to filtering out "." and ".." from dir($path) ? 20:07
pmichaud that feels wrong to me. 20:08
moritz_ why?
pmichaud because they're entries in the directory, same as others.
moritz_ but how often to do you actually need to use them?
nearly all of my scripts that recurse throug directories, list them etc. contain separate logic just to ignore them 20:09
tylercurtis moritz_: once Path is implemented, I'm pretty sure it's specced to have a method for that.
PerlJam moritz_: give dir an adverb to ignore/include them
pmichaud I'd prefer there to be an adverb/option for including/excluding them, then.
but simply excluding them always feels wrong.
moritz_ would dir($path, :all) work for you? 20:10
pmichaud (and once there's an adverb to include/exclude, it seems like include/exclude should be a Matcher)
with a default that excludes . + ..
PerlJam moritz_: sure
tylercurtis misunderstood.
[Coke] if rakudo * is going to bundle the perl6 compiler, what is the R* executable going to be called? 20:11
moritz_ pmichaud: so getting them all would be dir($path, :!exclude)
[Coke] or is the R* executable the perl6 compiler executable?
pmichaud moritz_: no, that feels wrong to me
tadzik moritz_: what about this BUILD but?
* bug?
pmichaud [Coke]: the latter
moritz_ tadzik: it's a bug.
tadzik report it?
moritz_ tadzik: I think it's known, but I'm not sure
[Coke] any thoughts on making the default name for the installed exe "rakudo" instead of "perl6" ? 20:12
moritz_ tadzik: if you can't find it in RT; please report it
pmichaud moritz_: I think it should be dir($path, test => ... )
Eevee imo the executable should be "rakudo★"
[Coke] (but keeping perl6 as a copy (until we bother with making a symlink)
pmichaud i.e., "dir" has a built-in 'grep'
[Coke] Eevee: did you mean to have extra chars there?
moritz_ pmichaud: 'test' is not very specific about whether it includes or excludes 20:13
Eevee [Coke]: yes it's U+2605 BLACK STAR :P
pmichaud moritz_: it includes
just like with grep
tadzik er, searching for BUILD on rt is not a good idea
pmichaud [Coke]: I've considered switching the executable name to "rakudo", yes. 20:14
tadzik moritz_: rt.perl.org/rt3/Public/Search/Simpl...=submethod nothing looks like this one
moritz_ tadzik: then submit
tadzik moritz_: the bug is, that the class variables are not visible in BUILD?
pmichaud [Coke]: I've not been able to argue myself into it enough to overcome the inertia of "perl6".
moritz_ tadzik: that the attributes aren't initialized yet
tadzik alright 20:15
pmichaud moritz/tadzik : should they be?
I thought BUILD was responsible for initializing them.
tadzik pmichaud: moritz_ told me so :)
pmichaud and then if left uninitialized, the default initializer grabs them
tadzik I only know that in Moose they alredy where. And it was the way to make some funny magic after an object construction
pmichaud but we can't say they should be initialized by default if BUILD is supposed to potentially leave them uninitialized
moritz_ pmichaud: I kinda thought the bless($candidate, %args) initializes them 20:16
pmichaud bless calls BUILD, yes.
moritz_ wait
it calls BUILD for every class in the hierarchy
pmichaud including the one you're in, iirc.
jjore_ hi, what method do I call to get the list of methods available on an object? 20:17
moritz_ and shouln't Mu.BUILD initialize all attributes, leaving custom BUILDs free from these chores?
tadzik moritz_: what's the preffered naming for perl6 modules again?
