6.2.10 released! xrl.us/hxnb | geoffb's column (/. ed): xrl.us/hxhk | pugscode.org | pugs.kwiki.org | paste: sial.org/pbot/perl6 | www.geeksunite.net
Set by autrijus on 10 October 2005.
stevan hey luqui 00:19
autrijus adamc00: yo
journal up. good night! 00:39
stevan good night
luqui hi stevan 00:51
stevan hey luqui 00:52
luqui you could also think of a class MyClass as a role, and use the property system to make the class object. That is, (Object but MyClass).new 00:54
er, Class but MyClass
stevan luqui: I am not clear how all that works
luqui wait, maybe it is Object
when you say $foo but bar, you are "reblessing" $foo into an anonymous subclass of its former class with the "bar" role mixed in 00:55
stevan yes, that is similar to eigenclasses
luqui you could use the same mechanism
stevan rebless == change-class-of 00:56
luqui sure
stevan luqui: I dont have properties implemented,.. eigenclasses are :)
wolverian this might be a stupid question, but how do I specify that a method's invocant is the class itself, not an instance of it? (ie a class method)
luqui well maybe you can go the other way then
use eigenclasses to implement properties :-)
stevan wolverian: that is what we are talking about actually
luqui: that may be the way to go :)
wolverian stevan, heh, I'm just a poor peasant looking for the right syntax :)
stevan wolverian: you say method foo (Class $class:) 00:57
luqui thinks that that way of doing it is a deceptive oversimplification 00:58
I would rather see something like classmethod...
wolverian right, me too.
stevan agree's with luqui
and wolverian
but I am not on the cabal *cough* luqui *cough*
luqui when larry wrote that, though, he was still thinking of declarators as sacred 00:59
now he thinks of them pretty much as compile-time functions
stevan we could be like Ruby
luqui so he might be more open to that approach nowadays
stevan method Foo.bar () { ... }
syntax is not my department though
luqui Yeah, not mine really either...
my syntax looks good at first and then looks worse and worse as you get used to it 01:00
Larry's seems to be the exact opposite... I guess that's why we trust him
oh...kay... you want ==> instead of ~>, crazy man, but we'll go with it 01:01
and now ~> looks too small for what it's doing
stevan hmmm, why do class methods need to be inherited?
does that really make sense?
luqui well... maybe 01:02
stevan are we not inheriting the behavior without the accompanying state?
luqui it depends on what you think about class interfaces
stevan vows never to use class methods again,.. they are just icky
luqui: how so?
luqui If a class C is a subclass of a class D, does that also mean that C (not an instance of, but the class itself) should behave like D
wolverian reading synopses, enums seem really overloaded to me.
luqui likes the "new" method quite a lot
stevan new is an instance method of Class 01:03
luqui wolverian, those are being torn down by theory theory however it works out
that is its primary focus
stevan it is not a class method at all :)
wolverian luqui, ah, nice!
luqui uh what?
you're saying that when I say Foo.new, I'm asking Class to do something, not the Foo object? 01:04
stevan Foo is an instance of Class
luqui yeah
stevan new() is an instance method of Class
luqui no
stevan so Foo (being an instance of Class) can call new()
luqui because you can override it
stevan yes
uhm
yeah, singleton method on the Foo 01:05
luqui righto
stevan and it is easily overridden
luqui so it's really more a method on the eigenclass
stevan yes
which shadows Class::new
luqui okay
stevan but you can still call SUPER::new since the eigenclass is a subclass of Class (ultimately) 01:06
luqui so the eigenclass idea is that there is no difference between a class method and an instance method on the eigenclass?
stevan not exactly
eigenclasses are methods attached to a particular instance
not to a class
however classes are instances,.. so you get 2 for the price of one 01:07
luqui eigenclasses are not methods (in response to "eigenclasses are methods..."
)
so what did you mean to say?
stevan sorry 01:08
eigenclasses are a means to an end
the end is singleton methods
which are methods attached to a particular instance
not to a class
luqui okay 01:09
stevan you can accomplish this easily with eigenclasses
luqui okay, so now back to the question: are class methods inherited?
stevan yes, should they be 01:10
luqui let's start from the top: what does it mean for a class *object* to be inherited from another
what kind of relationship holds between the objects?
stevan it is a subtype relationship
luqui let's try not to confuse classes and types 01:11
stevan each class has a list of superclasses
this means nothing on it's own though
luqui so the set of classes in the program forms a dag
stevan only when you either 1) create an instance or 2) dispatch a method call does it mean anything
uhm, yes it should be dag-ish 01:12
luqui has the urge to go back to mathematical fundamentals
but it might not get us anywhere here
stevan the class is an instance of class messes up the dag-ish-ness
luqui oh right, instance of itself
stevan uhm I was not aware there would be math on this test professor luqui
but that is a special case and can be ignored 01:13
so "its all a dag man" .....
luqui has found that when you come across a hard abstract question, rather than trying to decide the answer, you should rejig your thinking to make the answer obvious
which, at least in my theory and junction proposals, has found great solace in mathematical fundamentals :-) 01:14
aanyway...
stevan class == state + behavior
luqui a description of state 01:15
stevan state
luqui not the state itself
right?
stevan some kind of storage of values
stevan wonders if luqui has gotten to the TaPL chapters on objects 01:16
luqui no, I've not really been reading TaPL
stevan I have only skimmed them myself
luqui should I before continuing?
stevan nah
let me reviese my statement
object == state + behavior
luqui that sounds much better to me 01:17
and class isa object, no?
stevan yes
but that is Class not class
proper noun :)
luqui well, a class is an object
Class or MyMagicClass 01:18
stevan Class is an instance of Class which is a subclass of Object
luqui and a method is a function whose domain is a set of objects (which we call a "type")
stevan remove classes from the model for a moment, they are fussing things up
no, lets go back even further 01:19
luqui okay
stevan object == state + behavior
luqui behavior == ?
stevan think of an object as a set of closures, the functions being the behavior and the state being the bound free variables
luqui heh, like Class::Closure 01:20
stevan set is a bad word there,.. s/set/bunch/
yes, like Class::Closure :)
so,.. if we ignore the idea of classes completely,.. and just think of pure objects
state + behavior
now, lets add inheritence 01:21
stevan cracks open TaPL to make sure he doesnt start talking too much shit
luqui and we are asking whether two objects' behaviors are related to each other
wait--I'll go up my usual avenue. Let's not call them "class methods" but "type methods" 01:22
because you can never really say the name of a class, according to theory.pod
stevan ok class, please open your books to page 225 "Imperative Objects"
luqui a type is a set of objects
let me rant a bit, then I will
stevan ok
luqui and a type method is function with the domain of *sets* 01:23
with that bare definition, there is no requirement for methods to be inherited
but, as I did with roles, maybe there is an algebraic relationship that holds for type methods
like, if a type method is defined for a set, it is also defined for every subset of that set 01:24
giving us inheritance
then... does that make sense
stevan uhm, sort of... 01:25
do theories have state?
luqui Well, let's consider an example.
stevan, ceratinly not. but they can inject stateful functions into classes (just like roles)
that's how I'm thinking of it
stevan ok
luqui what's a good example of a type method? 01:26
how about new?
if you can create a new member A, then you can also create a new member of *any* subset of A
s/member/member of/
that..... doesn't seem to be true 01:27
in particular, you cannot create a new member of the empty set
stevan hmmm this is getting a little funky 01:28
luqui but more generally, creating a new A presumably gives you some predefined element of A. By saying that you can create a new member of any subset, that means you can create any object in that set
well, it's a brainstorm
but maybe "creating a new" object when it is a value type doesn't make sense 01:29
stevan A.new is not really inheirted
luqui (and my subset thinking only works on value types)
fair enough
what kind of class method would be?
stevan the canonical example is an instance counter
A.counter # 0
A.new
A.counter # 1 01:30
luqui interesting
stevan but this doesnt make sense to be inherited by B
luqui well, that particular method has a very interesting algebra
B is A
stevan B.counter should have it's own state,.. uncurrupted by A
luqui A.counter # 0
B.new
B.counter # 1
A.counter # 1 !
