»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.
Tene "Some of it's still broken, and a lot of it isn't done yet, but this is something that you can see, and try, and maybe use for some things. We use it ourselves, and want to share both the ideas and the product. We hope you like it, and hope you'll submit bug reports." 00:01
That's what I see it as.
masak Tene++
masak heads bedwards
00:02 masak left
lue goodnight masak o/ 00:02
E1SECONDLATE
pmichaud what are we using it for?
what are the key ideas that we think are particularly worth sharing?
lue We code rakudo with it :)
$_ = P6 of course
pmichaud yes, but what would we share with people who want to use p6, as opposed to create an implementation of p6? ;-) 00:03
lue well, everything in proto/pls , for starters..
00:03 masonkramer joined
pmichaud no, I mean as far as features go. 00:04
what do we think would entice people to take a look at Rakudo Star?
or at least make them say "oh, that's really cool. I should take a look someday."
TimToady p5 people, or normal people?
pmichaud normal, in this case
lue *thinking*
pmichaud well, open-source leaning people if nothing else :) 00:05
Tene ARGV -> signature on MAIN -> usage statement
makes for nice little things on the command line
lue Well, meta/hyper/reductions ops [come on brain, more]
pmichaud grammars and parsing, obviously.
Tene Hm, 'sec, lemme ask my interested-but-not-yet-using friends.
diakopter everyone (Perl and non-Perl folks) will of course demand to know when a "stable release" is expected, and as usual an unqualified "patches welcome" response will raise hackles and spark derision; I recommend addressing that question, and carefully.
pmichaud diakopter: sure, I'm addressing that. got lots of stuff on that question. 00:06
diakopter oh
pmichaud diakopter: but even after I address that question, there's still the question of "why is it important?"
Juerd "When it's ready" and "patches welcome" are good ways to get rid of your audience, possibly for good, indeed.
pmichaud Rakudo Star is all about saying "it's ready for some people". 00:07
and also "still needs work"
but that's not the part I'm struggling with.
(or just now addressing)
to a larger audience, the question is "what should Perl 6 (and Rakudo Star) mean to me?"
diakopter how does it fit into the ecosystem of PLs
pmichaud yes, some people will see "not stable" and say "okay, not for me." That's fine.
diakopter so you need a PL framework or category system in which to place it 00:08
pmichaud diakopter: just need to be able to describe it in terms of something they're already likely to know
Juerd pmichaud: The future of multiparadigm programming, in an actually usable state. This piece of software is going to the next language all the other languages are going to steal from :)
pmichaud Juerd: that's a good meme.
lue a lot of code is greatly reduced to one line due to the insane number of operators.
Juerd So if you value your job or hobby, you will *need* to know about Perl 6. 00:09
Tene The metamodel is pretty great. OO in Perl 6 is very nice.
Juerd Given the pace of open source development, waiting until a stable release is out means you'll be having waited too long.
TimToady "Everything you liked about your other languages except for the bad parts." :) 00:10
pmichaud "A language with completely new bad parts." :-P
Juerd TimToady: People do tend to like the bad parts though :)
pmichaud scnr
if others come up with ideas, just post them to the channel 00:11
I think I'll try to draft something tonight that we can react to and edit
Juerd pmichaud: Frankly, perhaps the best way to convince people nowadays might be "because you can" 00:12
00:12 lestrrat is now known as lest_away
Tene I expect that grammar and OO stuff is going to be the most-visible. 00:13
pmichaud and strong typing
Juerd At 600 CPM my typing is already pretty strong :)
pmichaud Juerd: it's even stronger in perl 6 :)
Juerd pmichaud: Wow, isn't that great!
pmichaud your 600 cpm go a lot farther in perl 6 :)
diakopter Perl 6: The densest Perl yet 00:14
TimToady Perl 6, Almost a Black Hole. :)
Juerd "Everyone loves six"
"We're open 42/7"
pmichaud TimToady: I'm sure you've been asked this question before, though -- what do you emphasize to people who ask "Why Perl 6?" Or does nobody dare to approach you with that question anymore? ;-)
TimToady "It'll take the other languages ten years to catch up, and then we'll be another hundred years ahead." :) 00:15
well...
"because it's not there" except is almost is now
lue
.oO(P6 will been stolen from. We know, we went to future with it.)
 00:16
Juerd Heh
lue can't wait for the day for Rakudo to run itself :) 00:17
diakopter er
Tene lue: looked at yapsi?
sorear (people LIKE the Perl 6 metamodel?!)
lue yes I have. I'm talking about doing something like perl6 perl6.pl6 :) 00:18
jnthn lue: Same. :-)
lue: Some hacking to go before that though. ;-)
lue I type perl6, and it runs a shell script containing "perl6 perl6.pl6"
pmichaud sorear: I like the Perl 6 metamodel. A metamodel is good when you don't have to think about it to use it.
lue (Yes it's recursive, I'm waiting for quantum computing) 00:19
sorear quantum computing doesn't help with recursion
at all
pmichaud it's boxes with (dead|alive) cats, all the way down.
lue just kick the computer, and random events make it work! \o/ [so excited] 00:20
sorear pmichaud: I come from Moose land, the metamodel there is far better
jnthn sorear: Don't confuse the current implementation with how it's eventually meant to be.
lue (R* is quantum computing ahead of schedule)
TimToady though I would be interested to see a list of Moose's betternesses 00:22
jnthn Same. :-)
Rakudo's meta-model guts are in for a significant, well, gutting, though. :-)
lue rakudo: my $a = 2; pir::setprop('rw',$a); pir::setprop('ro',$a); say "$a is OK!"
p6eval rakudo bbacb7: OUTPUT«===SORRY!===␤The opcode 'setprop_p_p' (setprop<2>) was not found. Check the type and number of the arguments␤»
jnthn pmichaud: s/strong typing/gradual typing/
pmichaud setprop wants three arguments 00:23
00:23 dduncan joined
jnthn pmichaud: The combination of static and dynamic is the win here, with specifying more static stuff giving you more wins. 00:23
(on performance and checking)
but the dynamicness letting you just hack stuff up
pmichaud jnthn: +1 00:24
jnthn I guess in a sense, Perl 6 gives you a language you can prototype in, and then easily refactor towards production code, with potential to be safe and fast.
TimToady you can scale from quick hack to enterprise without hitting a brick wall
lue rakudo: my $a = 2; pir::setprop($a,'rw',1); pir::setprop($a,'ro',1); say "$a is OK!"
p6eval rakudo bbacb7: OUTPUT«===SORRY!===␤The opcode 'setprop_p_p_p' (setprop<3>) was not found. Check the type and number of the arguments␤»
pmichaud +1
jnthn Also, Perl 6 re-visits a lot of assumptions that other languages have just taken for granted.
lue such as? 00:25
pmichaud also, Perl 6 is designed to be future-compatible
jnthn "Why can't I just write if $foo == 1 | 2 { ... }?" In Perl 6, you can.
pmichaud or even better
"Why can't I just write if 1 < $foo < 2 { ... }"
TimToady such as "Obviously if I wanted a parser I'd use a parser generator like yacc"
jnthn That also.
lue (Perl6: Do whatever the hell you want.)
pmichaud we even have coded the notion of "whatever" :-P 00:26
lue we also feature more pronouns than any spoken language :)
TimToady such as, obviously the catch blocks go after the block in question 00:27
00:27 isBEKaml left
TimToady such as, there is only one dispatcher 00:27
00:27 eternaleye_ left
jnthn I find the combination of signatures and functional-style pattern matching quite exciting too 00:28
lue afk
pmichaud multidispatch.
TimToady such as, assuming that types are only good for strong typing 00:29
jnthn In many senses, in Perl 6 I find myself writing heavily multi-paradigm code without actually realizing I'm writing multi-paradigm code.
Because they feel like a coherent whole rather than just a bunch of features.
sorear one thing that I still haven't totally figured out how to do in Perl 6 is unboxing
you might annotate all your functions as working on Int, but that's not enough information to use native types, because someone might subclass Int 00:30
jnthn Int is not a native type though
sorear (also, our types have 2^N + 1 values)
no. it's not.
replacing non-native types with native ones is one of the most important optimizations in HLL compilers 00:31
00:31 eternaleye joined
sorear what should this do? 00:31
(more)
sub a { my class B { }; my &c = BEGIN sub { for ^4 -> $d { augment class B { method e { $d }; }; }; }; my $f = B.new; c(); say $f.e }; a; 00:33
00:40 eternaleye_ joined, eternaleye left 00:45 simcop2387 left 00:46 skids joined
jnthn sorear: I'm not sure what the BEGIN really changes in that. 00:46
(in that we don't run the sub until runtime anyway) 00:47
(and iiuc the augment takes place a BEGIN time or so with or without it.) 00:48
Tene jnthn: how does augment take place at BEGIN time if the sub isn't run intil the end? 00:49
c();
jnthn Tene: Same way that sub foo { class A { } } results in class A existing even if foo never gets run 00:50
Tene ah
jnthn iiuc what TimToady mentioned the other day, my classes are no different from our ones, other than they get installed in the lexpad. 00:51
And not in the package
It'd just be anonymous classes that are interesting.
Tene nodnod
jnthn needs sleep 00:53
night o/
sorear putting BEGIN on a sub prevents it from being cloned 00:54
00:54 lkk- joined 00:58 simcop2387 joined 01:04 lest_away is now known as lestrrat 01:06 ashleydev joined 01:07 eternaleye_ left
lichtkind good night 01:15
01:15 lichtkind left, elmex_ joined 01:16 elmex left, elmex_ is now known as elmex 01:29 masonkramer left
pmichaud hugme tweet rakudoperl Rakudo now has autovivifying arrays and hashes 01:45
ENOHUGS 01:46
01:49 eternaleye joined 01:50 lkk- left 01:54 kid51 joined 02:00 buubot joined
lue hugs pmichaud manually 02:04
02:05 Chillance left, buu joined
lue I'm wondering, 02:06
> my @a = 1,2,3; @a[3..5] := @a[0..2]; say @a.perl
[1, 2, 3]
is this a binding error or a slicing error?
pmichaud I doubte that you've implement list binding. :-) 02:07
also, note that @a[3..5] is Nil.
Also, I doubt there's a list-binding operator. 02:09
(I could be mistaken on this last point.)
lue if := doesn't do it already, it should (maybe invent a :::= if neede :D) 02:10
ooh! let's see if this works...
> my @a = 1,2,3; [:=] @a[3..5], @a[0..2]; say @.perl 02:11
Lexical 'self' not found
pmichaud rakudo: say [+] [1,2,3]
p6eval rakudo 55acd1: OUTPUT«3␤»
pmichaud rakudo: say [!=] 1 02:12
p6eval rakudo 55acd1: OUTPUT«1␤»
lue well, assign.pir checks for nil assignment, maybe := should check for nil binding?
pmichaud assign.pir checks for nil being assigned as a value to a target 02:13
in @a[3..5] above, Nil *is* the target.
lue ah, well, nvm then :)
pmichaud (and rakudo doesn't implement it correctly yet anyway)
lue rakudo: my @a = 1,2,3; say @a[3..5].WHAT
p6eval rakudo 55acd1: OUTPUT«Parcel()␤» 02:14
lue rakudo: my @a = 1,2,3; say @a[3..5]
p6eval rakudo 55acd1: OUTPUT«Any()Any()Any()␤»
lue What interests me is that binding to Nil doesn't die, but it doesn't work either. 02:15
pmichaud I did just say that Rakudo doesn't handle it correctly yet. 02:16
and by "doesn't work", I don't know what "does work" would mean. :-)
lue Ah.
EAMBIGUOUSSPEC 02:17
02:20 lkk- joined
lue Heh, when it comes to getting := to do what I want, it feels like dividing by 2. I keep getting closer to 0 each time, but never quite there :) 02:22
> my @a = 1,2,3; @a[$_+3] := @a[$_] for 0..2; say @a.perl 02:23
[1, 2, 3, 1, 2, 3]
not the best way, but it works.
02:24 tylercurtis joined 02:25 sftp left
lue I get the feeling there's no easy way to deal with Nil being the target in binding right now ( esp. considering it doesn't work right :) ) 02:28
02:29 skids left, duzy` joined 02:30 olgagirl joined 02:31 tedv joined 02:32 olgagirl left 02:35 agentzh joined 02:39 plobsing joined 02:44 sorear sets mode: +v buubot
dalek kudo: f8dde0a | pmichaud++ | docs/spectest-progress.csv:
spectest-progress.csv update: 507 files, 33927 (84.4% of 40197) pass, 19 fail
 02:45
sorear oh
apparently hugme is supposed to run on feather*2*
sorear has root on feather1 02:46
sorear pokes moritz_
ok, I *do* have an account on feather2 02:47
but I don't seem to have a sudoers line 02:48
02:57 lkk- left 02:58 ashleydev left 02:59 eternaleye left 03:15 ash_ left 03:21 kid51 left 03:22 eternaleye joined 03:38 skangas joined 03:39 eternaleye left
dduncan with all this talk about how Parrot is supposed to support one hosted language invoking another, are there any existing examples of this, where both languages are Parrot hosted and both languages existed before Parrot? 03:41
PerlJam dduncan: you mean like calling ruby from perl or vice versa?
dduncan yes
or to make this a little more on topic, any examples of Perl 6 code invoking code that is neither part of the Parrot toolkit (NQP, PIR, etc) nor is Perl 5? 03:44
examples that run
03:45 ashleydev joined
PerlJam Someone (jnthn?) had some examples like that at one point, but I don't know where they are right off. 03:45
dduncan okay
PerlJam Ah ... this is what I was thinking of blogs.gurulabs.com/stephen/2009/05/...ading.html 03:46
Tene That's me.
PerlJam yeah, sorry to confuse you with jnthn ;) 03:47
Tene It did work in the past, and then things got changed out from under me twice, and I gave up on trying to keep it maintained. I hear that it's working to some degree again these days, but afaik none of the langauges other than rakudo support it anymore, if even rakudo does, so it's not really useful.
