6.2.3 is airborne! | pugscode.org <Overview Journal Logs> | pugs.kwiki.org
Set by autrijus on 12 May 2005.
svnbot6 r3422, putter++ | Tweaked and unDisabled two t/rules/from_perl6_rules/ files. 00:00
r3423, mugwump++ | It's the Unicode ââ£â®âââªâ«â®â¼ 00:06
putter mugwump: dont forget generalized union/intersection! :) 00:15
mugwump: ping? 00:20
has anyone seen anything on, i'm not sure what to call it, "anti duck typing"? "aggressive multimethods"? eg, "I, a user, don't care if class C has method m. I "use Picky;", and if the picky folks havent approved C's m, I don't even want to be able to see it.". ? 00:47
sproingie eh?
generic functions?
got a usage example? could probably tell you the jargon. 00:48
putter the idea is to completely control what a method name means in a lexical context. which means hiding any external multimethods, "unapproved" object methods, etc. 00:57
sproingie so it would actually dynamically hide methods from another class. well that certainly has little precedent, so it doesnt have a name, i'd imagine 00:58
but any reasonably reflective metamodel or generics system could do it
with perl5 you could use Hook::LexWrap to override C's methods 00:59
meppl gute nacht 01:01
good night
putter the motivation is that as folks move to p6 and do a _lot_ more "domain specific language"-style programming, CPAN'ish classes crashing the party with their own little method-api-languages are going to stick out like a sore thumb. 01:02
meppl: 'night 01:03
meppl ;) 01:04
putter controling ones own naming means being able to frob everyone elses, even objects which are just wandering through.
use RubyMethodApisForAllCoreObjects; ;) 01:07
sproingie there's always going to be a clash of cultures insofar as modules go 01:08
this is the story with CPAN now
i think you'd probably want a module that adapted another, not mutated it 01:10
modules that pollute global namespace will be unpopular just as they always are
mugwump use Set-(Any)-cpan:FOO; 01:13
putter mugwump: huh? 01:14
mugwump don't like namespace pollution? set your perl to use a different default module namespace (the cpan:FOO bit) 01:15
putter: what did you mean by "generalised union/intersection", putter ? 01:16
s/,.* //
putter eg, 22C2 01:18
mugwump n-ary intersection?
putter intersection over a set of sets 01:19
mugwump well, for a start, because I don't have a non-unicode-operator implementation :)
putter (it was mainly just an excuse to throw in more unicode...;)
mugwump true
I mean, ⊨ 01:20
sproingie intersection over a set of sets would be a reduction
dont know what the math notation for reduce is, if there is one
mugwump would it? sounds sufficiently vague to mean just about anything to me ;)
sproingie eyes his irc program. no unicode support, and it's a windows app? bah
putter re, namespace, say I want to do some set algebra, but I want all my ops to be lazy. you provide a lazy union, but your "union" method isnt. I would like to "use Notation::Set::AllLazy;" and be able to use your implementation, without directly using your names. 01:22
Someone other than me can keep track of all the implementations and the characteristics of their assorted methods. 01:23
sproingie sounds like a classic problem of writing an adaptor 01:24
putter I can simply say "In this block of code, I want to use this method language/vocabulary, with these implied semantics, so "use" it." 01:25
sproingie that'd be a pragmatic module, most of those have lexical effect 01:26
s/effect/extent/
putter re adaptor, but the combinatorics are all wrong. the space is too highly dimensional. so you would have to use adaptor generators, which can be handed around indistinguishably from their wrappees... not going to happen. 01:27
{ use Way1; ... { use Way2; ... } ...} adaptor classes arent going to do it. 01:28
pragmatics are closer...
mugwump How is that different from a Role? 01:30
putter sproingie: sorry, but yes. adapters _do_ address the same problem. their just so... not good enough. ;) (see previous assorted exclamations of "software sucks";)
roles are still under the control of the class creator. the idea is the class user should be able to muck with how the class appears to her. 01:31
loosens the tie between "they created an implementation" and "they created an api". 01:32
mugwump ok, I understand, I think 01:33
What you want to be able to do, is compare Roles based on their definitions and not on their 'is-ness'
so, you could define a role, and so long as the thing that implements it is a superset of the role in all regards, then it can be used 01:34
but if they break the interface it won't work
putter reads the paragraph again... 01:35
mugwump like, a version of .does() that performs a deep method signature check, not just checking that the role is linked
putter signature checks cant do semantics. as a silly example, one might "use OnlySeeO(n^2)AndBetterMethods;"... 01:38
In an extreme case, some object's class has methods m and n, but in some section of code, I want m to mean n and visa versa. 01:39
what triggered all this musing was looking at Set.pm, and thinking, here we go again, short human-friendly method names, hello years of pain. 01:43
;)
sproingie different people will write different interfaces. and since most perl6 modules won't use generics, you won't get much chance to override them seamlessly 01:44
best you can do is rewrite or adapt, for the most part. i think a general case solution would be something of a software engineering silver bullet 01:45
putter mugwump: no criticism of Set.pm intended. its nifty. my reaction was just "here is something nifty and... pure/non-hairy/simple, and even here we are still going to get burned". 01:49
sproingie: might be sufficent to have "undef &foo;" and a multi sub foo() is run_in_preference_to_object_methods. 01:51
anyway, back to work. 01:53
mugwump: thanks for the module namespace pointer... 01:54
to rephrase an old objection to smalltalk, "+ is not a method, and if some object thinks it is, I wish to be able to successfully disagree in my own code" ;) 01:56
'night folks 01:57
taki hi all
what mean /i in this syntax ([0-9a-f:]+)/i)
02:00 sbkhh is now known as Odin-
robkinyon hello? 02:45
robkinyon has a new chat client ...
svnbot6 r3424, mugwump++ | Add shiny junction-based Set implementation 02:50
putter just for closure, I mention that having groveled over S12, it looks like the games described are already possible. ;) 02:51
mmm, new Set...
stevan mugwump++ # Junction based Set is verrrrry cool 03:08
mugwump glad you like it. all that hash key nonsense goes away too 03:10
stevan very nice 03:23
svnbot6 r3425, putter++ | Fixed typo in trig.t.
mugwump It's a shame that you can't remove items from a junction easily 03:24
eg, that remove could be something like;
$:members = $:members & none($to_remove); 03:25
but it doesn't work like that :)
stevan that would be nice
svnbot6 r3426, mugwump++ | !($x =:= any(@x)) â ($x =:= none(@x)) 03:33
stevan mugwump: I followed your advice on the eval-into-pugs-object-space 04:02
svnbot6 r3427, Stevan++ | Perl::MetaModel - experimental version of compile() to complile classes to the Pugs object space; compiled classes support meta() and (my version of) isa(); more to come tomorrow,... must,.. get... sleep :)
mugwump choice
stevan I am too sleepy to do it right, but that is a proof of concept for now :)
I think I will create a Perl::Meta::Complier tomorrow which does it better 04:03
(and I think the meta() and isa() functions are not correct per-S12 either) 04:04
but I need sleep, so it will have to wait another ~8 hours
mugwump My gut feeling is that the introspection stuff in S12 is only very lightly considered 04:05
stevan Smalltalk Metaclass model ++
mugwump: yes, I reviewed it earlier and felt the same way
autrijus thinks our docs/ should eventually be added onto the end of S12
not yet of course 04:06
anyway must sleep
mugwump 'night!
stevan adios &
svnbot6 r3428, putter++ | Added atan(_) and fixed broken atan(_,_) tests. 04:30
eric256 did the journal just die? 04:31
svnbot6 r3429, putter++ | Fixed typo in Perl-MetaModel/README. 04:38
eric256 is in love with making his own operators... ;) 04:42
anyone know of a precompiled parrot for windows? or where i should look? 05:10
google to the rescue 05:13
hmm or not. only one i found is too old.../me continues the search.....seems kinda sleepy around here 05:15
f0rth I've searched but found nothing 05:16
svnbot6 r3430, eric256++ | Added examples of overloading. Should be fleshed out a bit but it's a start.
eric256 www.jwcs.net/~jonathan/perl6/ <-- apears to work. but i'm still downloading
ack.. somewhere i stopped reading and downloaded pugs. lol. 05:17
f0rth :p 05:18
mugwump what does the .perl method do? 05:21
eric256 looks like i'm out of luck for now. and i was just hoping to be able to split a string. ;) 05:22
officialy? i dunno. but it returns a fairly intuitive representation of the data. it is entirely possible that i shouldn't be using it though
mugwump oh, right. probably a pugs-specific hack 05:24
castaway_ .perl does a data-dumper like dump.. I think its p6 05:25
05:25 castaway_ is now known as castaway
eric256 dev.perl.org/perl6/doc/design/syn/S02.html 05:26
mugwump I stand corrected.
cool
eric256 i'm learning to use google to search the online documents 05:27
isn't % mod? 05:45
mugwump sure 05:46
or symmetric difference on Sets :)
( $s1 \ $s2 ) U ( $s2 \ $s1 )
eric256 any ideas why my $t = $M % $N; causes a sub to fail to compile 05:47
lol nevermind. i overroad % in postfix...i think that messed up the infix version. going to test
hmm. i would have thought you could have an operator mean one thing in infix and another in postfix... 05:50
svnbot6 r3431, eric256++ | Some minor additions 06:16
obra hm. That commit message could be better :/ 06:19
svnbot6 r3432, mugwump++ | Express some core Perl entities as Perl code; this should be instructive/educational in developing the MetaModel. Rough, rough sketches.. 06:22
brentdax And *that* is why you don't put "disconnect" right under "part", and why you put an "are you sure?" prompt on said "disconnect" command. 06:26
svnbot6 r3433, scook0++ | Really flaky hack for munge_haddock.pl that lets Haddock 07:00
r3433, scook0++ | document all the private functions.
autrijus rehi! 07:02
<- reading Ecma-335 spec about CIL
theorbtwo Morning, autrijus.
autrijus ponders Pugs.Compile.CIL just for fun.
theorbtwo You have a very strange definition of fun, autrijus. 07:05
CIL is the .NET VM, right? 07:06
castaway Do .JVM while you're at it? ;)
autrijus well, seeing how the codegen is already there in Modrian and Haskell for .NET
nothingmuch morning 07:09
scook0 theorbtwo: you wrote munge_haddock.pl, right?
theorbtwo Yeah. 07:10
scook0 I changed it so that it also nukes long-style module declarations
theorbtwo I took a quick look at that. 07:11
Looks sane.
scook0 It /barely/ works, but I think it's good enough for now
(my 2nd ever Perl5 program :-) 07:12
theorbtwo I'm not 100% with it at present.
scook0 Are you aware of the problem that explicit module export declarations cause?
Normally, if you just write "module Foo where", all functions in the module are considered public 07:13
theorbtwo The haddocks only show the exported stuff, since that's what the haddock author thinks they're for.
scook0 which is not what we want, right?
Because we're using it for internal documentation, not library documentation 07:14
nothingmuch autrijus: batch pingback
theorbtwo Exactly.
scook0 so what my code does is:
first, it scans through until it finds the module declaration
it extracts the actual module name 07:15
autrijus nothingmuch: yes?
bah. Mondrian doesn't seem to run on Mono
scook0 and prints out "module Foo.Bar where"
nothingmuch tell me when you feel like continuing the internals chat
autrijus and the rotor license is not acceptable
autrijus suspends thinking about Pugs.Compile.CIL
scook0 then it chucks away lines until it hits /where$/, which marks the end of the old declaration 07:16
theorbtwo Works for me.
scook0 I just thought somebody other than myself should know about it, in case it breaks
theorbtwo nods.
I only see what looks like odd corner-cases where it'll break badly, though. 07:17
(Modules trying to export functions named where, or people using really strange formatting.)
scook0 I had to keep tweaking it until it worked with all our existing source
Yeah, that's why it's so flaky 07:18
But hopefully nobody will be silly enough to trigger them...
autrijus nothingmuch: I need to run for $work. will ping you when I get next round of tuir 07:25
tuit, even
&
svnbot6 r3434, autrijus++ | * Upgrade to Syck 0.55 from «why the lucky stiff».
nothingmuch ciao
autrijus btw, I like how www.mono-project.com/Main_Page is organised. 07:27
brentdax MediaWiki...heh, funny ow the same things pop up everywhere.
autrijus it looks professional without much customisation
well, you can't do much customisation with it anyway. 07:28
;)
brentdax Munging MediaWiki has convinced me that PHP is evil, incidentally. And that I'd like to write something nice in Perl 6 so I don't have to use it anymore. 07:29
autrijus port Kwiki to Perl6? :) 07:30
Perl6: Spiffier than Spiffy!
probably just need to port Spoon to Perl6
brentdax The problem with Kwiki is that it suffers from Ugly Wiki Syndrome.
autrijus everything else would follow
svk.elixus.org is ugly? :)
brentdax Symptom: it doesn't pretend to not be a wiki.
autrijus it's not exactly beautiful, but seems workable 07:31
a Mediawiki skin would be useful as well.
brentdax Oh, yes, it's workable. You just wouldn't want to make Wikipedia in it.
autrijus sure.
brentdax I found out that Mom has used Wikipedia to research her books. She didn't even realize it was maintained by the general public.
(She writes historical novels set in England.) 07:32
autrijus nice!
castaway then write a Kwiki::Theme module or something (assuming there arent any already. ,) 07:33
brentdax But when you look at even the SVK site, there are a lot of awkward things about it. Like camel-case links, for example, although that's not the only one.
autrijus verily, verily.
really & 07:34
brentdax Heh, kay.
theorbtwo G'work.
b6s 15:41 <@mjhsieh> b6s: /join #pugs /me crys WE NEED Net::Socket, PLEASE!!
b6s flee
brentdax Is there any sort of file I/O in Pugs, or have you guys been waiting till you have some idea what it'll look like? 07:35
castaway right, but Kwiki doesnt insist on CamelCase links at all.. (tho it would be nice to be able to turn them off completely)
castaway runs off to $work 07:40
Aankhen`` PHP is a language with a serious identity crisis. 07:41
brentdax How so?