A.new
B.counter # 1 01:31
A.counter # 2 !
stevan something like that could not be default behavior
luqui right
stevan it would mean calling A::counter though B would have an implied next METHOD 01:32
luqui anyway, nothing is really falling into place
stevan cause class methods suck camel balls
luqui which makes me think that there is no fundamental reason to inherit class methods
stevan except that you could in p5 :)
luqui right, historical reasons
okay, off theory
justatheory what?
luqui what practical value would it be to inherit class methods
what practical value would it be *not* to inherit class methods? 01:33
stevan bad class design would be alive and well in p6 :)
luqui you can't stop that
wolverian after using Java I'm tempted to ask what practical value are class methods?
stevan to not inherit simplifies the meta-model
wolverian: exactly!
luqui wolverian, good question
maybe Foo.bar is not a class method 01:34
stevan especially since Class isa Package and you can always stash plain ole subs
luqui since you can't actually refer to a class
but just a sub in the Foo package
uh, yeah, right, we just said the same thing :-)
stevan that is how we are storing Foo class state
class Foo { our $.bar; } is the same as package Foo { our $.bar } 01:35
luqui wolverian, did you see *any* example of a class method in java that made sense to be a class method?
wolverian luqui, no.
stevan the only examples I have seen are for things like the File class
luqui it might be worthwhile just to kill them altogether
stevan but I dont think those are *good* examples
wolverian luqui, well, some did, given the restrictions of Java
but would not in the context of Perl 6
luqui right, just as a poor excuse for a function
however, some methods really do make sense on the class 01:36
stevan luqui: if you think of class methods as singleton methods which override or add to the instance methods of Class, they dont seem as bad
luqui from a cognitive perspective
stevan, right
wolverian luqui, any examples?
luqui and it's probably safe to think exactly that way
stevan but methods attached to an instance,.. are not inherited 01:37
luqui wolverian, Foo.num_instances
wolverian luqui, right. true.
luqui stevan, righto
so...
so be it?
"class methods" are not inherited
stevan welllll
ponder this one 01:38
actually nevermind
stevan just realized his example is moot in the context of BUILD
luqui I think inheriting class methods would not be an inheritance issue at all 01:39
and the only reason you'd want to do it is for scoping
stevan I would guess that any behavior you might accomplish with class methods could be just as easily accomplished with either
1) Role composition of singleton methods
2) subclassing Class
luqui stevan, sure, but Joe Shmoe has no idea what the heck subclassing Class means
maybe even role composition 01:40
stevan 3) getting a good book on OO and stop writing procedural modules ;)
luqui say you're from Java
wolverian oh, isn't BUILD a class method?
luqui maybe that kind of scope makes sense...
stevan luqui: then you will have been sworn off of class methods a long time ago
luqui wolverian, no, it is an instance method
stevan wolverian: no, BUILD is a submethod on the instance
luqui submethod, there's an icky corner of the object model 01:41
stevan luqui: actually it's not,.. they are quite clean
wolverian hm. .new, then?
luqui what does it accomplish?
stevan wolverian: new is an instance method of Class
luqui wolverian, right, but that has no inheritance relationship
wolverian stevan, right.
stevan submethods are good for BUILD and DESTROY,.. thats about all really
inheriting BUILD and DESTROY are not usually what you would intend 01:42
wolverian instance method of Class == class method?
luqui and they're not even good for that
stevan especially if they are all run by BUILDALL and DESTROYALL anyway
wolverian: instance method of Class == instance method of Class 01:43
they look like class methods,.. but they are not
because Class itself is an instance
wolverian argh, destroy all self-referential semantics.
stevan LOL
luqui: I will ponder this some more,.. I think it might be a sane proposal 01:44
luqui please do 01:45
stevan it will likely be meet with resistence
luqui object model is important :-)
stevan, I'm not so sure
stevan however, if we can counter all points with examples,.. it should fly
wolverian oh wow, I think I understand the difference now. that was a maze.
stevan needs to take some cold medicine, watch some mindless TV and sleep 01:47
I will likely p6l this tomorrow
adios &
wolverian I have class in two hours; 5am now.
stevan++ # bye! 01:48
02:54 JAPJuggler is now known as radjuggler
eric256 whistles in the dark 02:55
eric256 whistles in the dark some more 04:06
luqui although I'm having a wonderful time I'd rather be doing what you're doing 04:09
eric256 lol
luqui catamorphisms are SO COOL 05:55
obra cataWHATs? 06:00
luqui catamorphism 06:07
you know how fmap is generalized map (from lists to arbitrary whatevers)
catamorphism is generalized reduce
and anamorphism is generalized something, I'm just not sure what 06:08
justatheory Bah 06:27
I'm having trouble building GHC 6.4.1 on Tiger. Do I need GCC 3.3, still?
justatheory knocks on the channel 06:31
Is this thing on?
Khisanth bzzzzzzzzzzzzzzz 06:34
justatheory Hrm 06:35
integral #haskell can be more awake 06:46
PerlJam justatheory: I think the awake ones here are also sans Tiger. You need to find the appropriate Mac-heads. 06:48
justatheory I thought that all geeks were a Mac-heads these days.
integral sure, just cheap ones with Panther =) 06:49
justatheory oh 06:50
well, it's probably really a GCC 4 issue rather than Tiger.
At least, I expect so.
integral are you using a precompiled binary? 06:51
justatheory no
They only have darwinports now.
They used to have .dmg files, but I guess not anymore. :-(
I don't use darwinport.s
or any ports system. I have an allergy.
integral err, really? *sigh* I guess no one put the .dmg on the website. It was announced on the ML
justatheory For GHC? No, just darwinports. 06:52
I was trying to install pugs with CPANPLUS.
integral no, it was announced on the mailing list. I've got the damn email in front of me: www.uni-graz.at/imawww/haskell/GHC-6.4.1.pkg.zip
justatheory That's the dmg? Gool.
integral my mistake, it is a .pkg
justatheory downloads that
Same diff, AFAIC 06:53
s/AI/IA/
integral++ # Thanks for the link to the .pkg! 07:14
integral bows 07:16
justatheory piers++ # "Anyhow, chromatic…" 07:18
nothingmuch oh my *GOD* rindolf is an idiot 07:19
justatheory rindofl? 07:20
rindolf?
integral nothingmuch: has he done something new, or just the same testing + meme stuff? 07:21
nothingmuch the Test::Shlomif::Harness
nothingmuch is embarraced to know him IRL
integral I just feel sorry for the other people who run that perl-meme site
justatheory notes that compiling Perl6::Pugs is causing serious lag…
nothingmuch i really don't see how someone has the tact to make changes like the code Andy quoted 07:22
justatheory Oh, rindolf == shlomi fish?
nothingmuch yes
furthermore, his code is insultingly stupid
justatheory doesn't think that he's been insulted by other people's code, except in comments
piers++ # "Your summarizer wasn't entirely sure what they were talking about." 07:23
nothingmuch =)
perlmeme.org/tutorials/cgi_script.html (also insulting... i suspect it's his work too ;-) 07:24
luqui what's wrong with that? 07:25
nothingmuch it's how to start a CGI script
assuming the user knows pipes, and env variables
and stuff like these
QtPlatypus Yes and the problem is? 07:26
justatheory use CGI::Carp qw(fatalsToBrowser); # NO!
nothingmuch QtPlatypus: the issue is that instead of teaching how to write a CGI the script gives an example, and says what it does, but not what the things it does mean
which is good for a reference, but this is not a tutorial 07:27
integral it's what people want though...