PerlJam: Quite honored, actually. Maybe you should apologize to jnthn instead. :)
PerlJam heh
dduncan any thought as to whether a modern analogy might be working in time for Rakudo Star? ... specifically of Perl 6 calling other languages 03:48
[Coke] I don't think that's on the short list. 03:49
Tene dduncan: I expect it'll only happen if an HLL Interop Champion shows up to drive it. That was my job for quite a while, but I haven't been available in any reasonable way for quite a while. 03:50
[Coke] (not that I'm someone who'd know. =-)
Tene It shouldn't be that much work for someone comfortable with parrot-hosted languages, but it's not going to happen unless someone who cares shows up to do the work and get the other languages in line, etc. 03:51
dduncan well maybe I'll work on it then ... in a few months
Tene That'll be after the first release of rakudo start, but it would still be great to have someone working on it. :) 03:52
dduncan since I'm creating a Parrot hosted language, I'd like to call it from Perl, and perhaps vice-versa
Tene I'd be more than glad to try to guide you through it.
dduncan thank you ... anyway, my schedule has it that I don't expect to start before October, but that isn't too far off
Tene If it turns out to be a pressing feature that needs work before then, that would probably mean that someone else now cares enough to work on it, so that's probably fine. 03:53
03:54 hejki joined
dduncan on a tangent, something I'm wanting to start working on in August is a test suite for my new language which is intended to be used to test multiple compilers of my language, and I'll be looking at the Perl 6 suite for guidance 03:55
Tene Certainly not a bad place to look.
dduncan I also intend to be working on no fewer than 4 distinct implementations ... or maybe 3 ... one over Perl 5, one over Perl 6, one as a native Parrot hosted language, and one as a Postgres procedural language ... though probably the last 2 can be one and the same 03:56
so its useful to have a common test suite from day one
the first 2 are expected to be easier as they'll just generate Perl code that does the same thing, and the Parrot one is expected to be the "real" one 03:57
where I have to do more work myself
similarly, I'd be following the Perl 6 example of making much of the language self-hosting, so I only have to do it once for all implementations, by default 03:59
that is, mostly self-hosting 04:00
04:03 ashleydev left
dalek ecza: af32687 | sorear++ | (4 files):
Implement the possibility of metaobject cloning
 04:04
ecza: 81429e4 | sorear++ | setting:
Revise metaobject API for new guts
04:05 Guest99304 left 04:07 tedv left 04:23 quietfanatic joined
quietfanatic How much of Perl 6 grammars can I currently use in Rakudo? 04:23
because I'm getting some inexplicable syntax errors. 04:24
Wait, no I see the error 04:25
sorear quite a lot of them - enough to parse Rakudo Perl 6 itself 04:26
dduncan that's good 04:28
quietfanatic Cool
How do I parse a file with a grammar?
I'm trying slurp($file) ~~ grammar_name 04:29
but it's returning 0
tylercurtis Grammar.parsefile(filename, :actions(Actions.new))
I think.
quietfanatic (where grammar_name contains a rule TOP {...})
That's a little...verbose
but I'll try it.
wait, it requires actions of some sort? 04:30
sorear no
tylercurtis No.
sorear named args are always optional
(in standard functions)
quietfanatic so just parsefile, eh? I was hoping smartmatch would be smart enough for that
sorear a grammar is fundamentally a kind of class. slurp($file) ~~ grammar_name is an isa test
also, grammars are intended for fairly complicated languages 04:31
if your grammar is small enough that you care about saving ten characters, just use a regex
quietfanatic okay, I see. 04:32
That makes sense, actually.
tylercurtis Alternately, if you really want to use slurp, Grammar.parse(slurp($file)). That's probably what Grammar.parsefile does internally, anyway.
TimToady well, it also sets the filename for error messages 04:33
quietfanatic Just putting the rules bare in the file gives a bunch of "useless declaration of has-scoped rule..." warnings
and then can't find the rule.
TimToady grammar {...}
quietfanatic I guess I have to my them all.
TimToady grammar MyGrammar {...}
just put them in a grammar declaration 04:34
quietfanatic So then I'll use a grammar after all.
TimToady hard to have methods without a class
quietfanatic It's not returning true even with a trivial rule. 04:35
rule TOP { ^ .* $ }
TimToady rakudo: grammar Foo { rule TOP { ^ .* $ } }; say Foo.parse("stuff").perl 04:36
p6eval rakudo 55acd1: OUTPUT«Match.new(␤ # WARNING: this is not working perl code␤ # and for debugging purposes only␤ from => 0,␤ orig => "stuff",␤ to => 5,␤)␤»
TimToady that matched 04:37
04:37 duzy`` joined
quietfanatic Well my file is not matching. 04:38
I've tried parsefile and parse(slurp())
PerlJam quietfanatic: you can always nopaste the code and input
04:39 duzy` left
quietfanatic Alright, let's see to that then... 04:39
TimToady you're not trying to run it from the command line are you?
quietfanatic wait where's the nopaste?
TimToady: Of course, why would that matter?
TimToady shell quoting and interpolation 04:40
quietfanatic Oh, no, I misunderstood what you meant.
My grammar is in a file.
say slurp(@*ARGS[0]) works, but grammar.parse(slurp(@*ARGS[0])) does not. 04:41
PerlJam quietfanatic: I tend to use gist.github.com
sorear quietfanatic: rule TOP { ^ .* $ } is not going to match anything that starts with whitespace 04:43
TimToady troo
s/rule/token/
sorear use "token" unless you actually want the spaces in your pattern to be taken semi-literally 04:44
quietfanatic . doesn't match whitespace?
sorear " " matches \s* possessively
TimToady rules don't backtrack, and ^ won't match the beginning if the implicit <.ws> matched first
sorear "^", seeing that it's not at the beginning of the string, fails to match
quietfanatic Oh fooey 04:45
sorear and drop the ^ and $. grammars, unlike naked regexes are always anchored
quietfanatic I see.
Well, that fixes it. 04:46
I do remember that 'rule' means sigspace, but I never figured out why 'rule' does and 'token' doesn't
in terms of the word choice I mean.
Why 'rule' for sigspace? 04:47
TimToady tokens are typically used to match small bits that exclude whitespace, and rules are typically used to call into subrules that want to marshall the matching of multiple "words"
quietfanatic I see, but 04:48
TimToady of course, when you match something like an expression in parens, the parens are a token on the outside, but parse a rule on the inside
quietfanatic there are times when I want to disable sigspace in a regex, and then it looks odd having a token contain rules. 04:49
TimToady happens all the time for any kind of circumfix
quietfanatic I just don't think the terminology matches the meaning well.
The instance this time is matching a closing } with the same indentation as the opening { 04:50
sorear you are free to turn off sigspace in the middle of a rule
quietfanatic adding sigspace makes that not work.
sorear rakudo: say ("a b" ~~ rule { a b }) 04:51
p6eval rakudo f8dde0: OUTPUT«a b␤»
sorear rakudo: say ("a b" ~~ rule { a:!s b })
p6eval rakudo f8dde0: OUTPUT«␤»
sorear rakudo: say ("ab" ~~ rule { a:!s b })
p6eval rakudo f8dde0: OUTPUT«ab␤»
sorear or simply
rakudo: say ("a b" ~~ rule { ab })
p6eval rakudo f8dde0: OUTPUT«␤»
04:52 TiMBuS joined
quietfanatic I'll just stick with 'token 04:52
'
tylercurtis It matches it to some extent, because many programming languages' grammars care about exact whitespace in their tokens, but only care about some amount of whitespace or no whitespace in their rules. And the distinction is pretty simple to remember once you learn what they mean in the context of Perl 6.
TimToady that's fine, especially if you're gonna be fiddly with the whitespace
sorear although sometimes I miss the phase separation in p6 04:53
quietfanatic I can see that *usually* token means token and rule means rule
but it bothers me that it's almost a literal meaning but actually is not at all.
...that was worded weirdly 04:54
I just don't like it that the terms don't mean exactly what they say they mean.
TimToady that's always a problem when you're dealing with concepts that don't map well onto the lexicon 04:57
quietfanatic I don't know, everything else in Perl 6 is usually pretty self-explanatory in its naming. 04:58
tylercurtis quietfanatic: No name is self-explanatory. A name can, at best, be self-explanatory with a certain background knowledge and in a certain context. Rule and token aren't quite self-explanatory to someone who knows what rules and tokens are in the context of grammars in general, but once you understand what they mean in the context of Perl 6, their association with the meaning of "rule" and "token" in general is a helpful mnemonic. 05:09
05:10 jaldhar left
quietfanatic That's a point, I suppose. 05:13
tylercurtis comb is similar; it's not a standard function across many programming languages, but it's a simple name that's easy to remember and the thought of "combing through" something is a helpful mnemonic.
Many of Perl 6's names for things are not things I find self-explanatory, but once I learn what they mean, they're easy to remember. 05:15
quietfanatic But comb doesn't obviously mean anything at first.
token and rule, in informal usage mean a small thing and a large thing
but in implementation they mean :!s and :s.
which is not the same thing at all. 05:16
TimToady you're missing an important distinction 05:18
quietfanatic So when I see the term token, I'll have to wonder whether they mean a small unit of parsing, or a regex without sigspace.
sorear I don't know where you come from, but here, in informal usage, a token is a small round metal thing representing a commodity
quietfanatic informal programmer usage, I mean. 05:19
TimToady precisely *because* tokens don't have sigspace, they participate as alternatives in lexers, and hence are tokens in the usual (technical) sense
leading whitespace tends to disqualify a rule from participating in a lexer
sigspace is merely a means to an end 05:20
or a not-end, in this case
05:20 skangas left
tylercurtis Really, in languages in which whitespace only matters to separate tokens, rules and tokens mean exactly what they should mean for that language. 05:20
quietfanatic Huh. I guess I must concede the point due to my head spinning too much.
TimToady even in English, the distinction between "word" and "phrase" is unclear 05:21
for instance, "the King of England's hat"
the 's wants a word on its left, so 'the King of England' functions like one temporarily 05:22
quietfanatic What bothers me isn't the ambiguity in distinction, it's that there's two possible distinctions that aren't the same.
TimToady if things like that bother you, either you should or shouldn't be a language designer. :) 05:23
quietfanatic Well, perl has always been a very interpretive language. 05:24
TimToady most words in natural language are very imprecise at the best of times
quietfanatic But the words 'token' and 'rule' in Perl 6 are very precise.
TimToady yes, and they're just about the best words we could find that match the abstract concepts 05:25
quietfanatic in their meaning, I mean.
TimToady and Perl 6 tends to emphasize the abstract meanings over the concrete
quietfanatic But their literal meaning isn't abstract in any way.
TimToady given/when vs switch/case
huh?
quietfanatic one means sigspacem the other doesn't
*,
TimToady that's its concrete meaning 05:26
but there are ramifications on the abstract level
quietfanatic whereas picking the terms 'token' and 'rule' overlays abstract meaning on it that I think is too confusing.
I'm trying to think of something it's like... 05:28
well my brain is not operating on analogy mode tonight. 05:29
tylercurtis I think TimToady is saying that when choosing a name for something, the pool from which you can draw(which mostly consists of natural language words, because no one would even have a clue what "figel Foo { zabez TOP { <statement> }; gix statement { 'bar' } }" meant without being told in advance, and they'd have difficulty remembering even then), consists of things with imprecise meanings that rarely correspond exactly to what you really mean.
05:29 skangas joined
tylercurtis E.g., the mathematical concept of a function rarely coincides with the general meaning of the word, and neither is identical to most programming languages concepts of a function. 05:30
quietfanatic tylercurtis: True. I just am of the opinion that in this case there is too much disconnect between the abstract and the concrete here.
TimToady about the best you can do is minimize the gratuitous overloading in a particular context; at one point "proto" meant about four different things in perl 6
quietfanatic Oh yeah I remember that :) 05:31
Well I'm not changing anybody's mind, I just wanted to cast my dissent.
tylercurtis Of course, there are some times when this can go wrong. E.g., structs in C++ not being what a C programmer coming to C++ would expect at all(normal objects, just with a different default visibility for members). 05:32
quietfanatic Well, normal objects are just structs underneath, right?
tylercurtis C structs don't have member functions, can't inherit(well, you can manually do something similar), will never have a vtable unless you explicitly implement one, etc. 05:34
Or constructors or destructors.
TimToady Perl stole BEGIN and END from awk and made them mean something rather different 05:35
ingy greetings
TimToady o/
ingy how do I findbin in rakudo/p6? 05:36
tylercurtis Anyway, I should be getting to bed. quietfanatic, why must you start interesting conversations that I get drawn into? :P Good night, folks.
sorear run("which")?
TimToady um, use FindBin? 05:37
quietfanatic gn tyler
ingy :\
oh?!
TimToady leastwise t/spec/S16-filehandles/dir.t uses it
ingy where do I findfindbin? 05:38
TimToady I don't know if rakudo runs it at all though
05:38 tylercurtis left
TimToady looks like there's one in ./ext/FindBin/lib/FindBin.pm that was probably used by pugs 05:39
yeah, I don't think rakudo will grok m:P5// 05:41
sorear did pugs grok m:P6//?
quietfanatic Are lookbehinds NYI in rakudo?
TimToady not really
ingy TimToady: where is this ext?
I don't see it in rakudo src
TimToady the upgs repo 05:42
*pugs
05:42 gabiruh joined
ingy do we use that anymore? 05:42
TimToady parts of it, but maybe not that part :)
still, should be pretty easy to port to rakudo 05:43
ingy like what parts?
TimToady like the specs and the tests all live in the pugs repo
so does STD
ingy hrmm
TimToady and several other implementations
rakudo: say "foo" ~~ / <?after 'oo'> / 05:45
p6eval rakudo f8dde0: OUTPUT«Method 'after' not found for invocant of class 'Cursor'␤ in <anon> at line 11:/tmp/5jXKpIznTR␤ in 'Cool::match' at line 2309:CORE.setting␤ in 'Regex::ACCEPTS' at line 5240:CORE.setting␤ in 'infix:<~~>' at line 401:CORE.setting␤ in main program body at line
..11:/tmp/5jXKpIznTR␤»
TimToady quietfanatic: it would appear not
sorear lookbehind has to be implemented at /some/ level for .ws to work
you might need to drop down to pir though 05:46
quietfanatic I'm not gonna try to fiddle with that then. 05:47
TimToady that only needs one char of lookbehind
ingy is github.com/audreyt/pugs up to date?
TimToady so probably cheats with substr or some such
quietfanatic Being unable to override ws makes writing a line-based grammar really hard.
sorear ingy: wrong pugs repo 05:48
that's the repository which contains pugs
we're talking about the repository called pugs
quietfanatic I'll have to switch to semicolon-based. That'd be more better anyway probably.
ingy sorear: url? 05:49
sorear yow, I broke STD
svn.pugscode.org/pugs/
Can't locate object method "_SUBSUMEr" via package "main" at Niecza/Grammar.pm line 1171.
TimToady how did you get an object blessed into main? 05:50
sorear I misspelled a :lang
and apparently CORE::bless thinks undef = PL_defstash
yay for wrong defaults
TimToady huh
dduncan in my mind's experience, writing a parser is a lot easier if we don't have to keep track of lines, and all whitespace outside string literals is the same 05:51
TimToady you'll note that in P6 the type is now the mandatory invocant of bless :) 05:52
ingy sorear: they pretty similar. is there a git mirror of the svn repo?
I remember svn checkout of pugs taking a very long time. :\ 05:53
dduncan by "same" I mean "differences aren't significant"
sorear ingy: yes but it's pretty stale iircx
TimToady dduncan: but the highest goal of writing a parser is not to make it easy to write, but easy for the user to use. Easy to write is nice as a bonus, of course. 05:54
anyway, even the languages that say "all whitespace is the same" often treat 0-width whitespace as significantly different from 1-or-more width 05:55
dduncan for my own purposes in writing a parser, the only useful thing about paying attention to which whitespace you have is to give useful error or diagnostic messages
sorear ingy: on the plus side, now that pugs is gone, checking out pugs is faster. (confused yet?) 05:56
dduncan by "the same", I mean different line-endings versus tabs vs spaces ...
