Pugs t-shirts www.cafepress.com/pugscode | Pugs 6.2.9 released | pugscode.org | pugs.kwiki.org | paste: sial.org/pbot/perl6 | www.geeksunite.net
Set by stevan on 15 August 2005.
svnbot6 r6387 | putter++ | pugsbugs/parsing_do_my.t: do{my $a = 3; $a} yields Undeclared variable: "$a" 00:01
fglock_ hi 00:47
Pugs has few people in my timezone span :( 00:51
dudley what timezone? I often feel the same way. (GMT -6) 00:53
fglock_ GMT-3
I have some 'List' code that I think belongs to Prelude 00:54
but I'm not sure I can place it there without breaking something 00:55
dudley Throw it in, there's always "svn revert" :)
bbiab 00:56
fglock_ How can I load Pugs without the precompiled Prelude (I mean, recompiling Prelude)
dudley fglock_: if you turn it off in config.yml, does it recompile? 01:06
I thought I heard someone mention that a while back, but I've never tried it.
fglock_ (trying) 01:07
dudley: it should work, but it requires that Pugs be recompiled 01:15
?eval (1,2,3).ref 01:16
evalbotzy ::Array
fglock_ is there a way to create a 'List'
dudley ?eval ::Array.isa(List) 01:22
evalbotzy bool::true
fglock_ (!) 01:24
?eval my $x=List.new; $x.ref
evalbotzy ::List 01:25
fglock_ he
dudley whoa, pugs just segfaulted
gantrixx so if an child object inherits all methods from the parent, can the methods be overwritten in the child or must they be defined in the parent as submethods? 01:32
fglock_ gantrixx: the child class can override methods 01:34
dudley I thought submethods weren't inherited... 01:44
fglock_ submethods aren't inherited 01:46
dudley Oh, you were saying define them as submethods so they won't be inherited. Nevermind.
fglock_ (or maybe I didn't understand the question, sorry) 01:47
I found that I can redefine 'Lazy' in Prelude. I don't know if this will break 'Array' yet 01:48
dudley Why do that in the Prelude? 01:49
I mean, why do you need to redefine 'Lazy' ?
fglock_ Lazy is not defined yet - it is actually a non-lazy list 01:50
I've written it as a module, but I'm trying to integrate it back in Pugs
dudley Ah. Right. 01:51
fglock_ I can't have Haskell to call Lazy.new without messing with Pugs source, but I can redefine '..' to create a lazy list. 01:53
dudley Why not mess with Pugs source? It just sounds like it would be more at home in the Pugs core than the prelude. 01:57
But then again, I have no idea what I'm talking about ;-) 01:58
fglock_ Prelude is portable - it gets compiled to PIR/PIR2, so it's one less thing to port to JS, P5, etc. 02:02
dudley Good enough for me 02:04
fglock_ native implementations would be more optimized 02:06
cool - (1..Inf).shift works!! 02:12
I'll try to make it in one line for the evalbot 02:13
?eval class Lazy{has Code $.cstart}multi infix:<..> (Int $a,Int|Num $b){my $i=$a;Lazy.new(cstart=>sub{my $tmp=$i;undefine $tmp if $tmp>$b;$i++;$tmp});}multi method shift (Lazy $x:) returns Any {return &{$x.cstart}();}say (1..Inf).shift; 02:28
evalbotzy 1 bool::true
scook0 ?eval my $Foo::3; $Foo::3 = 'hello'; say $Foo::3 02:44
evalbotzy hello bool::true
scook0 ?eval my $Foo::::::bar = 'goodbye'; say $Foo::::::bar 02:48
evalbotzy goodbye bool::true
fglock_ I can't make '@a= **(1..10)' work (flatten lazy list) - I get 'Unknown syntactic construct: Syn "**"' 02:52
it was declared with 'multi prefix:<**> ( Lazy $x ) { $x.flatten }' 02:54
scook0 fglock_: it's parsing as a Syn, rather than as an App...
fglock_ what is a Syn? 02:55
scook0 fglock_: (difficult to explain) 02:56
something that can't be represented by a sub call
the point being that your multi doesn't even get considered
(weird that there's no implementation for it though) 02:57
fglock_ I'll try another name - I can fix it later
scook0 fglock_: try going into Parser.hs, line 1121
, preSyn "* **" 02:58
fglock_ I can't recompile pugs here - too little ram :(
scook0 oh, ok
fglock_ I did it once - took 2 hours 02:59
scook0 I wonder if there's a test for it? (probably not)
fglock_: make unoptimised is faster
I have 256MB, and unoptimised is at least tolerable (though still not fun) 03:00
svnbot6 r6388 | putter++ | JavaScript-FrontEnd/ARuntime.pm: the (very) humble beginnings of a JavaScript-on-p6 runtime. 03:06
scook0 ?eval my $::x = 3; say $::x 03:08
evalbotzy Error: unexpected ":" expecting word character or "::" Empty variable name
dudley what is Syn short for? 03:12
scook0 "Syntactic Construct" 03:13
dudley thanks
03:21 autark is now known as jp-autark
arcady wow, soon we'll be able to run javascript on perl6 on javascript 03:21
fglock_ redefining Lazy breaks Array, because @a=1..10 is the same as @a = $lazy, which creates an array with one element 03:22
scook0 arcady: let me know when Perl 6 runs on an infinite stack of turtles :) 03:23
fglock_ good night - time zone overflow :) 03:30
scook0 ?eval $?PUGS_VERSION 04:00
evalbotzy \'Perl6 User\'s Golfing System, version 6.2.9, August 3, 2005 (r6378)'
svnbot6 r6389 | scook0++ | More Parser refactoring:
r6389 | scook0++ | * Introduced ruleDelimitedIdentifier to factor out common code
r6389 | scook0++ | Also removed prefix-** from the list of Syn parsers, since it
r6389 | scook0++ | was giving fglock trouble, and doesn't seem to be implemented
r6389 | scook0++ | anyway
gantrixx If one object in one file inherits from another object in another file, do I have to do a "use ParentClass" ? 04:16
scook0 gantrixx: I think so 04:18
(otherwise how else would it find it?)
gantrixx yes, it does
ok, I think I know the problem. In order to use attributes from the parent class I have to use the $self.parentattrib correct? 04:23
svnbot6 r6390 | scook0++ | * Updated Parser docs to reflect refactoring
r6390 | scook0++ | * Removed some dead code (killed by refactoring)
gantrixx or is that out of date now?
because it says *** Undeclared variable: "$self"
scook0 well, $self isn't a special variable name
try $?SELF 04:24
gantrixx what really is $?SELF?
scook0 (or explicitly naming your invocant)
inside a method, $?SELF is the object that the method was invoked on
like C++/Java's 'this'
gantrixx so do I also use @?SELF.attrib for an attribute that is an array? 04:25
scook0 no, because that would imply that SELF is an array 04:28
svnbot6 r6391 | putter++ | JavaScript-FrontEnd/ARuntime.pm: a few more bits.
scook0 you would just use $?SELF.attrib
gantrixx so $?SELF is a reference? Just like we would us $self in Perl5? 04:29
except this is built in
scook0 gantrixx: I think so 04:30
you can also use $self, provided you mention it in your method parameter list (like in P5)
method foo($self: $x, $y, $z)
gantrixx ok, why does $SELF have a ? in it (i.e $?SELF ) 04:31
scook0 I don't think people are actually supposed to use $?SELF directly (though they can if they want to)
gantrixx: short answer: because it's special
gantrixx so is $_
scook0 to avoid polluting the normal identifier namespace
gantrixx but it isn't $?_
scook0 hmm 04:32
(maybe $_ is very special) 04:33
either way, having to type $?_ would kind of defeat the point of having $_
gantrixx I think that is one feature of Perl5 that I've heard complaints about. Perl5 is such that the code can become very cryptic. 04:34
scook0 S02 mentions $? variables 04:35
(but curiously, S12 never mentions $?SELF) 04:36
I don't think P6 will make it harder to write cryptic code
but it should make it easier to write non-cryptic code
(provided the person reading the code has actually bothered to learn the language) 04:37
that's probably the main reason people say P5 is "hard to read"
gantrixx P5 can be hard to read, depends on the programmer 04:38
scook0 yeah 04:39
some of that is because of the language, some of it is because of the programmer
P6 is trying to remove the language-induced difficulty
gantrixx two things that can make P5 hard to read: 1) you can use cryptic symbols like $_ and 2) there is more than one way to write the same statement
I had to maintain some code from a guy that wrote things in a very funny way. It always had me refering to the Camel book to figure out exactly what he was doing. It would have been soo much easier if he followed conventions. 04:40
scook0 reading retarded code is never fun
I recently had to help someone try to read some nasty C++ 04:41
gantrixx One collegue was saying that is one of the strengths of Python, you don't have the flexibility to be retarded
scook0 gantrixx: you'd be surprised at what people are capable of 04:42
gantrixx I'm not a Python programmer
doesn't seem to be a big calling for it in the consulting biz, but there are a lot of open source guys using it these days
scook0 but it is true that Python, in general, makes it harder to write unreadable code
(but not impossible)
I mean, if you took the C++ program I was talking about 04:43
and transliterated it into Python
you'd still have no idea
eventually I gave up on even trying to understand the code
and just re-wrote it myself in P5
(my second-ever P5 program!) 04:44
gantrixx In my experience it is more important to be "wordy" and explicit in your code (making it easier to maintain for the sucker who inherits it) than to try to be tricky and compact with everything
I've rewritten code that I couldn't understand before also
scook0 well I didn't so much re-write it as re-implement it from scratch 04:45
putter autrijus: re compiling javascript to javascript... for completeness I'll note that while js on p5, and js on haskell, might actually be useful, the real motivation for starting the js frontend was really to get the first p6 language "pass-thru"? started. ie, js on p6 on PIL2JS. ;)
scook0 the problem with being 'wordy' 04:46
is that sometimes you end up with a lot of code that is easy to read line-by-line 04:47
but just as incomprehensible as a whole
gantrixx well yes, there is such a thing as over-kill
scook0 that's the problem I run into with Python and Java 04:48
the languages prevent you from doing complex things
but the complexity has to go somewhere
so your program ends up much more difficult to follow
putter Java has nifty code policy checkers and transformers. P6 will probably need them too.