luqui Joe Schmoe would rather copy a template and tweak it than write one by hand
luqui knows this after a year of teaching OpenGL :-)
QtPlatypus nothingmuch: Your worried that it will encouage cargo cult programing.
nothingmuch no, it's more than that 07:28
no one ought to use CGI for HTML
it's just there for backwards compatibility
why encourage that?
luqui nothingmuch, what should you use instead?
assuming little knowledge of perl's workings, of course
nothingmuch HTML::Template is simple, clear, and powerful enough
luqui should check that out the next time he generates HTML... i.e. never :-) 07:29
nothingmuch anyway, nevermind
integral nothing wrong with HTML combinators per se though, just that they're in the wrong module
nothingmuch integral: remember please that this is the first impression someone is going to get 07:30
people whove been coding for 2 months don't appreciate pretty printing combinators
luqui "Perl is a language for getting your job done" # Camel book
if it works faster than any other language, then it makes a good impression
anyway... 07:31
luqui backs off
integral nothingmuch: sorry, I'm just trying to be balanced, rather than ripping rindolf to shreds continually just because rindolf is rindolf.
justatheory likes TT and Mason
nothingmuch integral: fair enough
nothingmuch backs down
integral particulary when for once it's not just rindolf's site.
nothingmuch enough for now
luqui backs up
justatheory does the holky polky
justatheory turns himself around 07:32
geoffb Grrr. Damn SBC incompetent ADSL management . . . . 07:39
*yawn* 07:44
OK, time to crash (into bed), methinks
nothingmuch ciao geoffb 07:49
07:49 Grrrr is now known as Gruber
svnbot6 r7554 | luqui++ | Solved the GADT problem in the Notes section. 08:39
Juerd Quick poll: what are your opinions about getting rid of .() altogether? 10:47
I believe it is unnecessary and confusing for people new to the language 10:48
autrijus $blah() instead? 10:50
$blah .()
or do you mean we rid of code dereference in function application, which is insane? 10:51
bbiab... 11:07
rafl autrijus: ping 11:58
What puts the include-dirs: src/perl5 into Pugs.cabal? 12:25
I mean, where does the string 'src/perl5' come from?
autrijus rafl: util/build_pugs.pl 13:05
(I'm in train station, wireless going off range at any point)
rafl autrijus: Yes, I already found it, but it doesn't matter anymore, I guess.
autrijus ok
rafl autrijus: Why I pinged you: Why does ghc-pkg need to access all include-dirs when registering a package? 13:06
autrijus: We have src/syck, etc. in there which won't get installed.
autrijus oh hm. I have no idea really.
rafl autrijus: So you need a pugs src dir to register with ghc. That's bad, isn't it?
autrijus not _that_ bad as it is usually registered from install time
but not for you I surmise
rafl Indeed.
autrijus can you bring it up with SyntaxNinja or other Cabal people on #haskell?
or on the cabal list
for I need to run now. be back in a couple hours... 13:07
rafl Before I do that: Is there a way to remove the non-absolute paths from include-dirs and still keep compiling work?
OK, bye!
autrijus maybe. or you can absolutify all of them
in build_pugs
rafl Well, won't help. A /home/rafl/projects/debian/.../pugs/src/syck/ won't exist for every user. 13:08
We would need to install the includes.
autrijus nod. 13:10
&
rep ops 13:33
rafl get's ignored in #haskell, yay.. 13:41
iblechbot: ping 13:46
washu does anyone know how to create a byte array from string in perl (like java getBytes?) I'm messing around with unpack, but it doesn't seem to do the job. 13:51
rafl washu: Maybe you want to join #perl. This channel is about perl6/pugs development. 13:52
PerlJam washu: in perl6 it would be @bytes = $string.bytes; # probably :)
washu #perl is not showing up on my list. I thought it was gone. Thx. 13:55
PerlJam: that would be nice =>
PerlJam still isn't clear if it's been blessed that .bytes(), .chars(), etc. return a list in list/array context. 13:56
washu: and when you show up on #perl, I'll help. 13:57
eric256 does pdcawley_ do the weekly summary? 13:59
PerlJam I think Piers and Matt are still tag teaming. 14:07
eric256 ahhh.. from his comment yesterday i thought he was. no matter. just thought it was funny that the they used brane instead of brain. ;) 14:08
rafl ingy: ping 14:26
ingy: I have a problem with Spork 0.20: Can't locate object method "slides_directory" via package "Spork::Config" at /home/rafl/.local/share/perl5/Spork/Command.pm line 44, <DATA> line 1. 14:27
Juerd autrijus: No, getting rid of it, as in: no replacement 14:34
autrijus: Well, no postcircumfix replacement anyway 14:35
autrijus: I haven't really thought about unnamed calls yet
rafl ingy: OK, fixed it. I missused it. 15:12
ingy :)
rafl ingy: Can you tell me how it calculates the size of the images? 15:13
ingy: I have a picture with 200x400 px here which gets streched to 350px width so it doesn't fit the screen anymore.
ingy rafl: just say {image: url 200} 15:14
rafl ingy: Great, spork++; # It should really be in Debian. 15:15
ingy otherwise the width will be set to 350
or whatever the default width is
rafl ingy: Thank you. 15:16
Do you know what nwp (the guy who want's to maintain it) does currently?
ingy no
rafl :-( 15:17
Jooon he is in the #kwiki channel currently, although there is not much activity there 15:19
svnbot6 r7555 | iblech++ | * ChangeLog: Typo fix. 15:23
r7555 | iblech++ | * PIL2JS:
r7555 | iblech++ | * jspugs.pl: Fixed the display of the paths to the various external
r7555 | iblech++ | programs/files jspugs uses if $*CWD is not "$srcdir/perl5/PIL2JS".
r7555 | iblech++ | * P5 Prelude::JS, P6 Prelude::JS::Operators: Division by zero and module zero
r7555 | iblech++ | raise exceptions now (instead of doing what JS does, namely returning
r7555 | iblech++ | Inf/-Inf or NaN).
eric256 can you set $*IN to chomp automaticaly yet? 16:08
hmmmm 16:16
svnbot6 r7556 | eric256++ | Updated animals.p6 with new language features. 16:24
rafl autrijus: I'll do a talk on perl6/pugs next month. Can you please tell me what you told the audience in the 4 slides following 'But First ...' in your apocalypse talk? 16:26
svnbot6 r7557 | eric256++ | hangman.p6 - some more general pugs-ification 16:42
r7558 | eric256++ | Examples\Games\*.p6: Updated input methods to chomp(=$*IN) 16:48
r7559 | iblech++ | * t/builtins/strings/string_interpolation.t: "&func .()" and "&func. ()"
r7559 | iblech++ | should (and doesn't in current Pugs) interpolate.
r7559 | iblech++ | * examples/games/animals.p6: my %hash = {...} creates a hash with one elem (key
r7559 | iblech++ | is {...}, value is undef). Also see t/var/assigning_refs.t.
r7560 | iblech++ | PIL2JS: Prelude::JS::Operators: Implemented ~& ~| ~^, using a higher-order 17:00
r7560 | iblech++ | function which operates on JS sourcecode. :)
r7560 | iblech++ | Note: We need more tests for the bit operators (especially for the edge cases,
r7560 | iblech++ | e.g. "foo" ~& "", "\foo" ~& "\0" etc.). But, as I don't know what the correct
r7560 | iblech++ | result should be, somebody else has to write the tests. :)
17:02 cm_ is now known as cm
pdcawley_ Yes, I still write the summary, or at least, I wrote the last one. And branes was deliberate. 17:07
PerlJam I think I'm going to have to killfile Thomas Sandlass. Reading his posts to p6l is just too tiring. 17:09
cm :-) 17:11
eric256 hehe....any ideas why my pugs programs don't seem to be getting @ARGS filled? 17:12
PerlJam eric256: example?
eric256 if i do... @ARGS.perl.say; (as the only part of a program.) then do 17:13
pugs example.p6 test
it prints []
not sure but i think pugs is sucking up the arguments...
broquaint Works fine here. 17:14
eric256 this is on windows
lemme test on feather
broquaint Ah, this is on feather.
eric256 definitly works fine on feather...gonna try the exact same file..maybe i did something dumb 17:15
lol
PerlJam works fine on my local non-feather box too :) 17:16
eric256 gtg i'll try later
PerlJam (of course I'm running linux)
cm what's feather :) 17:22
17:23 typester is now known as sleepster
svnbot6 r7561 | iblech++ | PIL2JS: Prelude::JS::Operators: Implemented +^ and ~^. This means PIL2JS passes 17:30
r7561 | iblech++ | 17/17 of bit.t. :)
autrijus cm: it's feather.perl6.nl, aka perlcabal.org, Juerd's server that provides free shell access to p6 developers 17:31
cm thank you
autrijus iblechbot: where is it written that "&x .()" or "&x. ()" should interpolate? 17:32
iblechbot: we don't do that for array/hash either and I think it makes little sense
(and a casual glance in S02 did not reveal that syntax...)