jjore_: ^methods
TimToady Mu runs first, not last
moritz_ jjore_: $obj.^methods
jjore_ o_O
moritz_ or $obj.^methods(:local), if you don't want parent attributes
20:17 oZ] joined
jjore_ I guessed .methods, .meta.methods, and .META.methods but wouldn't at have ever arrived /there 20:17
thanks
moritz_ TimToady: right, that's why I thought that a custom BUILD already sees initialized attributes 20:18
jjore_ Oh, I don't see ^methods in the list of methods
moritz_ jjore_: that's because it's a method on the meta class
or syntactic sugar for that 20:19
tylercurtis rakudo: say Int.HOW.methods
p6eval rakudo 3370f0: OUTPUT«too few positional arguments: 1 passed, 2 (or more) expected␤ in main program body at line 22:/tmp/gmJgdBhs_q␤»
moritz_ rakudo: say Int.HOW.methods(Int)
p6eval rakudo 3370f0:
..OUTPUT«ACCEPTSfrom-radianspredatanhexpcosecBridgesinhto-radiansWHICHacotanrootsComplexlnsecaseccotanhcislogsechlog10randtruncatesinsqrtasinIntBoolcoshsuccsignasinhacosechabsrealsceilingunpolarfloorasechRatroundacoshacosecNumcosechcotanatan2tanRealatancosacostanhNumericperlStracotanhtrim-t…
tylercurtis rakudo: say Int.^methods
moritz_ $obj.^foo is short for $obj.HOW.foo($obj)
mathw mutters and mumbles about action methods
p6eval rakudo 3370f0:
..OUTPUT«tanhatanhcosecNumWHICHperlrootstanlog10from-radianscospredacosecsinhasinhNumericcosechrealsStrasecacotanhfloorsechACCEPTSasechrandtruncateroundsqrtasinBoolcoshexpsuccacoshsigncotanatan2ComplexseccislogsinIntBridgeto-radiansabsacosechacotanlnceilingcotanhunpolaratanRealacosRattanhat…
jjore_ ah 20:20
mathw okay
it's re-read spec time
moritz_ because the meta class of Int handles all the objects, and differnt objects can have different methods mixed in
tylercurtis Ah, right.
jjore_ Is it weird that .HOW doesn't show up either?
tadzik moritz_: what's the preffered naming for perl6 modules again?
moritz_ rakudo: class A { method x() { } }; my $y = A.new; say ~$y.^methods(:local); $y does role { method new_metho() { } }; say ~$y.^methods(:local)
p6eval rakudo 3370f0: OUTPUT«x␤new_metho␤» 20:21
moritz_ tadzik: foo-bar or perl6-foo-bar
tadzik ok
pmichaud .HOW is more of a macro than a true method.
(although I am surprised it doesn't appear in Rakudo :)
moritz_ I'm kinda surprised that x doesn't show up in the second .^methods
jnthn moritz_: I'm not. 20:22
moritz_: Mixing in a role derives an anonymous sub-class and mixes into that.
moritz_: If we just replaced methods rather than doing that, you'd not be able to defer up to the original class from the mix-in, for example. 20:23
moritz_ jnthn: ah, so x is one level away
jnthn moritz_: Yes :-)
moritz_: So if you mix in an x and the original class had an x, you can still get hold of both.
:-)
20:24 _macdaddy joined
moritz_ proposes :*foo to mean :foo(*) 20:25
... not really :-)
jnthn :*|
masak *.O
pmichaud proposes :*| to mean someone just got punched in the nose.
moritz_ std: :::foo; class foo { };
p6eval std 31774: OUTPUT«ok 00:01 114m␤»
moritz_ \o/ STD.pm parses colonpairs with :: sigil
masak pmichaud: wouldn't that be :*( in most cases? :P
moritz_ rakudo: my $x = :::foo; class foo { }; 20:26
p6eval rakudo 3370f0: ( no output )
20:26 timbunce joined
TimToady um, no 20:26
moritz_ rakudo: my $x = :::foo; class foo { }; say $x.perl
jnthn masak: Depends if you were sad about being punched or just disappointed, I guess.
p6eval rakudo 3370f0: OUTPUT«undef␤»
TimToady anything done with :: is hardwired; it's not a sigil
moritz_ what is this undef thing?
20:26 perimosocordiae joined
masak jnthn: I think I missed the Unicode codepoints "SAD ABOUT BEING PUNCHED IN THE NOSE" and "JUST DISAPPOINTED ABOUT BEING PUNCHED IN THE NOSE" :) 20:27
pmichaud and "HAPPY TO HAVE A CLOWN NOSE"
aka
:*D
masak pmichaud++ 20:28
that karma point was for the mental images alone :)
lue
.oO(CIRCLE THAT USED TO BE SQUARE)
 20:29
TimToady ▢
pmichaud wouldn't that be "CIRCLE THAT IS ABOUT TO BE SQUARE"?
lue SQUARE THAT USED TO BE CIRCLE
masak lue: was it punched in the nose as well?
lue Yes. It just has an ose now :) 20:30
mathw \o/ got Form successfully not using {*
{*} #= key stuff
pmichaud \o/
mathw thanks to the Book
jnthn mathw: yay!
tylercurtis :●D HAPPY TO HAVE A CLOWN NOSE FACING THE READER
mathw which hinted at me how to pass up the ast from a subrule match in a manner that doesn't drive me insane 20:31
masak "Perl 6: HAPPY TO HAVE A CLOWN NOSE"
mathw although it's still not entirely elegant
pmichaud sadly, Camelia does not have a clown nose.
masak and people say we're bad at marketing...
dalek kudo: 2783660 | moritz++ | src/core/IO.pm:
dir() now defaults to filtering out . and ..