Aankhen`` It desperately tries to be Perl and not be Perl at the same time.
brentdax lol
Aankhen`` I really dislike it. :-) 07:42
brentdax I don't blame you.
When I start working in it, I start thinking like it's Perl, but it's not, so I make lots of little mistakes and write myself into algorithmic corners. 07:43
Aankhen`` Heh.
The PHP community is bloody great.
brentdax Like...the only way to make an anonymous function is by passing a string with your code to create_function. But that's really equivalent to an eval(), and has a runtime penalty.
Aankhen`` If you ask someone for help with Perl, they'll help you find a place to learn to do it.
If you ask someone for help with PHP, they'll give code to copy and paste, and you'll be right where you were before you asked. 07:44
brentdax Heh.
Aankhen`` I guess people might like the PHP approach since it means less work for them. 07:45
I like the Perl approach since I want to keep improving as a programmer.
brentdax Well, only theoretically. In the long run it's not very nice.
Aankhen`` Indeed.
It's a very short term view of the situation.
The most messed up thing I ever heard of was an IRC bot for an MMORPG I play -- it was a real heavy-duty bot, with lots of nifty features -- which numbered thousands of lines of... wait for it... PHP. 07:47
brentdax Ouch. 07:48
Another channel I hang out in has a guy who's trying to write a PHP bot. I keep thinking "if you like pain that much, why not just use INTERCAL?" 07:49
Aankhen`` Websites written in PHP are usually laughable. Command-line PHP is definitely a case of a square peg in a round hole. PHP and IRC feels like the spawn of...
:-)
ROTFL.
People just seem to think that PHP is the solution to everything. 07:50
SM_ax while it is solution to nothing? :)
Aankhen`` Well, that's my opinion. :-)
brentdax One thing I think is interesting about Perl 6 is that if you can convince Perl that your entire code file is a giant literal string, it has much of the inside-outness of PHP.
Aankhen`` But I suppose "if it works for you" applies there. 07:51
brentdax >> Inside-outness?
brentdax ``<html>{ 1+1 }</html>``, instead of ``print "<html>", 1+1, "</html>"``. 07:52
Aankhen`` Ahh.
SM_ax Hmm, I've never seen a good PHP code or a good piece of software in PHP
Aankhen`` SM_ax >> One good PHP app is vBulletin.
WordPress is another.
(I don't like the language, but I'm not biased against the apps)
brentdax PHP's strength, I think, is that it makes it *really* easy to start small and grow from there. 07:53
Aankhen`` Yes.
But the problem is that it encourages many bad practices.
And that whole magic quotes thing... *shudders*
brentdax (Granted, once you've grown it falls on its face, but by then you're stuck.)
SM_ax brentdax, for some people it is weakness :)
brentdax Dear God, yes. That's the most hackish thing I've seen in a language in a long time.
Aankhen`` Heh, in PHP you have to STAY small.
brentdax MediaWiki's pretty big... 07:54
Aankhen`` Otherwise you just end up with a ton of spaghetti code.
I've seen quite a few large PHP projects.
Uuuugly.
"Code is poetry" just doesn't apply. :-( 07:55
brentdax 55.5kloc in includes/, which is most of the code. It's not too hard to find the section of the code you need to mess with.
Aankhen`` Hrm, I haven't seen the code for MediaWiki.
The greatest thing about Perl 6, IMHO, is pattern matching. 07:56
brentdax Yes. I actually mentioned that on wikitech-l, that when PHP-on-Parrot and Perl 6 are available, we can replace the creaky multi-pass substitution "parser" they currently have. (Heck, Perl 6 isn't even really necessary for that, as long as there's a library for PGE access.) 07:57
Aankhen`` Yeah.
PGE as a library would be so amazing.
Heh... PCRE seems so outdated compared to PGE. 07:58
Um.
Compared to Perl 6's pattern matching, rather.
brentdax Heh.
That's one of those parts of Perl 6 that I wanted *now* when I read about it.
Aankhen`` Me too!
Blasted Micro$haft Winduhs won't let me link Parrot into Pugs. >_>
No pattern matching for me. >_< 07:59
brentdax Ack.
Windoze--
Aankhen`` (--Windoze)--
brentdax Windoze -= Inf;
Aankhen`` Windows == -Inf;
brentdax And maybe when Ruby-on-Parrot's available, we can steal Rails. And there will be much rejoicing. 08:01
Aankhen`` Hehe.
Parrot++
PGE++
brentdax Parrot'll be great because it'll let me use all those other languages without having to actually *write* in them. 08:02
Aankhen`` I'm not such a great fan of Ruby-on-Rails, though, after I read their solution to the whole non-idempotent GET links thing.
Indeed.
brentdax (Of course, with the Rubyometer continually rising, we might as well be writing in Ruby.)
What's that?
Aankhen`` The solution? 08:03
brentdax Yes.
Aankhen`` One sec.
I'll have to dig up the link.
wolverian besides, we have Catalyst on CPAN. 08:04
why need Rails? :)
brentdax ...why don't people tell me about these things? 08:05
Aankhen`` brentdax >> community.moertel.com/ss/space/star...05-05-08/1
wolverian I discovered Catalyst only a while ago, too.
it is quite nice.
Aankhen`` In principle, it was a good solution, but now you're going to have most Rails users putting forms within forms within forms within forms.
brentdax Oh, God, it's the GWA thing. 08:06
Aankhen`` Yeah.
brentdax I wnated to kick *so* many faces in when people started complaining about that.
wolverian Aankhen``: there is no other solution to it.
Aankhen`` Yep.
wolverian >> I know, they did the best they could. 08:07
I just wasn't particularly happy with it. :-P~~~
wolverian why not?
Aankhen`` The whole idea of creating another form to put the button within... 08:08
So if you have: <form>...(button code goes here)...</form>, you end up with: <form>...<form>...</form>...</form> 08:09
wolverian is there another way to do it?
Aankhen`` No, like I said, they did the best they could.
The only other way would be to educate all their users.
Aankhen`` sniggers.
wolverian ..and they would have to do the exact same thing.
Aankhen`` Bleh, Pugs still refuses to link with Parrot. 08:10
:-(
brentdax What does it do? 08:12
pasteling "Aankhen``" at 203.101.19.91 pasted "Compiling Pugs with PUGS_EMBED=parrot fails" (30 lines, 1.1K) at sial.org/pbot/10245 08:14
Aankhen`` autrijus told me to use MSYS.
I don't understand it very well though.
brentdax Looks pretty nasty... 08:15
Aankhen`` I did download MinGW... all those files being looked for do exist, with the name libfoo.a, for example, libole32.a
Is <ws> = \s* or \s+? 08:35
brentdax It Depends. 08:36
<ws> is the same as the whitespace in a :w rule.
Aankhen`` On what?
Ah.
brentdax (iirc)
Aankhen`` Oh well, \s* does just fine.
brentdax Yeah, S5: / <?ws> / # match whitespace by :w rules 08:38
Aankhen`` OK, thanks.
autrijus brentdax: re: 08:43
"Is there any sort of file I/O in Pugs, or have you guys been waiting till you have some idea what it'll look like?"
we have open(), accept(), listen(), say, flush, etc
all sorts of things you can reasonably expect :)
there's plans for piped open and io() as well.
$work &
brentdax Alright, good, thanks.
Aankhen`` If a rule in a grammar invokes another rule, does the rule being invoked have to be defined before it? 08:46
brentdax Presumably mutually recursive rules are legal... 08:47
Aankhen`` Okay.
Is there any list of builtin rules, like <alpha>? 08:54
10:54 Aankh|Clone is now known as Aankhen``
svnbot6 r3435, iblech++ | * Usual svn properties and EOL at EOF. 11:25
r3435, iblech++ | * Updated svnlog2graph.pl's scaling as we're close to r3500. :)
scook0 returns 11:27
11:44 chady is now known as chady_
castaway ruffles Limbic 12:00
Limbic_Region is in the middle of his morning ritual of catching up on the news castaway
Limbic_Region hugs castaway anyway then wanders back off
castaway :) 12:01
autrijus greetings! 12:09
autrijus is off-$work
castaway grats :) 12:10
plus6 right? 8pm? 12:11
autrijus yup
I like at least using some Perl 5 at work... it keeps me in perspective :)
castaway indeed 12:12
castaway wonders which new shiny functionality we're going to get tonight 12:13
autrijus it's the second half of the week
so it's likely I'll move into bugfixing.
castaway mm, makes sense ,)
autrijus cleaning the smoke stack, so to speak 12:14
castaway *g*
Limbic_Region yeah, it is a never ending cycle - bufixes lead to refactoring, refactoring lead to neat new ideas, neat new ideas lead to more bugs 12:16
autrijus o/~ and it moves us all / through despair and hope / through faith and love o/~ 12:17
castaway tosses theorbtwo.perlmonk.org/kwiki/index.cgi?VBFAQ at Limbic :)
Limbic_Region castaway - I haven't gotten to the Yahoo Mail section of my morning routine 12:18
I am going to see what I can do about devoting a couple of hours a week to the project though
castaway *g* Ok
cool
well the link wasnt in the mail anyways, since I just created it today 12:19
castaway pokes bro. 12:52
Limbic_Region castaway - almost (but not quite) finished with my morning routine 12:55
svnbot6 r3436, iblech++ | Added (OO) Net::IRC::JoinOnInvite and Net::IRC::SeenExt roles.
Limbic_Region can I /msg you some initial thoughts?
castaway sure, go ahead 12:57
stevan morning all :) 13:07
svnbot6 r3437, autrijus++ | * remove primitive infix operators from being inserted into the
r3437, autrijus++ | user-defined level of the precedence table; previously it was
r3437, autrijus++ | breaking "or" and "and".
Limbic_Region salutations stevan 13:10
stevan morning Limbic_Region
Limbic_Region autrijus - I often see PAR questions go unanswered (or at least not answered well) at the Monastery - any general advice since I know nothing about PAR? 13:14
he did say he was off $work right? 13:15
castaway he did
hmm, I got advised to ask the mailing list, who then told me to search the list archives..
(but I havent gotten around to it yet) 13:16
Limbic_Region *shrug* - I just don't like to see good questions go unanswered 13:17
castaway yeah me either.. I guess there are not too many PAR experts lurking in the monaster 13:21
yeah me either.. I guess there are not too many PAR experts lurking in the monaster 13:23
bingbong, Odin 13:25
Limbic_Region Is there a Dumper yet for Pugs? 13:26
castaway .perl ?
$mydatathingy.perl gives you a data::dumperish output, iirc 13:27
Limbic_Region ok - good enough for me
castaway (duh, if I want to get mail today, I should probably restart fetchmail..) 13:32
Limbic_Region I think I just found a bug 13:35
castaway grats :)
Limbic_Region my %hash = map -> $_ { $_ => undef } 1..10; # works as expected 13:36
my %hash = map { $_ => undef } 1..10; # doesn't work at all
I thought -> $_ was assumed if not explicitly stated?
castaway too
Limbic_Region anyone else around to confirm/deny? 13:37
castaway Mine takes too long to build.. I could try lilith.. :)
scook0 pugs: cannot cast from VList [VInt 1,VInt 2,VInt 3,VInt 4,VInt 5,VInt 6,VInt 7,VInt 8,VInt 9,VInt 10] to Pugs.AST.Internals.VCode 13:38
Limbic_Region scook0 - I know that's what it says when it breaks
I am asking if breaking is proper behavior or not
IOW - does map require -> $var
castaway Limbic, I get: pugs: cannot cast from VList [VInt 1,VInt 2,VInt 3,VInt 4,VInt 5,VInt 6,VInt 7,VInt 8,VInt 9,VInt 10] to Pugs.AST.Internals.VCode
ah, hmm 13:39
castaway dives for the docs
Limbic_Region is going to go find out if this copy of HOP he loaned out is available to be borrowed back
if you find that it is optional ($_ is assumed) - let me know and I will write a test for it
scook0 IIRC, {...} is just a special notation for ->{...}
castaway hmm, no map in S04 13:40
scook0 I think { $_ ... } means { $^_ ... } which means -> $_ { $_ ... } 13:41
castaway A bare closure without placeholder arguments that uses $_ (either explicitly or implicitly) is treated as though $_ were a placeholder argument:
$func = { print if $_ };
$func("printme");
broquaint I thought map need to be called with a colon if you wanted to use a bare closure (at this stage) i.e map:{ ... } LIST;
scook0 Seems to me like it's just a pugs bug, not deliberate behaviour
Limbic_Region broquaint - you very well may be right 13:42
castaway That implies to me (from S04) that it should work as Limbic expects..
Limbic_Region I haven't assimilated enough to know for sure yet
scook0 that might explain why it's expecting 1..10 to be a VCode
broquaint I think it should work as L~R expects, but I don't know if Pugs supports that style yet.
Limbic_Region well - it is broken either way
so it is a bug
just not sure how to test yet 13:43
scook0 Maybe it thinks { $_ => undef } is a hash that it's mapping over, so it assumes that 1..10 must be the mapping function
Limbic_Region depending on which way is right
castaway I dont recall seeing "map" mentioned in any of the synopses yet
broquaint { $_ => whatever } hashifies IIRC
Limbic_Region my %hash = map:{$_ => undef} 1..10;
castaway hmm, yes.. somewhere mentions that => implies hashiness.. 13:44
Limbic_Region actually, IIRC what it says is
scook0 my %hash = map { $_ => undef; } 1..10
Limbic_Region {} is always a block
scook0 works fine
Limbic_Region unless
castaway A bare closure without placeholder arguments that uses $_ (either explicitly or implicitly) is treated as though $_ were a placeholder argument:
$func = { print if $_ };
$func("printme");
Limbic_Region it looks very much like a hash
castaway oops
On the other hand, anywhere a term is expected, a block is taken to be a closure definition (an anonymous subroutine). If the closure appears to delimit nothing but a comma-separated list starting with a pair (counting a single pair as a list of one element), the closure will be immediately executed as a hash composer.