PerlJam those shouldn't interpolate IMHO
I too think it makes little sense
autrijus ditto 17:33
autrijus ponders
PerlJam I thought the rule was that it would interpolate only if the sequence was contiguous (i.e. no intervening whitespace) 17:34
autrijus yeah
iblech Hello everyone 17:35
rafl: pong
autrijus: It should *not* interpolate. This is what the test tests :)
autrijus: I.e. "&func .()" should eq '&func .()', but "&func.()" should eq ~func() 17:36
theorbtwo I like pb&j. (But not with lemon jelly!) 17:37
fglock__ iblech: I had to interrupt a 'make smoke' with ctrl-C - is there a way to look at the results?
iblech fglock__: As far as I know, no :( As tests.yml is only written at the very end
cm "Cannot export Parrot_string_find_digit: symbol not defined [..]" -- any idea? :( 17:38
autrijus iblech: ok, my fault for only looking at the commitlog :) 17:39
cm: how did you trigger that?
cm trying to mingw32-make parrot, as described by wiki.kn.vutbr.cz/mj/index.cgi?Build...ActivePerl
iblech autrijus: No, my fault for forgetting the small suffix "n't" (shouldn't instead of should) :)
cm autrijus, this happens when creating parrot.exe 17:40
autrijus, i am the latest code from subversion (../trunk)
autrijus leo__: your ground here :) 17:41
cm: did you do a really fresh build?
cm autrijus, I guess.. I just checked out the code from subversion, then ran Configure.pl & started make'ing 17:42
autrijus cm: send details to [email@hidden.address] or visit irc.perl.org #parrot, can also work
cm: nodnod. the much-needed varreg refactoring is in place, so I expect the trunk to be somewhat fragile
cm autrijus, i'll try #parrot first, thanks
autrijus, is that the variable frame refactoring thing leo was talking about on perl6-i?
iblech cm: Yep. 17:43
svnbot6 r7562 | iblech++ | * t/operators/binding/subs.t: Added a =begin unspecced guard around a subtest. 17:47
r7562 | iblech++ | * PIL2JS: Prelude::JS::Array: &splice should destroy possible bindings of the
r7562 | iblech++ | array elements when splicing. This cuts down test failures of
r7562 | iblech++ | t/operators/binding/arrays.t from 3 to 1.
leo__ cm: recent Parrot? 17:49
cm leo__: latest svn. mdiep_ solved my problem, [19:45] <mdiep_> cm: I believe parrot.def needs to be edited -- those functions were recently removed
leo__ yup excatly
cm leo__: I guess it's too trivial to send a patch? :)
leo__ I can't test on Win, so sill send a test please 17:50
argh, lagging connection - sorry
cm leo__, where is parrot.def generated? 17:51
leo__ it should have a comment at the top - else grep inside config for the failing symbol 17:52
# over to #parrot
svnbot6 r7563 | iblech++ | * t/pugsbugs/smartmatch_autovivifies.t: Use a P5 instead of a P6 regex object. 18:05
r7563 | iblech++ | * PIL2JS: More squashing of minor bugs
r7563 | iblech++ | * Prelude::JS::Array: 3..2 results in the empty list now (previously, it
r7563 | iblech++ | evaluated to (3)).
r7563 | iblech++ | * Prelude::JS::Array: %hash.sort works now (the result is a list of pairs.
r7563 | iblech++ | Dunno how/whether it's specced).
iblech Hm... {42}.arity is 1 in current Pugs, and there exists a test for this. Why is this so? (In PIL, {42} is a block taking an *optional* $_, and optional params don't count in the .arity, right?) 18:07
wolverian should .arity ever be a simple scalar? 18:11
iblech There was a thread about that, but there was no final ruling, IIRC 18:12
wolverian right. 18:13
xinming_Beijing hmm, Will pugs need ghc 6.6 in the future after ghc 6.6 released? :-/ 18:14
autrijus xinming_Beijing: most likely yes, when 6.6 is deemed stable enough 18:15
xinming_Beijing I sometimes got the complie progress failures. And I use GHC 6.5...
So, Now, I will try GHC 6.4.1 instead...
autrijus well... you are on your own with 6.5.x :)
6.4.1 is recommended; 6.4 is possible
GeJ I guess so, autrijus was involved in a hackathon at IFCP to add features that would ease pugs implementation IIRC 18:16
svnbot6 r7564 | iblech++ | * New t/builtins/arity.t: Added four very very simple tests for &code.arity and
r7564 | iblech++ | referred to a thread on p6l (which didn't contain a final ruling).
r7564 | iblech++ | * t/subroutines/bare_block_with_dollar_underscore.t: Extremely minor cosmetical fix.
autrijus GeJ: yeah, but more importantly, GHC 6.6 has much better SMP support and saner GADT
gaal is GADT pronounced "gadget"? :)
obra How many changes in recent GHC are driven by pugs? 18:17
PerlJam autrijus: and transactional memory?
wolverian oh, hell, VB9 has STM?
what will I ridicule now? 18:18
autrijus PerlJam: STM is already in 6.4 18:19
theorbtwo I only really meant for .airty to be a first step toward a prototype method that returned the complete prototype.
autrijus obra: two or three syntax things mostly
PerlJam wasn't sure how well it was supported
autrijus only one deep thing that I had not had tuits to finish it
PerlJam: it works quite well but not yet SMPized 18:20
theorbtwo SMP support is getting to be important now that HT is common.
Khisanth theorbtwo: so there is some sort of &func.signature ?
GeJ more important as all the manufaturers tend to multi-core I guess 18:21
theorbtwo Not that I'm aware of; I never got further then .airity.