gantrixx So far P6 syntax seems pretty nice. I start a new P5 contract on 8/29. It will be a shame to go backwords. 04:50
scook0 the main reason I became interested in P6 was that I was always interested in Perl as a concept
putter Picture the most horrible p5 code you've seen, then add macros, unicode, junctions, etc. It will be quite easy for the tasteless to create something _truely_ unreadable.
scook0 but I had heard some nasty things about it in reality
gantrixx I'm interested in P6 because I'm interested in a long life for Perl. I'd like to see Perl as much of a staple as C/C++ 04:51
scook0 putter: but these features also make it possible to write more readable code, /if/ you're trying in that direction 04:52
gantrixx long life is a big selling point for many clients 04:53
scook0 gantrixx: I think P6 will have a long life -- look how long it's been going for already :)
putter scook0: yes. but p6 does provide better tools for writing incomprehensible code. so a question is, what are the concequences? will CPAN need a stylecheck barrier? will corporations need to see a readable cannonical form to avoid them discarding p6 as unmaintainable? what other issues does it raise...? 04:56
scook0 putter: interesting points... 05:00
I think there was some mention of 'policy modules'
putter that would be neat. 05:01
scook0 so you could ensure your code contains no macros, no junctions and no unicode operators
gantrixx OK, here is a funny one. If I want to set attributes in a child class that it inherited from teh parent class, how do I do that?
scook0 just by saying "use FooCorp::CodingStandard"
putter and rules + PIL will allow some powerful games.
scook0 gantrixx: in a method in the child class? 05:02
I think you just do $?SELF.inherited_attr = val
(I think a lot of the fine details of OO are still up in the air, so some of this is me guessing) 05:03
gantrixx No I mean BlackJack::Card.new( name=>$name, suit=>$suit) but $.name and $.suit come from PlayingCard::Card with BlackJack::Card inherits
scook0 gantrixx: are you trying to set them explicitly or implicitly? 05:04
putter scook0: googling for perl6 policy.module didnt turn up much. any other phrases that come to mind? 05:05
gantrixx I'm trying to set them as you see above 05:06
scook0 putter: there's a mention of "policy" modules at www.perl.com/pub/a/2005/04/p6pdiges...50403.html 05:07
but that may not be what I'm thinking of
(It could be that I misinterpreted the term, and invented the concept myself) 05:08
but either way it's useful functionality
dudley scook0: I definitely remember hearing talk of it before, so it's not entirely in your head :) 05:09
putter scook0: thanks. :) 05:12
putter is reassured that our fearless leaders are sufficiently insane to make p6 very nifty...
dudley doesn't think the insanity of our fearless leaders was ever in doubt... 05:14
putter ;) 05:15
dudley Anybody here know Scala?
putter I skimmed it, but nothing more. 05:16
dudley I'm wondering if there's a way to work around Scala not allowing type synonyms at top level 05:17
putter arg, (non-html language specs)-- sorry, no idea. 05:19
end of day. good night all. & 05:20
dudley good night
gantrixx Is there a way to list all the attributes of an object? 05:21
scook0 gantrixx: you'd need introspection for that 05:24
it probably will be possible
but I doubt it's implemented
gantrixx I'm just thinking for debugging purposes it might be nice 05:25
you know to make sure it inherits what it was supposed to inherit 05:26
dudley For debugging and for IDE/automated refactoring stuff, too
gantrixx vi is my IDE
I'm a hammer and chissle guy 05:27
dudley emacs is mine
scook0 I like IDEs in theory, but in practice they tend to bug me
dudley I'm a thermonuclear jackhammer guy
scook0: Me too. Eclipse is incredibly annoying. 05:28
scook0 I'm content with any editor with a decent GUI and syntax highlighting
emacs is another thing I like in theory, but can't stand in practice
dudley It has so much potential and does so many things well, but the whole package just doesn't quite do it. 05:29
gantrixx gvim does syntax highlighting
scook0 If there was a version of emacs that followed modern GUI conventions, I'd be all over it
gantrixx it would be cool if gvim would implement tabbed editing
dudley emacs is pretty much the least of a near-infinity of evils
scook0 I'm using GEdit at the moment
gantrixx if vi = hammer + chissel, notepad = chopsticks 05:30
dudley I can't use emacs in a GUI. It drives me absolutely nuts.
scook0 gantrixx: notepad = one chopstick
it's 2005, people should at least be using editors that indent new lines for you 05:31
gantrixx I just don't see how notepad is even useful for anything
scook0 well, notepad is literally just a textbox with load/save dialogs 05:32
can't even handle unix newlines :(
gantrixx If you get an "error line 32", try using notepad to find the error, it's difficult 05:33
it doesn't wrap lines
it's a real hinderance
scook0 but it's the only plain-text-editor you get out-of-the-box on windows 05:34
dudley isn't there an 'edit' or something in dos? 05:35
mugwump edlin++
gantrixx when I'm forced to use a windows box, I usually spend a day loading the win32-gnu tools, gvim, virtual desktop manager, firefox, etc 05:36
mugwump I think they dropped it in about DOS 4.0 though ;)
gantrixx oh, and activestate perl5
scook0 'edit' still works on my XP box
in fact, unlike notepad it will automatically keep your indentation level when you go to a new line 05:37
mugwump mmm.... "edlin" was very similar to UNIX "ed", a fine editor when your /usr is offline
scook0 I had to use 'ed' the other day
in a shell script for a uni assignment 05:38
but I certainly wouldn't want to try regular editing with it
gantrixx it is a bit surprising that MS doesn't include some of the free stuff 05:39
what would it hurt if they offered BASH as a command line interpreter? 05:40
I really don't see how that would hurt there business strategy
scook0 speaking of which, I wonder what the 'Monad' shell thingy is like...
gantrixx: probably more of an ideological/cultural thing 05:41
nothingmuch morning 06:19
dudley morning 06:21
nothingmuch Devel::TypeCheck is out
or at least should be out soon 06:22
search.cpan.org/~bargle/
xinming Is there a method for Array which will help Join the array into string? 06:59
oops. just now, I asked this In #perl, And someone is sticking on my question with perl 5 grammer. :-)
I know as, .as('%03d', ', ') which will mean if you get the element of from a array "slot", it will be formated as `$(%03d),` But Do I have to join them myself? 07:00
* DeathWo1f has quit (Remote closed the connection)
arcady there is "join", isn't there?
xinming arcady: yes, What I want to know is, if there a "built-in" join for Array. 07:01
arcady what do you mean?
@a.map { .as("%03d") }.join(", ") or something like that 07:02
xinming arcady: ...
arcady: I mean, It will hold a string "internally"... 07:03
arcady okay, I'm not getting what you mean by that 07:04
xinming arcady: I can't turning the way of thinking sometimes. :-( 07:05
s/turning/turn/ 07:06
arcady so wait, what exactly needs to be holding a string internally? 07:07
scook0 what about [~] ? 07:08
arcady oh yes, that works 07:09
is there a >>.<< ?
scook0 arcady: huh? 07:10
xinming arcady: some time, If you don't wish to declare a scalar to hold the "slurped" array yourself.