$hashref = { "a" => 1 };
(that one)
broquaint dev.perl.org/perl6/doc/design/apo/A..._operators 13:45
castaway (which is further down in S04)
Limbic_Region someone confident they know the right behavior and is willing to write a test it would be greatly appreciated
personally, I am just going to wait until after I finish this project 13:46
broquaint If I write the tests will someone commit them?
Limbic_Region I am sure to find more bugs/questions
broquaint - yes - lots of people will be willing to commit them
castaway I'm guessing that the second statement narrows the first one, so its not a bug
Limbic_Region but becareful - it may land you a commit bit
castaway ,)
broquaint Good point, L~R :) 13:47
broquaint discovers a bug in his first line of code :/
Or is it? pugs -e "say (map {; $_=>1} qw/a b c /).perl" 13:48
castaway Im not sure pugs takes over the {; .. idiom
I think you want a : (somwhere)
broquaint I think autrijius said it did ...
Yeah, it does. 13:49
castaway ok, I stand corrected
scook0 well, I must be off
bye all
broquaint It seems that () inserts references to keys in that example. Which doesn't seem right. 13:50
nothingmuch how easily could the schwartzian transform be done by hacking itypes? what I mean:
my @array := map { ... } @original; 13:51
broquaint What does the 'I' stand for in ITypes?
nothingmuch bind_indices(@array, @orig);
@array.=sort;
# @orig sorted as well, due to element relocation being carried over to @orig
except that it's not a schwartzian transform anymore 13:52
partial binding of the container meta object
phooey
Limbic_Region hmmm - found my second piece of strange behavior already 13:55
likely this is just an unimplemented feature
broquaint How does one test a parsefail?
Limbic_Region sub foo (Int $bar is copy, @list is copy) { ... } foo( 'hello', 1..10); 13:56
nothingmuch broquaint: if it dies in an string eval too, then comment it out
and put a fail("FIXME: parse fail"); right before it
Limbic_Region notice the 'hello' is not of type Int
but it works just fine
when it shouldn't
nothingmuch Limbic_Region: is this happenning because of is copy?
broquaint Thanks, nothingmuch.
nothingmuch i can promise to look at it when I get home
but I doubt I can solve it 13:57
broquaint decides to recompile Pugs before writing any tests ...
Limbic_Region I don't think so nothingmuch
in fact - I know it isn't 13:58
try the following
nothingmuch i can't try now
got to hack C++
if i start with pugs i'll get distracted
broquaint Poor, poor, nothingmuch ...
nothingmuch because I don't really feel like doing the C++
Limbic_Region pugs -e "sub foo (Int $bar) { say $bar; } foo( 'hello' );"
iblech I actually think Pugs' behaviour is correct -- the Str is automatically morphed into an Int
(Which happens to be 0, in this case)
Limbic_Region no 13:59
that's not what it does
nothingmuch i think it's bad for it to do that even if it did
oh my!
"hello" + 1 should not work
Limbic_Region I would be fine if that is what it did iblech
but it doesn't
it says hello
nothingmuch but see autrijus's thread on '1.28' * '2.56'
iblech Limbic_Region: oh...
nothingmuch i sort of know my way around bindparams, i'll try to look at it when I get back, Limbic_Region 14:00
Limbic_Region so I am just keeping track of oddities I encounter
will report them as I encounter them
nothingmuch test test test!
Limbic_Region but I don't intend to personally take action until I am finished
nothingmuch this info tends to get forgotten
Limbic_Region anyone along the way is happy to add tests
nothingmuch - see what I just wrote
nothingmuch are you writing this stuff down?
Limbic_Region yep 14:01
little notepad next to me
do it everytime I code p6
nothingmuch okay, so when I get back if you haven't already tested, I'll ask you to show me a list of bugs
Limbic_Region it is so new to me, I need to remind myself of things
k
nothingmuch back to work now
PerlJam Limbic_Region: Isn't that a policy decision? Sometimes you want strict type checking, sometimes you want autoconversion, and sometimes you just want what you want no matte what you wrote (the behavior you're seeing now)
Limbic_Region PerlJam - I think when you want it to "just work" you don't explicitly define the type 14:03
I think if you explicity define the type you want some kind of checking
I am pretty sure it was initially a difference between Int and INT and int
PerlJam Limbic_Region: yeah, I agree. I'm just being devil's advocate for a minute :)
Limbic_Region where 1 would convert for you but not remember (cache) 14:04
and 1 wouldn't convert at all
let me try the other two variants to see if either does what I want
nothingmuch Int is an object, int is autoboxed
Limbic_Region INT works
or rather
INT doesn't parsefail 14:05
int parsefails
Int doesn't parsefail
none of them do the right thing
grrr
Limbic_Region gets back to work
should .. have higher precedence than - ? 14:09
0 .. $foo - 3;
that doesn't work like p5 14:10
you have to add parens
0 .. ($foo - 3);
castaway try it without spaces? 14:11
$foo-3
(something like that, iirc)
eric256 pugs -e"my $foo = 5; print for 1 .. $foo-1" 14:12
prints 4
as do any variations on spacing.
castaway mm, leastways its on purpose, I thihnk :)
PerlJam which version? 14:13
Limbic_Region uses a binary build and has no way of knowing revision
eric256 6.2.3 (r3428) .... freshly built last night
Limbic_Region www.jwcs.net/~jonathan/perl6/ 14:14
PerlJam eric256: well it worked on an older revision that I happen to have handy (r3119_
Limbic_Region mine was built this morning at around 1 AM
eric256 if you just rung pugs Limbic_Region they fixed it so it shows the correct revision number
Limbic_Region eric256 - pugs -V has never worked on any of the binary builds I have got from Jonathan 14:15
unless it was fixed since 1AM this morning
I think it is a problem with svk versus svn (but I am not sure - hearsay actually)
eric256 no need for -V... just start its interactive mode and the splash screen has it
they fixed the svk vr svn thing a day or two ago
Limbic_Region no it doesn't
eric256 ??? realy?
Limbic_Region it says the exact same thing as pugs -V 14:16
which is to say it says 0 for revision
I get a version number only (6.2.3)
eric256 maybe he uses SVK to get it? weird thought because that jsut started working just a day or so ago. wonder if one part got fixed and broke something else. ;) 14:17
Limbic_Region dunno - I am not worried about that
eric256 either way i can't figure out how that code would print just 4 unless it is trying to do length(1..5) - 1
Limbic_Region I am worried that PerlJam is saying that 0 .. $foo - 3; used to work and doesn't anymore
means we didn't have good enough test coverage
anyway - have written it down on my list 14:18
moving on
eric256 or someone changed the specs so that now it is correct and before it was wrong.. which seems to be what castaway was hinting at
PerlJam eric256: I think it's doing something like (1,2,3,4,5) - 1 and the list is evaluating to the last element
Limbic_Region grrr - naked blocks as looping constructs aren't available in p6 anymore are they - grrrr
PerlJam Limbic_Region: nope, they need to have a "do" prefix 14:19
eric256 i think thats suppose to be a loop prefix
loop { ... } while <<expre>>;
PerlJam eric256: I think Limbic_Region was referring to how you can say { ...; redo } and it just works (in perl5) 14:20
Limbic_Region it's alright - it really should never have been written the way it was anyway
eric256 yea... i thought what i read last night said to use a bare loop for that...i guess that would make an infinite loop though 14:21
Limbic_Region infinite loop is what I want
integral hmm, is there a &? to get the current loop?
eric256 loop {} works then
PerlJam Limbic_Region: oh, then loop { ... } is it.
integral s/loop/block/ ?
Limbic_Region with a last to break out when appropriate
PerlJam - yeah, but it really doesn't make much sense - I was shortcutting the p5
integral { ...; &?CURRENT_BLOCK() } # ∞ loop using tailcalls
Limbic_Region it boils down to this loop { ... next if <some condition>; .... last; } 14:23
which means it may potentially run 1..Inf times
s/potentially// 14:24
eric256 whould loop { ... } while ( <<condition>> ) continue { ... } work?
svnbot6 r3438, Stevan++ | Perl::MetaModel - more work on the Perl::Meta::Compiler class; currently it only compiles properties of classes and some basic built-in methods; some tests for it too
PerlJam eric256: I don't think continue blocks exist any more. 14:26
It'd be more like: loop { ... NEXT { ... } } while $condition; 14:27
eric256 i was just reading that section... didn't get down to the closure traist, didn't dawn on me that it would apply to loops
PerlJam okay, r3437 ... $ pugs -e 'my $foo = 5; print for 1..$foo-1' 14:28
1234
Limbic_Region how do I do an array slice ? @foo[ @bar ] ? 14:29
iblech yep 14:30
Limbic_Region found another bug 14:50
svnbot6 r3439, putter++ | Switched some parsefails into evals in t/rules/from_perl6_rules/.
Limbic_Region pugs -e "my $i = 3; my @array = 1..5; @array[ --$i ] ; say $i"
$i should be 2 14:51
but it is 1
unless I missed something about predecrement operators?
iblech No, you're right, I think.
Limbic_Region port p5 to p6 code is a great way to find b0rk stuff 14:52
PerlJam again, on r3437 ... 14:54
$ pugs -e 'my $i = 3; my @array = 1..5; @array[ --$i ] ; say $i'
2
(but write tests anyway if that's your aim :)
Limbic_Region how new is 3437?
PerlJam very new. 14:55
Limbic_Region ok - so I will summarize a list of bugs when I am done
PerlJam svnbot just mentioned a commit by Stevan that put us as 3438
Limbic_Region all should get tests regardless
PerlJam s/as/at/
Limbic_Region but it is good to know some have already been fixed
PerlJam oh, and there's one from putter
Limbic_Region having tests ensures things don't get unfixed 14:56
PerlJam Limbic_Region: make sure you have the latest version whenever you do get around to doing something. It may be that tests have already been added.
(not that more tests would be a bad thing :)
Limbic_Region well - I am in a bit of a predicement WRT testing ATM 14:57
would be real nice if there was a way to grep/search svn online 14:58
Limbic_Region needs to talk to Jonathan about including the tests in the binary build he makes
hey PerlJam - would you mind running something for me under your build? 14:59
PerlJam idly wonders if google is indexing svn repos around the world.
Limbic_Region: what do you need run?
Limbic_Region that way I will be able to keep the syntax clean
will nopaste current working version 15:00
and then will nopaste what I think should work if there were no bugs
PerlJam okie
Limbic_Region perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
pasteling "Limbic_Region" at 129.33.119.12 pasted "Iterative Combinations (bug workarounds)" (27 lines, 793B) at sial.org/pbot/10249 15:01
Limbic_Region first ensure that works
pasteling "Limbic_Region" at 129.33.119.12 pasted "Iterative Combinations (fewer bug work arounds)" (26 lines, 770B) at sial.org/pbot/10250 15:02
PerlJam [1, 2] [1,3] [1,4] etc.
Limbic_Region then try that
PerlJam *** Error: Modification of non-creatable array value attempted
on the second one just after it output [1,5] 15:03
Limbic_Region ok - let me try one more version - testing only 1 bug at a time
pasteling "Limbic_Region" at 129.33.119.12 pasted "Iterative Combinations (1 bug assumed fixed)" (26 lines, 774B) at sial.org/pbot/10251 15:04
Limbic_Region this one only assumes 1 bug has been fixed
? 15:05
PerlJam same error
Limbic_Region ok
odd since you said pugs -e 'my $i = 3; my @array = 1..5; @array[ --$i ] ; say $i' prints 2 15:06
oh well, I will leave the workarounds in
more important to move on at this point
PerlJam wanders off to do some work
putter seen stevan 15:22
mugwump: ping 15:24
iblech Hm... my $hash = { key1 => "baz", key2 => "baz" }; say "Hallo $hash<key1>" doesn't currently work as expected 15:35
svnbot6 r3440, putter++ | Added "has Perl::Type $paramType;" to Perl-MetaModel/lib/Perl/Param.pm.
r3441, putter++ | Added "Split value and implementation types in Perl-MetaModel/lib/Perl/Param.pm.
Limbic_Region iblech - doesn't giving it a hint by say my $hash = hash { ... } help? 15:36
iblech iblech: The problem is the output -- it outputs ~$hash (all keys and values, tab and \n delimited) and then "<key1>" 15:37
Limbic_Region errr
that should have been does
iblech Where it should output only "baz"
Limbic_Region not doesn't
stevan putter: pong 15:40
Aankhen`` :-o 15:41
autrijus >> I just saw your name on the CPANPLUS AUTHORS list.