 20:32
mathw only two of seven test files now failing!
lue .u FFD0
phenny U+FFD0 (No name found)
mathw I think I'll commit this
lue
.oO(should have said EYE OF THE BASILISK)
pmichaud okay, off to register for oscon and nomhunt 20:33
bbiaw
lue goodluck o/
rakudo: say "\xFFD0"
p6eval rakudo 3370f0: OUTPUT«￐␤»
mathw and my cat's come to give me a congratulatory nose rub 20:34
just the encouragement I need to figure out what's wrong with the other two files 20:35
tadzik github.com/tadzik/perl6-Config-INI 20:36
moritz_ tadzik++
tadzik suggestions/patches/criticism welcome
(see also Acme::Meow :>) 20:37
20:37 _mpu left
moritz_ looks nice 20:37
PerlJam tadzik: want to add it to Rakudo*?
tadzik it's missing sections
PerlJam: once I make it finished, sure
20:38 takadonet joined
PerlJam doesn't understand "finished" :) 20:38
tadzik hmm
moritz_ btw the last patch to rakudo shows a nice use case for none()
tadzik nopaste.snit.ch/22177
20:39 takadonet_ joined
tadzik nice 20:39
masak chromatic is writing interesting things about Little Languages. www.modernperlbooks.com/mt/2010/07/...uages.html -- he's kind of building up to Why We Want Perl 6.
PerlJam tadzik: wiki.github.com/rakudo/rakudo/whats...nto-rakudo Look just above "Add more here :-)" 20:40
mathw Best practice question: should libraries with errors be using 'die' or 'fail'? 20:41
tadzik PerlJam: thanks :)
oZ] Would anyone be willing to point me at any discussions regarding threading in rakudo?
PerlJam oZ: The discussions would be very brief. There is no threading in rakudo :) 20:42
ingy greetings
moritz_ oZ]: it's not yet implemented, and blocks on parrot not supporting threads for high-level languages
(unless a brave soul ports rakudo to another backend) 20:43
oZ] I'm aware it doesn't yet exist -- just wondering about what thought has gone into it and if it's a goal for later. :)
ingy would anyone be interested in previewing my log post?
*blog
PerlJam oZ]: or do you mean threading as in autothreading ala junctions?
ingy gist.github.com/483546
PerlJam ingy: sure!
moritz_ takes a look
oZ] Nope -- I mean, open a thread, throw something in there for the thread to do. ;)
20:43 rblasch_ joined
masak mberends: ping 20:44
moritz_ it's planned, yes
but not any time soon
PerlJam heh "short ten year wait"
oZ] Oh.
tylercurtis oZ]: perlcabal.org/syn/S17.html although I don't know if anyone has much if any of that implemented, so it'll probably change drastically between now and being implemented.
moritz_ unless somebody takes it up, and implements it
ingy sorry about the badly formatted gist
one sec
moritz_ downloads the raw thing, and puts it into vim
oZ] Any pointer on the parrot issue you're talking about? 20:45
moritz_ ingy: please remove that 1.0. It's not going to be the official version number
oZ]: just a sec...
ingy moritz_. PerlJam: regist: gist.github.com/483559 20:46
moritz_ oZ]: trac.parrot.org/parrot/ticket/757 20:47
masak phenny: tell mberends when I remove (Cool-)Str.bytes, t/spec/S32-io/IO-Socket-INET.rakudo fails 10 tests. sure enough, src/core/IO/Socket.pm contains the only use of .bytes in the whole Rakudo code base right now. I'd like to confer with you re what to do with it.
phenny masak: I'll pass that on when mberends is around.
20:47 _macdaddy left
PerlJam ingy: there's another Rakudo * 1.0 later in the doc 20:47
oZ] moritz_++ thank you very much for your help! 20:48
20:48 literal left, avar left
moritz_ oZ]: also if you search for 'threads', you'll find that there are quite some patches for it, but they stay unapplied 20:48
PerlJam oh, you didn't get rid of the first 1.0
20:48 rblasch__ joined
moritz_ oZ]: because nobody feels responsible for them, and nobody really understands it 20:48
masak reviving Ponie at some point might be a very special kind of fun.
ingy PerlJam, moritz_: 1.0 removed
moritz_ oZ]: there's hope though. Currently a student works on a new threading system for parrot. Talk to 'Chandon' on #parrot (irc.perl.org) 20:49
oZ] Ooh, OK. Thank you again.