This is the problem I ever encountered... hmm, But, From my poor Programming experience. This might not be necessary really... :-) 07:11
arcady scook0: apply a method to each element of an array 07:14
scook0 arcady: there's @foo Ā».bar
(don't think its implemented yet)
which will call .bar on each elem of @foo
xinming my @ary; @ary.as("abc: %s"); my $slurped := @ary.scalar; And then, Every time you push a element into the array, you won't need to join yourself... and also, Every time you modify the one of the elemts in the array, The $slurped varaible will get the new result. 07:15
arcady that's... way too much magic that is happening there, I think
scook0 I would recommend something like: 07:17
arcady scook0 I think they've decided on not having >>(op) or (op)<< and just having >>(op)<<
otherwise it's just unparseable
scook0 arcady: really? when was this?
xinming: instead of having a scalar, you could have a sub 07:18
my &slurped = { @ary.scalar } 07:19
(or whatever)
then when you call slurped()
it has the most up-to-date value
arcady ?eval (1,2,3)>>+<<2
evalbotzy (3, 4, 5)
arcady ?eval (1,2,3)>>+2
evalbotzy Error: unexpected ">" expecting term postfix, operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
xinming ?eval (1, 2, 3) >>+ 2
evalbotzy Error: unexpected ">" expecting operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
arcady it may not have been decided but it is the current state of things 07:20
scook0 ?eval -Ā« [1, 2, 3]
xinming ?eval (1, 2, 3) <<+ 2
evalbotzy -3
Error: unexpected "<" expecting operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
arcady ?eval -<< (1,2,3)
evalbotzy (-1, -2, -3)
scook0 ?eval (-Ā« [1, 2, 3]).say
evalbotzy -3 bool::true
arcady whoa
scook0 oh, oops
arcady ?eval 2 +<< (1,2,3)
evalbotzy Error: unexpected "<" or "+" expecting operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
arcady oh 07:21
nothingmuch morning
arcady that makes sense
scook0 opĀ« is for prefix ops
arcady morning
yes
scook0 Ā»op is for postfix
Ā»opĀ« is for infix
arcady so which one is .?
scook0 arcady: . is postfix 07:22
or postcircumfix
(can't remember)
arcady no, definitely not postcircumfix
that's like []
xinming ?eval my $lval; sub get { return $lval }; sub set () is rw { $lval }; set = 100; my $t = get; $t.say;
evalbotzy 100 bool::true
xinming ?eval my $lval; sub get { return $lval }; sub set () is rw { $lval }; set = 300; my $t = get; $t.say;
evalbotzy 300 bool::true
scook0 I don't think the . itself is considered an operator 07:23
I believe the whole .meth(args) is considered postfix
nothingmuch it isn't
arcady oh
nothingmuch scook0: you're right
nothingmuch forgot the dvi-vga converter at home =( 07:24
no second monitor today
scook0 nothingmuch: awww
nothingmuch yeah
scook0 dual-monitors is great fun
nothingmuch is very very spoiled with his montiro
monitors
xinming so, anyone here can give ma example to compose these two functions into one?
nothingmuch a 17" powerbook + 17" samsung LCD 07:25
xinming and, also, without declaring a my $lval outside.
nothingmuch the samsung is not the best monitor in the world, but it's pretty good
scook0 xinming: you mean 'as' and 'join'?
nothingmuch xinming: 'state $lval;'
xinming scook0: No...
nothingmuch but that'll only work for one sub
xinming ?eval my $lval; sub get { return $lval }; sub set () is rw { $lval }; set = 300; my $t = get; $t.say;
evalbotzy 300 bool::true
scook0 oh, sorry, I didn't see your evalbot stuff
nothingmuch ?eval sub value is rw { state $lvalue; $lvalue }; value = 100; my $v = value; $v.say; 07:26
evalbotzy Error: Can't use readline() or getc() in safemode.
nothingmuch ?eval sub value is rw { state $lvalue; $lvalue }; value = 100; my $v = value; $v
evalbotzy Error: Can't use readline() or getc() in safemode.
nothingmuch huh?
?eval my $lvalue; sub value is rw {$lvalue }; value = 100; my $v = value; $v
evalbotzy Error: Can't use readline() or getc() in safemode.
nothingmuch wtf am I doing wrong?
xinming what does the keyword state do? It seems it doesn't appear in synopsis.
nothingmuch xinming: state is like static in C
scook0 it keeps its value between sub calls 07:27
nothingmuch where in perl 5 you said 'my $var; sub routine { $var ||= $initial_value; ... }' you can now say 'sub routine { state $var = $initial_value }
xinming nothingmuch: If so, The space of the var for state will never be freed, right?
nothingmuch hmm 07:29
maybe in a closure, once the closure is freed so is the value
xinming In this case, I think the sub should be declared with my or our...
or my 07:30
nothingmuch yeah, if you say 'my sub routine { state $x = some_func() }' then $x should be reset at scope exit, i would guess
xinming nothingmuch: But this hasn't been supported in pugs. :-) 07:32
nothingmuch i don't think pugs has reached this level of detail yet
xinming nothingmuch: What's null key really in perl 6? 07:34
nothingmuch null key?
xinming nothingmuch: ( => 'value' )
nothingmuch that's a syntax error, i think 07:35
xinming And how to I access it? or in perl 6, I won't be parsed.
nothingmuch you can say undef => 'value'
then it's a pair with the left side being undef
if your hash's shape is told to store more than just strings, than you can say my %hash = (undef => 'value');
and then just %hash{undef};
xinming ?eval ( undef => 'value' );
evalbotzy ('undef' => 'value')
nothingmuch oh wait, that autoquotes 07:36
xinming ?eval ( => 'value' );
evalbotzy Error: unexpected ">" expecting term
xinming ?eval { undef => 'value' );
evalbotzy Error: unexpected ")" expecting operator, term postfix, postfix conditional, postfix loop, postfix iteration, ";" or "}"
xinming ?eval { undef => 'value' };
evalbotzy {('undef' => 'value')}
nothingmuch ?eval Pair.new(undef, 'value')
evalbotzy Error: Must only use named arguments to new() constructor
nothingmuch damnit, now to find Pair.new
scook0 ?eval (undef) => 'value'
evalbotzy (undef => 'value')
scook0 woo!
nothingmuch scook0++ 07:37
scook0 you just have to circumvent the parser's cleverness
in fact, you'd have to do that for any bareword-function-call 07:38
xinming ?eval my %h = { (undef) => 'value' }; my $b := %h{undef}; $b = "test"; %h.kv;
evalbotzy ('', 'test')
scook0 ?eval sub bob {3}; bob => 'value'
evalbotzy ('bob' => 'value')
xinming Is an empty string undef? 07:39
scook0 ?eval sub bob {3}; (bob) => 'value'
evalbotzy (3 => 'value')
scook0 ?eval ~(undef)
evalbotzy ''
xinming or, is (undef) an empty string
scook0 when you stringify (undef), it turns into ''.
?eval defined('') 07:40
evalbotzy bool::true
scook0 ?eval defined(undef)
evalbotzy bool::false
xinming scook0: How to I preserve the undef ?
scook0: the example I used above
scook0 xinming: you would need a hash that can accept keys that are not strings 07:41
at the moment, it automatically turns all key values into strings (I think)
no easy way to preserve the undef at the moment
xinming scook0: So, the hash in perl, will hash key not accept the non-string object by default? 07:42
scook0 xinming: I think that is the default
you will be able to specify hashes that can have other values
but it's not implemented
xinming scook0: hmm, I know, we can use my %hash is MyHash syntax. 07:43
scook0 xinming: you can also use my %hash is shape(Any) -- see S09
now that I think about it, I'm not sure whether or not hashes should accept undef... 07:44
but that would be getting into not-specced-yet territory anyway (re: typed undefs) 07:45
xinming ?eval "my $a; $a = 5; $a.say".eval; 07:54
evalbotzy Error: Undeclared variable: "$a"
xinming will this be valid in perl 6?
nothingmuch xinming: use '' 07:55
?eval 'my $a; $a = 5; $a.say'.eval
evalbotzy 5 bool::true
xinming ...
forgot the qq and the q. ;-)
nothingmuch what you wrote is like ("my " ~ $a ~ "; " ~ $a ~ " = 5; " ~ $a.say).eval 07:56
xinming nothingmuch: yes, I know, I just forgot to use '' instead of ""
nothingmuch wants lunch to arrive 07:57
xinming wants money to fall from the sky. 08:01
webmind that'll be bad for inflation 08:02
:)
xinming knows the it's bad for inflation, so xinming might want beauty instead... 08:04
s/the// 08:05
hmm, May I ask a non-perl6-related question here? 08:13
autrijus you just asked one 08:15
and the answer is yes
xinming blogs.msdn.com/danlehen/archive/200...46259.aspx 08:16
and search for the string
That is truly one of the most amazing things
in the comment.