That was a pleasant discovery. :-) 15:43
svnbot6 r3442, iblech++ | Like in Perl 5, die "...\n" now doesn't output the file position etc. where the 15:55
r3442, iblech++ | error has happened.
putter stevan: I did some toe-in-the-water fiddling with Perl::Param. Added has Perl::Value paramValueType, paramImplType. Thoughts? 15:56
s/Thoughts\?/Any feedback welcome./ 15:57
stevan putter: to be honest, that is mugwumps territory 15:58
I am just an implementation monkey :)
Forth__ hum...have no idea to write a pugsh 15:59
putter ah, ok. :) thanks. 16:00
svnbot6 r3443, autrijus++ | * We have OUTER:: scope now.
r3443, autrijus++ | * Closure (VCode) now captures over the entire Env, instead
r3443, autrijus++ | of just the envLexical slot. This is currently only used to
r3443, autrijus++ | restore the original OUTER:: scope when calling a subroutine,
r3443, autrijus++ | but once we switch to oleg's CC_2CPST for Eval monad, this will
r3443, autrijus++ | get us rubyish coroutines (and serialised subcontinuations) for free.
stevan ahhh ,... my fingers are starting to code Perl 6,.. even when I am writing perl5 :P
putter stevan: ;) yes. the effect needs a name. also, after typing "perl" for the order-hundreth time to run pugs, I gave up and created aliases. 16:04
stevan :)
putter: I just discovered that ~ does not concat strings very well in perl5 :P
and Perl5::Object.new() does weird things :P
although I think really what I need is more coffee :) 16:05
putter oh yes. *laugh*
cognominal "rubyish coroutines": rubyish shouds like rubbish... 16:06
autrijus heh 16:07
rehi, btw
stevan hey autrijus :)
cognominal hi
autrijus just woke up -- I somehow fell unconscious ;)
stevan did you implement ::OUTER while you were sleeping? 16:08
autrijus yeah, I dreamed about it
so I woke up and coded it in
stevan :)
I did that with the Meta class compiler last night
too much smalltalk before bed :)
autrijus that's nice 16:09
stevan: deriving Ord is just implementing <=> in perl 16:10
stevan ok
autrijus Typeable is guaranteed -- it's having a .ref
stevan cool
autrijus why are you using lots of ~ ? 16:11
stevan autrijus: I am planning (in my head) a basic implentation of Roles using the Meta::Model classes
autrijus: bad habit of mine
autrijus 'foo ' ~ $foo.bar() ~ ' bar baz ' ~ $baz.baz()
stevan it helps me organize it visually
autrijus so you find "foo $foo.bar() bar baz $baz.baz()" less readable? 16:12
stevan no not less readable
autrijus in that case, I won't touch the code then
ohh
stevan but I like to indent and format things
autrijus nod
stevan (code as ASCII art)
autrijus cool
stevan I never studied Comp Sci, I went to art school
code aesthetics are important to me :) 16:13
autrijus *nod* and we love your figlets
stevan :)
autrijus so, there's Roles
what else would you like Perl::MetaModel to have?
stevan oh 16:14
autrijus is looking for a time point to snapshot that directory and backport to haskell
stevan I was going to try to make it compile the class with roles
using the meta-model
because the compiler right now just builds a string and evals it 16:15
so I am thinking I can add the Role understanding to the compiler maybe
(just for fun)
and see if I cannot prototype Roles that wat
s/wat/way/
it will just create a plain Pugs Class
autrijus that sounds intriguing and powerful 16:16
stevan when ultimately the meta-model needs to understand this stuff I think
autrijus yup
stevan however I still need to make the compiler deal with methods first
right now it just creates a meta() (which returns the Perl::Meta::Class instance) 16:17
and an isa() which points to the meta-class isATypeOf for now
I would do more, but I am starting to get behind in $work so ... 16:18
autrijus nod 16:19
svnbot6 r3444, iblech++ | warn() implemented. 16:22
autrijus iblech++ # core hacking in earnest
iblech My Perl 6 p6bible needed warn(), but it wasn't yet implemented, so... :) 16:24
putter stevan: I just realized the metamodel typing still needs junctives, hierarchical, *and* subtypes... anything else come to mind? 16:25
stevan oh,.. it needs much more that that too I am sure
:)
putter ;)
autrijus iblech: I'm so looking forward into installing p6bible in p6 by default :) 16:26
iblech :)
putter yes, please. and I contribute a script which merges S\d\d.pod to S.pod, making it easier to search...
autrijus it seems that the Eiffel XSLT app guy succeeded into getting PIR as his scripting language. 16:27
we need a libpugs :)
(so that those xslt functions can be written in perl6)
it may just be a simple case of a few "foreign export", really.
it looks like GHC has a libghc mode, too. hopefully it will land to 6.4.1 16:28
putter hmm... use Eiffel;... but probably have to create an additional level of (very light weight) threading...
autrijus that will magically enable Inline::GHC
jhorwitz prays for libpugs 16:31
autrijus jhorwitz: you'll find it useful? :)
it will however not be based on parrot at first
but you can surely link both in :) 16:32
jhorwitz don't be so coy... ;-)
compiling pugs -> parrot w/o running pugs or pugscc would be most excellent. 16:33
autrijus discovers www.volker-wysk.de/hsshellscript/ 16:34
putter Ha, "Plugging Haskell In" www.cse.unsw.edu.au/~dons/hs-plugin...lugins.pdf has an interpreter "Plugs". :) 16:35
iblech autrijus: [+] no longer works, as the fix from earlier today filters non-userdefined ops. Userdefined ops do work 16:37
autrijus iblech: hey. if I define infix Z, I don't get [Z] 16:38
svnbot6 r3445, iblech++ | die() and warn() no longer harddie with "Prelude.last: empty list" if no error
r3445, iblech++ | message was specified, but output "Died." or "Something's wrong." (like Perl 5).
autrijus iblech: you know how or why?
iblech: ugh. as for [+], I think the true solution is to switch Prim.hs into adding prec info 16:39
iblech autrijus: s/Z/q/ and it works. This seems to be a weird bug I was unable to trace
autrijus and derive the big table in Parser.hs
based on precedence metadata
so we don't end up with duplicates
iblech I think so, too 16:40
autrijus you think you're up for it? :D
autrijus is amazed by iblech's lambda-fu
iblech I'm amazed, too ;) -- But now, I don't think I can do that yet
s/now// 16:41
autrijus ok. 16:43
I'll do it then.
btw, you aware that
[<] 1,2,3
is subtle broken?
I'm not sure what to do
arguably the op1Fold need to probe for fixity
I mean associativity 16:44
and fold accordingly
iblech Saw that bug some minutes ago... [<] 1,2,3 seems to reduce to (1<2)<3 instead of 1<2<3
putter Hey, the May 2005 Haskell Communities and Activities Report is out. Probably mentioned already, but always worth reading. www.haskell.org/communities/
autrijus yeah because op1Fold is just leftward
putter: ooh, thanks for the info
hm, I found a way to easily do linked list consing in Perl6 16:45
[=>] 1..10;
and just remap .key to head, .value to tail
osfameron Larry wrote he was hoping no-one would notice => for consing.. 16:47
can you really use => in [] ?
theorbtwo That is very evil, autrijus.
iblech Of course, => is a infix op like +, *, etc., too
theorbtwo osfameron, it's (mostly) just an infix operator, so why not?
autrijus and it's foldr 16:48
as God^WLisp intended
because => is right assoc
osfameron so [,] is a noop?
svnbot6 r3446, iblech++ | Added test for [=>].
autrijus osfameron: yeah
iblech wonders what other interesting [some_infix_op] combinations exist... 16:49
putter iblech: are you going to add [,] to reduce.t, or shall I? :) 16:52
iblech putter: :) You can do it 16:53
I'm adding the next one then :)
putter k. *much laughter*
PerlJam why would you use [,] ? 16:55
Are you trying to make perl into lisp?
stevan autrijus: the meta-compiler handles methods now :) 16:56
I had a flash of inspiration
back to $work now :)
autrijus whoa.
stevan it is kind of primative now though
I do not deal with arguments and method signatures and such 16:57
(args just get passed in as a *@args)
iblech Is for =open(...) {...} supposed to work? (Pugs hangs currently) 16:58
Limbic_Region what's the p6ism for my $count = keys %hash; ?
svnbot6 r3447, Stevan++ | Perl::MetaModel - meta-compiler now handles methods (and method inheritance too); added C<can(Str) returns Bool> to built-in methods; see the t/50*.t for more info
Limbic_Region .count ?
iblech my $count = +%hash, IIRC 16:59
my $count = +keys %hash will work, too
Limbic_Region that works
thanks
iblech but my $count = heys %hash will give you an arrayref to the list of keys
s/heys/keys/
autrijus iblech: I fixed [+] etc 17:00
it may fix [Z] accidentally as well
iblech autrijus++ # yay! :) Please tell me what the problem with [Z] was, then :) 17:01
iblech food &
autrijus iblech: nvm it's still bad 17:04
svnbot6 r3448, autrijus++ | * [+] rescued
autrijus iblech: the bug was 17:07
ruleApply was after ruleLit
and [Z()] is a lit.
Limbic_Region autrijus - I am working on a small project that I think will get new people on board 17:14
svnbot6 r3449, autrijus++ | * [Z] now works
Limbic_Region has uncovered 4 bugs in the process (2 of which may already have been fixed)
and it is all because I write bad code 17:15
;-)
broquaint Bad code makes for good tests! 17:19
Limbic_Region also looks like a good candidate to profile pugs and optimize where appropriate
broquaint - actually bad code makes for good "improve this code" contests
I do need a way to determine if a list of sorted numbers is in a sequence of 3 or more 17:20
i.e. 5,6,7
putter r3450 (reduce.t) is untested. sorry, I have to unbreak syck. 17:21
autrijus unbreak syck? 17:23
how is syck broken?
(I imported 0.55)
putter syck.h defines some macros using malloc,etc, and memset, etc, but doesnt #include stdlib and string.h. I patched this before locally, but I'm afraid pushing it upstream is still on my todo list (somewhere...). Mea culpa. 17:26
svnbot6 r3450, putter++ | Added [,] to reduce.t. And reordered some tests.
r3451, autrijus++ | * oops, reverted the precedence, only allowing reduce
r3451, autrijus++ | metaoperator to parse before literals
Limbic_Region thinks he might have found another bug 17:32
svnbot6 r3452, ninereasons++ | work around unexpected behavior 17:38
vcv-- netsplit, ouch 17:41
autrijus ninereasons: adding "motd.p6 - " as the prefix of commit log, and/or explaining the mishevariour, would be appreciated... (and makes it easier to fix for) :-) 17:42
(since that looks like a genuine bug) 17:43
iblech autrijus: re. Found a bug: ("abc"|"def") ~~ rx:Perl5/.../ works as expected (returns a junctive match object), but my $a = "abc"|"def"; $a ~~ rx:Perl5/.../ returns only one 17:47
autrijus: re. Found a bug: ("abc"|"def") ~~ rx:Perl5/.../ works as expected (returns a junctive match object), but my $a = "abc"|"def"; $a ~~ rx:Perl5/.../ returns only a non-junctive match object 17:48
autrijus weird. 17:49
svnbot6 r3453, putter++ | r2706 reapplied (syck.h required #includes). And this time, upstream has been notified.
autrijus but then again, not weird really, because ~~ is rw!
($a|$b) ~~ s:P5/x/y/;
I wonder what does that do.
autrijus is confused
Limbic_Region without parrot on my Win32 box, what is the fastest way I can run code (options wise) - or am I stuck with what I got? 17:50
iblech I think: It should try to s:P5/x/y on both $a and $b, and return a junctive object which may evalute to true|false, if, for example $a didn't contain a "x" but $b did 17:51
autrijus Limbic_Region: you can tell autrijus to send you a copy
Limbic_Region a copy of parrot?
autrijus of binary parrot build
yeah
Limbic_Region or a pugs.exe linked against parrot with the appropriate library 17:52
clkao *yawn*
autrijus I hadn't got msys linkning to work.
Limbic_Region well - I kind of want the code to be slow
autrijus judging from Aankhen``'s report, nobody did :)
I think extern parrot for now
Limbic_Region I just want to find out if something worked (which will take quite a while)
autrijus is the surest bet
in PGE?
Limbic_Region nope
I will nopaste what I am working on
autrijus er, what do you want parrot for then
sure 17:53
Limbic_Region keep in mind that it is intended as a "how much can you improve this" contest
to get people interested
autrijus autrijus.org/tmp/parrot.tgz btw.
uploading
done
win32 vc6.0 build
vcv-- vc6? ick :) 17:54
Limbic_Region parrot nopaste
autrijus the only version perl is known to build correct ;)
Limbic_Region perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
autrijus s/correct/correctly/
mmm. 17:55
Aankhen`` autrijus >> I didn't try MSYS.
autrijus ponders eval_scheme support
vcv-- perl wont build on vc2002/2003?
autrijus Aankhen``: oh, you didn't? then somebody else did
Aankhen`` Atleast...
I just installed MinGW.
autrijus oh ok.
pasteling "Limbic_Region" at 129.33.119.12 pasted "My Idea for a "Sharpen your P6 skills" contest" (128 lines, 3.6K) at sial.org/pbot/10253
Aankhen`` vcv-- >> Visual Studio?
autrijus vcv--: yes, but then an extra dependency on VC's libc will be generated
vcv-- oh. 17:56
autrijus vcv--: and the "free" version is iirc unusable
at least no success reports has been found
vcv-- right
Limbic_Region specifically, I want to test the right-jack and runs code
clkao karma vcv
Aankhen`` The free version sucks. -_-
autrijus karma vcv
karma for vcv
hrm, no workee :)
vcv-- ill have to try getting it to work on the free version
theorbtwo Seen hasn't been working either.
autrijus vcv--: I tried... lots of porter tried
all failed
that's after multihundred metabytes of SDK download
vcv-- shame :( 17:57
autrijus so it's painful and unrewarding
Aankhen`` $karma vcv
:-(
Limbic_Region autrijus - would you mind running that code I nopasted
vcv-- i know someone on the visual studio team, indirectly. he hates perl though, so dont think i could get any info out of him
Limbic_Region and give me a line that either has a Jack in the first 4 cards or a hand that has a sequence of 3 or more 17:58
Limbic_Region needs to verify the scoring is working
btw - it is going to take a while
autrijus why does this need parrot?
Limbic_Region for speed
autrijus hm? speed? 17:59
Limbic_Region actually - anything that will speed it up will be appreciated
except re-writing it to be more efficient
Aankhen`` How do I link Pugs against an external Parrot?
Limbic_Region that is what the contest is for
autrijus I don't think parrot codegen handles this
Limbic_Region ok
autrijus Aankhen``: you do nothing :)
Limbic_Region so is there anything that can make the same code run faster?