Khisanth just thought that might be useful for the lazy documentation writer :)
theorbtwo That was my thinking as well.
xinming_Beijing Perl 6 will be compiled to bytecodes. So, I wonder, If the "binary" will contain the comments. And also, I wonder If perl will also do the same in perl 5, hmm, just like. `cat file.p6 | perl`, and if you use `perl file.p6`, perl 6 will automatically call the compiler and generate the file.p6b, handles regeneration of the file as python does... hmm, python will do this if IIRC... 18:27
maybe not comments, I mean the document part of the source code. 18:28
Khisanth why? will compiling in p6 be a lot slower than p5? :) 18:31
autrijus xinming_Beijing: good question. larry's decree was that by default the source is stored with the compiled binary
Khisanth: possibly, yes :)
Khisanth: if you ask for it, that is 18:32
Khisanth If I ask perl to be slower?
autrijus Khisanth: if you want to trade compile time for runtime.
think -O0 vs -O2 vs -Ofun (the latter will nopaste your program to IRC and send back amusing reviews) 18:33
xinming_Beijing autrijus: How about the document part of the source code? they are useless "most" times.
autrijus xinming_Beijing: but %=DOC (or %=POD, or whatever it's called today) will hold them
xinming_Beijing: so they are stored as well by default. I imagine there will be strip programs
and/or options 18:34
xinming_Beijing I'd prefer 3 choice... automatically, strip, keep :-)
automatically will strip if it's not used, otherwise keep. :-P 18:35
autrijus it's hard to detect if it is used or not, but well, sure, you can apply heuristics 18:36
"heuristics", aka "does not work too well in general but you get what you ask for"
Khisanth not "works most of the time"? :) 18:37
geoffb -Ofun article has gotten up to rank 1163 in the O'Reilly weblog ratings.
geoffb sets a personal goal to break 1000
stevan geoffb: how do I rank it? 18:39
geoffb (For reference, my previous best was 3908)
cm autrijus, can I build pugs without hs-plugins (some build problem on win32) 18:40
geoffb stevan, I think it's based on page hits
stevan keeps refreshing his browser window
geoffb (That was not a call for spurious hits. :-)
I meant, I'm setting a personal goal to write something cool enough that it naturally breaks into the 3-digit range
autrijus cm: sure, it is not mandatory 18:41
geoffb practice, practice, practice I guess
stevan sets up a script to cause all his companies servers to hit geoffb's article
xinming_Beijing autrijus: hmm, I think, for the code which directly used the %=DOC will be detected. though, except some codes which is running in some tricks.
geoffb stevan, LOL
autrijus xinming_Beijing: *nod* that is the idea, yes 18:42
xinming_Beijing stevan: while true; do wget my.blog.url/ && sleep 5s; done;
stevan: this might help... But I don't know if it will really hit the counter. :-)
autrijus I think "ab" works better in general
stevan just hired a roomful of offshore programmers to hit geoffb's blog 24/7 18:43
geoffb LOL
Nothing like a good laugh in the morning to really make the day better . . . .
stevan autrijus: did you see my discussion with luqui in the backlog, re: class methods are useless 18:44
xinming_Beijing stevan: hmm, could you please give me the blog-url?
stevan geoffb: the way I see it, -Ofun is largely responsible for keeping pugs moving at it's current pace 18:45
autrijus stevan: yes, and I think it is mostly sane.
stevan xinming_Beijing: see the topic
geoffb stevan, DEFINITELY
autrijus stevan: I'm wading thru theory.pod for the _third_ time this week
stevan geoffb: because after all, perl6 isn't *that* cool on it's own
autrijus: yes, I kind of glazed over when he starting bringing theory.pod into it 18:46
xinming_Beijing can only find the paste bot.
autrijus ...well, it used to be *that* cool back in the RFC process...
geoffb huffs at stevan for daring to note that Perl 6 might not be the coolest thing since the sandwich
stevan autrijus: true
autrijus ...but I wasn't there :)
stevan geoffb: yes, but even sandwiches have to be fun :) 18:47
geoffb heh
stevan I mean peanut butter is fine,.. but when you add jelly... then you got a party!
geoffb :-)
.oO( What would the RFC process have been like if autrijus was involved . . . ? )
 18:48
stevan thinks there would have been a lambda keyword for sure
in unicode of course
geoffb And type systems would have been at least 10 RFCs by themselves 18:49
xinming_Beijing capture the output from other program will be the part of perl or the part of lib? 18:56
autrijus luqui: hi.
xinming_Beijing: qx// is part of perl6 as specced
geoffb Not having that in core would be on my list of "no longer Perl" things 18:57
luqui hi autrijus 18:58
right, perl still needs to be -e-friendly
autrijus, do you know about the Rec type?
autrijus luqui: in p6 context? no. 19:04
it's not in any spec iirc
luqui no, not in p6 19:06
in haskell
the catamorphism thingy 19:07
autrijus er, I don't understand 19:08
Rec is part of the various extensible record proposals
luqui data Rec f = In (f (Rec f))
autrijus ah, banana in space 19:09
yes.
luqui raises an eyebrow when he hears about the space banana
autrijus aka Mu.
luqui anyway, I'm trying to define List using Rec 19:10
autrijus luqui: the banana papers are interesting reads
luqui data LF a lf = Nil | Cons a lf
but I can't use cata on it unless it is a functor (in a)
and I can't seem to make it a functor 19:11
and #haskell is no help ;-)
ajs luqui, et al.: Keep in mind that "in core", as Larry has pointed out, is a murky phrase in P6. It might well be in a library or available through a call to a Parrot primative (though I expect that Perl has too many semantic extras for it to be pure Parrot)
luqui ajs, sure 19:12
autrijus luqui: "From Hindley-Milner Types to First-Class Structures"
luqui has been avoiding arguing about "core" for ages now
ajs So, I fell off the planet for a while... has anyone taken over S29, or should I be jumping back in and reviewing the current state of the world?
autrijus ajs: hi!
ajs autrijus: indeed, nice to be "back" 19:13
PerlJam ajs: S29 needs to somehow become "official" IMHO
(no, no one has picked up the S29 ball that I know of) 19:14
autrijus luqui: something like: instance Functor lf => Functor (LF lf) where fmap _ Nil = Nil; fmap f (Cons lf x) = Cons (fmap f lf) (f x) 19:16
ajs Ok, cool. I'll jump back in and see what the state of the world is. When I left it, there was a lot of work to be done turning email, details from other synopses and the like into tect
er text
luqui autrijus, of course
but lf is the *second* argument of LF
autrijus you want type lambda, but you can't get type lambda, because haskell's type expansion is predictive 19:17
luqui and if I make it the first, then Rec (LF a) doesn't work
autrijus so maybe using newtype to flip them?
luqui well, I tried that, but that seemed to run into other problems that I didn't understand 19:18
I'll try it again
autrijus maybe you want Morror and MLF. hm.
Morrow, even.
luqui: hm, have you read the banana papers? 19:19
luqui no
luqui is completely academic-papered-out
geoffb resembles that remark 19:20
autrijus ok. they are: "Bananas, lenses, and barbed wire", "Bananas in Space", "Dealing with Large Bananas", "Revisiting Catamorphisms Over Datatypes with Embedded Functions"
geoffb o/~ One of these titles just doesn't belong . . . 19:21
autrijus geoffb: lol, exactly my thought
luqui: I've read that newtype or a wrapper data type can be made to work, but I'm not exactly the one you want on technicalities 19:22
luqui mmkay
autrijus maybe you want oleg :)
(in ICFP, invited talk, "the next ML", the speaker lists as one of the major direction for ML: "Sexier types but without Oleg") 19:23
geoffb ??
autrijus it's a compliment of sorts
geoffb: okmij.org/ftp/README.html -- oleg is behind some extremely weird type magick in haskell land 19:24
geoffb [added to yet-unread firefox tabs ...] 19:25
luqui got it! 19:26
I was thinking about it backwards
autrijus including that zipper filesystem thing, "true" ocaml style OO system in haskell, logic transformer that adds prolog-like search ability to arbitary monads, types parameterized not with types but with numbers, delimited continutations, and assorted other things.
luqui it actually did have to be a functor in lf :-)
autrijus luqui: ah. :)
...and whatever benign language features introduced, oleg can poke holes in it and abuse it into something truly startling. 19:28
sort of like a cross between damian and abigail.
luqui what is oleg? 19:29
*who?