20:50 avar joined
tylercurtis oZ]: parrot.org/blog/836 for more info and Chandon's GSoC. 20:50
moritz_ ingy: btw you're welcome to implement JSYNC on top of "my" JSON::Tiny grammar
20:51 literal joined, rblasch_ left
PerlJam ingy++ interesting post. I like the no-nonsense aspects of it :) 20:52
ingy: btw, have you learned ruby yet? I find it weird that you like python more than perl 6.
tadzik I find it weird you like both. They feel completely antagonistic to me 20:54
cxreg anyone at oscon that could provide a ubuntu cd or know where to find one?
moritz_ ingy++ nice blog post 20:55
tadzik sends a call to arms for the local perl mongers list, call to write perl6 modules
20:56 justatheory joined, Kodi joined
Kodi masak: ping 20:56
phenny Kodi: 16:51Z <supernovus> tell Kodi nice work on the Temporal stuff!
masak ingy++ # nice post 20:57
Kodi: *pong*
Kodi masak: I dropped day-name from DateTime 'cause it ain't specced. I changed stftime.pm appropriately, but maybe novus's patch conflicted with that.
jjore_ You know, I think I really do like that sigils in ruby denote scope. That's nice coming from perl where there's nothing visible except convention like uppercase variables being constants or class data
masak Kodi: ah. that means that the spectests are in the wrong. 20:58
Kodi: in that case, I think I can commit your patch.
moritz_ jjore_: well, Perl 6 has adoptedthat partially
masak jjore_: there's twigils in Perl 6.
moritz_ with twigils
Kodi masak: No, I if I remember correctly, DateTime.t doesn't try to use a day-name method. [more] 20:59
PerlJam apparently everyong thought the same thing when reading jjore_
moritz_ $*foo being a dynamic variable, $!foo an attribute, $^foo a formal parameter etc.
jjore_ we have scoping twigits?
oh, hmm
Kodi masak: It's just that day-name is called by strftime, and it shouldn't be.
cosimo oh hai, #perl6. Q: is .pm6 the "cool" extension now? I thought we wanted to keep .pm
Kodi strftime should have a day-of-week name array and use that.
That ought to fix the failing tests. 21:00
cosimo but i see increasing amounts of .pm6 files
moritz_ cosimo: it seems some projects have moved to .pm6, due to having to ship both p5 and p6 modules
I don't really care
21:00 paul1234 left
masak Kodi: applying your patch caused failing tests as shown in my email. 21:00
Kodi: you lost me after that in your above explanation. 21:01
cosimo moritz_: ok, i'll keep using .pm for now, then :)
Kodi masak: Sorry. Here's what I'm trying to say: the call to DateTime.day-name, if I recall correctly, doesn't happen in any of the test files, it happens in strftime.pm. So strftime.pm is what needs fixing.
21:02 takadonet left
Kodi The tests are correct in thinking that strftime should support names of weekdays. 21:02
cxreg is :rw supposed to be a valid open argument?
masak ah.
moritz_ cxreg: yes, but NYI
masak Kodi: is it possible you could prepare a patch with (1) s/Calendrical/Dateish/ and (2) the strftime.pm issue fixed? 21:03
21:03 kensanata left
masak Kodi: that'd greatly simplify my apply work. 21:03
cxreg moritz_: due to seek issues or something?
21:03 Su-Shee left, takadonet_ left
moritz_ cxreg: probably just because nobody got around to implement it yet 21:03
Kodi masak: Sure, I can do that tonight. I just need to run a participant (psychology experiment) and have dinner first.
masak ++Kodi 21:04
Kodi: take your time.
moritz_ wonders if #perl6 is just a big psychology experiment
PerlJam Where does HTTP::Daemon live? In web?