Aankhen`` Parrot is required for PGE, right?
autrijus Aankhen``: just set parrot in path.
Aankhen`` Ah. 18:00
autrijus PATH, that is
Aankhen`` I was wondering what I did wrong, heh.
autrijus Limbic_Region: yes, work on parrot codegen :)
Limbic_Region ok - so I guess I just need to be smarter than the average bear
seed my test with desired data instead of waiting for the iterator to get to it
vcv-- autrijus: i know you said its painful and unrewarding, but i took that as a challenge. look what you got me into! 18:01
PerlJam Limbic_Region++ It's great that you're writing so much perl6 :-)
autrijus vcv--: it's not worth it... :)
but hey, if you insist, I can't stop you can I 18:02
wolverian does anyone have a link to something that discusses the definition of reduce as used in perl6? wondering if I could show a friend that it's more mathematical than what other languages use.
vcv-- i have a deep down hatred for anything vc6 related, and i wish to tinker with parrot..so :p
wolverian: the best thing i read was damian's 9 use examples of it 18:03
autrijus wolverian: WhatIsAFold
wolverian vcv--: sure. I'm just wondering specifically about the mathematical definition, since people have been saying the perl6 reduce conforms to that.
vcv-- oh, gotcha :)
wolverian autrijus: thanks.
autrijus wolverian: haskell.org/hawiki/WhatIsaFold
Aankhen`` vcv-- >> Parrot works fine.
You just can't link it against Pugs. 18:04
Or perhaps it was vice-versa. *looks sheepish*
autrijus same thing... link is twoway :)
vcv-- well whatever it is.. im going to get my hands dirty
autrijus wolverian: but that's reduction for lists
theoretically you can also fold over trees
Aankhen`` Yeh, but in this case, Parrot is being linked into the Pugs executable, right?
autrijus or even function types! 18:05
svnbot6 r3454, autrijus++ | * added a line of hint in Makefile.PL that says parrot.exe needs
autrijus Aankhen``: yes.
svnbot6 r3454, autrijus++ | to be placed in PATH for external parrot to work. Prompted by Aankhen.
wolverian autrijus: mm. right.
autrijus: that page doesn't really say if mathematics generally consider reduce/fold to take a list sub or a binary sub, which was my point, I think. :) (put in perl6 terms) 18:06
vcv-- binary sub? 18:07
wolverian vcv--: reduce { $^a + $^b } ...
theorbtwo wolverian: I don't think mathematics generally considers there to be much of a difference between a binary function and a general n-ary function.
vcv-- oh right
wolverian theorbtwo: right. :)
Limbic_Region yay - it appears to work 18:08
vcv-- im surprised by how many terms i dont know that i see you guys on the p6 mailing lists use
autrijus foldl :: (a -> b -> a) -> a -> [b] -> a 18:09
foldr :: (a -> b -> b) -> b -> [a] -> b
that's the two kind of folds haskell understands... both comes with a base case
vcv-- Then autrijus comes along with his Haskell and confuses me even more
ninereasons autrijus - stale reply: I wasn't sure whether it really was a bug, and that's why I was vague. But I'll work on being more expressive :)
autrijus ninereasons++ # cool 18:10
PerlJam vcv--: why are you surprised? The denizens of p6l are always using obscure terms ;-)
autrijus vcv--: well, the haskell here in foldl is really quite straightforward :)
vcv-- Heh.. true. surprised isnt an appropriate word to use i guess
Aankhen`` Yey! 18:11
vcv-- autrijus: yeah.. i just havent wrapped my brain around it yet. my brain doesnt like to learn with conventional methods. ill figure it out eventually :)
Aankhen`` autrijus++ # Thanks for adding that.
autrijus vcv--: its translation in Perl6 is somethinglike: sub foldl (&code<::a, ::b> returns ::a, ::a $head, ::b @rest) returns ::a
vcv-- nods head 18:12
autrijus see the WhatIsaFold link above to see why it's defined like that :)
Aankhen`` "&code<::a, ::b> returns ::a" is specifying parameters and a return type for the code block?
autrijus yes. 18:13
Aankhen`` Neato.
Would that work in Pugs?
autrijus eventually. not yet
Aankhen`` (mine is still on `nmake`, I haven't even got to `nmake test` :-P)
autrijus we don't do quantified type parameters yet
Aankhen`` OK.
autrijus type variables, even.
I'm not sure, in general, how it works in perl6. maybe everything is runtime checked ;) 18:14
Aankhen`` nods like he understands.
s/\.$/ the distinction./
autrijus oh. the distinction is that whether it can validate the &code<Int> against something that is actually accepting an Int at compile time 18:15
PerlJam autrijus: I don't understand how agressive perl6 will be towards type checking.
svnbot6 r3455, putter++ | eval_is()d [,] and [=>] in reduce.t.
autrijus PerlJam: oh, I think it will be checked.
Aankhen`` Ah.
autrijus I'm just not sure how early is the check.
PerlJam: it looks like everything is until runtime.
wolverian autrijus: I think it might be &foo:(::a, ::b) nowadays, but it's underspecced, really.
PerlJam autrijus: sometimes I think it should check what it can at compile-time and defer everything else to runtime.
autrijus wolverian: yeah.
Aankhen`` Windows XP's Remote Desktop looks pretty neat. 18:16
PerlJam But then, as you say, it looks like everything is runtime
Aankhen`` And really useful.
vcv-- Hm. I know there was talks of one guy wanting an IDE made specificall for Perl6. Maybe ill get myself into that. by writing it in perl6 of course :)
autrijus PerlJam: "everything it can" is looking liek not much, unless we apply mad inference-fu
vcv--: nice
Aankhen`` Perl 6 will be all but impossible to parse. :-(
Maybe you could just rip it out of Pugs though. ;-)
autrijus Aankhen``: that's actually the plan
PerlJam Aankhen``: It's a good thing we'll have perl6 to parse it! ;-)
autrijus just apply it to a regex. 18:17
vcv-- im a gui nazi. thats my specialty
autrijus a rule, that is
Aankhen`` Oh yeah, I forgot... there's going to be a builtin grammar for it, isn't there?
osfameron and hooks!
Aankhen`` Excellent􏿽xE9.
Aankhen`` takes back his uncalled-for whining.
PerlJam Aankhen``: built-in, user redefinable, fully introspective grammar
autrijus and you can even derive that rule by pretty-printing the active "Perl6" grammar object from your current lexical scope!
vcv-- Aankhen``: i thought so too.. but the foundation of it is what makes it so simple yet powerful. i think its revolutionary :)
autrijus what a fix!
putter fyi, [=>] and [,] arent actually working, and [//] @array where my @array = (undef, undef, 3, undef, 5); nonterminates.
osfameron is really excited about a refactoring browser as good as IDEA for Perl...
osfameron actually slobbers a little 18:18
Aankhen`` I love pattern matching.
I drooled when I read A05.
Limbic_Region ok - writing perlmonks journal entry now
putter that's as of r3453
vcv-- autrijus already put reduce [] into action in pugs??
autrijus Limbic_Region: url? :)
vcv--: er, sure.
pugs> [*] 1..10
3628800
vcv-- you amaze me 18:19
PerlJam vcv--: If you dream it, odds are autrijus has already implemented it ;)
autrijus thank you.
&product = sub { &prefix:<[*]> o (&infix:<..>.assume(1)) }
mmm pointless perl6.
even more pointless than pointless haskell. 18:20
and it's .assuming.
(the use of lots of the "o" operator from fp.pm is called "point-free", aka "pointless", style.)
putter autrijus: re eval_scheme, it just so happens I have (most of a) scheme to p5 self-hosting transliterator... ;)
autrijus putter: oh wow! 18:21
vcv-- How about a parrot to x86 assembly convertor? :)
autrijus there is HScheme, too
which is why we can get eval_scheme for free
PerlJam vcv--: we call that a JIT ;-)
autrijus vcv--: it's already there :)
it's a full ContT monad with call/cc support
vcv-- no way 18:22
so it would be possible to compile to a native executable with little to no overhead?
autrijus [not|autrijus]~/work/parrot/jit$ ls
alpha/ arm/ hppa/ i386/ ia64/ mips/ ppc/ sun4/
vcv--: right, and you know, it's already there too!
# www.parrotcode.org/docs/native_exec.html 18:23
vcv-- I'm at a loss for words
PerlJam vcv--: go on ... ask for something else that you think is far fetched!
autrijus (there is a reason why pugs is targetting parrot.)
vcv-- Perl: I got nothin'! 18:24
putter re hscheme, looks like it has some hscheme.sourceforge.net/issues.php . community report lists it as "stalled"...
Limbic_Region autrijus - not done yet
will post when it is finished
autrijus cool
vcv-- "This method wraps the VM runtime engine and a precompiled Parrot program into a single binary."
aww..see..thats cheating!
autrijus but it's jited :) 18:25
PerlJam cheating? I say it's all according to plan
vcv-- Maybe one day there will be true compiling with *no* overhead :)
dont get me wrong, im happy with the JIT, its very impressive :) 18:26
autrijus vcv--: if you look for ansi C codegen, I don't think it's parrot's goal
although you're certainly welcome to help out Pugs's _other_ backend, which can one day compile to ansi C with no runtime system :) 18:27
(Perl6->Haskell->Grin->Jhc)
vcv-- That I will consider
putter given the Scheme's... err, minimalist standard, one really wants a scheme implementation factory class. thus, use Scheme::Guile;. otherwise snarfing code from the world is just too hard.
autrijus putter: you know my Inline::MzScheme? :)
vcv-- If not, I will possibly invest in some programmers who could do it
autrijus however you lose interop with other parrot languages. :-/ 18:28
putter yes. 'fraid I havent used it though.
Aankhen`` Does a `grammar { ... }` declaration work? 18:29
autrijus Aankhen``: no, not yet.
Aankhen`` Darn.
putter I should have said "use Scheme::Transliterating::Guileish"; One problem with the Inline:: naming scheme is that one might want Guile external common, Guile external dedicated, Guile internal common/dedicated, Guile compatibility package on other foo, etc.
autrijus I'm not sure qualified rules even work. 18:30
in PGE, that is
vcv-- `` isnt still system()?
Aankhen`` Qualified rules?
vcv-- >> `...` is just my way of differentiating code from the rest of the text. :-)
autrijus /<Foo::bar>/
vcv-- Ahh... ok :)
Aankhen`` autrijus >> Ah.
autrijus (which is what you normally want grammar for)
Aankhen`` Well, if grammars don't work... :-)
vcv-- >> I tried |...| but then I figured since I'm using Markdown syntax for most stuff anyway, I might as well do it for code too. :-P 18:31
autrijus what do you want grammars for?
Aankhen`` |...| looks uglier anyway.
autrijus >> groups-beta.google.com/group/perl.p...55a5a65b9f
PerlJam autrijus: that's just a subrule. subrules work.
autrijus subrule?
oh.
you mean if I install Foo::bar by qualified name.
PerlJam right.
autrijus but what if <bar> is called in the same grammar package? 18:32
oh. I need to tell PGE the things in scope.
vcv-- Aankhen``: i was looking at that. considered giving it a shot
autrijus and also initialise it with builtin rules.
PerlJam: hey, you know if there's a list of builtin rules? if not, can you start assembling one or find people to? :)
vcv-- That XML parser doesn't involve DTDs though.. right? right?
Aankhen`` Heh, I got most of the productions converted already, vcv--.
vcv-- DTDs are the devil :(
Aankhen`` Well. 18:33
PerlJam autrijus: I think Damian posted the beginnings of such a list to p6l once.
Aankhen`` I'm gonna check out some of the P5 XML parsing modules.
PerlJam certainly you can mine the As, Es, and Ss to find ones that are probably built-in
autrijus I'm trying to delegate :)
PerlJam I know, I just don't have time so I'm delegating back :) 18:34
autrijus putter++ # notifying why
Aankhen`` You could start with the common ones -- <sp>, <ws>, <ident>, etc.
And then add the POSIX rules -- <alpha>, <digits>, etc.
autrijus Aankhen``: can you be the delegate? :D docs/quickref/rules will be an excellent place
just list the names
putter re grammars/rules... there is a great deal of red in the smoke...
autrijus with definitions if you can conjure up, otherwise leave a XXX
Aankhen`` Mmm, I'll try.
You want the list formatted in any particular way? 18:35
autrijus Aankhen``: no, just the usual quickref format
Aankhen`` OK.
autrijus (minimal formmatting the way you see fit)
Aankhen``++
Aankhen`` BRB.
autrijus "grammar" is parsed now, btw.
vcv-- delegate something easy to me ;)
svnbot6 r3456, autrijus++ | * parse for "grammar", as suggested by Aankhen.
PerlJam Aankhen``: <ident> <null> <before ...> <after ...> are a few more I can think of right off but you'll have to check S05 or A05 for their defs
autrijus vcv--: sure... proofread docs/quickref and maybe add something into it 18:36
PerlJam wait a second ... I may have a list in my notes from when A05 first came out ...
PerlJam looks
autrijus vcv--: oh, are you not a committer? then give me your email and add yourself to AUTHORS
as the first assignment :)
vcv-- whereat?
theorbtwo In alpha order by first name. 18:37
PerlJam ah yes ... Aankhen`` look at the bottom of this page: lighthouse.tamucc.edu/ScottDuff/Perl6Apocalypse5
vcv-- er.. which docs?
theorbtwo First "vcv--" Last (CPAN) utf-8
autrijus or just "vcv" :)
i.e. give me an address I can send invitation mail to; after getting the invitation, use the newly acquired password to check in to svn.openfoundry.org/pugs/ 18:38
vcv-- matt @ weakmind . org
autrijus the quickrefs are in docs/quickref/*; the AUTHORS file is in the base dir.
invitation sent. welcome aboard!
vcv-- ok. thanks :)
Limbic_Region autrijus - www.perlmonks.org/index.pl?node_id=458728
Aankhen`` PerlJam >> OK, shall I throw that into a quickref?