I *do* consider the presence vs absence of whitespace significant
zero-width meaning absence 05:57
TimToady dduncan: on the other hand, vertical whitespace looks different to the user, so is psychologically valuable to distinguish sometimes
in particular, the desire for extensibility has driven some of the P6 whitespace decisions, such as no ws before postfixes, and } at end of line always ends the expression 05:58
and of course, Perl has had here-docs for ages 05:59
which very much distinguish \v from \h whitespace
dduncan on a tangent, I found it useful with my own language to format comment strings like other data strings, in that they have a non-whitespace delimiter at both ends, rather than just at the start and treating the end-of-line as the end delimiter 06:00
the Perl analogy would be having a trailing # in addition to a leading one
TimToady both forms can be useful
06:01 uniejo joined
TimToady and pod is also keyed to vertical transitions 06:01
dduncan yes 06:02
well, its an exercise in trade-offs ... I went for something that I perceived was 90% as easy to write by a user but was 5+ times easier to parse 06:04
dalek ecza: 1c21454 | sorear++ | (4 files):
Implement quoted identifiers, some more class stuff
 06:05
ecza: 68a676c | sorear++ | (4 files):
Implement ClassHOW injection and bootstraping (3rd time)
ingy TimToady: do you know about $*PROGRAM_NAME?
that's all I wanted
:\
Well I guess I learned a little along the way. 06:06
dduncan about strings, another design decision I made, which I'm not aware of any other language doing, is disallowing literal occurrences of the string delimiter within the string, even with a leading backslash ... rather the same char is represented with something else ... 06:07
as a result, when you encounter an opening delimiter, you know that the next occurrence is its pair, and you don't have to test for leading backslash or whatever
TimToady sounds like you're making it easier for you than for your user 06:08
06:09 lafille joined
dduncan well they still have to escape somehow, and I just have them escape differently 06:09
granted, Perl 6 has the matching braces thing, so often you don't need escapes of the same thing 06:10
but practically every other language doesn't do that AFAIK, and I'm comparing with them in this case
TimToady but how do they know what that other character is?
ingy: $*PROGRAM_NAME seems to work here, as long as it's in a file
dduncan first of all, there are only a small number to remember ... 2-4 ... and its documented 06:11
06:11 lafille left
TimToady what if your string has all of those? 06:11
dduncan mainly, the differences are single-quotes, double-quotes, and backslashes ... everything else looks as you're used to
except for the odd removal ... 06:12
sorear dduncan: you appear to be reinventing the Perl 5 quote-like operator parsing engine 06:13
read "Gory details of parsing quote-like operators" in perlop, spare yourself some time
dduncan actually, no
all Text literals use single quotes, period
all Name literals use double quotes, period 06:14
same as in SQL
so in a Text (Str of Perl 6)
only the single quote and backslash need escaping
other things may optionally be escaped, but don't have to
so no quote-like operator 06:15
TimToady why does the backslash need escaping if you're not using it to backslash single quote?
dduncan this is Text: 'hello world' ... Name: "Hello World" ... Comment: #hello world#
if you have a literal backslash 06:16
one moment ...
a relevant url: search.cpan.org/dist/Muldis-D/lib/M...g_Literals 06:17
sorear goes to read up on MROs
dduncan keep in mind, I'm going for now for a 90% solution, and my primary competitor is SQL 06:18
in a Text literal, a single-quote may be escaped as either \a (apostrophe) or the same \c<> format that Perl 6 has 06:19
and a backslash, with a \b 06:20
moritz_ Perl 6 uses \c[...]
dduncan these rather than \' and \\
yes, okay
I think I meant \c followed by a delimited unicode name or num
06:20 justatheory left
dduncan so, 'this isn\at something' is a Text literal 06:22
basically just the \a, \b etc are different from common practice, the rest is more the same
especially, the \t, \n, \f, \r are the same as convention 06:23
dalek ecza: ee41c0d | sorear++ | (3 files):
Don't pollute global syml/ with our data
sorear looks like someone needs to be exposed to nibblers.
moritz_: hugme is AWOL
dduncan \a or \b used to mean "alarm"/"bell" but I think hardly anyone uses those
sorear no, \b is backspace, not bell 06:24
(and I've used it)
dduncan sure, okay, but its used rarely isn't it?
sorear very much so
BinGOs has used
06:24 hugme joined
dduncan well one can still use the \c form for those, or even include them literally 06:24
sorear I still think you're actually making this more complicated when you think you're simplifying
Perl6 regex was designed, among other things, to make processing strings with escapes easy 06:25
dduncan sure, but I'm also targeting implementations not over Perl 6, in addition to Perl 6 06:26
TimToady recognizing \' is *just as easy* as recognizing \a, and isn\'t is a lot more readable than isn\at
dduncan well, maybe I'll change that, but at the moment I don't see this issue to be a problem in practice 06:28
TimToady even isn''t is better than isn\at
well, as long as you don't expect this to be used for English text, you'll be fine... :P 06:30
seriously though, one of the P5 things we ran away from screaming is multi-pass parsing, so I wouldn't feature it 06:32
it just comes back and bites you later
dduncan I certainly don't feature it
a key feature of Muldis D is that it can be single-pass parsed 06:33
TimToady then there's little point in excluding the delimiters from recognizable constructs in the middle
dduncan understanding the meaning of any character only requires having read the prior ones
you make a point there 06:35
TimToady what if Unicode in their infinite wisdom decides to add a character name containing one of your quote chars?
dduncan you mean, having something other than latin letters and spaces? 06:36
TimToady they already have parens in there
anyway, it doesn't buy you much to exclude them, and doesn't help readability, in my estimation 06:37
dduncan so, if I change to using \' and \\ etc then you would consider this to be a huge improvement in useability?
TimToady certainly a great number of C/C++/Java programmers will think so 06:38
dduncan fyi, while the language is single-pass parseable, I imagined it might be easier to do multiple passes in practice ... my current design could make the tokenization phase easier ... then again, that may be antiquated thinking 06:40
so I'll make this design change now to conform to more common practice and hope for the best 06:41
TimToady nods, probably off to sleep...
dduncan I'll just do the \' though, no '' ... that one looks like 2 strings side by side 06:42
thanks for the feedback 06:43
07:20 Mowah joined 07:25 Ross joined 07:54 mberends joined 08:04 plobsing left
sorear class Foo { } ; (BEGIN Foo.new).isa(Foo) # this = 0 according to my current understanding of perl 6 08:05
moritz_ why should it?
if that were the case, 1.isa(Int) would also be false 08:06
because 1 is kinda like Int.new(1)
at compile time 08:07
08:07 mikehh left
sorear well, Foo, as an our object, is instantiated more than once 08:07
first it's instantiated against the protopad, producing Foo#1
the BEGIN returns an instance of Foo#1 08:08
on each execution of the closure, Foo is bound to the current environment, producing Foo#2, etc
moritz_ that's the first time I've heard of the "protopad"
sorear so it's really (Foo#1.new).isa(Foo#2)
moritz_ why should the Foo's resolve to different objects? 08:09
I don't see any closure closing over Foo
it's not a variable
sorear because they close over different lexical environments
moritz_ ... and?
Foo is an 'our' symbol, as you've said yourself
sorear my $x; class Foo { method bar { $x } }
moritz_ why should it be different in different instances of a closure? 08:10
rakudo: class Foo { }; say (BEGIN { Foo.new}).isa(Foo); 08:11
p6eval rakudo f8dde0: OUTPUT«===SORRY!===␤too few positional arguments: 1 passed, 2 (or more) expected␤»
moritz_ rakudo: class Foo { }; say (BEGIN { Foo.new}).^isa(Foo);
p6eval rakudo f8dde0: OUTPUT«===SORRY!===␤too few positional arguments: 1 passed, 2 (or more) expected␤»
moritz_ rakudo: class Foo { }; say (BEGIN { Foo.new}) ~~ Foo;
08:11 cjk101010 joined
p6eval rakudo f8dde0: OUTPUT«===SORRY!===␤too few positional arguments: 1 passed, 2 (or more) expected␤» 08:11
sorear moritz_: what should (class { method foo { $_ } }).foo for ^2 return? 08:12
I assert that the answer is 0, 1
moritz_ sorear: I have no idea, but I don't see how it is related to the previous question 08:13
sorear classes close over their lexical environment, just like the methods they contain
moritz_ sounds kinda sane
08:13 pnate left
sorear moritz_: a class is a closure, so each time it is cloned it produces a new value 08:13
rakudo: my sub foo() { say "Hi" }; BEGIN { foo; } # If rakudo had working protopads it could run this 08:14
p6eval rakudo f8dde0: OUTPUT«===SORRY!===␤Could not find sub &foo␤»
sorear pugs: my sub foo() { say "Hi" }; BEGIN { foo; }
p6eval pugs: OUTPUT«Hi␤»
sorear that's what protopads are
moritz_ waitwaitwait
cloning a closure only clones the lexpad
not the entire thing
so there's still a a relation between the clones
sorear a relation, yes 08:15
but not ===
moritz_ I don't think .isa needs to consider the cloned parts, does it?
sorear I had taken it for granted .isa would use ===
but now that you say that - I am not so sure
thanks
moritz_ then it wouldn't work for (3 but True) ~~ Int 08:16
which would be kinda surprising
IMHO
but I think I'm not the right person to discuss it with
sorear well, (3 but True) has a class of all(__ANON__, Int, Cool, Any, Mu) 08:17
so it could match Int regardless
but - the cloned parts of a class or role affect only implementation, not interface, so there's no reason for isa and does to consider them
moritz_ all(Int, Any) ~~ Int is false
so it needs to do a bit smarter matching :-) 08:18
TiMBuS does Any ~~ Int work? 08:29
moritz_ rakudo: say Any ~~ Int 08:30
p6eval rakudo f8dde0: OUTPUT«0␤»
TiMBuS i guess that's the problem
moritz_ not really 08:31
it's a feature
TiMBuS of course
moritz_ I guess that changing all() to any() would do 08:32
TiMBuS rakudo: say any(Int, Any) ~~ Int
p6eval rakudo f8dde0: OUTPUT«0␤»
TiMBuS ahhh 08:33
moritz_ now *that's* a problem
08:33 thebird joined
TiMBuS is there a "great big table of smartmatching" to say what works and in what order? i recall seeing one but I think it was for perl 5 08:35
sorear right
perl 5 has a Great Big Table
perl 6 smartmatching is very simple
moritz_ TiMBuS: S03 has such a table
sorear A ~~ B = B.ACCEPTS(A) 08:36
moritz_ but it's not all implemented
TiMBuS I see
moritz_ sorear: not quite
first $_ is set to A
sorear A ~~ B = (let $_ = A in B).ACCEPTS(A)
moritz_ which explains how 'a' ~~ .uc is supposed to work 08:37
perlcabal.org/syn/S03.html#Smart_matching
08:38 pnate joined 08:39 masak joined
masak oh hai, #perl6 08:39
moritz_ lolitsmasak 08:40
fun question: what happens if you smartmatch against a Match object?
masak er. 08:41
moritz_ btw I'm kinda surprised that the fallback for smartmatching is === and not eqv
masak aye.
moritz_ since most cases of ~~ test structural things, not object identity
08:42 clintongormley joined
TiMBuS i think i've ran into this issue! and most of the time I wanted Match to be stringified 08:42
dduncan TimToady, sorear - I've now committed the design change on char escape seqs we discussed - utsl.gen.nz/gitweb/?p=Muldis-D;a=co...h=afd38e57 08:43
08:44 dakkar joined
dduncan ... and also the prior commit lays groundwork, since I had used double-backslash for something else which had to be altered first so not to conflict 08:44
moritz_ I'm curious, what did it mean before?
dduncan moritz_, was that question directed at me? 08:47
moritz_ dduncan: yes
dduncan okay, the old meaning was related to "unspace"
the prior commit - utsl.gen.nz/gitweb/?p=Muldis-D;a=co...=e25af94a2 08:48
essentially, I was using the same grammar token, a backslash pair surrounding optional whitespace, for 2 distinct purposes
one inside strings to split them over lines, and the other like Perl 6's unspace but without (initially) embedded comments 08:49
optional whitespace means a double backslash could be in a string literal
I split the concepts so that the one inside the literal is now called 'splitter' instead of 'unspace' and the whitespace is mandatory
08:50 Ross left
dduncan in the future I'll probably update my 'unspace' to have embedded comments, but I omitted those for now for simplicity 08:50
er, hm, just realized another thing I have to fix ... 08:53
08:53 meppl joined 08:54 thebird left
dduncan the "\ " in a string is now ambiguous ... 08:55
08:57 mikehh joined
dalek ecza: 7292187 | sorear++ | (3 files):
cleanup .method, implement dup@, start Mu
 09:05
ecza: 8c57de0 | sorear++ | (4 files):
Nicer syntax for boxing in NIL
ecza: 5135541 | sorear++ | (4 files):
Bootstrap more methods. Add integers to NIL. NEW CLASS: Mu
sorear it's finally starting to come together 09:07
dduncan now that's fixed - utsl.gen.nz/gitweb/?p=Muldis-D;a=co...;h=6756404 09:11
sorear out. 09:12
dduncan ditto
I mean ...
dduncan ditto
09:12 _mpu joined 09:22 mberends left 09:44 Ravi joined 09:45 Ravi left 09:50 quietfanatic left, masak left 09:54 dduncan left 10:03 lestrrat is now known as lest_away 10:04 jaldhar joined 10:10 thebird joined
jnthn Afternoon, #perl6 10:17
10:18 pmurias joined
moritz_ did you see the new don knuth talk? it's hilarious 10:19
pmurias moritz_: link? 10:20
moritz_ river-valley.tv/media/conferences/t...Don-Knuth/
and if that times out, try a mirror: onlineflashplayer.net/?url=http%3A%...-Knuth.flv
10:23 skids joined 10:27 Ross joined
pmurias YOU_ARE_HERE works in mildew :) 10:30
moritz_ /me IS THERE
10:36 dakkar left
hghgd "the bell is also part of the logo" :) 10:38
moritz_ :-)
10:46 agentzh left, dakkar joined 10:58 Trashlord joined 11:06 szabgab left, john__ left, literal left 11:09 szabgab joined, john__ joined, literal joined, ascent_ joined, mdxi_ joined, fda314925 joined, TimToady joined, farmer.freenode.net sets mode: +o TimToady
dalek p-rx: d01b490 | pmichaud++ | src/HLL/Compiler.pm:
Refactor interactive mode context handling slightly.
 11:14
p-rx: 2fb1606 | pmichaud++ | src/stage0/ (3 files):
Update bootstrap files.
11:14 Backglass joined 11:15 hghgd_ joined
Backglass who can help me using regex? 11:15
hejki prepares for collective facepalm 11:16
11:19 hghgd left 11:22 Guest7606 left, Guest7606 joined, Guest7606 is now known as pragma_ 11:25 wallberg joined 11:31 wallberg left
pmichaud pmichaud@plum:~/rakudo$ ./perl6 11:35
> my $a = 5;
5
> say $a * 3
15
arnsholt Shiny!
pmichaud > role A { method x() { say 'x' } }; class B does A { }; B.new.x 11:36
x
moritz_ Perl 6 regex? 11:38
dalek kudo: 5f40d37 | pmichaud++ | src/Perl6/Actions.pm:
Rename some setup variables to be more accurate.
 11:39
kudo: b0cd2b8 | pmichaud++ | src/ (2 files):
Add a $MAIN parameter to unit mainline (signals that the unit is the mainline).
kudo: c9aa0e9 | pmichaud++ | src/Perl6/Actions.pm:
Refactor !YOU_ARE_HERE to use outer_ctx.
kudo: 3d0a9c2 | pmichaud++ | src/glue/run.pir:
Refactor !UNIT_START.
kudo: 114971b | pmichaud++ | build/PARROT_REVISION:
Bump PARROT_REVISION.