Kodi masak: And then we'll have working Date and DateTime in Rakudo *! Hooray!
moritz_ somwhere on gitorious,iirc
21:04 Mowah left
masak yes, gitorious. 21:04
PerlJam looks
21:05 oZ] left, hercynium joined
masak rakudo: module M {}; say 42 ~~ M 21:06
21:06 M_o_C left
p6eval rakudo 278366: OUTPUT«Method 'ACCEPTS' not found for non-object␤ in 'infix:<~~>' at line 397:CORE.setting␤ in main program body at line 22:/tmp/fC9POxlNFX␤» 21:06
masak submits rakudobug 21:07
mathw++ for discovering it.
moritz_ what should it do?
is M a type?
masak print "0\n"
moritz_: I'd expect it to be an object, given that everything is, if pressed.
ingy moritz_: I plan to implement JSYNC on your code, of course.
21:08 Kodi left
moritz_ \o/ 21:08
ingy PerlJam: Python is nice and boring
:)
moritz_ is anybody *not* getting passing TODOs in t/spec/S32-str/uc.rakudo?
ingy Rakudo has too much "excitement" ;)
PerlJam ingy: oh. In that case, I can see quite well how you would like Python over Perl 6 :-) 21:09
masak moritz_: I've got them today. haven't paid attention to whether they've been consistent.
Tene ingy: but you don't put spaces after twigils... 21:11
pugssvn r31775 | moritz++ | [t/spec] unfudge some passing TODO tests, and see if we get failure reports from any platforms
masak moritz_: not consistent. I just ran uc.rakudo again, and didn't get passing TODOs. 21:13
21:16 patspam left 21:19 paul1234 joined
ingy Tene: *! ;) 21:23
21:23 perimosocordiae left
masak ingy: did you just punch Tene in the nose? :P 21:23
ingy: that kind of Pythonista behaviour isn't tolerated around here... :P 21:24
Tene ingy: eh?
masak poor Tene is all dazed now.
21:25 gbacon left 21:27 macroron left
ingy blogs.perl.org/users/ingy_dot_net/2...ready.html 21:27
how do I get this post linked to blogs.perl.org front page? 21:28
tadzik it appeared 21:32
21:33 felliott left
jnthn ingy++ # nice post! 21:33
21:38 sftp left, Guest62932 left, skids left
masak looking closer at the Str.bytes use in src/core/IO/Socket.pm, I think it not only *can* be changed into Buf.bytes, but *should* be. still, it'd be nice to talk to mberends++ before making the switch. 21:46
god natt, #perl6.
tadzik o/
21:46 masak left
jjore_ Hi, my runtimes vary *alot* depending on the contents of the files it is parsing. Anyone want to gesture at this to guess why? diotalevi.isa-geek.net/~josh/100720...rakudo.txt 21:51
pugssvn r31776 | dolmen++ | [S32/Temporal] Make DateTime immutable.
21:52 mreed joined, mreed is now known as markjreed
dolmen ingy++ 21:52
markjreed Where in the Synopses is the new hash subscript syntax discussed? Is there any difference between %hash<foo> and %hash<<foo>> (or %hashB+fooB; 21:53
Ok, so much for Unicode in this irc client. :)
So far I've done a cursory search over S02, S03, and S09.. 21:54
jnthn markjreed: I'd guess S02 or S09.
pmichaud back
dolmen phenny, tell masak I have committed my Temporal spec patch
phenny dolmen: I'll pass that on when masak is around.
jnthn markjreed: Ah. :-)
21:57 rblasch_ joined, pmurias left
jnthn jjore_: Those are...interesting. 21:58
jjore_: There's no correlation with filesize? 21:59
ingy greetings
jnthn But I guess not looking at just how big the differences are.
pmichaud: diotalevi.isa-geek.net/~josh/100720...rakudo.txt may be interesting to you when you get chance.
jjore_ Oh! that's what I meant to add to the report, file size.
21:59 hercynium left
jjore_ and whether the pattern succeeded or not 22:00
jnthn pmichaud: Wonder if it's hitting some expected pathological backtracking case.
*unexpected
*unexpectedly
gah
:-)
22:00 rblasch__ left 22:01 mce_ro joined
jnthn afk for a little 22:01
jjore_ fwiw, I'm merely running that program against the perl5 source. I wrote it last year at use.perl.org/~jjore/journal/39233 22:02
markjreed Based on the Apocalpyse and Exegesis, I would expect it to be in S02, but if it's in there, I don't see it.
tadzik there will be more then one Rakudo Star? 22:03
pmichaud tadzik: Rakudo Star is a series of distribution releases, not a single release.