search "That is truly one of the most", I wonder if there are really funny comments written in English there. :-) 08:17
autrijus it's mildly funny yes :)
xinming autrijus: You can understand Chinese. and There are many such comments written in Chinese. So, I wonder If there is such comment which written in English. Except this one. 08:18
大åøˆēš„čƝēœŸå¦‚ā€œå¤§éŸ³åøŒå£°ę‰«é˜“ēæ³ā€ļ¼ŒēŠ¹å¦‚ā€ę‹Ø开äŗ‘é›¾č§é’å¤©ā€ļ¼Œä½æꈑē­‰ē½‘ę°‘ēœ‹åˆ°äŗ†åøŒęœ›ļ¼Œēœ‹åˆ°äŗ†ęœŖę„ļ¼ę™“天霹雳ļ¼Œé†é†ēŒé”¶ęˆ–č®øäøč¶³ä»„形容大åøˆę–‡ē« ēš„äø‡äø€ļ¼›å·«å±±č”Œäŗ‘ļ¼Œé•æę±Ÿęµę°“ę›“éš¾ä»„ęÆ”ę‹Ÿå¤§åøˆēš„ę–‡ę‰ļ¼é»„钟大吕ļ¼ŒęŒÆč‹å‘č©ļ¼ä½ ēƒ›ē…§å¤©äø‹ļ¼Œę˜Žč§äø‡é‡Œļ¼›é›Øéœ²č‹ē”Ÿļ¼Œę³½č¢«äø‡ę–¹ļ¼é€čæ‡ä½ ę·±é‚ƒēš„ę–‡å­—ļ¼Œęˆ‘ä»æ佛ēœ‹åˆ°äŗ†ä½ é¹°č§†ē‹¼é”¾ļ¼Œé¾™č”Œč™Žę­„ēš„伟å²ø英å§æļ¼›ä»æ佛 08:20
ēœ‹åˆ°äŗ†ä½ ę‰‹ę‰§å¦‚ę¤½å¤§ē¬”ļ¼Œå†™å¤©äø‹ę–‡ē« ēš„ę™ŗꅧē„žę€ļ¼›ä»æ佛ēœ‹č§äŗ†ä½ ęŒ‰å‰‘四锾ļ¼Œę±Ÿå±±ę— ę•°ēš„英ꭦ갔ꦂļ¼
autrijus xinming: yow. 08:22
autrijus rushes to /clear
xinming That's just copy and past...
...
autrijus xinming: I've read enough text that praises Chiang Kai Shek and Mao... please spare me further things like this :) 08:23
xinming autrijus: hmm, I think I would prefer seeing the finish of pugs rather than read these articles. :-) 08:24
autrijus riiight. 08:25
autrijus goes back to hack
Supaplex feeds xinming /dev/urandom until his $TERM goes *poof* 08:48
xinming Supaplex: Hmmm. Sorry... 08:49
Supaplex heh, tiz ok.
I use to enjoy meannacing peeps IRC sessions with chunks of /dev/urandom or beeps (^G aka ) hehe. 08:50
revdiablo heh, actually xinming's paste showed up fine in my client+terminal. looks kinda neat.
I even have a font with all the glyphs 08:51
xinming Supaplex: Do you wish to hear the sound of the kernel? :-)
Supaplex: cat /proc/kcore > /dev/dsp
Supaplex haha
xinming ?eval class C { has $.a = 1; }; 08:52
evalbotzy Error: unexpected "=" expecting trait, "handles", ";" or "}"
Supaplex I'm enjoying knobtweakers.net tracks too much to listen to the kernel atm.
xinming Can pugs use the all the perl 5 libs for now? 08:53
maybe most... Infact.
autrijus xinming: only if you link against perl5 when building pugs, and there are known problems with certain XS modules like DBD::Pg. 08:56
Supaplex what a shame 08:58
autrijus indeed; which is why we are working toward compiling to perl5 code. 09:00
then you can work with any perl5 code.
xinming running perl 5 code is "a bit" too slow... 09:06
autrijus xinming: for what purposes?
xinming autrijus: hmm, I just now try to use "use CGI", And It seems it will slowly finsihed... :-) 09:07
autrijus: well, I know it's not perl 6 really, I just do some practising. :-) 09:08
autrijus xinming: oh... but it's perl6's CGI
if you want to use perl5 CGI, you need to say
use perl5:CGI;
xinming In fact, use CGI works also. at least, run at the same speed. 09:09
autrijus: by the way, Is there any difference between fun_call.say and fun_call().say ?
?eval f { 'test' }; f().say; 09:10
autrijus xinming: if &fun_call is in scope
evalbotzy Error: No compatible subroutine found: "&f"
autrijus then no, no difference.
xinming ?eval sub f { 'test' }; f.say;
evalbotzy 'test'
xinming ?eval sub f { 'test' }; f().say;
evalbotzy test bool::true
xinming autrijus: But in fact, here we got the different result. 09:11
autrijus wow... a bug.
write tests :)
f.say is somehow parsed as f(say()).
xinming autrijus: :-) Well, I will record all the bugs I now encountered. And will write the test after I "finished" learning perl 6. 09:13
scook0 (wow, the parser is just chock full of redundancy...) 09:18
xinming In perl 5, If you say "use Data::Dumper", But In perl 6, it doesn't works, so what the differences between "use Data::Dumper" and "use Data:Dumper" are? 09:19
scook0: well, that's better than nothing. :-)
scook0: for me, At least I can learn how to use perl 6 programming in "real" world.
scook0 xinming: ooh, that's a good one :) 09:20
and I'm working on getting rid of some of it
xinming hmm, Ok, Just go ahead, I will try to figure out the difference between Data::Dumper and Data:Dumper myself. :-) 09:21
masak how come practically all of my tests fail? 09:23
xinming ?eval my $var =3; $var *= (5, 9); $var.say; 09:28
evalbotzy 6 bool::true
xinming hmm, Why It is 6 please?
revdiablo ?eval +(5,9) 09:29
evalbotzy 2
integral is it the , or the () that is the list constructor? 09:30
scook0 integral: , is the list constructor
integral ok
scook0 but because of precedence, most of the time you need () also
hence the confusion
xinming ?eval my $var =5; $var *= (5, 9); $var.say;
evalbotzy 10 bool::true
integral I thought the precedence had been "fixed" in perl6?
ie , is now above = ?
scook0 integral: some people want that 09:31
but last I heard it was still the same
integral oh, ok. I thought it was a done deal ages ago
scook0 it really should be changed though
iirc, the only reason to stay the same was so you can do parallel processing in C-style loops 09:32
(which is a pretty bogus reason, if you ask me)
integral which are now decidedly not C-style now ;)
xinming ?eval my $var; say $var; 09:34
evalbotzy bool::true
integral xinming: what did you mean when you said "use Data::Dumper" didn't work in perl6? 09:35
xinming integral: Sorry, I was wrong. :-) 09:40
integral oh
xinming integral: But Is there any difference between using : and ::?
integral :: separates namespaces, but : separates attributes 09:41
xinming hmm, who update the CGI.pm in blib6/lib/ by the way? 09:55
Just try 'use CGI;'
svnbot6 r6392 | scook0++ | * Merge some more parser redundancy 10:13
r6392 | scook0++ | * Oops, ruleDelimitedIdentifier shouldn't use lexeme identifiers
r6392 | scook0++ | * ruleDelimitedIdentifier now matches leading delimiters
10:39 autark is now known as jp-autark
xinming ?eval NaN 10:41
evalbotzy NaN
xinming ?eval +NaN
evalbotzy NaN
xinming ?eval "test'~NaN
evalbotzy Error: unexpected "t" expecting block, "\\", "$!", "$/", "$" or "\""
xinming ?eval ~NaN
evalbotzy 'NaN' 10:42
xinming ?eval 'test' ~NaN
evalbotzy 'testNaN'
coral ?eval int(NaN) 10:47
evalbotzy -269653970229347386159395778618353710042696546841345985910145121736599013708251444699062715983611304031680170819807090036488184653221624933739271145959211186566651840137298227914453329401869141179179624428127508653257226023513694322210869665811240855745025766026879447359920868907719574457253034494436336205824
coral it's negative now?
oh, that was Inf. n/m
arcady ?eval int(Inf) 10:53
evalbotzy 179769313486231590772930519078902473361797697894230657273430081157732675805500963132708477322407536021120113879871393357658789768814416622492847430639474124377767893424865485276302219601246094119453082952085005768838150682342462881473913110540827237163350510684586298239947245938479716304835356329624224137216
xinming I think, Perl will be a language which will stop people from going deeper... :-S 11:01
anyone here would explain this for me? `'test' ~NaN` `"test'~NaN` 11:03
scook0 ?eval "test" ~ NaN 11:08
evalbotzy 'testNaN'
scook0 ?eval 1/0 11:09
evalbotzy Error: Illegal division by zero
xinming in fact, ~ should be a infix:<~> But, with default "infix" operator, there should be no space before or after the operator. 11:10
user defined infix:<~> must have space...