PerlJam Aankhen``: sure. 18:39
Limbic_Region I also want to talk to you about the 5 possible bugs I found
everyone is invited to join in the Perl6 challenge btw
autrijus Limbic_Region++
Limbic_Region perlmonks.org/index.pl?node_id=458728
autrijus Limbic_Region: sure. which 5?
PerlJam Aankhen``: Wait, you mean the URL or the stuff on the page? Better to put the stuff on my page in the docs so that people can hack on it.
autrijus (if they are made into failing tests, I'll get around to fix them sooner or latter)
Limbic_Region well - I say possible because I am not sure of behavior 18:40
autrijus the map {=>} thing is not a bug.
it's a misfeature.
but I think it's gonna stay.
sad but true... write {=>;}
i.e.
Limbic_Region not sure I understand
autrijus map { $_ => $_; }
Limbic_Region my %hash = map { $_ => $_ }; #doesn't work 18:41
autrijus map { $_ => $_ } @foo; # illegal
yes.
because that's like saying
my %hash = map \%another_hash @array;
Limbic_Region ok - so -> $_ is necessary?
autrijus you can use a semicolon.
PerlJam autrijus: is {=>} related to [=>] ? ;>
autrijus my %hash = map {$_ => $_;} @array;
that works.
PerlJam: ;)
PerlJam: <<=>>>
Limbic_Region my %hash = map -> $_ { $_ => $_ } @array;
also works 18:42
autrijus yup but more verbose.
both works
Limbic_Region ok - so the ; is the succinct way to go
Limbic_Region is intentionally disqualifying himself
when you define a sub parameter type - is there supposed to be any type checking
autrijus currently it's only used in MMD. 18:43
Limbic_Region sub foo (Int $bar) { say $bar } foo( 'hello' )
autrijus it will be used for coercion later maybe
Limbic_Region ok - so that IS an unimplemented feature then?
iblech Hmm... »=>« actually has a meaning (surprise! :)): my @array_of_pairs = @keys »=>« @values
autrijus i.e. coercing 'hello' into 0
Limbic_Region right - fine by me
autrijus but I'm not sure when coercion stops and when castfail begins
Limbic_Region but leaving it as hello isn't the right thing
autrijus no it's not. problem is I don't know what righ things is 18:44
Limbic_Region I thought it was the difference between Int and int
autrijus I asked on p6l
Limbic_Region 1 was allowed to be typecast but wouldn't remember (cache) result
one was a flat out failure
autrijus Larry gave some sort of answer -- but I think it needs to be reraised
AtnNn how do you type << and >> on us layout?
Aankhen`` PerlJam >> Doesn't <ident> only match a character that would be valid within an identifier, not the whole identifier?
autrijus Limbic_Region: maybe you need to reraise that on p6l
sub foo (Int $bar) { say $bar } foo( 'hello' ) # this
Limbic_Region ok - so I will hold off on #2 until we know WHAT to be testing for 18:45
autrijus I don't think it's been answered before
Limbic_Region and will instead raise it on p6.l
Aankhen`` And don't worry, I am putting the stuff from the page in the quickref, not just the URI. :-)
autrijus Limbic_Region++
also ask (int $bar) :)
Limbic_Region right
I intend to explain "my understanding was...."
svnbot6 r3457, iblech++ | Added the usual svn properties and EOLs at EOFs to the apparently unstoppable
r3457, iblech++ | efforts of Stevan++ on ext/Perl-MetaModel/. :)
r3458, iblech++ | Added a preliminary p6explain.
r3459, ninereasons++ | add t/pugsbugs/unhashify.t | adding a pugsbugs test to demonstrate that when a hash ((key => 'val')) is passed to a sub, it becomes a List \((key => 'val')).
ninereasons a little more descriptive, autrijus ? ;-) 18:46
Limbic_Region and then ask what the currently accepted behavior is
iblech AtnNn: <Compose> <Greater than> <Greater than> and <Compose> <Less than> <Less than>
Limbic_Region ok - on to #3
putter Aankhen``: re <ident>, nope, see t/rules/Disabled/from_perl6_rules/stdrules.t
Limbic_Region it may have been fixed already
Aankhen`` Alright.
Limbic_Region 0 .. $foo - 3;
PerlJam Aankhen``: I think it matches a sequence of characters that would be valid in identifiers, but don't take my word for it, read the As, Es, and Ss, and ask p6l if there is a lack of clarity or consensus.
Limbic_Region that didn't do the right thing in the version of pugs I have that Jonathan compiled last night
AtnNn <Compose>? 18:47
Limbic_Region 0 .. ($foo - 3); # was necessary
it may have already been fixed though
PerlJam tries to concentrate full on work.
iblech AtnNn: On Linux, you can declare some unused key to act as a compose key. Don't know about other OSes, though
Limbic_Region: Already fixed :)
Limbic_Region in any case - the parens should not be necessary right autrijus?
ok - so needs a test to ensure it doesn't get unfixed iblech ?
or does it already have a test?
autrijus pugs> my $foo = 10; say 0..$foo-3; 18:48
01234567
I don't recall a test
but when that was brokenm
lots and lots of other tests fail
iblech No, there's no test for it currently, IIRC -- do you want to write it or should I? :)
autrijus like shortcircuit.t
of course, more tests are always welcome 18:49
Limbic_Region iblech - even if I write it I can't commit
so if you have the tuits
iblech Ok, will do
autrijus it looks like iblech is gifted with a rich supply of tuits :)
Limbic_Region on to #4
putter Aankhen``: Damian uses <ident> in his recent useful capture post www.nntp.perl.org/group/perl.perl6....uage/20985
autrijus ident ==> <\w+>
also found in pge's demo.pir
putter yes. 18:50
Limbic_Region pugs -e "my $i=3; my @array = 1..5; @array[ --$i ] ; say $i"
that prints 1 for me and not 2
I have heard that it too is fixed already
autrijus but that gives numeric letter as valid ident
Limbic_Region but I am not sure about a test
autrijus it's 2 here.
the great infix checkin last night surely broke a lot
(it's still very much worth it) 18:51
Limbic_Region ok - so a test to ensure unfixing is in order?
autrijus sure
Limbic_Region ok
last 1
autrijus Limbic_Region: latest pugs btw -- autrijus.org/tmp/pugs.zip 18:52
just the .exe
so quite small
er, wait, wrong link. never mind that :)
autrijus will regen
Limbic_Region wasn't going to fix stuff up anyway
autrijus oh ok.
Limbic_Region that's what the contest is for
autrijus cool
last one?
Limbic_Region while ( @combo = $next() ) {
should I be able to do 18:53
putter Aankhen``: note that ScottDuff/Perl6Apocalypse5 is a bit out of date. eg, :u1 etc are renamed.
Aankhen`` OK.
vcv-- aw. the parrot makefile assumed i had perl in c:\perl\bin when activestate installed it in e:\perl\bin >:/
Limbic_Region while ( @combo = sort { $^a <=> $^b } $next() ) {
I ended up making the @combo = sort ... the first line of the while block as a work around
18:55 _metaperl_ is now known as _metaperl
autrijus hm 18:55
putter what is more critical path, metamodel or rules?
Limbic_Region ok - sounds like a good question to have asked
svnbot6 r3460, iblech++ | Added a test testing &infix:<..>, &infix:<...> and their precedence.
autrijus Limbic_Region: you should. that's a bug 18:57
not sure where the bugs lies, though.
I'd appreciate a minimised test case
Limbic_Region autrijus - will do 18:58
actually - will write it in my journal tonight - might not get a test for it till tomorrow
Limbic_Region is kinda burnt out
autrijus Limbic_Region: that's fine, take care! 19:01
svnbot6 r3461, iblech++ | reduce.t -- Skipping a hanging test and unTODOed a now working test :)
vcv-- uh oh. nmake is stuck at:
c1 : warning C4349: /Gf is deprecated and will not be supported in future versio
ns of Visual C++; remove /Gf or use /GF instead
putter mugwump: ping 19:05
stevan putter: I think rules is probably more critical now 19:06
(to answer a 10 minute old question)
svnbot6 r3462, ninereasons++ | t/data_types/hash.t, add tests after t/pugsbugs/unhashify.t, as an exercise against the bug demonstrated there, which changes a Hash into a List when passed to a sub.
stevan putter: the meta-model stuff may or may not end up being part of the perl6 port of perl6 19:07
(probably not, at least is its current form)
putter: however, that said, I would love for you to help on the meta-model. But I really see rule support as more important since without rules we cannot have a perl6 port of perl6 19:08
Aankhen`` Anyone know the difference between :overlap and :exhaustive? 19:10
theorbtwo "aaa" ~~ /a*/ gives aaa, aa (twice), and a (three times) with one, and only aaa, aa, and a with the other.
19:11 _metaperl is now known as metaperl
theorbtwo (One requires unique starting positions, the other does not.) 19:11
Aankhen`` Ah.
Hrm.
Subtle.
Thanks. :-)
putter stevan: ok. thanks. I was thinking that autrijus sounded like he was in wait mode on classes for some level of progress on meta. meta as spec rather than meta as impl.
vcv-- Does anyone know where i can define the GHC path in the makefile for pugs? 19:12
stevan putter: I dont think autrijus is waiting on it 19:13
autrijus putter: oh, no, it's just I'm trying to avoid dup work :)
stevan LOL
autrijus I'm /wi36
er.
ww
anyway.
autrijus ponders the sanity of maintaining 40 irc subwindows 19:14
stevan sanity--
theorbtwo You are the west wing?
autrijus theorbtwo: hm?
theorbtwo ww = west wing. American TV show.
vcv-- stupid makefile! >:/
autrijus so. where was I. yes, classes. I'm in bugfix mode for rest of this week
so it'd be cool if next week I can avoid re-research and apply Perl::MetaModel back 19:15
but if there's missing spots, I'll fill it up :)
stevan I think much of the Perl::* level stuff was actually just perl6 versions of Haskell stuff 19:16
vcv-- odd. did a search for "assert_ghc" in all files in c:\source\pugs\ and nothing came up 19:17
autrijus stevan: right, I'm thinking about Meta level
stevan autrijus: what parts are the most useful for you? I will try to concentrate on that
autrijus stevan: the Meta stuff, and in particular Role
stevan is trying to use recursion with his methods too, to make it easier to backport
autrijus don't worry about that :) 19:18
stevan autrijus: ok, I will work on the Role stuff then
autrijus I've got forM defined :)
(and forM_)
stevan autrijus: I like recursion better anyway :)
so much cleaner
autrijus cool :)
putter I was thinking meta-level isATypeOf, types and subtypes needed fleshing out...
autrijus recursion++
stevan putter: I agree
the isATypeOf was really just so support isa(Str) 19:19
beacuse we dont have type dispatched multi-methods :)
vcv-- gives up and throws computer out window 19:20
autrijus we don't?
stevan autrijus: last I checked we didnt
autrijus oh, we have typedispatched multisubs
but not multimethods
putter are "hierarchical types" subtypes? they are type constrains, but I havent seen them refered to as such.
stevan see t/oo/methods/multi.t (i think)
autrijus is that the case?
vcv-- autrijus: Where can I define the GHC path for the makefile for pugs? its driving me crazy trying to find it :(
autrijus putter: I have no ida
idea
stevan putter: I am not familiar with them,.. which Synopsis is it?
autrijus vcv--: GHC
vcv--: just the GHC env
putter checks... 19:21
autrijus or put them in PATH
vcv-- heh.. duh. thanks, im not thinking today.
stevan putter: if you want to work on the types stuff though, that would be very very very cool.
Corion Does anyone know if Tye McQueens idea of making require() return a factory was seen by @Larry ? (perlmonks.org/index.pl?node_id=424374)
putter S12.pod:=head1 Types and Subtypes S06.pod:=head2 Hierarchical types 19:22
autrijus stevan: indeed
stevan Corion: nothingmuch kind of does something like that 19:23
his packages always return __PACKAGE__ instead of 1;
AtnNn pugs> [**](2,3,4) -> Internal error?
Corion stevan: Ah - I'd like to see that in Perl6, so I'll ask nothingmuch too ...
stevan produces a similiar efffect I think
Corion stevan: Basically, returning __PACKAGE__ is the minimum you need, yes. 19:24
Limbic_Region autrijus et all - use.perl.org/~Limbic%20Region/journal/24776
will work on post to p6.l now
Corion (backlogging) - I build Perl with the free MSVC7 compiler, but I haven't built Parrot with it. Is there an all-in-one source package that I can try without investing any time other than "gunzip; untar; nmake" ? 19:25
stevan putter: i see what you mean now
Array of Array of Int
stuff like that
Implementation types look interesting too 19:26
iblech AtnNn: Thanks for reporting, fixed, committing in a minute :)
AtnNn :) 19:27
Corion Gah. I'm teh stupid. I left in some unfinished patches from yesterday and now wonder why the build breaks ... 19:28
Limbic_Region message sent
vcv-- Corion: you're the windows guy right?
Corion vcv--: "The" guy is a bit much, but yes, I use Windows (and try to make sure that Pugs works on Win32) 19:29
vcv-- thats what i meant, sorry :p
do you use vc6 then?
Corion vcv--: No, the free VC7 compiler
theorbtwo Corion doesn't give himself enough credit -- he's the guy. 19:30
Corion theorbtwo: You forget gaal, he's doing much Win32 stuff too
vcv-- oh really? i thought they were saying it didnt work with the free vc7 compiler
theorbtwo I didn't realize gaal used win32. 19:31
vcv-- or maybe it was linking parrot and pugs together?
putter stevan: I wonder just how costly typechecking may become... you call a sub of one argument, and suddenly your off crawling down a list of junction types with assorted pattern matching constraints... recursively...