spectest-progress.csv update: 507 files, 33926 (84.5% of 40171) pass, 2 fail Failure summary: S32-temporal/DateTime.t 15 - epoch at 1970-01-01 01:01:01 S32-temporal/DateTime.t 16 - as Str 1970-01-01T01:01:01+0100 S32-temporal/DateTime.t passed 18 unplanned test(s)
moritz_ pmichaud: btw [Coke] has asked to link to the nqp-rx revision in the parrot commit that updates the nqp-rx bootstrap
wich I kinda like, for easier bisecting 11:40
pmichaud moritz_: okay, wfm
time for some sleep -- bbl 11:41
moritz_ good night 11:42
jnthn pmichaud: Wow!!
pmichaud++
moritz_ pmichaud++ indeed 11:48
Juerd Why is the executable named perl6, not rakudo? 11:53
moritz_ hysterical raisins? 11:54
jnthn I think there's a precedent of people typing language name rather than implementation name in many cases. I'm pretty sure it's "java" or "javac" whichever underlying compiler/vm implementation you have. 11:59
moritz_ otoh C compiler names differ 12:00
szbalint would mess up a whole lot of shebang lines too
12:00 proller left
jnthn Though perhaps perl6 wants to be an alias to rakudo and then people can choose which implementation their perl6 is aliased to as well as easily get at different impls. 12:00
takadonet morning all 12:01
12:02 proller joined 12:08 pmurias left 12:10 pmurias joined 12:12 sftp joined 12:14 bluescreen joined 12:15 bluescreen is now known as Guest43164 12:26 ruoso joined 12:31 duzy`` left 12:32 Trashlord left 12:34 Trashlord joined 12:36 Backglass_ joined 12:39 Backglass left, Backglass_ left, Backglass joined 12:42 pmurias left 12:43 Backglass left 12:49 rgrau_ joined 12:53 Mowah left 12:57 Mowah joined 13:02 IllvilJa joined 13:09 charsbar_ joined 13:11 charsbar left, rokoteko left, rokoteko_ joined, BinGOs_ joined 13:12 tadzik joined 13:13 araujo left, spinclad_ joined 13:14 araujo joined, sunnavy_ joined 13:15 TiMBuS left, szbalint left, sunnavy left, LionMadeOfLions left 13:17 BinGOs_ is now known as BinGOs, BinGOs left, BinGOs joined
[Coke] my preference would be that we get a rakudo, and have a switch (hardwired to on for now. =-) that installed a copy / link as perl6 13:17
13:17 nsh_ joined
Juerd That's a good idea 13:19
13:24 szbalint joined 13:27 TiMBuS joined, athomason joined 13:35 jaldhar left 13:41 hans__ joined 13:50 Ross left, tylercurtis joined 13:54 nicePerl6 joined 13:55 nicePerl6 left, hans__ left 13:59 uniejo left
pmichaud good morning, #perl6 14:16
mathw oh hai pmichaud
jnthn morning, pmichaud 14:17
pmichaud hugme: tweet rakudoperl Rakudo's REPL now remembers lexical variables across input lines 14:19
hugme hugs pmichaud; tweet delivered
mathw \o/ 14:20
pmichaud++
TiMBuS yes. yes!
jnthn pmichaud: I see your roles lexicals fix worked too 14:21
14:21 perlygatekeeper left
jnthn pmichaud: I'm not sure I understand why. :-) 14:21
But I'm happy it's fixed. :-)
pmichaud actually, I don't think it worked.
I think it must be an artifact of something else going on.
need tests. :-)
if it did work, it's because I somehow changed the order of initialization :-) 14:22
jnthn rakudo: role R { method m { say 42 } }; class C does R { }; C.new.m
p6eval rakudo 855ca6: OUTPUT«Could not find sub &say␤ in 'm' at line 11:/tmp/uC4AK36C_m␤ in main program body at line 11:/tmp/uC4AK36C_m␤»
jnthn pmichaud: That's the classic test case.
(not sure if p6eval is up to date)
pmichaud right. I don't know why it worked for me.
but:
TiMBuS rakudo: any(Int) ~~ Int
p6eval rakudo 855ca6: ( no output )
TiMBuS rakudo: say any(Int) ~~ Int
p6eval rakudo 855ca6: OUTPUT«0␤»
TiMBuS ^should that work
pmichaud rakudo: say 42; { role R { method m { say 42 } }; class C does R { }; C.new.m }
p6eval rakudo 855ca6: OUTPUT«42␤Could not find sub &say␤ in 'm' at line 11:/tmp/c3KgwWOEid␤ in main program body at line 11:/tmp/c3KgwWOEid␤»
pmichaud hmmm 14:24
anyway, for some reason it "worked" that one time for me in the repl. not sure why.
rakudo: say Int ~~ Int
p6eval rakudo 855ca6: OUTPUT«1␤»
jnthn TiMBuS: I don't think so. 14:25
It's a type check
pmichaud TiMBuS: I'm not sure how junctions are supposed to autothread there.
jnthn They can't.
Otherwise auto-threading won't work.
TiMBuS the reverse works fine
jnthn Because the type check won't fail, causing the signature not to bind, causing auto-threading.
TiMBuS Int ~~ any(Int)
jnthn Right
But that's normal :-) 14:26
Smart-match is not symmetric.
In that case it does Junction.ACCEPTS
Which threads through the things in the junction.
14:32 alester joined 14:33 perlygatekeeper joined 14:34 perlygatekeeper left 14:37 plobsing joined 14:38 macdaddy joined, macdaddy is now known as Guest84429
jnthn pmichaud: confusing bug is confusing. 14:40
> role R { method m { say 42 } }; class C does R { }; C.new.m
Could not find sub &say
however
> role R { method m { say 42 } } 14:41
R()
> class C does R { }; C.new.m
42
pmichaud right
14:41 patspam joined
pmichaud it probably has to do with the way the REPL now captures lexicals from one line to the next 14:42
in fact 14:43
jnthn oh!!
Actually
I think it's to do with the fact that we run the body of the role at the time the "does" statement happens
And that is in a :load :init
In the case where it's split over lines, the role body block has had change to somehow get captured by the mainline body instead. 14:44
moritz_ speaking of loading times... why is module loading deferred to :load :init time?
jnthn moritz_: It shouldn't be - unless it's in a pre-compiled module and that's the earliest it can happen 14:45
But we should be loading them when we parse a use statement otherwise.
Otherwise we can't stub lexicals into the lexpad
moritz_ rakudo: use Test; BEGIN { @*INC.push: 't/spec/package/' }
p6eval rakudo 855ca6: ( no output )
moritz_ rakudo: use Test; BEGIN { @*INC.push: 't/spec/packages/' }
p6eval rakudo 855ca6: ( no output )
moritz_ locally that failed... 14:46
maybe fixed by now
jnthn I did it in the REPL and it works OK
moritz_ oh wait
rakudo: use Test; BEGIN { @*INC.unshift: 't/spec/packages/' }
jnthn no error with -e either
p6eval rakudo 855ca6: ( no output ) 14:47
moritz_ hm
it used to do the 'use' too late
and then find t/spec/packages/Test.pm
and fail during loading
but it seems to be fixed now
14:49 masak joined
pmichaud yes, I want to switch BEGIN so that it's an actual phaser, instead of having its own separate way of doing things. 14:52
masak pmichaud++! # REPL memory
dalek kudo: c89a843 | pmichaud++ | src/ (2 files):
Remove IN_EVAL() function, $*IN_EVAL dynvar.
 14:57
kudo: 17654af | pmichaud++ | docs/ROADMAP:
Mark REPL task as done in the ROADMAP.
kudo: 7bbb35f | pmichaud++ | docs/spectest-progress.csv:
Merge branch 'master' of github.com:rakudo/rakudo
15:00 Backglass joined
masak Backglass: hi! 15:00
15:04 tadzik left
pmichaud is $x.map({ .key => .value + 1}), ('a' => 6), 'map on pair works (=>)'; 15:04
IllvilJa concludes that "programmers are unreasonable" after reading a blog post by Ovid...
pmichaud how are we calling this one? Should the block passed to .map end up being a hash composer or no?
moritz_ pmichaud: block, because it uses $_ 15:05
pmichaud ...and the same would be true with { 'a' => 3 + $_ } ?
moritz_ yes 15:06
pmichaud my $hash = { 'a' => 3 + $_ }; # fail.
15:06 plobsing left
moritz_ (the tests might not be consistent in some cases) 15:06
right.
dakkar hm. has anyone seen TimToady in the last few days? 15:07
TimToady nope
moritz_ no mirrors around?
pmichaud Tim... who?
dakkar hi!
TimToady: I'm just chasing down the last few speakers for the YAPC::EU
can I count on you for a 40- or 50-min keynote? 15:08
TimToady sure
moritz_ \o/
pmichaud has anyone noticed that flights to pisa from the u.s. are on the expensive side, and have weird times? ;-)
actually, getting *to* pisa is pretty easy, it's getting back that is weirdish. :)
TimToady we're flying from Barcelona...
pmichaud ...because you'll happen to be there already, or because that ends up being easier than the other options? 15:09
moritz_ pmichaud: Munich <-> Pisa is no problem either way
with Lufthansa
dakkar TimToady: could you log into the website and tell it that you are attending? then I'll create the talk and you can edit it at your leisure (relatively speaking…)
conferences.yapceurope.org/ye2010/
pmichaud dakkar: I noticed that my NQP talk was accepted (yay) -- would you prefer the Perl 6 talk instead?
dakkar pmichaud: we already have a few P6 talks, NQP seemed more interesting from a general point of view… 15:10
pmichaud dakkar: wfm 15:11
[particle] perl 6 is old news. nqp is the new hotness :)
15:13 Backglass left
TimToady dakkar: okay, I'm registrationalized. 15:14
dakkar TimToady: great, thanks 15:16
conferences.yapceurope.org/ye2010/talk/2953 ← here's your keynote, make something up :) 15:17
pmichaud dakkar: I think he always does. :-)
15:17 Backglass joined
dakkar that's why we like him :) 15:18
TimToady probably something on the order of "What makes Perl Perl?" this time
15:18 perlygatekeeper joined, LionMadeOfLions joined, Backglass left
masak rakudo: Any.substr(0, 1) 15:18
p6eval rakudo ddf224: OUTPUT«Method 'substr' not found for invocant of class ''␤ in main program body at line 11:/tmp/8hDhgTiUk_␤»
masak I think I read somewhere that this is expected nowadays.
and I think I agree.
moritz_ right
masak rakudo: substr(Any, 0, 1)
p6eval rakudo ddf224: OUTPUT«Method 'chars' not found for invocant of class ''␤ in 'Cool::substr' at line 2121:CORE.setting␤ in main program body at line 11:/tmp/BLXjHiaPx8␤» 15:19
masak but that... why does it even call something?
masak submits rakudobug
pmichaud maybe want a type constraint on the first arg.
moritz_ it should type-constrain substr() to Cool
masak moritz_: that's what I think too.
pmichaud (or invocant, if it's an "is export" method)
masak AFAICS, the sub is exported from Cool.substr
so it should have an implicit typing on the invocant. 15:20
pmichaud right
moritz_ shouldn't the exporting add a constraint automatically?
moritz_ too slow to type
TimToady hmm, if substr calls chars, we can't use it to test a cat string to see if there are more chars with $cat.substr($pos,1)
15:20 Backglass_ joined
masak I'd actually like to ask for &substr the sub to work on Any, if that's at all possible. 15:20
TimToady Any is not Cool
moritz_ might be a pecularity of the implementation though
masak it could just coerce it to an empty Str or somthing.
pmichaud the use of .chars is undoubtedly a think-o. 15:21
lots of times people contributing code run to .elems and .chars without asking if there's a lazier way to do it
moritz_ maybe it was to avoid "can't take substr out of range" or so
TimToady well, one way or another, we need a primitive that lets us work through a string without committing to finding its end
moritz_ ... and please don't let it be split() :-) 15:22
pmichaud my $len = $length // self.chars;
bah.
masak I need to delimit this ticket somehow. I think the bug is that the sub is not constrained to Cool.
but I'll include the other things too for posterity.
moritz_ masak: that's right
pmichaud TimToady: we're safe there, though -- if $length is provided we don't call .chars 15:23
(we break a little later on, but that's fixable)
TimToady perhaps cat strings have "false ends", and there's just an extra loop around when you think you've run out that increases it
jnthn Note that Rakudo doesn't put type constraints on invocants properly yet 15:24
So if the method is exported it won't type-check it's invocant unless you are explicit about what it should check just yet. 15:25
A few days ago, it didn't even do that though... :-)
pmichaud right
moritz_ jnthn: for every NYI feature you implement, we find two more to do :-)
jnthn Great, I'll be trying to implement Perl 6 for the rest of my life! :P 15:26
dalek kudo: 894e793 | pmichaud++ | t/harness:
Fix bug with --long and --stress option handling.
moritz_ jnthn: the good news is that you'll get used to it :-)
jnthn pmichaud: Is --long everything?
.oO( at this point, --even-longer might be more apt ;-) )
 15:27
TimToady it just means the community must keep doubling to keep up with what needs to be done
moritz_ hopes that when the community expands, the code developer pool also expands 15:28
TimToady we must all be resigned to becoming a smaller percentage of the Perl 6 community :)
pmichaud jnthn: --long is default
basically, a test can be marked as #long or as #stress
by default, #long tests get run (this was the bug -- they weren't being run)
by default #stress tests are not run 15:29
so:
jnthn Is there a make quicktest?
pmichaud yes.
jnthn aha!
\o/
pmichaud make quicktest does --long=0
and skips the S05-mass/rx.t and S32-trig tests
(and any other tests we deem to be "#long" in t/spectest.data)
jnthn Is it kosher to mostly run make quicktest and then commit if it's clean, if it's unlikely the patch woulda broken either of those? 15:30
pmichaud make spectest doesn't add any options, so we get the long tests but not the stress tests
make stresstest does --stress=1
which includes those tests that we deem to be "really long" :-)
moritz_ jnthn: I'll tell you when you broke something :-)
pmichaud jnthn: yes, it's kosher
use best judgement here
jnthn
.oO( testing...without the pork! )
 15:31
pmichaud: I'll try ;-)
pmichaud Since adding those, I've been tending to do "make quicktest" on local commits, and then a "make spectest" before push.
jnthn *nod*
pmichaud anyone have further comments on the discussion from irclog.perlgeek.de/perl6/2010-07-06#i_2523937 ? (about "why Rakudo Star"?) 15:32
I'll be drafting some text later this morning
(unless I get sidetracked)
rakudo: my @array = <1 2 3>; say ?(@array[0,1] ~~ List); say @array[0,1].WHAT; 15:37
p6eval rakudo ddf224: OUTPUT«0␤Parcel()␤»
pmichaud does @array[0,1] have to be a *List*?