Steve_p Haskell guy, isn't he?
autrijus luqui: Oleg Kiselyov
Steve_p checks autrijus's journal
Of course, just asking autrijus is faster :) 19:30
autrijus Steve_p: also scheme guy and prolog/c/c++/perl etc guy
sub fixpoint { &{$_[0]}(@_) }# Yep, it does look like @#&$! 19:31
Steve_p feels that pain
autrijus # ref: okmij.org/ftp/Perl/Scheme-in-Perl.txt
PerlJam idly wonders if mjd knows oleg 19:32
geoffb I know Damian's rep. Don't know abigail's. 19:33
Is Oleg a genius or just wierd?
autrijus geoffb: abigail is known for exploiting illdocumented perl5 bugs into japhs. 19:34
geoffb ah!
autrijus er, I mean, features.
geoffb: oleg is a genius who is also just weird.
geoffb heh
Well, I think the world needs more of those
Is he a prof like Damian used to be? Or just has a lot of time on his hands? 19:35
autrijus I think he is working on a day job but apparently does have lots of time on his hands.
geoffb nod 19:36
PerlJam geoffb: peruse perlmonks for posts from Abigail :) 19:39
geoffb PerlJam, oh dear, something else to read. 19:40
geoffb turns to look at his bookshelves
Oy
Someday . . . .
Those that fail to learn history are doomed to repeat it. Those who learn history don't have time to. 19:41
autrijus short lyf; long craft, lerne;
geoffb autrijus, nodnod 19:42
geoffb constantly has problems explaining to people the massive amount of self-guided continuing education necesary to stay vaguely current in this craft
autrijus it's easy to stay current if your river is narrow enough ;) 19:44
PerlJam geoffb: I've decided to mostly content myself with vaguely knowing what's going on by occasional participation.
autrijus unfortunately, PL is a huge river.
geoffb PerlJam, nod. I do that for some things. 19:45
autrijus hm, based on the things luqui are working on, he is probably at the stage of "Post-doc Haskell programmer" now 19:52
soon he'll progress into "Tenured professor"
# ref: www.willamette.edu/~fruehr/haskell/...ution.html
Juerd s/files/questions/ 19:53
(I'm not sending a followup to three lists :)
)
PerlJam Juerd: WRT the pugs repository ... is it automatically updated (i.e., as changes show up on svn.openfoundry.org/pugs, they quickly appear on feather) ? 20:00
autrijus PerlJam: the svk remo in my home dir is synced every 60 sec 20:04
PerlJam I wonder if it would be worth it to get a similar deal for parrot. 20:07
autrijus nod 20:08
and of the main perl6 repo, those are in //mirror/perl6/doc/trunk
if someone wants to set up a parrot mirror alongside it in either //mirror/parrot using robrt's bootstrap, or by hand, please feel free -- my ~/.svk/local is a+wrx 20:09
PerlJam *really* needs to push himself to start using svk more
autrijus finishes reading perl5/* 20:10
...excluding PIL2JS which would take another half day.
...but including theory.pod over many times 20:11
my brain is threatening to overload, so I'm going to take a shower and try to journal down something. 20:12
eric256 any ideas why -n as an ARG doesn't work? 20:43
ie....pugs example.p6 --n works, but pugs example.p6 -n does not 20:44
by works, i mean @ARGS.perl.say outputs --n in the first one, but nothing in the second
autrijus obviously, bug. write a test :) 20:45
geoffb autrijus, is your top journal intentional?
eric256 okay...just checking. ;)
geoffb It looks like a thread reply . . . .
eric256 has doubts about what is obvious and what is not. ;)
if i'm in a for @ARG { is there any way to get the next element in the list ? ... would be very helpfull ;) 20:46
PerlJam eric256: for zip(@ARG,0..) -> $a, $i { ... } 20:47
er, that should be 0...
eric256 zip hangs on infinite lists currently
PerlJam eric256: alternatively, you could use for zip(@ARG,[@ARG[1..@ARG.elems],undef]) -> $curr, $next { ... } 20:48
(assuming I've got the syntax right)
eric256 i've been using for 0 .. +@ARGS - 1 -> $i{ 20:49
to get inxeds
indexs even
PerlJam well, that'll work too
Oh, I guess mine has an obiwan error
eric256 @ARGV Y 0... -> $elem, $index will be a welcome idiom....if someone overloads Y to mean Zip 20:50
;)
why don't we just use regular old Y agian?
fglock__ eric256: "Y" works as zip 20:51
iblech eric256: for @ARG.kv -> $i, $a { my $next = @ARG[$i + 1]; ... } 20:52
eric256 iblech: nice
any ideas where or how i would test that autrijus? 20:53
iblech eric256: Probably t/pugsrun/
Juerd PerlJam: It appears to be something with public write access 20:54
eric256 yea i was looking in there to see..all examples seem related to actual pugs command lines...guess i can create a script that jsut @ARGS.perl.say and compare the output.
Juerd PerlJam: There is no daemon syncing, but if you "svk up" once, the local repository is updated for all users
PerlJam: Perhaps autrijus can provide you with more details
PerlJam keeps forgetting that .kv is blessed upon @arrays
Juerd PerlJam: Since the pugs building script uses the local svk stuff, it is updated at least once every 15 minutes 20:55
iblech PerlJam: It is :) And .pairs .keys .values is as well
autrijus geoffb: no, I think use.perl is confused 20:56
eric256 so at least i could just do @ARGV.keys and get the indexes with no -1 crap
autrijus I blame slash. rm'ing
geoffb: thanks for the reminder btw.
geoffb autrijus, ah. and np.
autrijus it's a reply to luqui's "conflicted" 20:57
svnbot6 r7565 | iblech++ | * t/unspecced/evaluation_order.t: Added some more tests
r7565 | iblech++ | (e.g. (die "first")[die "second"] etc.)
r7565 | iblech++ | * Pugs.Eval: Fixed evaluation order in array and hash access.
r7565 | iblech++ | * PIL2JS: PIL::PApp: Fixed evaluation order of (die "first")(die "second").
eric256 does Test.pm have any builtin handling for running commands and capturing output? 20:59
or should i jsut follow the pugsrun examples?
autrijus eric256: follow pugsrun for now, although I'm surprised there were no qx tests 21:00
so some qx tests would be appreciated too
meppl is there an description to use c++-moduls or is it possible?
autrijus meppl: no-one has attempted it before 21:01
meppl okay
thx
autrijus meppl: if you want to use p6 to script C++, try spidermonkey
and see if they have ways of C++ scripting
so we can use 'pugs -C JS' to manipulate them 21:02
IIRC adobe did exactly that
but not sure if there are code for it
autrijus proceeds to import 3.5G of songs into the shiny iPod Nano thing. 21:03
meppl no, but
thank you very much
;)
autrijus :)
(to qualify: "exactly that" = JS scripting C++ via spidermonkey. not "use p6 to script acrobat") 21:06
although it'd be nice
Juerd does not even have 3 GB of music 21:09
PerlJam Is there an rpm for SVN::Simple::Edit? 21:12
Or anyone have a clue why I'm getting this error on my FC4 box: 21:16
Can't load '/usr/lib/perl5/site_perl/5.8.6/i386-linux-thread-multi/auto/SVN/_Delta/_Delta.so' for module SVN::_Delta: /usr/lib/perl5/site_perl/5.8.6/i386-linux-thread-multi/auto/SVN/_Delta/_Delta.so: undefined symbol: svn_swig_pl_get_current_pool at /usr/lib/perl5/5.8.6/i386-linux-thread-multi/DynaLoader.pm line 230.
(and others like it)
integral is it linked against the right version of the subversion libraries? (eg. did you upgrade svn but not SVN::whatever?)
PerlJam AFAIK, I'm current with all rpms for FC4. 21:18
(not that the subversion bindings to perl can't be broken though)
integral hmm, FC4 doesn't have some bug tracking system you can check for this stuff? 21:19
pasteling "stevan" at 67.186.136.119 pasted "Is this correct usage of but" (18 lines, 320B) at sial.org/pbot/13627
svnbot6 r7566 | iblech++ | * t/syntax/list_quote_whitespace.t: skip test on non-vanilla-Pugs (eval() is needed). 21:21
r7566 | iblech++ | * t/subroutines/loopscope.t: Added two more tests dealing with an explicit $_.
r7566 | iblech++ | * t/subroutines/dollar_underscore.t: Added a comment why PIL2JS fails on this
r7566 | iblech++ | test and why PIL2JS might actually be correct.
autrijus stevan: doesn't the BUILD want "next;" ? 21:27
stevan autrijus: no, the BUILDALL will take care of that
I am trying to make a class which counts it's instances 21:28
the classic class method example 21:29
you can do it easily with a subclass of Class
but I thought maybe it would work with Roles and but as well
nothingmuch: ping 21:30
nothingmuch pong
stevan stupid adium ....