Corion theorbtwo: gaal is trying to keep msys alive :) 19:32
stevan putter: that is true
svnbot6 r3463, iblech++ | Fixed [**] and added a test for it, thanks to AtnNn++ for reporting.
Corion vcv--: Maybe Parrot doesn't work with MSVC7. I haven't tried Parrot yet.
vcv-- Yeah. autrijus said pugs wont link with parrot if you use the free vc7 compiler 19:34
which i was planning on seeing if i could figure out why in the near future 19:35
Corion ... maybe they just haven't tried Hard Enough :)
... but I don't want to spend hours chasing down various libraries, just to find out that one of these libraries doesn't build. So I want an all-in-one package there ...
vcv-- Yeah 19:36
Limbic_Region find coding in p6 very draining
vcv-- You've never coded in c have you? :P 19:37
Limbic_Region vcv-- very little
but it really isn't p6
it is any language that I am unfamiliar with
Aankhen`` vcv-- >> It won't link even if you use the expensive paid version.
Limbic_Region c qualifies 19:38
vcv-- Aankhen``: must..figure..out..why
Limbic_Region does feel a sense of accomplishment after getting something running though (p6 or c)
Aankhen`` vcv-- >> sial.org/pbot/10245
vcv-- the free version uses the asme compiler and linker.. just lacks the ide and various tools
Aankhen`` Apparently it needs a bunch of libraries which GHC can't find.
The free version has version 1.50 of nmake.
VS.NET 2003 has version 7.10. 19:39
vcv-- thats it? i could probably figure those errors out!
Corion Aankhen``: Wrong. Microsoft (R) Program Maintenance Utility Version 7.10.3077
Aankhen`` The free version will die at some point while compiling parrot, saying that the command line is too long.
Corion >> What's your point?
Corion (with my free version)
Aankhen`` Hrm.
That's weird.
I downloaded nmake and got 1.50.
Corion Aankhen``: Yes. That's the nmake you can download separately. 19:40
Aankhen`` I also downloaded the toolkit.
Corion The free version of MSVC7 includes nmake 7.10
Aankhen`` Still had 1.50...
Corion Aankhen``: You have the old one in the path then
Aankhen`` Perhaps.
vcv-- Yeah.. just checked. it does use 7.10
Aankhen`` I think I was a little confused about paths at the time.
OK. 19:41
stevan hey chromatic :)
Aankhen`` I stand corrected. :-)
Well, sit, really.
:-P
chromatic Hi there.
autrijus greetings chromatic-san. 19:42
chromatic I have a Haskell-ish question.
vcv-- E:\Program Files\Microsoft Visual C++ Toolkit 2003\lib\oldnames.lib
right there
autrijus chromatic: I have a Perl6-ish question
vcv-- is that all that was wrong with the vc7 free compiler??
autrijus chromatic: but you get to ask yours first :)
chromatic I was reading through t/oo/magical_vars.t last night and tried to add in $?CLASS support. It looked pretty easy, but I'm finding Eval.hs a bit confusing. I think I know where it should go, but the eval monad required to change the pad confuses me and I can't quite get it right. 19:43
It looks easy still; can I have a hint?
autrijus sure, you have 3 free hints
Aankhen`` vcv-- >> I have it too, but it is not found.
autrijus chromatic: it's in Eval.hs, you need to modify getMagical
vcv-- put "E:\Program Files\Microsoft Visual C++ Toolkit 2003\lib" in your PATH var maybe? 19:44
chromatic Ah, I was looking at applyExp.
vcv-- -E+C
autrijus and it involves a constSym.
that's all.
Aankhen`` It's in a different location, which is in the path.
chromatic Hm, I was considering refType of the invocant. Where can I look up the current class?
autrijus envPackage, surprisingly. 19:45
chromatic Alright, let me find where that is.
Limbic_Region would like to kick Rob (dragonchild) sometimes
autrijus but that is perhaps for $?PACKAGE
Aankhen`` autrijus >> Can I send you a first draft of docs/quickref/rules?
autrijus although I confess I'm not sure of the difference 19:46
chromatic Is there no current distinction between packages and classes?
autrijus Aankhen``: er, you're not committer?
Aankhen`` No.
vcv-- thats definately odd Aankhen``. i dont even get to that part of the compile, or else id try to help more
autrijus Aankhen``: hey. your email please
Aankhen`` Do you think you could look over it once first, though?
autrijus chromatic: no, there is not, and I'm a bit unsure about how they interace. 19:47
chromatic: I mean, a class is a package, right?
s/interace/interact/
but I need to ask my perl6 question ere I forget it
chromatic Go ahead.
autrijus in "Array of Int", the Array role is parameterised by the Int class
stevan mugwump was thinking that Class isa Package
autrijus or is it the Array class?
but anyway, say I have a userdefined "Set" 19:48
how do I specify it can be parameterised with "Set of Int"?
end of question.
chromatic I don't think anyone's decided if Array is a class or a role. I'm not sure it matters.
It probably should be a role, though.
Is Set a class or a role?
autrijus is "Set of Int" the same as "Set[Int]"?
chromatic I *think* so.
autrijus I don't know, I'm merely asking what the "of" thing is operating on.
then all containers needs to be roles 19:49
parameterising over its element type
vcv-- Aankhen``: i just ran perl Makefile.PL; nmake; on my system and it worked fine...
autrijus (which is what Fortress does with its traits system)
vcv-- er wait
Aankhen`` vcv-- >> You had PUGS_EMBED set to parrot?
And PARROT_PATH set to the location of a built source tree?
vcv-- yeah.. dont think so. lemme work on that
Corion autrijus: I have a Haskell / Pugs question, later :)
chromatic I don't think they *have* to be roles, but they just have to be parametrizable. 19:50
iblech chromatic: Are classes parameterizable? Thought only roles can be parameterized
Aankhen`` How can you set the path that GHC looks in for .lib files?
autrijus chromatic: I think larry ruled classes never take type params.
Corion Aankhen``: -I or -i
Aankhen`` OK. 19:51
Er... no environment variable?
chromatic Ah, I didn't realize that.
Corion Aankhen``: Not that I know off. www.haskell.org has some documentation on it. ghc --help isn't helpful
Aankhen`` I'm looking at the docs.
chromatic I can't see it not operating on the Set then, Autrijus.
vcv-- Aankhen``: mind if i PM you and ask about following the same steps you did to see if i can reproduce the error? ive never tried to link parrot and pugs before 19:52
Aankhen`` vcv-- >> Sure thing.
chromatic It's clearly making something more specific than an unparametrized Set.
autrijus chromatic: then how come Hash is only parameterised over one type
and the other (key) type needs to be passed over another shape() trait? 19:53
chromatic Default behavior of a hash? I can only guess.
autrijus I mean, usually Hash is defined like this
chromatic @Larry[0] is the only one who really knows what he has in mind here, I think.
autrijus role HashTable[::KeyType, ::ValueType] {...} 19:54
but in perl6 it's like
role HashTable[::ValueType] { does shaped; }
(or is it?)
anyway, it's really strange and I can't reason about it.
I'll think about it some more before p6l'ing it :) 19:55
chromatic Yeah, I can't figure out why a normal hash needs shaping.
autrijus another quick question. say I want to implement ruby yield().
I know the coro syntax is still up in the air.
but one can implement coro using cont.
problem is, what is syntax for cont? ;) 19:56
chromatic callcc is definitely out; it's not ugly enough.
autrijus if neither are specced, I'll need to improvise.
Pugs has &?CALLER_CONTINUATION
but it's by definition an escape continuation
chromatic Yeah, but how great is the difference between a calling continuation and a return continuation? 19:57
autrijus I'm not sure how to expose full cont, coro, or the like.
there's none, that's why I chose the CALLER syntax
it's clearer than &?CONTINUATION
which is ambiguous
&Continuation::CallCC 19:58
chromatic &?COME_FROM_CONTINUATION and &?GO_TO_CONTINUATION? :)
autrijus &Continuation::Shift
&Continuation::Reset
is not ambiguous either
&Continuation::Yield
so maybe go with that?
chromatic I do like yield.
autrijus with lower case anyway
&Continuation::yield
use Continuation <yield>;
chromatic That implies coros to me though. 19:59
autrijus right, then we need to figure a way to mark a call as a coro call.
Limbic_Region my favorit subject
autrijus (or define the language such all call is potentially a coro call.)
Limbic_Region ;-)
chromatic I like the latter.
autrijus I understand that luqui dislikes the latter. 20:00
chromatic He's young; he'll adapt.
autrijus ooh.
wolverian doesn't parrot give us that for free, too?
Aankhen`` How do I pass parameters to GHC using Makefile.PL? 20:01
chromatic Nearly free, yes.
Limbic_Region autrijus - after further conversations with him, I don't think that is actually the case
Luke sees a distinction between gather/take and coroutines
the former being already defined (mostly) and the latter undecided on
autrijus Limbic_Region: er, yes.
Aankhen`` Nevermind, I think I found it.
Limbic_Region what he doesn't like about certain coroutine proposals is the mutability of arguments 20:02
chromatic Hm, I can't seem to make envPackage work. Doesn't it need a getState call? But that's out of scope here.
Limbic_Region but some form of a coroutine he is fine with
autrijus chromatic: getState is only in parser monad
chromatic: in Eval monad it's called ask
Limbic_Region or at least that's how I interpreted our last conversation
autrijus chromatic: you want
asks envPackage
the type of it is (Eval String) 20:03
now try turning it to (Eval (Maybe Val)) :)
chromatic I have to modify envPackage now?
Aankhen`` Argh.
autrijus chromatic: hm? are you still in implementing $?PACKAGE ? 20:04
(and $?CLASS)
Aankhen`` I pass in a directory containing oldnames.lib to GHC but it still won't find the file.
autrijus chromatic: "asks" is a read-only operation
chromatic Just $?CLASS at the moment, but $?PACKAGE is almost free after that.
autrijus "asks envPackage" is defined as
do { env <- ask; return (envPackage env) }
chromatic I have so far: 20:05
getMagical "?$CLASS" = constSym $ asks envPackage
(Just guessing on the purpose of $ there based on other examples) 20:06
iblech chromatic: I think what autrijus meant was that you have to make $?CLASS return a Class (Just Val) object (or undef (Nothing)), not a String
autrijus right. but that doesn't work
chromatic Yep, big type errors.
autrijus see
constSym :: String -> Eval (Maybe Val)
asks envPackage :: Eval String
you can't pass in a (Eval String) to a function expecting a String.
wolverian what on earth is an Eval String? :) 20:07
autrijus you have to first turn that (Eval String) into a String.
chromatic Ahhh, I wish GHC showed the patterns it tried to apply on error.
wolverian (I still need to learn Haskell.)
autrijus wolverian: Eval String is something that calculates a string.
wolverian: it's not a string; it's an action that, when performed, returns a string.
you can perform an action using
result <- action
syntax.
wolverian autrijus: right. a promise?
autrijus so in chromatic's example here:
do
package <- asks envPackage 20:08
constSym package
will work.
but it's maybe easier to write it as
constSym =<< asks envPackage
where =<< has this type:
wolverian hmm. I would want to inline the <- (being the perl coder I am)
chromatic Ahh, I tried to apply it too soon.
wolverian oh. neat.
autrijus (=<<) :: (a -> Eval b) -> (Eval a) -> Eval b
that is, it accepts two actions 20:09
performs the right one first
then put the result to the left one
returning its final result.
(look at the a and b types)
wolverian that hurts my brain.
iblech is happy as he had the =<< line in mind before autrijus typed it :)
autrijus of course, timtowtdi, so you can also write
asks envPackage >>= constSym
and it will also work
as >>= has the type:
(>>=) :: Eval a -> (a -> Eval b) -> Eval b 20:10
chromatic I kept thinking constSym had String -> String
Aankhen`` autrijus >> Shall I send the draft in the morning, then?
autrijus Aankhen``: what's your email?
Aankhen`` I /notice'd you. 20:11
[email@hidden.address]
wolverian autrijus: an (Eval String) is basically a Code returns Str?
autrijus I missed it
wolverian: well yes.
wolverian autrijus: hmm. okay. thanks!
Aankhen`` G'night for now.
autrijus wolverian: but I'd be misleading you. 20:12
(Eval String) means, more formally:
an action that, when performed
may cause I/O side effects
and/or cause STM transactions in memory
wolverian autrijus: right. but perl6 defaults the other way.
chromatic Eval is a monad? 20:13
autrijus within which you can always look up things in your lexical (global) environment
wolverian (that is, you explicitly mark something as safe, not the as monad)
(safe = atomic)
s,the,,
autrijus and within which you can always invoke continuations to return early
this action, when fully performed, gives you back a String.
chromatic: yes.
all Perl 6, as evaluated by Pugs, is in the Eval monad. 20:14
there will also be CompilePIR and CompileGHC monad.
the plan is to unify the three under a class 20:15
so when you write
asks envLexical
in interpreter mode it looks up the lex in memory
in PIR compiler mode it generates the neccessary opcode to retrieve lex pad
and in GHC compiler mode it loads the current binding in topic
etc.
chromatic Hmm, envPackage seems wrong here.
autrijus so the code in Eval.hs can be shared.
chromatic: it's fit for $?PACKAGE 20:16
chromatic Definitely not for $?CLASS though; it's giving 'main';
autrijus chromatic: $?CLASS likesly wants something else. maybe a test first? :)
yeah, because it default to main
chromatic Let me fix up the $?PACKAGE tests in here and I'll commit the $?PACKAGE part.
iblech $?CLASS is already tests in t/oo/magical_vars.t 20:17
chromatic Actually yeah.
iblech s/tests/tested/
autrijus k
chromatic I don't know what the difference is between a class and a package; I don't know if there should or shouldn't be one.
autrijus and I'll make envPackage actually update when you call across package scope
chromatic Oh, okay. I'll add the tests then.
autrijus chromatic: hearing this from one of @Larry is alarming
chromatic It's a special array. 20:19
As the index increases, comprehension decreases.