Parcel feels more appropriate 15:38
moritz_ pmichaud: shouldn't you be the one answering the question? :-)
pmichaud well, it feels funny to require Parcel, too.
moritz_ would require Positional
pmichaud I'll switch to Positional for now. 15:39
15:40 timbunce joined
masak ok, so the &substr thing was one of two things that had broken in GGE since last time I got it up to date. the other thing was that $obj.?nosuchmethod now gives Nil, not Failure. 15:41
+1 on that, by the way.
but how does one test for Nil-ness in a nice way?
pmichaud ~ Nil seems to work.
er, ~~ Nil
TimToady or ~~ () is the same thing
pmichaud rakudo: sub foo() { return; }; say foo ~~ Nil;
masak right.
p6eval rakudo ddf224: OUTPUT«Too many positional parameters passed; got 1 but expected 0␤ in 'foo' at line 11:/tmp/SAbgTooWgg␤ in main program body at line 11:/tmp/SAbgTooWgg␤»
pmichaud rakudo: sub foo() { return; }; say foo() ~~ Nil; 15:42
p6eval rakudo ddf224: OUTPUT«1␤»
masak for some reason I settled on 'eqv Nil' in my code.
pmichaud eqv Nil also works.
TimToady: do we allow "list binding"? e.g., @a[1,2] := ($x, $y);
(please say "no")
TimToady we can say no for now 15:43
pmichaud we don't have a list prefix form of := at the moment anyway.
15:43 zostay left 15:44 zostay joined, Ross joined, envi^home joined
TimToady I think if Cat ~~ Stringy, then perhaps any Stringy needs to be able to handle * as a length, and have a Stringy.reify($pos) that returns $pos max $actualchars 15:47
tylercurtis pmichaud: I can't seem to get your attention in #parrot, so I'll ask here. Would you mind giving me, off the top of your head, a guess as to which PAST::Node subclass attributes might be useful to traverse for PAST::Walker/Transformer/etc.? Mainly, I'm looking for things like viviself, control, loadinit, etc. that can contain other PASTs. 15:48
pmichaud tylercurtis: I think you named them all already. :)
TimToady and substr would say my $realchars = $stringy.reify($offset+length)
pmichaud let me re-check.
tylercurtis: viviself, vivibase, control, loadinit 15:51
that's all I see at the moment.
tylercurtis pmichaud: thanks.
pugssvn r31567 | pmichaud++ | [t/spec]: Update some tests on slices, resolve RT #63014. 15:53
masak tylercurtis: any comment on gist.github.com/464349 ? I hope I'm not creating too much downstream chasing for you with your LLVM compiler. 15:56
pmichaud rakudo: 'a' ~~ /<alpha>/; for %($<alpha>) { say $_ } 16:00
p6eval rakudo ddf224: ( no output ) 16:01
16:08 Backglass_ left
tylercurtis masak: I like that change, actually. I'd rather you not rather worry about changing SIC on my account, anyway. Thanks for the warning, though. Not only will it be helpful to know that environment is going away when I implement blocks, it's given me an idea about how to implement them. 16:08
16:08 pragma_ left
masak tylercurtis: cool. 16:09
TimToady I am coming to the conclusion that ~~ should autothread, and relegate the highly specialized Junction testing to .^does(Junction)
masak tylercurtis: yes, I very much like the change too. I hope to find some quiet hacking time to write a new runtime for it.
TimToady we've run into that several times in recent days, and I think the autothreading will be much more useful 16:10
that is, the arg to ACCCEPTS should be Any, not Mu
16:10 quietfanatic joined
TimToady *CC 16:10
quietfanatic rakudo: say [[4, 5], [6, 7]].map: {.perl}
p6eval rakudo ddf224: OUTPUT«[4, 5][6, 7]␤» 16:11
quietfanatic isn't map supposed to recurse into nested structures?
moritz_ nope
wolverian no.
moritz_ only hyper operators do
quietfanatic hm. 16:12
pmichaud TimToady: that might be a bit of a problem, since .ACCEPTS is what we use for type checking and multidispatch 16:16
TimToady: is it sufficient if ~~ expects a lhs of Any ?
so that ~~ autothreads, and .ACCEPTS doesn't?
16:17 pragma_ joined
TimToady well, that precludes a simple desugaring implementatoin 16:17
16:17 pragma_ is now known as Guest52181
pmichaud we can try it and see. 16:17
(.ACCEPT autothreads)
TimToady why are you using .ACCEPTS and not the more basic .does?
pmichaud sorry, "type checking" being more generic than just does 16:18
16:18 hghgd_ left
pmichaud e.g. sub foo($x where { ... }) { ... } 16:18
TimToady so, constraint checking?
pmichaud right
cono rakudo: say $*VM<config><osname> 16:19
p6eval rakudo ddf224: OUTPUT«linux␤»
TimToady okay, lemme think about ACCEPTS s'more while I'm out walking
pmichaud sure thing.
TimToady did you see my .reify proposal above? 16:20
16:20 Ross left
pmichaud I did. 16:20
masak swimming &
16:20 masak left
pmichaud I'm waiting for a better definition of cats before jumping on it. :-) 16:20
16:20 ruoso left
pmichaud but it looks to me like we're just reinventing or copying the laziness model for strings :) 16:20
TimToady
.oO(you're never quite sure how many lives a cat has remaining...)
 16:21
pmichaud (which is fine)
TimToady well, it's positing that the Stringy abstraction can have an underlying impementation that pulls from a lazy list, at least
whether that turns into Ropes Of Unknown Size is anyone's guess 16:23
walking &
pmichaud avoids the fire swamp.
rakudo: my Int $b = "4" - 3;
p6eval rakudo ddf224: OUTPUT«Type check failed for assignment␤ in '&infix:<=>' at line 1␤ in main program body at line 11:/tmp/luNImBS3qe␤»
16:24 Ross joined, dakkar left
pmichaud rakudo: my @a = 1,2,3; say @a min 4; 16:26
p6eval rakudo ddf224: OUTPUT«1␤»
pmichaud rakudo: my @a = 1,2,3; say +@a min 4;
p6eval rakudo ddf224: OUTPUT«3␤»
16:29 thebird left
pmichaud rakudo: say "README" ~~ :f 16:29
p6eval rakudo ddf224: OUTPUT«0␤» 16:30
pmichaud rakudo: say "/etc/passwd" ~~ :f
p6eval rakudo ddf224: OUTPUT«1␤»
pmichaud rakudo: say "/etc/passwd" ~~ :s
p6eval rakudo ddf224: OUTPUT«1␤» 16:31
pmichaud rakudo: say "/etc/passwd" ~~ :d
p6eval rakudo ddf224: OUTPUT«0␤»
pmichaud rakudo: say "/etc" ~~ :d
p6eval rakudo ddf224: OUTPUT«1␤»
16:35 ruoso joined, skids left 16:40 cdarroch joined, cdarroch left, cdarroch joined, ashleydev joined 16:42 ash_ joined 16:43 Guest84429 left 17:02 TiMBuS left 17:04 alester left, alester joined, alester left 17:10 timbunce left 17:13 timbunce joined
lue ohai o/ 17:16
17:16 clintongormley left 17:23 envi^home left
quietfanatic How do I get the modtime of a file? 17:23
There doesn't seem to be a defined way.
short of run 'stat $_' 17:24
17:24 timbunce left 17:27 hercynium left 17:28 patspam left 17:29 eternaleye joined 17:34 Backglass joined 17:35 clintongormley joined 17:36 dtwright joined 17:37 dtwright left 17:41 shade_ joined
moritz_ it seems to be missing indeed 17:46
probably something like $filename.IO.modification-time or so 17:47
17:47 rokoteko_ is now known as rokoteko, rokoteko left, rokoteko joined
ash_ $filename.IO.last-modified ? 17:50
17:50 shade_ left
ash_ (its a bit shorter) 17:50
17:50 shade_ joined
lue I thought there was a file test for that. 17:51
(like :e, :f, etc.)
TimToady file tests only return boolean in p6 17:52
quietfanatic rakudo: say '/etc/passwd'.IO.last-modified
p6eval rakudo 894e79: OUTPUT«Method 'IO' not found for invocant of class 'Str'␤ in main program body at line 11:/tmp/IcZRPcD3nD␤»
TimToady rakudo: say open('/etc/passwd').last-modified
p6eval rakudo 894e79: OUTPUT«Method 'last-modified' not found for invocant of class 'IO'␤ in main program body at line 11:/tmp/QxUgtP7sE0␤»
TimToady rakudo: say open('/etc/passwd').modification-time
p6eval rakudo 894e79: OUTPUT«Method 'modification-time' not found for invocant of class 'IO'␤ in main program body at line 11:/tmp/JtnVOJuCp3␤»
TimToady rakudo: say ~IO.^methods 17:53
quietfanatic rakudo: say open('/etc/passwd').^methods(:local)
p6eval rakudo 894e79: OUTPUT«ins print say open printf t slurp close eof autoflush lines get getc chop tanh split match exp acosh words cotan atan2 lcfirst uc sec cis log substr from-radians sprintf trans sin pred acosec subst sinh cosech abs acosech chomp index ceiling trim asec trim-leading unpolar
..trim-trai…
rakudo 894e79: OUTPUT«getcinsprintsayopenprintftslurpcloseeofautoflushlinesget␤»
quietfanatic rakudo: say ~open('/etc/passwd').^methods(:local)
p6eval rakudo 894e79: OUTPUT«eof autoflush lines get getc ins say print open printf t close slurp␤»
quietfanatic what does t do?
rakudo: say open('/etc/passwd').t 17:54
p6eval rakudo 894e79: OUTPUT«0␤»
TimToady terminates the file :)
quietfanatic eheh :)
ash_ lol
t seems like and odd name for that functionality 17:55
TimToady rakudo: say open('/dev/tty').t 17:56
p6eval rakudo 894e79: OUTPUT«0␤»
lue I would expect eof
TimToady I dunno what it does 17:57
lue (or eot, if that's where .t gets its name)
.u EOT
phenny U+C5CD HANGUL SYLLABLE EOT (엍)
ash_ :t ? 17:59
jnthn multi method t() {
$!PIO.isatty;
}
lue so it checks if it's a terminal? 18:00
rakudo: say open('/dev/tty0').t
p6eval rakudo 894e79: OUTPUT«Unable to open filehandle from path '/dev/tty0'␤ in 'open' at line 4807:CORE.setting␤ in main program body at line 11:/tmp/cly5u62_cj␤»
quietfanatic rakudo: say "abc.abc" ~~ /^(.*)\.abc$/ 18:01
p6eval rakudo 894e79: OUTPUT«␤»
quietfanatic So why doesn't this work?
rakudo: say "abc" ~~ /^(.*)$/ 18:02
p6eval rakudo 894e79: OUTPUT«abc␤»
quietfanatic rakudo: say ".abc" ~~ /^(.*)\.abc$/ 18:03
p6eval rakudo 894e79: OUTPUT«␤»
quietfanatic rakudo: say ".abc" ~~ /^(.*?)\.abc$/
p6eval rakudo 894e79: OUTPUT«.abc␤»
quietfanatic rakudo: say ".abc" ~~ /(.*)\.abc$/
p6eval rakudo 894e79: OUTPUT«␤»
quietfanatic rakudo: say "abc.abc" ~~ /(.*?)\.abc$/
p6eval rakudo 894e79: OUTPUT«.abc␤»
quietfanatic rakudo: say "abc.abc" ~~ /^(.*?)\.abc$/
p6eval rakudo 894e79: OUTPUT«␤»
quietfanatic Lacking backtracking?
tylercurtis rakudo: say "." ~~ /\./ 18:04
p6eval rakudo 894e79: OUTPUT«.␤»
tylercurtis rakudo: say ".abc" /^(.*)?\.abc$/ 18:05
p6eval rakudo 894e79: OUTPUT«===SORRY!===␤Confused at line 11, near "say \".abc\""␤»
tylercurtis rakudo: say ".abc" ~~ /^(.*)?\.abc$/
p6eval rakudo 894e79: OUTPUT«.abc␤»
quietfanatic If I can't do that to recognize and remove a file extension I'll have to use rindex and substr. 18:07
pmichaud there's a bug in backtracking in captured expressions at the moment 18:08
but you should be able to do /^ <name>=[.*]\.abc $/
er
but you should be able to do /^ $<name>=[.*]\.abc $/
quietfanatic I'll try that
tylercurtis rakudo: "abc.abc" ~~ /^ $<name>=[.*] \.abc $/; say $<name> 18:09
p6eval rakudo 894e79: OUTPUT«abc␤»
quietfanatic That works. Thanks. 18:10
moritz_ stackoverflow.com/questions/3196539...l6/3197558
pmichaud regex backtracking is the last on my "big issues to fix for Rakudo #31" at the moment. 18:11
ash_ moritz_: isn't that because of closures?
pmichaud I got the other three out of the way this week. :-)
moritz_ ash_: that's what I've answered
pmichaud: for me, <foo> picking up lexicals, and escapes in char classes are also big issues 18:12
pmichaud moritz_: which is bigger?
i.e., rank the three. :-)
moritz_ lexicals, escapes, backtracking 18:13
I know others prefer backtracking though
pmichaud yeah, backtracking is the one that's likely to give the most pain. 18:14
anyway, it's likely to be (lexicals|backtracking), escapes
ash_ rakudo: say "abc.abc" ~~ /(.*?)\.abc$/
p6eval rakudo 894e79: OUTPUT«.abc␤»
pmichaud ash_: that ends up not helping either 18:15
ash_ sorry, i meant to do that in a /msg to p6eval
pmichaud basically, parenthesized captures and subrules act like they have a ':' immediately following them.
nqp-rx doesn't yet backtrack into them once they've succeeded
18:22 justatheory joined
Backglass Can someone help me to apply a regex to a variable? 18:26
TimToady what have you tried? 18:27
moritz_ rakudo: say 'abc 123' ~~ /\d+/ 18:28
p6eval rakudo 894e79: OUTPUT«123␤»
Backglass I'm using a module call WWW:Mechanize for make a regex in url
moritz_ that sounds like Perl 5 18:29
or is there a Perl 6 WWW::Mechanize already?
Backglass nop
is in perl 6
sorry
perl 5
moritz_ then #perl6 is not quite the right channel to ask
Backglass but help is a simple regex :S 18:30
diakopter simple to whom? 18:31
Backglass gist.github.com/467059
this is my script in perl
i want a url link this valeptr.com/scripts/runner.php?IM=0...d28110375. 18:32
but my script show all links in web
diakopter like moritz_ said, people here are focused on Perl 6
oh, this probably shouldn't be merely implied: and regexes are quite different in Perl 6 18:33
huf nah, they're different everywhere, you just made it readily apparent ;) 18:34
Backglass Does someone help me? 18:35
diakopter doesn't look like it, so far. I wouldn't bet on it 18:36
moritz_ Backglass: is there anything that prevents you from joining a perl 5 channel, and ask for help where it's appropriate?