PerlJam stevan: looks okay to me. Except I'm not sure about your use of "our" for $.count
stevan PerlJam: yes, that seems to be confusing
I am of the opinion that anything in a Role is dumped into the consuming class 21:31
nothingmuch PerlJam: i didn;'t agree either
PerlJam stevan: Also, remember that roles are composed into classes, so you may have potential conflicts with $.count, .count(), and .BUILD()
nothingmuch IMHO 'our $.count' is $Countable::count'
and 'has $.count' is assimilated as instance variable on the metaclass (that is, a class scoped variable, unique to each class)
stevan PerlJam: the "but" should avoid that,.. because "but" creates an anon-class first
nothingmuch PerlJam: IIRC roles that don't trust each other will be name mangled 21:32
so that $.count can be defined in any number of roles, and assimilated into the same class - the attr is private to the role
PerlJam stevan: yeah, but it's an anonymous class that's a clone of the original (which may have those members)
nothingmuch or maybe you use '$:count'
anyway, that shouldn't be a problem
PerlJam nothingmuch: oh, perhaps so.
nothingmuch the conflict should only happen with the method count 21:33
stevan PerlJam: I was under the impression is was a anon-class which is a subclass of the original actually
the result would be the same as the cloning
(thats how the eigenclass stuff works)
A but Countable is basically sugar for 21:34
::CountableA = class { does Countable; is A }
at least I think that is right :/ 21:35
so the count() would not conflict so much as it would override anything in A
eric256 autrijus: from back log....what is a qx test regarding pugsrun? 21:36
autrijus eric256: er, no, another test testing the qx// construct 21:37
eric256 ahh.. you had me pretty confused. hehe 21:39
nothingmuch javascript makes webdevelopment almost fun when it's encapsulated in e.g. HTML::Prototype
the results are so much more satisfying
autrijus eric256: sorries :) 21:41
eric256 hmmm -n is the only one that doesn't work. odd 21:42
leo__ stevan: (generated classes) ::CountableA - do you have any clue about the lifetime of such classes? 21:43
autrijus eric256: I suspect -p too
stevan leo__: I would expect that ::CountableA could be stored in the ::* package, which means it would live forever
or at least until the GC took it away
leo__ live foreve simplifies things for Parrot, GCing classes isn't really simple 21:44
stevan leo__: I think only when you do: my class { ... } do they need to be GCed 21:45
but that should be an instance of Class, and so clean-up-able like any other instance in the system
class Foo {} should be sugar for Class.new(:name<Foo>) 21:46
leo__ instances aren't a problem, but a Class get's a type number, registered into an array of type namess
stevan: nm - not so important know, but I've to consider that case eventually - thx for the input 21:48
svnbot6 r7567 | eric256++ | Added tests for passing args to script. Currently any pugs args are being caught even if they happen after the script name and should be applied to the script.
r7568 | eric256++ | Script needed for new Args tests.
PerlJam stevan: wouldn't it be logical for them to have the same lifetime as the variable to which they are "attached"?
stevan leo__: no problem :)
PerlJam: yes, but saying ::CountableA implies stashing in the global ::* package 21:49
eric256 i have no idea how to test qx since it doesn't work yet. ;)...or does it? 21:50
PerlJam wonders if there will be any destruction issues (guaranteed order for instance) with anon classes.
autrijus eric256: it doesn't work yet :)
stevan my ::CountableA = A but Countable;
I dont know if that is real syntax or not 21:51
PerlJam eric256: pretend that it works and write the test that way, but wrap it in an eval.
stevan: it seems reasonable-ish to me. I'd expect A to have :: too though.
eric256 PerlJam...i meant more that i don't know what to expect from it...just plain string that matches STDOUT of the test script? and do you have a suggested t directory?
stevan PerlJam: yes, that is probably correct :)
PerlJam eric256: oh, I'd do something like verify that qx(echo -n foo) is the same as "foo" or something like that. 21:53
I'm still unclear as to when the :: are needed and when they are not.
stevan PerlJam: yes, me too
eric256 PerlJam on win32 that equals -n foo though ;)
PerlJam eric256: well s/echo -n foo/pugs -e 'print foo'/ then :) 21:54
er, pugs -e 'print "foo"'
or whatever the appropriate quoting is for Win32 21:55
eric256 my point was that the quoting and everything would be different on each system.. that was part of my problem...but i can just make a generic script that returns what its called with, but i still have to find the pugs.exe which might or might not be in the path..... hince my original delima ;) 21:56
PerlJam eric256: so make a separate test for each platform. 21:57
integral has anyone else built pugs on feather recently? I've just tried and got a "PIL2JS needs the Class::Rebless module from CPAN", and /usr/local/bin/pugs is a about 200 revisions behind 22:00
autrijus integral: the "Rebless" thing is fixed within the 200 revs 22:02
eric256 integral ... there was a change in the compiling of pugs and i don't think it worked 22:03
wolverian has there been p6l discussion on adding .tail and .head (naming not important)?
eric256 i was actualy just reading Juerd's email on the list and find that my name got attached to that "yea!" lol. hopefully he is around
integral hmm, seems I needed to svk pull, not just svk update, so I'll try that now 22:04
eric256 .tail and .head would be easy enough to do no your own wolverian
wolverian eric256, yes.
eric256 sub head (@array) { return @array[0] }
wolverian eric256, that's not my question, though :)
eric256 lol. my point is they are simple enough to be in a list util library or soemthing 22:05
integral how efficient is tail? could it return something lighter weight than a copy of the array?
eric256 Juerd: ping
wolverian eric256, yes, I know they are. are they confusing enough to not put in the core? 22:06
eric256 ....try that agian. i didn't understand the question 22:07
wolverian eric256, oh: I kind of like the functional approach to manipulating list. thus .tail instead of [1...]. is that too confusing or unnecessary to put in the core? (ah, I said "not" on my previous line. sorry!) 22:09
integral if head works on lists, surely (@array) is the wrong prototype?
wolverian method tail { @_[1...] } 22:10
eric256 i would think then that maybe you want a library of functional helpers... i just don't think everything needs to be in the core. maybe i'm wrong, look at php before moving things to core
yes integral it wasn't a good example and for some reasons lists and arrays are the same to me regardless of how hard i try! lol
wolverian eric256, I don't want _everything_ in the core. :) but thanks, it's clear to me that you don't at least see any use for them in the core. that's good information.
integral here's a data point: GHC's Prelude just reexports stuff from elsewhere, so stuff should go into a module first 22:11
wolverian of course. with "core" I don't mean "don't modularise it!" :)
eric256 actualy i see a use for them..just not in the core. simpler is better. use Functional; at the top of your script would mean tons to someone reading it, gives them a visual que that the rest of your script is going to be more functional in nature. 22:12
wolverian eric256, that's a good point.