tadzik oh, nice 22:04
pmichaud looking at timing-rakudo.txt
22:04 justatheory left 22:05 timbunce left
pmichaud adding some judicious ratchets or 'cut' markers would likely speed things up a fair bit 22:05
for example, \N*:\n can be far more efficient than \N*\n 22:06
22:06 hercynium joined
pmichaud with \N*\n, the engine currently keeps a lot of backtracking points for the quantifier on \N* 22:07
markjreed ah. "(It also works to say %x<<foo>> as long as you realized it's subject to interpolation.)"
pmichaud markjreed: briefly, %h<x> is the same as %h{<x>} 22:08
and %h<<foo>> is the same as %h{<<foo>>}
markjreed S02. So my $foo="bar baz"; say %x<<$foo>>; should translate to say %x<bar baz>
pmichaud markjreed: correct. I think <<$foo>> isn't working in rakudo at present. 22:09
jjore_ fwiw, I think the perl5 engine ran against *all* the files in approx 3 seconds.
pmichaud rakudo: my $foo = "bar baz"; say <<$foo>>.perl
p6eval rakudo 278366: OUTPUT«"\$foo"␤»
pmichaud jjore_: that doesn't surprise me
jjore_ but thanks for the info. I don't know what gyrations the rakudo re engine is doing to make that happen
pmichaud jjore_: rakudo's engine is still fairly slow, and we get hit hard by gc performance issues in Parrot
markjreed ah, ok. totally apart from hash subscripting, Texas quotes aren't interpolating.
22:10 rgrau__ joined
pmichaud Texas quotes do interpolate. 22:10
markjreed ..but not in current rakudo.
pmichaud right.
markjreed We're violently agreeing, then. :)
pmichaud I wonder how hard it would be to re-implement that.
tadzik g'night #perl6
22:10 tadzik left 22:11 rgrau_ left
ingy blogs.perl.org/users/ingy_dot_net/2...ready.html (updated with my photo!) 22:12
pmichaud ingy: I enjoyed the post :)
ingy I <3 my hairdo
pmichaud: \o/
ingy buggers off to find TimToady 22:13
pmichaud has it been tweeted yet?
ingy pmichaud: should I tweet it?
pmichaud hugme tweet rakudoperl Ingy writes about Rakudo Perl: blogs.perl.org/users/ingy_dot_net/2...ready.html
hugme hugs pmichaud; tweet delivered
pmichaud it's tweeted now :)
ingy \o/ !!! 22:14
pmichaud++
ingy really buggers off to find TimToady
au|zzz ...and retweeted :)
22:14 au|zzz is now known as au|irc
pmichaud perlbuzz tweeted it also 22:15
au|irc ingy++ # lovely writeup
22:15 su_penguin joined
sbp "If your expection of Rakudo * includes blazing speed..." 22:15
expectation? 22:16
su_penguin Hello all. :)
jjore_ that 22:17
er,
ingy, (making fiends)++
tylercurtis rakudo: our $ingy += [+] 1..*; # for the Acmeist Perl 6 Regex plans. 22:19
p6eval rakudo 278366: ( no output )
markjreed Python nicer than Perl6? Heresy! :) 22:20
22:23 felliott joined 22:24 tylercurtis left, timbunce joined
ash_ has anyone made a camelia favicon? 22:25
ingy has located @TimToady / 2; 22:26
au|irc: thanks!
au|irc :)
ingy sbp: fixing... thanks 22:27
sbp :-)
22:27 rblasch_ left 22:28 justatheory joined
markjreed How can an Acmeist like a language whose motto is "There's Only One Way To Do It"? :) 22:28
ash_ rhetorically?
markjreed Perl6.org seems to have a Camelia favicon, ash_. 22:29
jnthn ash_: svn.pugscode.org/pugs/misc/camelia32.ico
ash_ ah, thanks 22:30
pmichaud oh dear
twitter.com/obra/status/18969296509
and reply at twitter.com/OReillyMedia/status/19000614690
ah well, "rise above" and all that. 22:31
22:32 mce_ro left, gbacon joined
ingy sbp: fixed 22:32
jnthn :/
pmichaud obra++ for nicely bringing it to oreilly's attention
ingy markjreed: their motto is a lie :)
markjreed Typical, really... 22:33
It's representative of an attitude, though. As the community has grown that attiude may have become less prevalent... 22:34
pmichaud I wonder how many (if any) lightning talks will be created based on that subject :) 22:35
jjore_ pmichaud: said earlier that \N*:\n could be more efficient than \N*\n. I didn't realize : was a metacharacter - a cut. Is it really? 22:36
22:36 felliott left, pyrimidine left
pmichaud it is 22:37
jjore_ woo
pmichaud it prevents backtracking into the previous item
jjore_ that must be why I wrote \: in the original regexp
markjreed Everyone wanted the colon, but the regex engine got it? :)
pmichaud well, all punctuation is meta in regexes
it's still likely to be slow, but you can save a *bunch* of gcables by not preserving backtracking points when they aren't needed :) 22:38
jnthn The regex engine really got the colon. There :, :: and ::: :-) 22:39
pmichaud for example, I would write
Local\ Variables\:\N*\n
as
'Local Variables:' \N*:\n
markjreed ok, outta here.. 22:40
22:40 markjreed left
pmichaud regrets that he did not bring his Windows flashcard with him 22:41
now I might have to learn/use gimp
jnthn Odd, I never have to remember one of those. ;-) 22:42
jjore_ thanks pmichaud. I'm retrying w/ your suggestions on doio.c which was approx 1 minute before.