[particle] head on over to #perl, they'll help you there
Backglass ok
thanks
I go to #perl channel
18:37 Backglass left
[particle] it's better to give options for more help, rather than simply refusing to help 18:38
dalek p-rx: 70f4eb7 | moritz++ | src/Regex/P6Regex/Actions.pm:
put AST generation for named assertions into a separate method
 18:44
lue (Go Germany...) 18:46
moritz_ "Schland" 18:47
lue: www.youtube.com/watch?v=JFd9JvyFRYo
lue s/Germany/Deutschland/ danke for reminding me :)
Aw, I can't haz flash :(
moritz_ clive $url; mplayer $generated_flv_file; 18:48
18:48 skids joined
moritz_ can I generate a PAST that does one things if a lexical exists, and something else if it doesn't? 18:48
oh noez, I can see another problem coming up 18:49
sub print(*@args) vs. <print> for printable characters 18:50
18:51 shade_ left 18:52 shade_ joined
lue rakudo: my @a; my @b; say @a[1..3].WHAT 18:57
p6eval rakudo 894e79: OUTPUT«Parcel()␤»
lue rakudo: my @a = Nil; my @b; say @a[1..3].WHAT 18:58
p6eval rakudo 894e79: OUTPUT«Parcel()␤»
19:03 cono left 19:04 shade_ left, shade_ joined 19:05 cono joined 19:07 pmurias joined
pmurias sorear: ping 19:07
19:09 timbunce joined, eternaleye left 19:18 timbunce left
lue > my @a = 1,2,3,4; @a[2..3] := @a[0..1]; say @a.perl 19:19
[1, 2, 3, 4]
well, at least now I know list binding is the issue.
(not Nil being incorrectly handled) 19:20
pmichaud lue: 19:21
15:42 <pmichaud> TimToady: do we allow "list binding"? e.g., @a[1,2] := ($x, $y);
15:42 <pmichaud> (please say "no")
15:43 <TimToady> we can say no for now
sorear pmurias: hi
good morning #perl6
sorear -> backlog
lue pmichaud: I bet the was a forced answer :) 19:24
s/the/that/
pmichaud it actually wouldn't be all that hard to implement, but it might be tricky to parse. 19:26
we'd have to do the same sorts of things for := that we're doing with = now.
anyway, "no for now" is the answer I'm going with. :)
sorear back
lue There is a way, which I pasted last night: 19:27
> my @a = 1,2,3; @a[$_+3] := @a[$_] for 0..2; say @a.perl
[1, 2, 3, 1, 2, 3]
tylercurtis rakudo: my @y = 0, 0, 0, 0; my $x; my $z; [+=] $x, @y, $z, 1; say ($x, @y, $z).join(' ')
p6eval rakudo 894e79: OUTPUT«Any() 0 0 0 0 Any()␤»
lue So I can use that until it works. I have been thinking about it (w/o looking at assign.pir), and the apparent issue is differently sized lists on either side. 19:28
methinks that would require spec clarity first, to be on the safe side. 19:29
tylercurtis Does Rakudo not support reduction of assignment ops? 19:30
lue [ good discussion for a not-world-cup time :) ]
pmichaud tylercurtis: what result were you expecting, ooc? 19:32
tylercurtis According to S03, "[+=] $x, @y, $z, 1" is equivalent to "$x += @y[0] += @y[1] += @y[2] ... @y[*-1] += $z += 1", so I was expecting it to output "6 5 4 3 2 1". 19:33
pmichaud seems reasonable. :-)
my $x; my @y = 0, 0, 0, 0; my $z; $x += @y[0] += @y[1] += @y[2] += $z += 1; say ($x, @y, $z).join(' ') 19:34
rakudo: my $x; my @y = 0, 0, 0, 0; my $z; $x += @y[0] += @y[1] += @y[2] += $z += 1; say ($x, @y, $z).join(' ')
p6eval rakudo 894e79: OUTPUT«1 1 1 1 0 1␤»
pmichaud rakudo: my $x; my @y = 0, 0, 0; my $z; $x += @y[0] += @y[1] += @y[2] += $z += 1; say ($x, @y, $z).join(' ') 19:35
p6eval rakudo 894e79: OUTPUT«1 1 1 1 1␤»
pmichaud however, looks like the result is 1, 1, 1, 1, 1 :-P
19:35 rhr left
ash_ rakudo: my 'foo' 19:37
oops
p6eval rakudo 894e79: OUTPUT«===SORRY!===␤Malformed my at line 11, near "'foo'"␤»
tylercurtis rakudo: my $x; my @y = 0, 0, 0; my $z; $x = @y[0] = @y[1] = @y[2] = $z = 5; say ($x, @y, $z).join(' ')
p6eval rakudo 894e79: OUTPUT«5 0 5 5 5␤»
lue rakudo: my $x; my @y = 0, 0, 0; my $z; $x = @y[0] = @y[1] = @y[2] = $z = 5; say "$x [@y] $z" 19:38
p6eval rakudo 894e79: OUTPUT«5 [@y] 5␤»
tylercurtis my $x; my @y = 0, 0, 0; my $z; [=] $x, @y, $z, 5; say ($x, @y, $z).join(' ')
lue be that an interpolation bug?
19:39 takadonet left
tylercurtis rakudo: my $x; my @y = 0, 0, 0; my $z; [=] $x, @y, $z, 5; say ($x, @y, $z).join(' ') 19:39
p6eval rakudo 894e79: OUTPUT«Any() 0 0 0 Any()␤»
tylercurtis rakudo: my $x; my @y = 0, 0, 0; my $z; $x = @y[0] = @y[1] = @y[2] = $z = 5; say "$x [@y[]] $z" 19:40
p6eval rakudo 894e79: OUTPUT«5 [0 5 5] 5␤»
tylercurtis lue: there you go.
lue ah. wonder why [@y] didn't interpolate though.
moritz_ because arrays and hashes only interpolate when there are trailing postcircumfixes. 19:41
pmichaud @-variables don't interpolate until suffixed with a postcircumfi.... right
that way we don't have to worry about [email@hidden.address] suddenly trying to interpolate @yahoo or @yahoo.com
19:41 Guest43164 left
lue rakudo: my $x; my @y = 0, 0, 0; my $z; $x = @y[0] = @y[1] = @y[2] = $z = 5; say "$x [{@y}] $z" 19:41
p6eval rakudo 894e79: OUTPUT«5 [0 5 5] 5␤» 19:42
lue :D
pmichaud yes, that's my preferred way of handling it :)
lue thinks of silly frenchman
"And your father smelt of elderberries!"
19:43 clintongormley left
tylercurtis -> recreational activities for a few hours. 19:43
19:43 tylercurtis left
lue what if your email is me@brac[]ket.com? 19:43
rakudo: say "what if your email is me@brac[]ket.com?" 19:44
p6eval rakudo 894e79: OUTPUT«===SORRY!===␤Symbol '@brac' not predeclared in <anonymous> (/tmp/mKSh2fK0Es:11)␤»
[Coke] then use single quotes.
pmichaud or backwhack the @ 19:45
or get a real email address
lue :)
Tene [] in domains?
pmichaud or live comfortably knowing that you aren't going to get caught up on some spam lists
[Coke] tene: no doubt the third great expansion of TLDs will include arbitrary unicode.
[particle] abc@def is a valid email address 19:46
lue they already allow Chinese (approved just recently)
[particle] so you could print [] directly afterward
i mean, it'd be a valid thing to do generally, not in perl 6
lue
.oO(∫upermath.net)
19:50 shade_ is now known as shade\ 19:54 rhr joined 19:56 justatheory left
moritz_ moritz.faui2k3.org/tmp/waffle.jpg # part of my dinner tonight :-) 19:57
19:58 Guest43164 joined
TimToady phone 20:00
jnthn moritz_: oooh, that looks tasty!
moritz_ jnthn: it was :-) 20:01
sorear Tene: iirc the only illegal characters in domain names are NUL, NL, and TAB
(bind 1 used a tab-delimited data file!)
Trashlord null
lue aah! Spain got a goal *sob* 20:02
jnthn lue: Yeah... :-/
Trashlord so, is it 1 - 0 for spain?
moritz_ well deserved, even if I don't like it
jnthn That won't get Germany any Klose to the final...
<ducks>
lue is currently engrossed in the world cup. Sorry for the intermittent appearance.
Trashlord er
I was hoping for a Netherlands - Germany final 20:03
and a German victory
ascent_ there will be Netherlands - Spain final, and Netherlands victory (yeah) :P 20:04
Trashlord Germany played very well throughout the entire tournament, they deserve to win for their high quality and very entertaining football, imo 20:05
PerlJam Trashlord: they deserve to win if they score more goals than spain ;) 20:06
20:06 Guest52181 left, Guest52181 joined
Trashlord or if they concede less ;p 20:06
cognominal is this the #perl6 channel? :)
moritz_ cognominal: not sure... :-)
20:06 Guest52181 is now known as pragma_
Trashlord it might be 20:06
jnthn I've been surprised...I expected Germany to do better than this again Spain.
moritz_ seems like #europe right now
PerlJam There's a reason it's called the "World Cup" 20:07
20:07 unik joined 20:08 eternaleye joined 20:09 macdaddy joined, macdaddy is now known as Guest5559
unik p6eval rakudo: multi sub fib(Int $x where 0) { 0 }; multi sub fib(Int $x where 1) { 1 }; multi sub fib(Int $x where { $x > 1 }) { fib($x - 2) + fib($x - 1) }; my $val = 0 .. 10; $val.map: fib; 20:13
p6eval unik: rakudo 894e79: OUTPUT«No applicable candidates found to dispatch to for 'fib'. Available candidates are:␤:(Int $x where ({ ... }))␤:(Int $x where ({ ... }))␤:(Int $x where ({ ... }))␤␤ in main program body at line 11:/tmp/Q0DlIQx48j␤»
20:13 lest_away is now known as lestrrat
unik p6eval rakudo: multi sub fib(Int $x where 0) { 0 }; multi sub fib(Int $x where 1) { 1 }; multi sub fib(Int $x) { fib($x - 2) + fib($x - 1) }; my $val = 0 .. 10; $val.map: fib; 20:14
p6eval unik: rakudo 894e79: OUTPUT«No applicable candidates found to dispatch to for 'fib'. Available candidates are:␤:(Int $x where ({ ... }))␤:(Int $x where ({ ... }))␤:(Int $x)␤␤ in main program body at line 11:/tmp/8oMXUeNYKx␤»
20:15 eternaleye left
PerlJam wonders if you can use a proto to define a default to OUTER::<$_> once 20:18
moritz_ unik: try $val.map: &fib
unik: otherwise fib() is called first, and the result is passed to map()
20:19 tadzik joined
moritz_ rakudo: multi sub fib(Int $x where 0) { 0 }; multi sub fib(Int $x where 1) { 1 }; multi sub fib(Int $x) { fib($x - 2) + fib($x - 1) }; my $val = 0 .. 10; $val.map: &fib 20:19
p6eval rakudo 894e79: ( no output )
moritz_ rakudo: multi sub fib(Int $x where 0) { 0 }; multi sub fib(Int $x where 1) { 1 }; multi sub fib(Int $x) { fib($x - 2) + fib($x - 1) }; my $val = 0 .. 10; say $val.map: &fib
p6eval rakudo 894e79: ( no output )
moritz_ uhm.
jnthn oddness. 20:21
20:21 Ross left 20:22 Ross joined
jnthn rakudo: multi fib(0) { 0 }; multi fib(1) { 1 }; multi fib($x) { fib($x - 2) + fib($x - 1) }; say fib(0); say fib(1); say fib(2); 20:22
p6eval rakudo 894e79: OUTPUT«0␤1␤1␤»
jnthn rakudo: multi fib(0) { 0 }; multi fib(1) { 1 }; multi fib($x) { fib($x - 2) + fib($x - 1) }; my $r = 0 .. 10; say $r.map(&fib); 20:23
p6eval rakudo 894e79: ( no output )
jnthn rakudo: multi fib(0) { 0 }; multi fib(1) { 1 }; multi fib($x) { fib($x - 2) + fib($x - 1) }; say (0..10).map(&fib);
p6eval rakudo 894e79: ( no output )
jnthn rakudo: multi fib(0) { 0 }; multi fib(1) { 1 }; multi fib($x) { fib($x - 2) + fib($x - 1) }; say (0..10).map(&fib).eager;
moritz_ WTBH?
p6eval rakudo 894e79: OUTPUT«Method 'count' not found for invocant of class 'Perl6MultiSub'␤ in main program body at line 1␤»
jnthn wtf
oh
unik p6eval rakudo: my $a = 0 .. 10; $a.map: { say $^a }
moritz_ but shouldn't say() be eager?
jnthn map + multi = boom? 20:24
p6eval unik: rakudo 894e79: ( no output )
jnthn moritz_: I thought that too
moritz_: Was just trying to tease out what's really wrong.
moritz_ map + multi == boom, right
jnthn Looks like map wants to .count
hugme: hug lue_sad 20:25
hugme hugs lue_sad
unik also, is it ok now that fib(N) where N is slightly above 19 eats loads of RAM and never seems to exit?
i just couldn't wait long enough for it to exit, fearing OOM killer will bork my workstation 20:27
moritz_ unik: yes, the recursive implementation tries to run 2**19 sub invocations
rakudo: say (0, 1, *+* ... *)[19]
p6eval rakudo 894e79: OUTPUT«4181␤»
moritz_ that's an iterative, more elegant version that doesn't blow all up
unik nice
moritz_ rakudo: say ~(0, 1, *+* ... *)[^10]
p6eval rakudo 894e79: OUTPUT«0 1 1 2 3 5 8 13 21 34␤» 20:28
moritz_ unlike haskell, Perl 6 can't easily do purity analysis and cache pure functions
unik it's stringified slice of a lazy list?
20:28 saaki left
moritz_ yes 20:31
20:32 justatheory joined 20:33 saaki joined
unik p6eval rakudo: my $a = 0 .. 10; ~$a.map: { say $^a } 20:34
p6eval unik: rakudo 894e79: OUTPUT«0␤1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤»
TimToady jnthn: that .count should become simpler when &fib is a proto wrapper 20:36
jnthn TimToady: Yeah
TimToady also thinking that perhaps we allow the form: proto foo ($sig); 20:37
where semicolon form means "write the body for me"
jnthn TimToady: I guess .signature is the signature of the proto too?
Eww
20:37 aindilis joined
jnthn I'd rather not do that unless we allow it as a general shorthand for { ... } 20:37
moritz_ I like that better than a magical empty function body
TimToady or maybe {*} 20:38
jnthn Would proto foo($sig) { ... } be an auto-written proto?
Or a stub?
pmichaud we can do count on multis -- just haven't implemented it yet.
TimToady I think it's best not to confuse stubs with auto-writes
pmichaud at least good enough to get past map. :-)
(and map shouldn't be using .count anyway.) 20:39
jnthn TimToady: Then probably <...> in regexes also wants to change.
pmichaud (it's a cheat for now)
TimToady yes, map should be using partial binding
moritz_ which isn't specced yet, right?
TimToady jnthn: I was going to change <...> anyway
20:39 rhr left
TimToady since {...} means the same, though now maybe we'll get {*} back to mean call the multies 20:40
jnthn proto foo($sig) { * }
or
20:40 masak joined
jnthn proto foo($sig) { {*} } 20:40
?
masak ahoy!
jnthn lolitsmasak!
moritz_ stares at a round tuit 20:41
masak moritz_: thanks for the link to the Knuth talk. I'd been wanting to see that talk.
TimToady just {*} in proto sub, { {*} } in proto regex
though I suppose we could allow the special form on regexen too
since * can't stand alone in a regex
moritz_ masak: you're welcome.
TimToady we might want to distinguish {*} from nextsame/callsame anyway 20:43
masak TimToady: can we make the spec change `slurp($filename)` -> `slurp($filename.IO)` now? I think it's a good idea, and would like to see it go into Rakudo before Star.
ditto `$filename.IO ~~ :e`, actually.