don't know if I am right. 11:13
scook0 ?eval "test"~NaN
evalbotzy 'testNaN'
scook0 (in your example, you mismatched " and ') 11:14
xinming ... 11:16
@_@
carelessness is a bad thing really. :-)
?eval my $a = 1 but false; $a.bit; 11:18
evalbotzy Error: No compatible subroutine found: "&false"
xinming ?eval my $a = 1 but not true; $a.bit; 11:19
evalbotzy Error: unexpected "t" expecting block construct, ":", term, term postfix, operator, "does", "but", "is", "^..^", "cmp", "<=>", "^..", "..^", "..", ";" or end of input
scook0 actually, if there's confusion between infix and postfix operators, the infix must have a space, and the postfix must not
(but I don't think any built-in operators will be ambiguous that way)
xinming ?eval my $a = 0 but true; $a.bit; 11:20
evalbotzy Error: unexpected ";" expecting term
xinming ?eval my $a = 0 but true ; $a.bit;
evalbotzy Error: unexpected ";" expecting term
xinming scook0: 1+5 here, IMHO this should be clear with "user defined" operator. just like, 1 + 5 will use the user defined operator. 11:22
and 1+5 will call the "perl 6" operator
scook0 no, I don't think so
what I was talking about is if you have: 11:23
infix:<++> and postfix:<++>
then you might have to write
$a++
or
$b ++ $c 11:24
but not '$a ++' or '$b++$c'
(I think that's right...)
xinming hmm, then, How to call the "default" postfix:<++>
scook0 xinming: not sure what you mean by "default" 11:25
are you talking about user-defined vs. built-in operators? 11:26
xinming scook0: yes, built-in operator.
scook0 hmm
normally I don't think you would replace the built-in ones 11:27
you would just add new versions for your own classes/types 11:28
(though you probably could change the built-in ones if you REALLY wanted to)
xinming scook0: hmm, by the way, Do you mean the operator can be... infix:<MINE_+> ? 11:29
or infix:<mine->
scook0 xinming: you want to be able to write `3 MINE_+ 4`? 11:30
or are you talking about `MyClass.new + MyClass.new` 11:31
xinming scook0: the first example.
if It's able to apend some letters to define a operator.
scook0 you can define new operators using whatever symbols you want 11:33
even unicode
xinming ?eval true;
evalbotzy Error: unexpected ";" expecting term
xinming scook0: thanks.
?eval true();
evalbotzy Error: No compatible subroutine found: "&true"
xinming ?eval true(1);
evalbotzy bool::true
scook0 ?eval sub infix<ā—Ž>($a, $b) { $a + $b + 1 }; 1 ā—Ž 1 11:34
evalbotzy Error: unexpected "<" expecting word character, "::", bare trait, subroutine parameters, trait or block
scook0 ?eval sub infix:<ā—Ž>($a, $b) { $a + $b + 1 }; 1 ā—Ž 1
evalbotzy 3
xinming :-) 11:35
?eval sub infix:<puls>($a, $b) { $a + $b + 1 }; 1 puls 1
evalbotzy 3
xinming ?eval sub infix:<puls>($a, $b) { $a + $b + 1 }; 1 puls 5 11:36
evalbotzy 7
xinming amazing. :-)
?eval sub func; 11:44
evalbotzy Error: unexpected ";" expecting word character, "::", bare trait, subroutine parameters, trait or block
xinming ?eval sub func{ };
evalbotzy undef
xinming ?eval sub func{ ... };
evalbotzy undef
xinming ?eval sub duplicate($n, *%flag, *@data) {...} 11:54
evalbotzy undef
xinming evalbotzy: sub duplicate($n, *%flag, *@data) {...}
evalbotzy: ?eval sub duplicate($n, *%flag, *@data) {...} 11:55
?eval -> $a { ... }; 11:57
evalbotzy sub {...}
xinming If I want to use role and grammar things, Do I have to use PUGS_EMBED="perl5 parrot" 12:04
mugwump you don't need either; parrot can be in your PATH 12:05
it starts a parrot subprocess and talks to it via pipes...
xinming mugwump: hmm, I mean, Do I have to compile parrot?
mugwump If you want to use Perl 6 rules, and not just PCRE, yes. 12:06
xinming mugwump: thanks
?eval sub fun will do { 'will'.say } { 'main'.say }; fun(); 12:07
evalbotzy Error: unexpected "w" expecting bare trait, subroutine parameters, trait or block
xinming ?eval sub function { caller; }; function(); 12:10
evalbotzy Control::Caller.new(('file' => '<eval>'), ('line' => 1), ('package' => 'main'), ('sub' => undef), ('subname' => undef), ('subtype' => undef));
xinming ?eval sub function { caller Sub; }; function(); 12:11
evalbotzy undef
iblech nothingmuch: You have mail :) 12:12
xinming hmm...
nothingmuch iblech: checking
iblech ?eval sub sub1 { caller Sub }; sub sub2 { sub1() }; sub2()
evalbotzy Control::Caller.new(('file' => '<eval>'), ('line' => 1), ('package' => 'main'), ('sub' => sub {...}), ('subname' => '&main::sub2'), ('subtype' => 'SubRoutine'));
xinming why it doesn't return function?
iblech Because it returns the *calling* functions
s/s$//
But there is no calling function, only the calling toplevel
xinming oh,ok 12:13
?eval my @x=[1,2,3;4,5,6;7,8,9]; @x[0..10; 1,0; 1...:by(2)] 12:22
evalbotzy Error: unexpected "[" expecting expression
xinming ?eval my @x=[1,2,3;4,5,6;7,8,9]; @x[:by(2)]
evalbotzy Error: unexpected "[" expecting expression
xinming hmm, what does :by(2) mean ? 12:23
QtPlatypus ?eval 1..10:by(2)
evalbotzy Error: unexpected ":" expecting digit, "_", fraction, exponent, term postfix, operator, "does", "but", "is", "^..^", "cmp", "<=>", "^..", "..^", "..", postfix conditional, postfix loop, postfix iteration, ";" or end of input
QtPlatypus Its supposed to mean that the range goes up by steps of 2
xinming QtPlatypus: So, ( 1..10 ) will become ( 1..20 )? 12:24
scook0 xinming: no, I think it's (1, 3, 5, 7, 9) 12:25
mugwump ?eval &infix:<..>(1, 10, :by(2))
evalbotzy Error: No compatible subroutine found: "&infix:.."
mugwump ?eval &Num.infix:<..>(1, 10, :by(2))
evalbotzy Error: Undeclared variable: "&Num"
mugwump ?eval &Num::infix:<..>(1, 10, :by(2)) 12:26
evalbotzy Error: unexpected ":" expecting "::", term postfix, operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
xinming is it equal @ary >>+<< 2 ?
mugwump alright, I give up on that approach :)
scook0 xinming: no, it means "count up in steps of 2" 12:27
(not that it's implemented)
mugwump ?eval (1..10).grep:{ $_ ^& 1 } 12:28
evalbotzy Error: unexpected ":" expecting term postfix, operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
xinming Oh, Ok, so, (1, 2, 3, 4, 5, 6, 7):by(2) will get a new @ary = (1, 3, 5, 7) 12:30
scook0 hmm
mugwump ?eval (1..7.5)
evalbotzy (1, 2, 3, 4, 5, 6, 7, 8)
scook0 I don't now how clearly :by() has been specced at this point 12:31
mugwump ?eval (1,3..10)
evalbotzy (1, 3, 4, 5, 6, 7, 8, 9, 10)
mugwump ?eval ( (1,3)..10)
evalbotzy (2, 3, 4, 5, 6, 7, 8, 9, 10)
xinming ?eval (1,3..10):by(2)
evalbotzy Error: unexpected ":" expecting term postfix, operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
scook0 it might even have been ditched...can't remember
xinming ?eval ( (1,5)..10) 12:32
evalbotzy (2, 3, 4, 5, 6, 7, 8, 9, 10)
iblech (1,5) is numified first, and, as it contains two elems, it numifies to 2 12:33
Compare this to:
?eval (1|5)..10
evalbotzy (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
iblech ?eval ((1|5)..10).perl
evalbotzy '((1, 2, 3, 4, 5, 6, 7, 8, 9, 10) | (5, 6, 7, 8, 9, 10))'
xinming iblech: I know what junctions do. :-) 12:34
?eval ( all(1, 5).. 10 )
evalbotzy (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
iblech xinming: Cool :)
xinming ?eval ( any(1, 5)..10 )
evalbotzy (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
iblech You need the (...).perl in evalbot
xinming ?eval ( any(1, 5)..10 ).perl
evalbotzy '((1, 2, 3, 4, 5, 6, 7, 8, 9, 10) | (5, 6, 7, 8, 9, 10))'
xinming ?eval ( all(1, 5)..10 ).perl
evalbotzy '((1, 2, 3, 4, 5, 6, 7, 8, 9, 10) & (5, 6, 7, 8, 9, 10))' 12:35
mugwump ?eval sub foo { }; &foo.perl
evalbotzy '\\sub {...}'
mugwump ?eval sub foo { }; &infix:<..>.perl
evalbotzy '\\sub {...}'
mugwump ?eval sub foo { }; &infix:<..x>.perl
evalbotzy Error: Undeclared variable: "&infix:..x"
mugwump ?eval sub foo { }; &infix:<..>(1,5)
evalbotzy (1, 2, 3, 4, 5)
mugwump ?eval sub foo { }; &infix:<..>(1,5,:by(2))
xinming by the way, Is there any difference between sub f { } and sub f { ... }
evalbotzy Error: No compatible subroutine found: "&infix:.."