eric256 of course that might be a miscue if you just want head and tail. ;)
wolverian we have import lists.. 22:13
eric256 true. /me anxiously awaits wolverian's example using his homebrew head and tail 22:16
wolverian hehe. a friend asked, that's all. :)
eric256 i would at least suggest an example implementation before bringing it up on the list. in developing the example you might become enlightened and have something more to share, if not you will have code to point at and say...this is why i think this way
wolverian I'm not sure how to specify a list invocant 22:17
oh, they turn into Arrays automatically, right? 22:18
eric256 integral implies not. ;)
wolverian right 22:19
but the array must be slurpy
and then it will allow itself to be called as $scalar.last too
Juerd eric256: pong
wolverian method *tail (*@foo) { @foo[1...] }
integral you want: tail(func()) to call func in list context, but @array is an item thing I thought? 22:20
Juerd eric256: pong
uh oh, duplicate pong! :)
wolverian integral, @array is context(...) I think
but that still allows this: "foo".tail 22:21
integral erk, I hate perl6.
wolverian wouldn't it be nice of Strings were Arrays.. :)
integral I suppose the only reason for stuff like is context is to sell P6BP where we find out that we shouldn't use it. :-P </troll> 22:22
wolverian anyway, anyone else have a candidate implementation? :)
eric256 hey juerd 22:26
do you need to linke to my auto directory so that people get the updated version or did you do that all ready? i would check but its mid build 22:27
Juerd eric256: ls -l /usr/local/bin :) 22:35
eric256 huh?
ahh 22:36
hmmm...it says its linked...but it still runs a different version
ahh. nm
okay if 15min or so pugs should be all better on feather 22:37
and now parrot should be working agian too 22:38
integral cool
eric256 well in 15 minutes they will both work agian +-15minutes ;)
integral++ for pointing that out ;) 22:40
integral try pugs on feather now... should work 22:46
integral yep, it does. thanks for that :-) 22:50
eric256 np 22:52
i run my own copy in my home directory so i didn't notice that one was outo f date 22:53
leo__ eric256++ # $ parrot --version ... (r9462) 22:54
eric256 scrolls through TSa's mail list message.....long...very long in a subject i'd already been lost on... 22:56
we need someone to summarize just the Sane Less Insane pair topic...maybe water it down a bit 22:57
leo__ eric256: I think TSa didn't get the subject 'Sane (less insane) pair semantics' 23:00
and the word sane seems to be missing in his dict
wolverian oh. method *tail ( *[ $head, *@tail ] ) { @tail } that oughta work, I think - or does it need 'is context(..)'? 23:01
eric256 did you try it wolverian?
wolverian no. :)
eric256 lol
?eval method *tail ( *[ $head, *@tail ] ) { @tail }; "hello world".tail
evalbot_7568 Error: unexpected "[" expecting parameter name
eric256 ?eval method *tail ( *$head, *@tail ) { @tail }; "hello world".tail
evalbot_7568 []
wolverian that should die, I think 23:02
eric256 i don't think you can define it functionaly like that. maybe you can. i dunno
?eval method *tail ( $head, *@tail ) { @tail }; "hello world".tail
evalbot_7568 Error: No compatible subroutine found: "&tail"
wolverian pugs doesn't support list unpacking apparently (or whatever you want to call it)
eric256 ?eval method *tail ( $head, *@tail ) { @tail }; (1..5).tail
evalbot_7568 Error: No compatible subroutine found: "&tail"
wolverian heh.
eric256 ?eval method *tail ( *@tail ) { @tail[1..] }; (1..5).tail
evalbot_7568 Error: unexpected "[" expecting word character, "::", term postfix, operator, postfix conditional, postfix loop, postfix iteration, ";" or "}"
wolverian @tail[1...] 23:03
eric256 ?eval method *tail ( *@tail ) { @tail[1...] }; (1..5).tail
evalbot_7568 ()
eric256 lol
dunno
wolverian that's.. interesting
eric256 5 oclock..time to bolt 23:04
later
wolverian wow, theory.pod looks nice 23:05
autrijus wolverian: it _is_ nice.
how to reconcile that with dynamism though, is part of what making my brain want to explode. :)
wolverian sure. it also makes me appreciate the separation of type and implementation declarations in Haskell 23:06
(although that's more generic than here) 23:07
23:07 Ikarus is now known as Ikarus[zleep]
wolverian I wonder if I'm the only one who prefers reading POD in vim over perldoc.. 23:08
mugwump I have set my emacs to use a big font for POD headings :) 23:09
wolverian I just like the colours :) 23:12
autrijus wolverian: I usually use pod2html :) 23:22
wolverian autrijus, that doesn't have colour, does it?
autrijus no, but it has variable sized fonts :) 23:24
(and hyperlinks)
stevan: dconway wants to know how "class F { method new { ... } } class G is F {}" works 23:25
dolmen is there a "Best Practices for Darcs" document for the Perl6 project ? 23:26
autrijus dolmen: er, uhm, no, and to be honest, our darcs mirror is just a convenience device for people who does not wish to install svn/svk
and as such provides very little best practices for the mirror -- no changelogs, bad tagging, read onliness 23:27
onlyness
dolmen oh, I thought it was the other way: svk the mirror
clkao grins
autrijus no, svn.openfoundry.org/pugs/ is the main
stevan: and in ruby, class methods like "def Person.blah" does affect the subclass in the usual C3 way. 23:31
er, I mean, regular-dispatch way.
(C3 is python.)
Juerd wolverian: You are not the only one. 23:33
autrijus stevan: I think parallel inheritance in the eClass chain does not look like avoidable 23:35
but maybe I'm missing something. hm.
wolverian Juerd, good to hear that.
svnbot6 r7569 | autrijus++ | * Checking in the "unadapted" version of "Larry was a mariner" 23:48
r7569 | autrijus++ | to facilitate brainstorming and discussions.
autrijus proceeds to send allison a committer bit. 23:49
Juerd autrijus: You mean there are still people without? :)
autrijus Juerd: you may be surprised :)
for example, guido has one
but larry had not
there is something very ironic about this. 23:50
Juerd Is this the Guido I think it might be?
autrijus yup.
geoffb Guido? Really? Has he ever committed?
Juerd Good to have him on board.
autrijus no, it was during ingy's oscon talk
larry invited guido
so I sent invitation to them both
guido took it; larry did not.
hmmmm.
Juerd Why did Larry not take it? 23:51
Did he specify?
autrijus I have no idea. separation of concerns?
he mentioned during hackathon that he wishes to be somewhat distanced from day to day implementations, just so he can judge proposals fairly
Juerd Seems counterproductive to me
autrijus well, I'm not sure about that. 23:52
Juerd I find that people who have actually worked with Perl 6 have a better mental map of what is sane and what is not.
autrijus of @Larry, luqui and chromatic are doing great and shipping things bothways nicely.
Juerd And someone who uses pugs, should practically be a committer, because they will find bugs :) 23:53
Speaking of mental sanity, I think I should stop trying to understand tsa and his posts. 23:54
autrijus mm, "mugs of pugs bugs" 23:55
Juerd He does usually have some good point hidden inside lots of complex misassumptions :)
autrijus ...that tugs hugs and hugs rugs...
Juerd It just takes a lot of effort for me to find them
autrijus er I mean, tugs jugs...
autrijus should clearly finish journaling and get some sleep. 23:56
Juerd wear ear plugs
autrijus mmm hlaghs
Juerd hmmm
pugs drugs. 23:57
*shrugs*
good nugs, eh, night
autrijus gugs nugs
geoffb night, Juerd
wolverian fu.. erm.. 23:58
PerlJam autrijus: would you happen to know why I get an error when trying to svk sync the parrot sourcet tree? I'm doing "svk sync //mirror/parrot -s 9463" and it's saying "HTTP Path Not Found: REPORT request failed on '/parrot/!svn/bc/0/trunk': path not found. 23:59
geoffb starts yet another profiling run at $work, with yet more profiling hooks turned on, hoping that this time he will find the info needed to make the problem clear
PerlJam I'm not sure why it appears to be trying to retrieve what looks like version 0
autrijus PerlJam: you want $svk for that, but I think you shouldn't mirror all revisions in general; Robrt should have provided a repodump
I mean, #svk

Powered by Cro and the Raku® Programming Language.

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

Thank you!