TimToady if we always start the multi dispatch with {*}, then nextsame can always refer to the outer
without playing lastcall tricks
20:45 spinclad_ is now known as spinclad
lue hai masak o/ 20:45
masak woo, the StackOverflow question that moritz_++ linked to (and answered) previously makes me happy. # stackoverflow.com/questions/3196539...l6/3197558
lue: \o
jnthn TimToady: So it's up to the proto if we then go deferring inheritance wise?
TimToady hmm, not sure about slurp, since slurp($string) would then be a no-op; was more concerned about the difference between "splitters" like words and lines
masak one, because it's a rather advanced question. two, because it's fixed in Rakudo already :) 20:46
PerlJam moritz, masak: what knuth talk?
masak PerlJam: grep today's logs for "Knuth".
TimToady yes, assuming that {*} can also be used internally as well as the whole body, you could redispatch with proto fib ($x) { {*}; nextsame }
20:47 patspam joined
moritz_ PerlJam: announcing the successor of TeX 20:47
jnthn slurp imo loses its usefulness if it can't just be given a filename
masak PerlJam: in it, he's revealing his new application, iTex <pingeling>
jnthn I see it more like open
TimToady well open-close
jnthn TimToady: Could work
TimToady: close doesn't take a file name :-P 20:48
TimToady: But yes :-)
TimToady: Are we going to parse {*} as a magical term?
masak what was the ultimate rationale for changing it? making words() and lines() more alike, no?
TimToady yes, but slurp isn't one of those 20:49
lue
.oO(work on NES emulator, or tackle the foreign code import mechanism...)
TimToady so Str.words and Str.lines always just split the string, and you have to use IO.words or IO.lines to read; however...
words() and lines() can still default to $*ARGFILES input 20:50
lue goes for foreign code.
moritz_ +1
(to TimToady)
masak this Backglass fellow who entered earlier has been chatting me up three times. I watched his script evolve gradually. I never gave him any specific tips, just pointed to CPAN modules that seemed appropriate. he grew increasingly insistent. 20:51
TimToady alternately, we have a default IO that you always say IO.lines or IO.words
where "IO" is negotiable
masak I should probably be more brusque from the start.
TimToady: I like just words() and lines()
TimToady I don't 20:52
jnthn masak: I love that in the dictionary one definition of that word is "ungracious harshness" :-)
TimToady because the () is required in a for loop
masak TimToady: will it still gobble the block if I forget the parens?
moritz_ std: for lines { }
p6eval std 31567: OUTPUT«===SORRY!===␤Function 'lines' needs parens to avoid gobbling block at /tmp/_LG5S8EOdK line 1:␤------> for lines⏏ { }␤Missing block (apparently gobbled by 'lines') at /tmp/_LG5S8EOdK line 1:␤------> for lines { }⏏<EOL>␤
..expe…
masak jnthn: oh, did I do the word thing again? sorry :P
jnthn :P
TimToady druther have something like: for slurp.lines {...}
masak jnthn: I should be less circumspect in my explications. 20:53
jnthn Not unless you want to the whole file
masak .oO(I accidentally the whole file )
TimToady: slurp.lines looks non-lazy to me, like jnthn said.
diakopter or more perspicuous 20:54
TimToady which is why I mentioned: for cat.lines {...}
the other day
moritz_ please don't make IO rely on lazy strings
jnthn What moritz_ said.
moritz_ it took us so long to get proper lazy lists and arrays 20:55
TimToady yes, well, what we really need is something short that means the IO of $*ARGFILES
moritz_ I don't want to wait another few years to do proper IO
diakopter but.. in theory, they're optimizable......
masak TimToady: we don't have a cat, so that's not an option Right Now(TM).
[particle]
.oO(something like the fish operator)
TimToady we want mumble.lines where mumble is something short for $*ARGFILES.IO
masak input.lines 20:56
lue
.oO($*AFIO ?)
moritz_ +1 to a term-like input
20:56 rhr joined
diakopter I like mumble 20:57
TimToady "IN" maybe
hmm, too much like $*IN
masak too similar to $*IN
TimToady: what TimToady said.
TimToady I am beside myself. 20:58
masak why am I suddenly seeing double?
TimToady ohoh
diakopter does a double gather
pmurias sorear: so we want viv to prepend Perl6:: to package names when translating p6 to p5?
lue (That's *my* TARDIS, TimToady)
masak I thought we fixed the cloning bug! :)
lue Looks like my biggest hurdle to foreign code is the spec. S21 makes it apparent :) 21:00
masak std: class A { method foo { $!b } }
p6eval std 31567: OUTPUT«ok 00:01 108m␤»
masak theoretically, one could detect direct references to nonexistent attributes, right? 21:01
TimToady nodnod
masak just checking.
21:01 diakopter sets mode: +v masak
masak for example, they're always wrong in a disconnected method. 21:01
diakopter hugme: hug masakbot 21:02
hugme hugs masakbot
masak purrs
jnthn masak: We will :-)
moritz_ rakudo: class Foo { method x { $!b } }
p6eval rakudo 894e79: ( no output )
masak jnthn: when?
jnthn masak: Probably after R* as part of object guts chnages. 21:03
moritz_ jnthn: speaking of attributes... is there any (hacky) way to set a parent attribute?
21:03 DarkWolf84 joined
jnthn moritz_: Eww 21:03
moritz_: From PIR?
masak DarkWolf84! \o/
moritz_ jnthn: Perl 6 or PIR
DarkWolf84 hello
hi masak
pmichaud if there's not a name collection, $!attribute works just fine. 21:04
jnthn setattribute obj, ['Parent';'Name'], '$!attr', value
pmichaud *collision
masak DarkWolf84: welcome back to the party! \o/
moritz_ jnthn: I want to write a Match.new... but the actual attribute is stored in Regex::Match
ah, thanks
pugssvn r31568 | kyle++ | Delete empty directories.
DarkWolf84 I have a little question 21:05
how to delete a var
masak DarkWolf84: how do you mean?
like, set it to its initial state?
moritz_ undefine($x)
DarkWolf84 undefine just makes $foo to be Any
masak $var = Nil; works too 21:06
moritz_ it's a bit wrong right now
DarkWolf84 but $var exists
PerlJam DarkWolf84: you mean you want to eradicate the var from whatever symbol table it happens to live in?
DarkWolf84 yeah
moritz_ no way
masak rakudo: my @a = 1, 2, 3, 4, 0; @a .= [@a]; say @a.perl
p6eval rakudo 894e79: OUTPUT«Method '!postcircumfix:<[ ]>' not found for invocant of class 'Array'␤ in main program body at line 11:/tmp/lVjyAC6JuO␤»
moritz_ existance of variables is a compile-time property 21:07
masak p6eval: what? :)
masak submits rakudobug
rakudo: my @a = 1, 2, 3, 4, 0; @a = @a[@a]; say @a.perl
pmichaud I hate to mention this (and I know it goes against tradition), but part of me now thinks that "slurp()" ought to do what lines() currently does.
p6eval rakudo 894e79: OUTPUT«[2, 3, 4, 0, 1]␤»
pmichaud i.e., it should return a list of elements, and we have something else that means "give us the string"
DarkWolf84 will a $var wich is Any be GC
pmichaud rationale: slurpy arrays and slurpy hashes are lists of elements
DarkWolf84 sorry for the bad english
masak DarkWolf84: no, it won't.
moritz_ setattribute self ['Regex';'Match'], '$!from', from 21:08
DarkWolf84 too bad
moritz_ The opcode 'setattribute_p_kc_sc_p' (setattribute<4>) was not found. Check the type and number of the arguments
masak DarkWolf84: easiest way to make a var go out of scope, is to end the scope :)
moritz_ oh, forgot a comma
DarkWolf84 yeah the local vars help
masak rakudo: my @a = 1, 2; @a.=[1]; say @a.perl
p6eval rakudo 894e79: OUTPUT«===SORRY!===␤Can not use .= on a non-identifier method call at line 11, near "; say @a.p"␤»
jnthn If you have some large data structure in a $var and you undefine it, what it used to hold will be GC'd. 21:09
pmichaud moritz_: won't .new initialize the attribute anyway without having to do something special?
21:09 pmurias left
masak I like this error message better. 21:09
DarkWolf84 that means now REPL will memory leack over time
masak right. the difference between variables and values.
pmichaud wonders about "REPL will memory leak..."
masak DarkWolf84: not any worse than the corresponding program would. 21:10
21:10 tylercurtis joined
moritz_ pmichaud: it does not 21:10
rakudo: say Match.new(from => 1, to => 5).perl
p6eval rakudo 894e79: OUTPUT«Match.new(␤ # WARNING: this is not working perl code␤ # and for debugging purposes only␤ from => Any(),␤ orig => Any,␤ to => Any(),␤)␤»
pmichaud moritz_: hmm, that seems odd.
DarkWolf84 I think there should be some way to GC the defined objects/vars 21:11
pmichaud DarkWolf84: undefine() will do that.
tylercurtis Hi.
moritz_ rakudo: class Parent { has $.bla }; class Child is Parent { }; say Child.new(bla => 5).bla
p6eval rakudo 894e79: OUTPUT«5␤»
pmichaud moritz_: I'm a little surprised by that.
jnthn I think that's spec.
moritz_ pmichaud: maybe it's because Regex::Match is a foreign object?
pmichaud I mean I'm surprised that it doesn't work for Regex::Match
jnthn oh 21:12
lue rakudo: my $a; my $a = Mu; say $a
pmichaud Regex::Match isn't all that foreign, though.
p6eval rakudo 894e79: OUTPUT«===SORRY!===␤Redeclaration of symbol $a at line 11, near " = Mu; say"␤»
jnthn pmichaud: Yeah, see your point.
lue rakudo: my $a; $a = Mu; say $a
p6eval rakudo 894e79: OUTPUT«Mu()␤»
jnthn is also a bit surprised.
pmichaud I mean, IIRC the default BUILD just walks through the objects attributes and does initializations.
a-ha
maybe the problem is that Regex::Match needs a BUILD
masak rakudo: multi fib(0) { 0 }; multi fib(1) { 1 }; multi fib($x) { fib($x - 2) + fib($x - 1) }; say (0..10).map(&fib).eager; # did anyone submit this to RT?
p6eval rakudo 894e79: OUTPUT«Method 'count' not found for invocant of class 'Perl6MultiSub'␤ in main program body at line 1␤»
jnthn pmichaud: oh, that makes sense. 21:13
moritz_ rakudo: say Regex::Match.new(from => 3).from
p6eval rakudo 894e79: OUTPUT«Null PMC access in type()␤ in main program body at line 11:/tmp/Rtki_Mk9Tc␤»
masak submits rakudobug
(the Null PMC access one)
moritz_ masak: don't. Regex::Match isn't spec
masak oh, right.
moritz_ retires
masak but shouldn't we eradi... oh, forget it.
pmichaud we should be able to get Regex::Match to disappear at some point, yes. 21:14
masak I meant the Null PMC access.
submitting tickets on those are reflex nowadays.
and I'm not sure "this class is not spec" is a justified excuse not to.
pmichaud well, if Regex::Match goes away, you won't be able to .new to get the NPA in the first place. :-) 21:15
masak I realize that. just like every other *bug* *fix*. :P
is there a "Regex::Match should go away" ticket in RT? 21:16
pmichaud no. But if you want to start filing those, I can give you a long list.
masak huh.
pmichaud there's quite a bit that's "visible" currently that perhaps ought not to be, or that isn't really part of the spec. 21:17
like many of our "helper subs"
masak nod
DarkWolf84 rakudo {my $a = 42; $a.say;}; $a.say
oops
rakudo: {my $a = 42; $a.say;}; $a.say
p6eval rakudo 894e79: OUTPUT«===SORRY!===␤Symbol '$a' not predeclared in <anonymous> (/tmp/iFdccWkSCe:11)␤»
DarkWolf84 yes
rakudo: my $a = 42; $a = Nil; $a.say; 21:18
p6eval rakudo 894e79: OUTPUT«Any()␤»
DarkWolf84 that's what I mean 21:19
$a will not be GC-ed
TimToady if "io" is the ingy-esque lazy IO thingy, then it's just: for io.lines {...}
masak DarkWolf84: *variables* aren't GC-ed. *objects* are. 21:20
21:20 wp6ft joined
lue DarkWolf84: If I may ask, why do want to delete a variable? 21:21
masak TimToady: it feelsn't like a straight line between $*ARGFILES.IO and that.
lue (do *you* want) 21:22
21:22 wp6ft left
DarkWolf84 masak: but aren't they leacking the memmory that way 21:22
21:22 shade\ left
DarkWolf84 s/memmory/memory/ 21:22
Any is an object isn't it? 21:23
masak DarkWolf84: ok, so "leaking memory" usually refers to allocating something and then never de-allocating it. just making sure we're agreed on that.
DarkWolf84 yeah the C pointers are that way 21:24
masak DarkWolf84: Any is an object. one which has plenty of live references to it in a Perl 6 program :)
for example, many routines in the setting have Any as a parameter type.
especially as it's the default type for parameters. 21:25
cognominal I don't find Knuth funny, he seems embiterred by all the new technology and marketting. Like if everything new was bad. Sounds similar to some perl5 crowd taking cheap shots at Perl 6.
masak cognominal: I didn't find it bitter. but he does put it in some well-needed perspective.
21:26 tadzik left
PerlJam I didn't get "bitter" from Knuth either. 21:26
DarkWolf84 masak, I think I've got the idea 21:27
21:27 Guest43164 left
DarkWolf84 $foo just points the Any() object when it's undefined 21:28
masak actually, I couldn't decide if an iTex living up to the spec in the presentation would be the Last Application Ever (and as such, fully desirable), or just Modern Tech Buzzword Overload. I kept flip-flopping between the two.