xinming hmm, maybe the yada operator. 12:36
scook0 xinming: sub f {} is just a normal sub that will do nothing
whereas sub f {...} will fail when you try to call it
xinming scook0: It's kind of "viusal" thing to user. :-) 12:37
scook0 ?eval sub f {}; f
evalbotzy undef
scook0 ?eval sub f {...}; f
evalbotzy *** ... - not yet implemented at <eval> line 1, column 8-11
iblech [PIL2JS] coro works now :) 12:38
scook0 iblech++
xinming ?eval sub f { }; sub f { }; 12:39
evalbotzy undef
xinming ?eval my $a; my $a;
evalbotzy undef
scook0 xinming: that shouldn't be allowed
xinming these 2 examples should all generate a error.
scook0: I know. ;-)
scook0 that's how you know pugs is incomplete 12:40
it doesn't have all of Perl 6's errors :)
xinming scook0: hmm, well, In fact, from these, We might write a more strict test to test perl 6 parser.
nothingmuch iblech++; # good ideas
sorry it took me so long
iblech scook0, nothingmuch: Thanks :) 12:41
nothingmuch: Sure, np
scook0 iblech: wrt. `use jsan:`
do you think it makes more sense to say:
`use jsan:Foo::Bar`
or
`use jsan:Foo.Bar`
iblech Foo.Bar 12:42
Because that's the official name of the modules
scook0 I guess
It just makes things more difficult :)
(only a little though) 12:43
iblech ah, yes
scook0 though it also makes sense to have the module-name-format dictated by the `lang:` prefix 12:44
I'll leave it as it is for now
it's hard to tell which parts of the pugs/perl5/jsan paths to merge together 12:46
and which to keep separate
iblech right...
scook0 (ooh, just realised that my code does some odd things) 12:48
nothingmuch i disagree 12:55
after you 'use jsan:Foo.Bar'
what do you get?
Foo.Bar.new?
or Foo::Bar.new? 12:56
scook0 nothingmuch: that was the other reason I thought of
iblech Hm
nothingmuch we can't support 'Foo.Bar.new' because some day a language which hassyntax errors for namespace separators will come along
IMHO we need to map the foreign namespace into the local one
iblech Currently, exported namespaces aren't autoboxed, so neither will work
nothingmuch prefixed by the domain
unless unambigouous 12:57
so Foo::Bar is OK
and jsan:Foo::Bar is also ok, if there is more than one Foo::Bar involved in the process
scook0 btw, can you say `use ::Foo`?
(I imagine you can)
nothingmuch scook0: for a relative name?
iblech And therefore, currently, you've to write $*JS.Foo.Bar.new
scook0: That should work, I think, and mean the same as you'd have dropped the ::s 12:58
scook0 nothingmuch: I'm not really fussed why
my only issue is `use pugs:::Foo`
maybe we say that you get leading :: only if you /don't/ use a namespace prefix 12:59
in any case, `use jsan:.Foo` is clearly bogus 13:00
which is what I'm about to fix
svnbot6 r6393 | scook0++ | * Disallow leading delimiters in ruleDelimitedIdentifier 13:05
scook0 eek, it's late -- bye folks 13:06
svnbot6 r6394 | iblech++ | PIL2JS: Coroutines work :) 13:10
r6394 | iblech++ | Unfortunately, Pugs and I have different opinions on how coroutines should
r6394 | iblech++ | work ;) I think a return() should reset the entrypoint, i.e.
r6394 | iblech++ | my $foo = coro { yield 42; return 23; die };
r6394 | iblech++ | say $foo(), $foo(), $foo(), $foo(); # 42 23 42 23
r6394 | iblech++ | Pugs' return does not reset the entrypoint. And another related problem:
r6394 | iblech++ | my $foo = coro { yield 42 }; # is really (in PIL2JS)
r6394 | iblech++ | my $foo = coro { return(yield 42) };
r6394 | iblech++ | say $foo(), $foo(), $foo(), $foo(); # 42 undef 42 undef
autrijus iblech: your semantics is correct. however, fall-off yield should not be put into return() 13:20
just as
sub { return 3 }
is not
sub { return (return 3) }
iblech Ah, ok 13:21
autrijus but I do agree explicit return should reset entry. does Coro.pm do that same?
iblech dunno 13:22
I look through the old p6l posts WRT coroutines currently 13:23
autrijus let me test
in p5's Coro it prints 13:24
42 42 42 42
iblech So it does not reset the entrypoint
autrijus because return() is considered rewind there
pugs currently is 13:25
42 23 23 23
I don't know... maybe p6l?
I havn't really been used yield() so I can't quite speak there 13:26
maybe ask some python or ruby folks :)
iblech starts irb 13:27
pdcawley Ruby yield isn't relaly the same as a coro yield. It doesn't return anything, it passes an argument to the implicit block and gets the return value.
autrijus yes, I understand that much.
pdcawley: btw, do you hear the pil2js passes all continuation.t tests? 13:28
pdcawley I did now. Whee!
iblech pdcawley: Right, just noticed...
autrijus t/unspecced/cont.t that is
which means it can probably work with sudo ku soon 13:29
pdcawley Whee! I shall have a play.
autrijus iblech said.uniq is not yet implemented, so it doesn't quite work yet 13:30
but junctions do iirc, so if that's the only blocking thing, we're close
pdcawley After I've finished watching _Dr No_
Easy enough to rewrite without uniq I think. 13:31
And as it stands is *desperately* inefficient. 13:32
svnbot6 r6395 | iblech++ | * t/unspecced/coro.t: return() should reset the entrypoint. 14:15
r6395 | iblech++ | * PIL2JS: PIL::PStmt, PIL::PVar, PIL::PStmts:
r6395 | iblech++ | sub foo { return 3 } is no longer sub foo { return(return 3) }.
r6395 | iblech++ | Similarily, coro foo { yield 3 } is no longer coro foo { return(yield 3) }.
xinming .doit(1,2,3) :{ $^a <=> $^b } 14:17
what does this do exactly please
iblech It calls doit with the argument 1,2,3 and the slurpy block { $^a <=> $^b } 14:19
You can declare such a method with:
method doit($a,$b,$c, *&slurpy_block) {...}
method doit(*@normal_args, *&slurpy_block) {...} will work, too, I think
QtPlatypus What does *&block mean? 14:21
I know what *@array means.
xinming QtPlatypus: It might be something link, grep {}, @ary... 14:22
iblech: So, Why this grammar exists is only because of we can write it in another way,right? 14:23
grep @ary : { ... };
iblech QtPlatypus: It allows you to specify a adverbial block, (map:{...}, grep:{...}, etc.)
QtPlatypus Arh 14:24
iblech xinming: I'm not sure I understood you correctly, but yes, it's because of TIMTOWTDI
xinming: That's disallowed, I think. There may not be whitespace between :{
xinming iblech: Oops. I forgot it. 14:25
iblech: can the :{ } syntax be written into ->{ } ? 14:27
iblech xinming: You mean something like @array.map:-> $a {...}?