pmichaud jjore_: I'm curious to see how much of a difference it makes.
jjore_ Now seeing 20ish seconds
jnthn Wow.
jjore_ for try in $(seq q ...);do...
jnthn pmichaud: ooc, does backtracking etc result in quite a lot of method calls too? 22:43
pmichaud jnthn: not so much
jnthn OK
pmichaud actually, it does
those are method calls I've been meaning to inline
yes, each backtracking point results in a method call at present.
jjore_ ok, well, re-running against the whole set. 22:44
pmichaud the code is designed so that I can inline some of those calls easily
jnthn OK
Grammars are one case where some of the stuff I'm working on for method dispatch performance improvements could *really* win us a bunch. 22:45
pmichaud very much agreed
jnthn Because all the dispatches are on the invocant.
And we know the type of it.
pmichaud I'm also very eager to re-look at sorear's work on dfa matching
jnthn pmichaud: Oh, that's one question I wanted to ask.
pmichaud: Are you using Multiple Inheritance anywhere in the grammar stuff?
pmichaud jnthn: I'm not aware of any off the top of my head. 22:46
jnthn OK
pmichaud if I am, it's likely easy to work around.
jnthn OK
I'm not saying we'd not support it there
jjore_ just eyeballing the times that are coming back, the program is seemingly 3x faster on all of the input files now
jnthn So much as it makes some opts harder.
jjore_ Just watched config.h which was 3m58.626s come in at 1m27.769s. 22:47
pmichaud jjore_: it may be faster still (and you can avoid the cut marks) if you put :ratchet right after the :i
oh, on second thought, probably not faster
jjore_ the re is already written to be amenable to only forward matching so I : cut everywhere I previously * quantified 22:48
pmichaud right
in that case you might want to look at "token" at some point.
token assumes forward matching (i.e., every expression has an assumed 'cut' after it) 22:49
:ratchet does the same thing
i.e., it says "assume ':' after each expression that follows"
jjore_ I may have made the program incorrect though. I said [ \N*:\n ]*:? but that is the one place which needs backtracking 22:51
pmichaud if backtracking is needed there, just don't put the colon 22:52
22:53 Patterner left
jjore_ yes, just went back to omit it. I had to tell #perl6 though because now I can't trust the 3x prior benchmark 22:53
it was kind of a fabulous number
pmichaud sure, make sense
*makes
afk for a bit
PerlJam should the R* dist contain the perl6 book?
pmichaud I think a reference to it is sufficient. 22:54
jjore_ oh that's nice. maybe approx 3x is still reasonable given config.h timing 22:55
22:55 stkowski left
ingy I need a name for the Acmeist P6R project. I was going to use pegex, but there's already a Scala based thing called Pegex on github! 22:56
22:58 Psyche^ joined
au|irc ingy: having a relatively obscure Paragraph Formatter named "par" never really stopped me coopting that name :) 22:58
22:58 Psyche^ is now known as Patterner
au|irc "pegex" is too good to pass :) 22:58
and just maybe kmizu can join your forces since you seem to be aiming the same goal... 23:00
23:00 sftp joined 23:10 ruoso left 23:11 dolmen left
sorear \o/ it's au|irc 23:14
au|irc o/ sorear :)
sorear has 2 unrelated programs named 'monop' in $PATH
jjore_ K, so like, embed.h is 2.77x. op.c is 2.64.