Tene I still don't follow what's happened with Knuth. Did he actually release something new, or not? I've seen a couple of links to a video, but I haven't had time to sit down and watch a video yet.
cognominal Except at the language level (objC sigh), I find the iStuff good. I wish free software facing end users to be half as good.
masak Tene: there won't be any doubts after you watch the video :)
Tene :) 21:29
21:32 unik left 21:33 Ross left
spinclad .u CHI 21:37
phenny U+2627 CHI RHO (☧)
masak as in "CHI RHO practor".
spinclad i thought the saddest thing (well, going to the opposite of FLOSS aside) was his caving on how to pronounce '\c[GREEK UPPERCASE CHI]'
masak spinclad: he failed to pronounce it 'x' a number of times after that himself :P 21:38
lue when I first read of iTex, I had to sit there for a moment.
spinclad but he seemed to keep forgetting to go with the vulgus mobile
21:39 silug left
cognominal masak++ # a good pun is a bad one. 21:39
lue I haven't seen the videos (and can't), but iTex sounds like a joke (based on a short description, of course)
masak cognominal: wasn't aware I made one. but thanks.
lue: you should also see the vid, to be completely sure. 21:40
spinclad CH R istian science: CHI RHO praxis
masak pmichaud: did that IO::Prompter module ever make it online?
lue if only I could... even if isn't flash, my poor old laptop can't handle good video. 21:41
[ would you like to make a donation? :) ]
masak I can give you a hug.
hugme: hug lue
hugme hugs lue and blushes
lue (I'm not a fan of Flash anyway) 21:42
21:46 alester joined, rhr left, ash_ left, LionMadeOfLions left, athomason left 21:47 shade_ joined 21:49 ruoso left, rhr joined, ash_ joined, LionMadeOfLions joined, athomason joined 21:53 rhr left, ash_ left, LionMadeOfLions left, athomason left
spinclad masak: re gist.github.com/464349, a sweet improvement. One note: "This will need to fall back to hash access [on] indirect lookup ($::'$a')": (more) 21:56
masak waits, hopefully, for more 21:59
spinclad seems to me this involves a static hash from strings ('$a') to accessors ($access = {[0, 0]}) and a way to apply them (fetch $access; store $access, $2)
21:59 rhr joined, ash_ joined, athomason joined
spinclad and the static hash goes in the static lexpad 22:00
masak hm.
that seems right. 22:01
spinclad (i don't expect this literal use of $access; i expect it would have to get compiled into a few operations each)
masak I will remember that for when we do indirect lookup.
slavik what is yapsi? 22:05
22:05 skangas left
masak slavik: it's a compiler/runtime I'm writing on my spare time, along with a bunch of other people -- notably snarkyboojum++ and isBEKaml++. 22:07
spinclad yapsi is Perl 6's first 'complete', 'official' implementation (see masak's release announcements: use.perl.org/~masak/journal/40429) 22:08
masak oh yeah, that too. :P
in the long run, Yapsi's PR staff is so screwed. :) 22:09
tylercurtis masak: Aren't you Yapsi's PR staff?
22:09 skangas joined
masak tylercurtis: d'oh! 22:09
DarkWolf84 that's a secret 22:10
:)
masak speaking of which -- download Yapsi today! github.com/masak/yapsi/downloads 22:11
you know you want to. :>
hm, I just noticed that the tags have 'tgz'/'zip' links. do I really need the download files too? 22:12
I guess they form a way to post-process the git repo a bit... 22:13
pugssvn r31569 | lwall++ | [IO,Str] regularize .lines to be more like .words when fed a string
r31569 | .lines will no longer interpret a string as a filename
masak TimToady++ 22:14
TimToady: why not just 'is export' on the method form? 22:15
TimToady er, copy/past from words?
masak same question for words, then :P 22:16
TimToady feel free, I'm geezernapping &
masak makes a guerilla edit 22:17
tylercurtis rakudo: (1,2,3 Z, 4, 5, 6)[0].perl.say
p6eval rakudo 894e79: OUTPUT«(1, 4)␤»
tylercurtis rakudo: (1,2, 3 Z 4,5,6)[0].perl.say 22:18
p6eval rakudo 894e79: OUTPUT«1␤»
22:18 kensanata joined
tylercurtis According to S03, Z is a shortcut for Z,. 22:18
22:19 hercynium joined
pugssvn r31570 | masak++ | [S32/Str] saved a few words and lines with "is export" 22:19
masak tylercurtis: that looks familiar. but feel free to submit to RT anyway.
tylercurtis: ah, here: rt.perl.org/rt3/Ticket/Display.html?id=73948 22:20
tylercurtis masak: Your RT-fu is impressive. 22:22
masak :) "Z," was unusually easy to search for.
the RT search thingie is idiosyncratic, but in predictable ways.
in particular, it has no metacharacters and doesn't care for word boundaries. 22:23
22:24 ash_ left 22:27 rhr left, rhr joined
tylercurtis rakudo: my @foo; push(@foo, 1, 2, 3); @foo.perl.say 22:28
p6eval rakudo 894e79: OUTPUT«[1, 2, 3]␤»
tylercurtis rakudo: my @foo; @foo ,= 1, 2, 3; @foo.perl.say
p6eval rakudo 894e79: OUTPUT«[1]␤»
tylercurtis S03: "@foo ,= 1,2,3 # same as push(@foo,1,2,3)"
masak well spotted. 22:29
masak submits rakudobug
tylercurtis++
22:31 justatheory left
tylercurtis rakudo: (1,2,3 Z, 4,5,6) >>+<< (7,8,9 Z, 10,11,12) 22:34
p6eval rakudo 894e79: OUTPUT«Sorry, sides are of uneven length and not dwimmy.␤ in 'hyper' at line 179:CORE.setting␤ in main program body at line 1␤»
tylercurtis rakudo: say (1,2,3 Z, 4,5,6).elems
p6eval rakudo 894e79: OUTPUT«3␤»
tylercurtis rakudo: say (7,8,9 Z, 10,11,12).elems 22:35
p6eval rakudo 894e79: OUTPUT«3␤»
22:37 athomaso1 joined 22:38 athomason left
masak TimToady: `Int $limit = *` typefails in Rakudo. was that a thinko, or is there something Rakudo should know about Int and Whatever? 22:43
TimToady: giving it a bit of thought, I think this is a case where $limit shouldn't be typed Int. as usual in these cases, it's more a question of a coercion than a typing. 22:44
jnthn It should fail 22:45
And I'd not type it
Also don't "as Int" it unless you want *.Int to happen though ;-)
*too 22:46
masak o.O
22:46 skangas left
masak I don't :/ 22:46
jnthn Then don't "as Int" it :-)
Default values are subject to type checking and type coercion. :-) 22:47
masak I've just written a small patch that effectuates TimToady's spec change. spectesting.
pmichaud I need to read that diff.
masak I didn't "as Int" it. :)
pmichaud (the spec change diff, not the patch diff)
masak :)
pmichaud yes. I think that one of the things that Perl 6 programmers will quickly re-discover is that strong typing can be very limiting. 22:48
Tene masak: so wait, did you "as Int" it, then? 'cos if you do that, .Int will happen.
masak there's definitely a way to overuse typing in Perl 6.
<masak> I didn't "as Int" it. :) 22:49
Tene: no.
in the end I used no coercion, and no typing, and I let the default be '*'. 22:50
22:50 _mpu left
masak the method I wrote delegates to Str.comb anyway, which likely does something sensible with $limit. 22:50
masak checks
er, Cool.comb 22:51
heh, Cool.comb also doesn't type/coerce its $limit param, and it seems to DTRT no matter what it gets. 22:52
22:52 skangas joined
Tene masak: did you try passing it a null pmc? 22:52
22:52 cdarroch left
pmichaud seriously considers trapping "Null PMC" exceptions and changing them to some colorful phrase instead 22:53
no, wait
even better
Null PMC exceptions should simply provide the url or script whereby masak++ can automatically post it to RT
masak :) 22:54
pmichaud so, Str.lines breaks a string into lines, similar to what IO.lines does?
Tene base64encode the stack trace, and print a url that will submit it to RT when clicked, of course.
pmichaud Tene: ...if only RT accepted submissions via the web....
masak what's the pir:: for getting a Null PMC again? 22:55
pmichaud so, it prints a url that goes to a bug proxy that converts it into an email and submits it to RT
frettled pmichaud: couldn't Null PMC exceptions automagically post to RT? :D
pmichaud masak: pir::null__P()
Tene It wouldn't be outside the realm of possibility to set up some kind of submission service on feather.
pmichaud frettled: that would just be evil.
frettled oh, befuddled lag
pmichaud: yes!
masak rakudo: say "foo".comb(/o/, pir::null__P())
p6eval rakudo 894e79: OUTPUT«Null PMC access in can()␤ in main program body at line 1:/tmp/QJCtpureJq␤»
Tene frettled: it's impolite to make internet connections without permission.
pmichaud and the submission server could also check to see if this particular Null PMC exception has been reported and not re-post it 22:56
masak Tene: seems it never reaches the .comb method.
frettled pmichaud: mmm
Tene: what if rakudo has a letter from its parents saying that it's okay?
pmichaud the rakudo will have been a naughty girl. 22:57
*then
masak is it wrong to find the spectests that need changing by changing the code and seeing what breaks? :) 22:58
jnthn masak: We could call it...Implementation Driven Development!
It's like TDD in the mirror.
masak 哈哈
DDT? 22:59
Development-Driven Development.
lue
.oO(Should Be Fine™ development)
 23:00
frettled lue++
masak Drive-By Development.
jnthn I Accidentally Development 23:01
frettled Development By Proxy
lue If it works great for us, it's a test. If it doesn't, it shouldn't be there :)
(IIWGFUIATIIDISBT) 23:02
masak jnthn: :D
tylercurtis Acronym-Driven Development
or.. ADD
masak Driven-Insane Development. 23:03
lue Why yes I DID
23:04 lestrrat is now known as lest_away
masak Didn't-I-fix-this-yesterday-Driven Development 23:04
lue aka Parodoxically-Driven Development
frettled Déjà-vu Development 23:05
jnthn masak: I think that's could "You-shoulda-used-TDD" ;-)
*called
masak :)
lue
.oO(Let's-Steal-From-The-Future-We-Did-Will-Do-It-Already Development)
frettled Bikeshed-Driven Development 23:06
lue [ tenses may be incorrect, please ignore that :) ]
frettled lue: don't be so tense
23:06 sawyer_ left
Arthur_TwoSheds_ What's this about a shed? 23:06
lue (it should've been Arthur_TwoSheds_Jackson) 23:07
masak the more rigorously I use TDD, the more I feel they settle into the role of a "sure, good, but you forgot this aspect of things" advisor.
frettled masak: perhaps it's time for DDT…
…Development-Driven Testing
masak which is a really good advisor to have when writing code.
frettled it's the new killer project management
masak frettled: only write one, and *exactly* one test after you've written some code? :P 23:08
lue RTD (Rapid Tuit Development)
sorear masak: Recommended reading for IRC helpers: www.shadowcat.co.uk/blog/matt-s-tro...a-bastard/
masak sorear: thanks. read it.
lue
.oO(OK, I've coded this awesome macro implementation for rakudo....
 23:09
23:09 masonkramer joined
masak sorear: I understand where he's coming from, but I'm not sure I have it in me to assume the role of a channel bastard myself. 23:09
sorear pmichaud: yes, that is the plan
lue ....I'll add this test to the macros tests: is 2+2, 4, "complex macro wizardry"; )
masak sorear: definitely not through privmsg, which has more of a booth-like feel to it.
sorear: the closest I got was tossing catb.org/esr/faqs/smart-questions.html at the fellow. 23:10
23:12 justatheory joined
lue
.oO(I should work more on my Time Traveller's Tenses Handbook before attempting to pull off that kind of thing again)
 23:13
frettled masak: Or, to quote someone who may very well have been Don Knuth in a sardonic mood: «if your code is correct, you don't need tests» 23:14
lue: Recommended starting point for that: www.google.com/url?sa=t&source=..._CoSaOfKag 23:15
artghle, sorry, I forgot about googles effin' way of handling URLs.
lue: en.wikipedia.org/wiki/James_while_J...he_teacher
23:15 justatheory left
masak frettled: sure. it's just a big 'if'. 23:15
frettled: I find tests let me lower my guard when coding, and get away with it. feels less tense. 23:16
frettled masak: which is why i kindof suspect that it was a sardonic remark :)
masak: I tend to add a lot of in-code tests, yet I somehow seem to slip quite often, leaving uncaught test cases, heh. 23:17
lue frettled: want me to link you to one big inspiration for this book? *evil grin*
sorear pmichaud: meant that for pmurias
frettled lue: Perhaps, I could need som help in getting to bed two hours ago :)
pmichaud sorear: I guessed that eventually. It's happened to me and pmurias a lot. :)
masak frettled: "in-code tests"? as in, mixing code and tests?
lue frettled: tvtropes.org/pmwiki/pmwiki.php/Main...nseTrouble *evil laugh* 23:18
frettled masak: yes, but not tests as in test-framework tests, since most of my code isn't about re-usable modules, but rather scriptish programs.
lue: uh-oh, TV Tropes...
lue (quite seriously, I was spending one late night browsing Time Travel Tropes when the book idea hit me)
23:19 sawyer_ joined 23:20 IllvilJa left
masak frettled: would be cool to have a testing framework that disrespected the reduced visibility of things like 'my' subs. :) 23:21
then they could be tested more easily.
lue
.oO(lexcial subs, the enemy to unfudged binding tests)
 23:22
masak also, private methods.
lue should really get to work on _something_. Stupid quantum feelings. 23:23
23:24 skangas left
frettled masak: ooh aah :) 23:24
23:24 IllvilJa joined
sorear lue: for someone who claims to be a hhgttg fan, you needed tvtropes to get the book idea? 23:25
lue I haven't read the first 5 of 3 in a looong time, and I only read 6 of 3 a little while ago. 23:26
23:27 sawyer_ left, sawyer_ joined
lue not only that, but at the time I had Doctor Who on the brain, which led me to the Time Travel Tropes in the first place :) 23:27
23:30 ashleydev_ joined, ashleydev left, ashleydev_ is now known as ashleydev 23:32 sawyer_ left, sawyer_ joined 23:33 rv2733 joined 23:40 tedv joined, masonkramer left, masonkramer_ joined, masonkramer_ is now known as masonkramer
masak heh -- every spectest file that has failed spectacularly (except for t/spec/S05-mass/properties-derived.rakudo, a usual suspect) has had the old 'lines' or '.lines' calls in it. :) 23:41
jnthn :-)
lue :) .oO(hrm, spectacularily not ok 63) 23:42
masak lue: yeah, like that. :)
pugssvn r31571 | Kodi++ | [S32] Dropped mention of quarters from Temporal.pod. I don't think quarter-numbering belongs in the core, since (as far as I know) there is no precise standard way of dividing calendar years into four. 23:43
masak hm, Kodi isn't on the channel.
seems dividing by triplets of months is sane enough for DateTime. 23:45
pugssvn r31572 | Kodi++ | [t/spec] Cleaned up DateTime.t a bit. (I hadn't meant to commit it yet.)
23:46 skangas joined, proller left
masak oh well, I won't miss quarters greatly. if no-one else kicks up a fuss about it, I won't either. :) 23:46
23:47 DarkWolf84 left, silug joined
jnthn Aww, I thought somebody was about to get hung, drawn and quartered ;-) 23:47
masak jnthn: well, can't do that after r31571, can we? :P
frettled can we settle for my 2 cents? 23:48
23:49 LionMadeOfLions joined
masak absolutely. 23:49
23:49 rv2733 left
lue rakudo if pir::getprop('ro', $target) { punish("severly") } 23:49
frettled masak: but, ye gods, does that mean I'll have to find when March ends and April starts all by myself? :D 23:51
23:51 silug left
masak frettled: that was your two cents? just checking. :) 23:52
frettled Yup, all out. Oh. D'oh!
masak frettled: you can still create a DateTime object for April 1, and check the day-of-year property on it.
in other news, we have too many trig tests. 23:53
jnthn masak: It's not a huge leap to realize that this is not going to be so simple for some years though. ;-)
lue wonders if any part of DateTime is suitable to (sim|em)ulate a CPU clock...
sorear STD.pm6 has lots of ¢ you may be able to borrow
lue: no
DateTime handles the insanity that is civil time 23:54
masak jnthn: it's not intrinsically harder for quarters than for months.
23:55 Psyche^ joined 23:59 Patterner left, Psyche^ is now known as Patterner
lue (not that the CPU clock is terribly important. I have to worry about VBlanking and mid-frame trickery and such.) 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!