That's disallowed
But
xinming hmm, without : 14:28
.doit(1,2,3) ->{ $^a <=> $^b }
iblech you can use placeholder argument if you want to name the params: @array.map:{ $^a + $^b }
Ah, that's disallowed, too
But .doit(1,2,3, -> $a, $b {$a <=> $b}) should work
xinming so, .doit( *@ary ) `@ary = ( 1, 2, 3, 10, 11 )` will pharsed like .doit( 1,2,3, -> 10, 11 { ... } ); right/ 14:30
iblech IIUC, no, * doesn't convert the argument types, so it would mean .doit(1,2,3,10,11). 14:32
You'd have to use
@ary = (1,2,3, -> ... {...}); .doit(*@ary)
xinming wonders if in emacs, that syntax highlighting for perl will pharsed by human instead of computer... 14:34
QtPlatypus xinming: What do you mean? 14:35
iblech xinming: $EDITOR could use Pugs to parse Perl 14:37
Meaning that it would also work with user-defined operators, macros, and grammar redefinitions
IIRC theorbtw1 was thinking of a Perl 6 syntax colorizer which used Pugs to parse the Perl 14:38
xinming iblech: yes, I mean that. 14:42
If you use pugs to do that, the "process" might all be slown down.
iblech Right, $EDITOR would have to optimize things a lot 14:43
xinming iblech: At least, I don't want elisp to call "pugs-parser" to handle the syntax highlighting. :-)
iblech So it only gives Pugs small regions to reparse
xinming pugs might be the "father" of perl 6. :-) 14:44
iblech Well, it's inevitable if you want user-defined ops, macros, and grammar redefinitions to be colored correctly
xinming iblech: do you use Emacs? 14:47
iblech xinming: No, vim 14:51
xinming iblech: I prefer Emacs to Vim... But For editing the config file, I prefer vim. 14:54
oops
$omeone prefer $var_1 to $var_2, In this grammar, is $var_1 better than $var_2 to someone please? 14:55
sub tmp (*$a, *%hask, *@ary ) { ... } 15:01
?eval sub tmp (*$a, *%hask, *@ary ) { ... }
evalbotzy undef
xinming ?eval sub tmp ( $a, *%hash, *@ary ) { $a.say; }; tmp( 101, a => 100 ); 15:05
evalbotzy 100 bool::true
xinming ?eval sub tmp ( $a, *%hash, *@ary ) { $a.say; }; tmp( 101, { a => 100} ); 15:06
evalbotzy 101 bool::true
xinming ?eval sub tmp ( ?$a, *%hash, *@ary ) { $a.say; }; tmp({ a => 100} );
evalbotzy a100 bool::true
xinming ?eval sub tmp ( ?$a, *%hash, *@ary ) { $a.say; }; tmp({ a => 100} ).perl;
evalbotzy a100 'bool::true'
xinming ?eval sub tmp ( ?$a, *%hash, *@ary ) { $a.say; }; tmp({ a => " 100"} ).perl; 15:07
evalbotzy a 100 'bool::true'
xinming { a => 100 }.say
?eval { a => 100 }.say
evalbotzy a 100 bool::true
xinming ?eval sub tmp (*%hash, *@ary ) { return undef }; tmp({ a => 100} ); 15:11
evalbotzy undef
xinming ?eval sub tmp (?$a, *%hash, *@ary ) { if $a { $a.say }; else return undef }; tmp({ a => 100} ); 15:12
evalbotzy a100 Error: No compatible subroutine found: "&else"
xinming ?eval sub tmp (?$a, *%hash, *@ary ) { if $a { $a.say } else return undef }; tmp({ a => 100} ); 15:13
evalbotzy Error: unexpected "i" expecting ";", statements or "}" reserved word
xinming ?eval sub tmp (?$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp({ a => 100} );
evalbotzy a100 bool::true
xinming ?eval sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp({ a => 100} );
evalbotzy a100 bool::true
xinming hmm, 15:14
how can we distinguish named parameter and slurp hash parameter?
iblech autrijus: my $num = 3; my $a = \$num; $num := 4; say $$a; # 4, right? 15:58
autrijus iblech: verily 15:59
?eval my $num = 3; my $a = \$num; $num := 4; $$a 16:00
evalbotzy \3
autrijus oh wow.
that is broken 'course :)
iblech (Writing tests for =:= and eqv currently) 16:01
autrijus excellent. 16:02
iblech .id operates on the container or the cell? 16:09
autrijus neither.
on scalar it operates on the value; on array/hash it operates on the array/hash object. 16:10
iblech ok
autrijus =:= is orthogonal to id
svnbot6 r6396 | iblech++ | * Pugs.Parser: Parse the new eqv operator. 16:23
r6396 | iblech++ | * t/operators/identity.t: Fixes and improvements -- =:= operates on containers.
r6396 | iblech++ | * t/operators/eqv.t: 38 tests for the eqv operator.
r6396 | iblech++ | * PIL2JS: Prelude::JS:
r6396 | iblech++ | * Minor fixes to &infix:<=:=> -- PIL2JS passes 35/39 of identity.t.
r6396 | iblech++ | * Implemented &infix:<eqv> -- PIL2JS passes 37/38 of eqv.t.
rmr hello everyone ! :-) 16:24
so what's going on here?
svnbot6 r6397 | iblech++ | New t/var/refs_point_to_containers.t: References should point to containers, 16:35
r6397 | iblech++ | not cells or even values (passes 12/12 on PIL2JS).
xinming hmm, maybe You all didn't see what I confused about, 16:52
how can we distinguish named parameter and slurp hash parameter?
?eval sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp({ a => 100});
evalbotzy a100 bool::true
fglock_ how can I declare a method for 1..10:by(2) 16:57
I tried multi infix:<..> ( Int|Num $a, Int|Num $b, Int|Num ?$by = 1 ) ...
iblech xinming: named paramaters and pairs are currently discussed on p6l, thread "Demagicalizing pairs" started by Luke 16:58
fglock_: I'd think so, but I doubt that that has been implemented yet. But better ask putter, he fixed :adverbial modifies recently 16:59
bbl &
autrijus discovers Symbol::Opaque 17:20
xinming many people will refer the IIRC Thing, what this is really. 17:21
autrijus luqui is quite insane.
xinming: "if I recalled correctly"
xinming: en.wikipedia.org/wiki/IIRC
xinming autrijus: thanks 17:22
putter How do we do goto LABEL in p6? 17:31
integral hopes it's faster than perl5 17:33
iblech putter: By using continuations? :) 17:34
putter: dunno. It might as well stay "goto LABEL"
autrijus yeah, that's probably that. 17:35
and surely it'd be invoking continuations? 17:36
autrijus.org/tmp/Devel-TypeCheck-0....hed.tar.gz
alias typecheck 'perl -MO=TypeCheck,-main'
bargle++ # nice work
putter ah well. I'll add an abstraction layer then. The JS spec uses "goto step 4" constructs. Given the size of the spec, the rate at which I was progressing with hand rewrites... it would take *days*. So it's time for some global text massage. 17:38
Devel::TypeCheck looked interesting. I only briefly skimmed the paper, but was left wondering why only three programs(?) were used for Experimental results. Oh, duh. Paper deadline of course. ;-) 17:40
autrijus yes :) 17:41
due in a week.
xinming hmm, Perl now reached a place which is the hell for "English", Just now, I watched my qwert keyboard. and try to find if there is any punctuation which isn't been used in perl... 17:46
Maybe not for English, Even In Chinese. I think It will get there also... 17:47
Khisanth already has!
xinming what does ` do in perl 6? 17:48
Maybe we can consider this one.
autrijus larry wants it to mean unit. 17:49
fglock_ what is the next milestone for Pugs?
autrijus fglock_: 6.28.0, full OO support.
fglock_ it means having the Perl6 version of the MetaModel? or in Haskell? 17:50
autrijus yes, or something like that.
putter I don't suppose anyone has gotten around to writing a Filter::Simple like thing for pugs yet?
autrijus putter: iblech checked in macro support a while ago.
filter-simple style macros.
putter yeah, but macros require arguments to parse. you cant say "and take the rest of the line". 17:51
fglock_ is there a todo list? 17:53
putter tries to avoid having thousands of little one-line blocks because he cant grab lines of code.
fglock_ or something I can do to help 17:54
there is not much I can do in the 'Lazy' front at the moment 17:55
putter or me. I'm only off play silly JS frontend games because I dont see a way to affect the critical path.
autrijus uh, the critical path is probably make "make p5smoke" work. 17:59
in the same fashion as "make jssmoke".
maybe some metaprogramming to fill in PrimP5
in a saner fashion
that is perhaps the #1 important thing atm. 18:00
xinming sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp( a => 1, q:h{ a => 100} ); 18:01
autrijus i.e. perl5/PIL-Run/lib/PIL/Run/PrimP5.pm
also, a shared inferface with jspugs.pl etc, or at least something similar, would also help a lot
xinming hmm, anyone here would tell me if this might be ok... hmm, IMHO, I think It is ok for me...
autrijus since I intend to hook in -BPerl5 and -BJS off them.
xinming: q:h means that "%x" is interpolated 18:02
xinming: it does not mean that you can write a=>100 and have it mean a hash
to do that, use
hash(a => 100)
er sorry 18:03
{ a => 100 } would do
xinming autrijus: I just wish to figure this problem. :-) because, if the you specified the named argument, You might not be able to have a hash key as the argument name, 18:04
fglock_ is there a way to define the primitives using the metamodel? (maybe by inserting methods in Int,Num...?)
autrijus xinming: if you said { a => 100 } that'd work surely?
putter one could clone jsbe. but sane requires objects. could be main difference between p5be and jsbe - good use of objects. doesnt seem much point otherwise. but pil2 doesnt do objects yet. full stop.(?) common interface can be done though.
xinming ?eval sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp( { a => 100} ); 18:05
evalbotzy a100 bool::true
xinming ?eval sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp( );
evalbotzy undef
xinming autrijus: *%hash should slurp all "Pairs" .
autrijus xinming: yes and you are looking at a bug :)
xinming But with the same syntax, the named argument can't be pharsed, 18:06
autrijus no the bug is simpler thant hat.