What's steeper than the inverse square?
autrijus threshold
;) 20:20
Corion autrijus: I have rawSystem() die on me with something that is not an ExitCode but something else. How do I trap exceptions in Haskell?
autrijus Corion: catch
Corion: see how tryIO is implemented 20:21
and grep for catch
chromatic Hm, undeclared variable $?PACKAGE, but I can't find any declaration of $?CLASS in the parser.
Corion (I tried to make tryIO work, but it didn't)
autrijus: Should tryIO work for me in that case? Or should I adapt tryIO to my needs?
Corion hunts tryIO
autrijus chromatic: just commit the code you have first? :)
(so I get to look at it)
Corion: catch it yourself I think
Corion: fallbcak to #haskell if you must :) 20:22
chromatic #3465 20:23
autrijus discovers p6explain
pugs> class Foo { method bar { say $?CLASS } }; Foo.new.bar; 20:24
Foo
as of r3465.
svnbot6 r3464, autrijus++ | * make envPackage actually update when entering a closure
r3464, autrijus++ | defined in another package.
r3465, chromatic++ | First stab at making $?CLASS and $?PACKAGE work. It compiles anyway.
autrijus chromatic++
"it compiles! it must work!"
chromatic It's Haskell. That's a good metric.
theorbtwo We have the option to warn when the impossible might happen turned off. 20:25
autrijus theorbtwo: you mean -Wall?
but GHC alone can make the impossible happen.
"easy things are hard, hard things are trivial, impossible things just happen"
chromatic "... once you stop seeing type mismatch errors." 20:26
theorbtwo -fwarn-incomplete-patterns 20:29
(OK, not turned off, but not turned on.)
chromatic Hmm, VCode has subEnv which carries the subroutine's package name, doesn't it? 20:33
autrijus yes.
chromatic: r3464 does that
(have you synced up to r3464?) 20:34
chromatic Uhh, I might be at 3463.
autrijus please svn up and it will magically work.
PerlJam magic++
chromatic What's maybe? 20:35
svnbot6 r3466, autrijus++ | * bikeshed: align die() and warn() with perl5.
autrijus chromatic: ask ghci :)
make ghci
or just "ghci"
maybe :: b -> (a -> b) -> Maybe a -> b
:t maybe
maybe :: b -> (a -> b) -> Maybe a -> b
(the ":t" is how to ask ghci about a function) 20:36
maybe takes three args
a default case to use when the candidate Nothing
a transformer to use if the candidate is Just a
the candidate
...and returns the applicable case.
chromatic Does the last case recurse? 20:37
autrijus "maybe" does not recurse.
if you want recursion, try "until"
chromatic Oh, I was reading it wrong. 20:38
Hm, now why does $?CLASS fail? 20:39
Aside from the fact that it probably shouldn't be a string.
autrijus pugs> class Foo { method bar { say $?CLASS } } Foo.new.bar
Foo
is that not correct?
(did you recompile?)
chromatic: ok, I'll have it yield a Type object. 20:40
chromatic I *think* that's right. 20:41
autrijus pugs> class Foo { method bar { say $?CLASS.ref } } Foo.new.bar
Type
chromatic Can't find anything in AES about it though. 20:45
svnbot6 r3467, autrijus++ | * $?CLASS is now a type. 20:51
r3468, iblech++ | * Added a test testing method references (t/oo/methods/ref.t).
r3468, iblech++ | * Added the usual svn properties to ninereasons++ t/pugsbugs/unhashify.t.
autrijus g'nite! 20:57
autrijus waves &
webmind nite
putter 'night 20:59
svnbot6 r3469, iblech++ | * Fixed t/oo/magical_vars.t. 21:02
r3469, iblech++ | * Made $?CLASS a Typoe object, too.
r3469, iblech++ | * Added $?ROLE to Eval.hs (same as $?CLASS currently).
r3470, iblech++ | Fixed [**] test.
ingy hola 21:11
brentdax ?que tal? 21:13
putter hey ingy. 21:17
ingy hi putter 21:18
mugwump ingy! 21:20
mugwump runs over and hugs ingy
ingy hi mugwump 21:23
stevan hey mugwump, check out Perl::Meta::Compiler 21:25
it constructs and evals perl6 objects now
Juerd autrijus: How does the prefix ? operator in your journal's example not clash with Bool context ?? 21:26
stevan Juerd: which one? 21:27
wolverian Typoe object :) (re: r3469) 21:28
stevan Juerd: I dont see ? defined in the journal, maybe the char is not displaying properly for you?
putter Juerd: fyi, autrijus bid goodnight 1/2 hr ago. 21:29
wolverian same here.
stevan hey putter, want to talk Perl::Meta::Type? I have given up on $work for the day :)
ninereasons iblech, what is the procedure for adding those "usual svn properties" ? 21:31
Juerd stevan: "?"
stevan: Hm, has to be then :) 21:32
stevan Juerd: what sentence is it in?
Juerd Hm, yes, it's a non-ascii char :)
stevan: It's an example of user defined operators, in a verbatim block.
wolverian I see infix:<Z> 21:33
(I mean, it looks like Z. in the newest journal.) 21:34
ninereasons the zorro operator
wolverian haha.
putter stevan: sure, though 21:43
iblech ninereasons: (couldn't sleep...) it's: 21:44
putter I'm not sure I know much more than I did earlier today. pursuing individual questions was too inefficient, so I started a "big sweep"... in progress.
iblech svn propset svn:eol-style "native" foo and
svn propset svn:mime-type "text/plain; charset=UTF-8" foo
stevan putter: bug sweep of ? 21:45
Juerd: ping (a perl6 question if you have a moment)
putter AES, p6l postings, etc.
stevan putter: s/bug/big/
putter: ok 21:46
ninereasons got it iblech thank you.
putter Faster than pursuing stuff like "so, you define a subtype ST where { defined }. my ST @array. what is @array[0]?" ;)
Juerd stevan: pong
stevan Juerd: I am trying to make this code: 21:47
pugs -e 'sub i ($f, *@args) { $f(*@args) }; i(sub ($a, $b) { say "a: $a\nb: $b"}, 1, 2, 3)'
print a: 1 b: 2
basically take flattened args,.. and then flatten them as I pass then to $f
21:47 Odin_ is now known as Odin-LAP
stevan so that $f's params work 21:47
Juerd Shouldn't that work as it is?
stevan is that possible?
Juerd: that is what I thought :) 21:48
it doesnt
Juerd I think that's a bug then.
stevan :)
cool
I will make a test
putter stevan: any idea?: "A role applied with C<does> may be parameterized with an initializer in parentheses, but only if the role supplies exactly one attribute to the mixin class:" 22:02
svnbot6 r3471, Stevan++ | Splaty args dont seem to work well, which makes the method compiler trickier; this tests the error, i will work on a workaround for now
stevan putter: I have to read more about role paramters 22:03
I dont remember much of their details
I assume too that any discussions we have will result in many p6l emails
putter ok. thanks. it seemed sufficiently bizarre to be worth ask about, rather than just adding to the pile for later. 22:04
cognominal can someone explain the beginning of limbic_region journal? 22:05
stevan putter: I will re-read S12 now actually, I need to really look over it in more detail than I have previously
cognominal why adding a semicolon at the end of a block makes a difference?
iblech cognominal: Because then it's surely a Code, and not a Hash 22:07
cognominal: { a => 42 } is a Hash
cognominal: But { a => 42; } or { ; a => 42 } or -> { a => 42 } is a Code
cognominal oki. 22:08
that will surprise many people
putter k
Khisanth perl needs some other brackets for hashes and code blocks :/ 22:10
mugwump finishes catching up on the day's backlog, very interesting 22:30
svnbot6 r3472, Stevan++ | Perl::MetaModel - comming minor changes
r3473, autrijus++ | * associativity-preserving reduction over left, right
r3473, autrijus++ | and chain infix operators.
r3474, autrijus++ | * allow addressing &infix:<**> by name.
r3475, autrijus++ | * "left" == "pre" in reduce
r3476, autrijus++ | * correctly implement chained ops like [>].
iblech YAY!
autrijus++ 22:31
Khisanth either the bot has a long delay or autrijus is coding in his sleep :)
iblech The bot has only a marginal delay, so he *is* coding in his sleep :)
mugwump must ... finish ... committing ... before ... daybreak..
iblech :D 22:32
elmex ...
Khisanth iblech: that explains his productiveness
elmex mygirlfriend++
svnbot6 r3477, autrijus++ | * &reduce always reduce from left to right, unlike [..]. 22:39
r3478, autrijus++ | * [...] now evaluates its arguments in slurpy context.
r3479, iblech++ | unTODOed the now suceeding tests of reduce.t, $autrijus++ for 1..Inf! 22:56
putter stevan: how goes? 23:00
mugwump: ping? 23:04
svnbot6 r3480, ingy++ | Perl 6 pragma for Perl 5
mugwump hey
putter btw, I added, lets see, what was it... paramValueType and paramImplType to Perl::Param. A toe-in-the-water effort. Any feedback welcome. 23:05
mugwump Did those symbols come from the Haskell source? 23:06
or did you just think they were needed :)
putter later. S mumble was rather explicit.
s/later/latter/
not the right thing? 23:07
mugwump the idea was that those classes are what you would get out of introspecting the Haskell source 23:08
does that make sense? 23:10
putter ah. not right thing then.
should have been in meta somewhere.
mugwump You are doing the right thing, but the names should be taken from the Haskell source where possible 23:11
putter ah, ok. (adding note to README...) 23:12
mugwump pugs is a prototype! so let's build the prototype Perl 6 Perl 6 compiler from the prototype Haskell Perl 6 compiler :)
Hopefully at some point there will be enough code there that a massive arc of energy joins the Perl 6 and Haskell code bases together 23:13
putter ;)
mugwump bzzzts
putter just Haskell? <putter looks dejected> I have grammars and sometimes runtimes for C, C++, Scheme, ... . Let's have a nuclear barbeque. 23:14
osfameron and javascript? 23:15
mugwump yes of course... because those languages fit somewhere in between the mathematical purity of Haskell and the irreverant linguistic chaos of Perl. They will be electrolysed in the process. 23:16
putter suddenly has images of electroshocking bacteria to get them to swap code. 23:17
pugs/p6/parrot/whatever as linguistic membrane perforator... 23:19
mugwump: nice metaphor. 23:20
just how does one metamodel "A given enum value can function as a subtype, a method, or as an ordinary value (an argumentless sub)."? >O 23:21
mugwump That's a good question 23:22
that comment is from AST/Internals.hs? 23:24
You see, usually when I've been working with metamodels, they have been fairly simple. Like, in T2 there are only four metaclasses - Class, Attribute, Method and Association 23:26
putter S12
mugwump oh
wolverian T2?
mugwump anyway, I had a hunch while starting on this that the metamodel that can describe itself to the language level needs the typing stuff there too
svnbot6 r3481, putter++ | Removed some inappropriate attributes from Param.pm. Added clarifying note to README.
putter makes sense. 23:27
kelan does map take adverbial blocks like grep? ie: %hash = map:{ $_ => $_ } @array 23:28
putter is a member of the church of Everything Should Be First Class.
mugwump so the fact that putting the metamodel into Perl 6 seems to be pulling the whole Pugs code base with it doesn't surprise me overly :)
putter :)
mugwump kelan: yes, that's right. But you look like you want a Set with that idiom :) 23:29
kelan that was just an example from limbic region's journal
he gave two other ways to write it, but left that way out
just making sure he didn't leave it out on purpose and i was missing something 23:30
knewt worries about crossing the streams 23:32
*g*
putter kelan: I don't actually see a test case for map:{...} @array . Care to create one? :) 23:34
kelan ive been thinking about asking for committer. how does it work with darcs? or can you only read with darcs and must use subversion for read+write? 23:35
or does darcs push work?
putter re, darcs, I'm unsure, but I believe it is ro. 23:40
re committer, we're in the middle of the ~4hr daily lull on #perl6. elsewhen, there are folks with committer committing bits... 23:42
re map:{ $_ => $_ } @array ... hmm. are you sure mugwump? can adverbs be in the middle of an argument list? 23:43
kelan does no one who works on pugs live in the eastern time zone? its always weird hours in here
mugwump @array.map:{ $_ => $_ } would be how I'd write it 23:44
kelan ah, maybe thats what im thinking of
mugwump is in +12:00
So it's Friday! :D
kelan heh 23:45
must be nice
putter so to sum up, map:{ $_ => $_ } @array is not valid. @array.map:{ $_ => $_ } or map { $_ => $_ ;}, @array 23:46
kelan well if it can't be in the middle, how about: %hash = map @array :{ $_ => $_ }; 23:48
revdiablo seen Limbic_Region
kelan pretty weird lookin
revdiablo hmm, what bot is that?
kelan !seen kelan
shrug
mugwump seen jabber?
doh!
jabbot, I mean :)
revdiablo [2005-05-18 04:18:52] * jabbot [gugod@irc.csie.ntu.edu.tw] has quit [Remote closed the connection] 23:49
doh
putter speaking of timezone...
stevan: perhaps compare notes tomorrow? 23:51
'night folks.
svnbot6 r3482, autrijus++ | * support I/O in Inline::Pugs.
stevan kelan: I am EST :) 23:57
kelan yay!
well
edt?
stevan edt? 23:58
I am in eastern US
kelan its day-light savings time right now
but there are parts of indiana that don't observe it i believe
stevan oh, I dont pay attention to that stuff :)
kelan hah
stevan kelan: where are you?
kelan nc
stevan mugwump: ping 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!