sorear jjore_: that looks like you meant #p6p 23:15
rokoteko ingy: f33l fr33 t0 l1st3n t0 my 3lit3 t4lk :)
sorear #p5p
rokoteko ingy: that mean I assume you are good. not only because someone named your nick is also good ;)
23:16 pnate left, colomon joined
sorear ingy: out of curiousity, where did that nickname come from? 23:16
23:18 mberends left
jjore_ sorear: er? I'm editing my perl5 code with a rakudo program 23:19
23:20 pnate joined
colomon jnthn++ # for posting my status report to #phasers 23:20
jnthn colomon: Thanks for thinking to send it along.
colomon: It was a good reminder of the grepgrepbug. 23:21
colomon it was a last minute thing before we left.
jnthn :-)
colomon and it would have been hell to type on my phone. :)
jjore_ very originally I wrote a perl5 program to edit perl5 but then for giggles ported it to ruby and perl6 for the experience points
sorear jjore_: Yes, I just woke up and didn't backlog far enough 23:23
jjore_ oh yeah, and I did it /last/ year in July 09 so it made a bit of retrospective to see if anything was different 23:24
from the perspective of that program, nothing seemed to have changed 23:25
The one 'TODO' I didn't try undoing was I'd originally tried a regexp back reference which didn't work. I tried all the others though. 23:26
Oh, though apparently s/// works but I didn't guess my way into working syntax so I kept .substr
jnthn jjore_: If you came back a year later and we didn't really break your program, that's at least sort of promising. :-) 23:27
pmichaud actually, surprising. :)
colomon very promising! :)
jnthn s/// should work 23:28
pmichaud afk for a short while
23:30 irclogbot_backup joined, irclogbot_backup left
jjore_ well, "Unable to parse postcircumfix:sym<{ }>, couldn't find final '}' at line 18" at diotalevi.isa-geek.net/~josh/100720/woota2.txt 23:32
23:32 mberends joined
sorear Would it be acceptable for *.perl to return a Cat? 23:34
23:35 su_penguin left
jnthn std: s{ x }{ "y" } 23:36
p6eval std 31776: OUTPUT«===SORRY!===␤Unexpected block in infix position (two terms in a row, or previous statement missing semicolon?) at /tmp/09lJeqD7m6 line 1:␤------> s{ x }⏏{ "y" }␤ expecting any of:␤ infix or meta-infix␤ quantifier␤Parse failed␤FAILED
..00:01 117…
jnthn std: s{ x } = "y"
p6eval std 31776: OUTPUT«ok 00:01 117m␤»
jnthn jjore_: Like that.
If you're going to use the bracketed form.
jjore_ but I don't know the contents of "y" until I get some capture variables
jnthn jjore_: The RHS there is actually closure-ized. 23:37
So you should be able to refer to them.
rakudo: my $x = "l0l1l2l"; $x ~~ s{\d} = 2 * $/; say $x 23:38
p6eval rakudo 278366: OUTPUT«Could not find sub &s␤ in main program body at line 22:/tmp/aI2KU7h4eu␤»
jjore_ Ok, refresh. However the eror is still "Unable to parse postcircumfix:sym<{ }>, couldn't find final '}' at line 18"
23:39 Kodi joined
jnthn rakudo: my $x = "l0l1l2l"; $x ~~ s[\d] = 2 * $/; say $x 23:39
p6eval rakudo 278366: OUTPUT«Use of uninitialized value in numeric context␤l0l1l2l␤»
jnthn :/
OK, it's still broken, it seems :-(
I know we fixed .subst recently
jjore_ So then $ostr ~~ s{...} = ...; *isn't* working then? 23:41
I can keep using .substr. It's ok.
jnthn jjore_: Seems not. :-( 23:42
Sorry, thought it was.
ingy jjore_: get used to it :)
jjore_ I would just prefer to adjust my syntax if that's all that prevented it
ingy it's The Way of the Camel
jnthn "Spits occasinally."
ingy :)
jnthn No point getting the hump though. :) 23:43
sorear it's awesome having enough foundations in niecza down that I can actually hack setting code in Perl 6 now
jnthn :) 23:44
Kodi phenny: tell masak gist.github.com/483792 I actually think I got the day-name stuff right the first time. At any rate, this time, DateTime-strftime.t works fine for me. If you get that problem again, try double-checking that you're using the right strftime.pm (or strftime.pir). 23:48
phenny Kodi: I'll pass that on when masak is around.
Kodi
23:51 cdarroch left 23:52 Kodi left 23:54 [Coke] left 23:56 amkrankruleuen joined
amkrankruleuen Hej :p 23:56
jnthn hej hej
23:56 Psyche^ 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!