?eval sub tmp (+$a) { $a }; tmp(3); 18:07
evalbotzy \3
autrijus ;)
named participate in positional bindings. that's wrong.
xinming ?eval sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp(3, 'a' => 3 );
evalbotzy 3 bool::true
xinming ?eval sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp(3, 'a' => 5 );
evalbotzy 5 bool::true
xinming You see what I means? :-)
autrijus I see, but +$a means it can't take positionals 18:08
so that's currently a bug that it could.
xinming autrijus: hmm, But there will be confliction, with named parameters and the "slurp hash keys"
maybe slurped hash. :-) 18:09
putter autrijus: did you mean common interface with runjs.pl? jspugs is the repl. crude-repl.pl already looks a lot like it (though jspugs doesnt actually eval expressions yet). could be made more similar, and to the pugs repl. 18:10
a runp5.pl would give you -B. 18:11
autrijus putter: yes, that. 18:14
putter ok. in progress.
xinming ?eval sub tmp ( $a, $b ) { "$a $b".say }; tmp( a => 3, b => 5 );
evalbotzy 3 5 bool::true
xinming autrijus: so, this works, a => 3 is a pair.
putter autrijus: still need something fglock_ can do to help. 18:15
autrijus putter: also, the extent of OO in PIL1 -- i.e. classless OO -- would be useful to support that in runp5.
xinming autrijus: What about the "slurped hash"? How to pass the a pair to that hash?
autrijus xinming: *%h takes whatever +$x not bound; similar to how *@a takes whatever ?$x not bound
xinming: any named things not explicitly bound to other args will go there. 18:16
dudley autrijus: Do you know if you can compile a Scala program to a .jar that will run standalone on a jvm? 18:17
autrijus putter: mm, maybe make mandel.p6 work on p5 :)
dudley: yes, you can.
xinming autrijus: Yes, So, If we have a key which name is the same as the "var" name, Then, It might pass wrong...
autrijus xinming: that is so.
xinming autrijus: hmm, So, I think We should make some point to avoid this.
dudley I've got a punielicious PIL2 repl in Scala, but if it can't run without a Scala runtime, I may go the Nice route...
putter ooooo. 18:18
autrijus oh wow.
xinming just like, the ( a => 'value' ) are bound to variable. and some_func( a => 'value' ) to explicitly pass it to the hash... 18:19
autrijus xinming: (h => { a => value })
will pass to %h. I think.
xinming autrijus: ok, fall back to my example above. 18:20
?eval sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp( a => 3, a => { a => 3 } );
evalbotzy 3 bool::true
xinming ?eval sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp( a => 3, a => { a => 5 } );
evalbotzy 3 bool::true
autrijus ?eval sub tmp (+$a, *%hash, *@ary ) { %hash }; tmp(hash => { a => 3 })
xinming :-(
evalbotzy {('a' => 3)}
autrijus :-)
xinming autrijus: your right.
:'(
putter autrijus: if you really want PIL1 p5be, I can transliterate PIL2JS. But there doesnt seem to be much point. I believe one would take a very different approach if one had PIL2 objects. It would be a throwaway. 18:21
autrijus putter: from my primitive runcore work, that is not so -- PIL2 is essentially typed and normalized PIL1
xinming ?eval sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp( a => 100 ); 18:23
evalbotzy 100 bool::true
autrijus putter: but it's not much transliteration as building the abstractions in p5 that unifies metamodel, intrinsics and ffi
iblech putter: At least in PIL2JS, the code to support methods is ~~15 lines
xinming ?eval sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp( {a => 100} ); 18:24
evalbotzy a100 bool::true
autrijus putter: especially FFI, which is orthogonal to the IL
xinming autrijus: is this example should return undef?
autrijus and I think p5 FFI is one of, if not the, most important thing for pugs 6.28.0
?eval sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp({b => 123}) 18:25
evalbotzy b123 bool::true
autrijus xinming: yes, undef.
xinming: both should be bound to @ary.
xinming: so tests welcome
xinming autrijus: I've said I will write tests. ;-)
autrijus cool
fglock_ what is FFI?
autrijus fglock_: foreign function interface 18:26
use perl5:DBI;
use jsan:Number.Roman;
xinming ?eval sub tmp (+$a, *%hash, *@ary ) { if %hash<a> { %hash<a>.say } else{ return undef }}; tmp( {a => 100} );
evalbotzy undef
autrijus xinming: that is correct, since it should go to @ary 18:27
fglock_ you mean to translate parameters and return values back and forth between P6 and native?
xinming autrijus: You said `{ a => 100 }` should go to array, so ` a => { a => 100 }` should go into %hash, right?
autrijus fglock_: yes... basically, how P6 should work with its host runtime 18:28
fglock_: that part is of course runtime specific
xinming: no, it should go to $a
xinming: b => {a => 100} goes to hash.
fglock_: it's similar to iblech's existing work on use jsan:Module
xinming autrijus: then, b here is something like pseudo... It will do nothing... 18:29
autrijus xinming: er no, b becomes the key in %hash
%hash<b><a> would be 100
xinming autrijus: How to generate a hash key named a? just like the result of %hash<a> would be 100? 18:30
autrijus fglock_: more concretely, get t/unspecced/p5/*.t to pass on runp5
xinming: sorry, I can't parse your question
xinming ?eval sub tmp (+$a, *%hash, *@ary ) { if %hash<a> { %hash<a>.say } else{ return undef }}; tmp( a => {a => 100} ); 18:31
evalbotzy undef
xinming ?eval sub tmp (+$a, *%hash, *@ary ) { if %hash<a> { %hash<a><a>.say } else{ return undef }}; tmp( a => {a => 100} );
evalbotzy undef
autrijus ?eval sub tmp (*%h) { %h<a><b> }; tmp(a => { b => 3 }); 18:32
evalbotzy \3
autrijus that's what I mean.
xinming ?eval sub tmp (*%h) { %h<a> }; tmp(a => 3);
evalbotzy \3
xinming This is what I mean,
?eval sub tmp (+$a, *%h) { %h<a> }; tmp(a => 3);
evalbotzy \undef
autrijus yes, that is correct behaviour.
?eval sub tmp (+$a, *%h) { %h<b> }; tmp(b => 3); 18:33
evalbotzy \3
autrijus and this is also correct.
xinming how to pass the pair `a => 3` into %hash rather than the argument.
autrijus ?eval sub tmp (+$a, *%h) { %h<a> }; tmp(h => { a => 3 });
evalbotzy \3
autrijus like that.
xinming yes,
?eval sub tmp (+$a, *%h) { %h<a> }; tmp(y => { a => 3 });
evalbotzy \undef
xinming ?eval sub tmp (+$a, *%h) { %h.keys }; tmp(a => 10, h => { a => 3 }, b => 3, c => 4); 18:35
evalbotzy ('a')
svnbot6 r6398 | iblech++ | Net::IRC: Updated the documentation of Net::IRC-OO and minor fixes to 18:42
r6398 | iblech++ | Net::IRC's documentation.
xinming ?eval sub tmp ($a, *%h, *%ary) { }; tmp( a => 1, h => { }, ary => [], x => 5, y => 6 );
evalbotzy undef
xinming autrijus: so, Is this example mean that the pair ` x => 5, y => 6 ` are all ignored? 18:43
s/Is/does/
autrijus xinming: I think so, and I further think it's correct 18:44
iblech Hm, I'd say that's an error
As all params should have a var which they can be bound to 18:45
autrijus iblech: and if there's (*%h, *%h2), it binds to h2?
xinming autrijus: In fact, That's why I wish to clarify... :-)
iblech That's my intuition, yep
xinming autrijus: I now caught up with you, But... :-)
autrijus okay, I think that's probably best.
note that in methods you get *%_ for free 18:46
so practically that error won't happen
unless of course you do (_ => ...)
iblech Why is this, BTW?
autrijus xinming: so test that it shoudl raise an error
rmr hi everyone
autrijus iblech: subclasses can then redispatch to super/next without fear of incompatibility.
greetings rmr 18:47
iblech autrijus: Ah, makes sense
rmr i'm a begineer of perl scripting language
:-)
xinming rmr: hi
autrijus nice. hopefully one day you will be a master of perl programming language :)
xinming autrijus: Which example you think will raise an error? 18:48
rmr autrijus: Thanks :-)
xinming mine?
rmr: In fact, I am on the way...
autrijus xinming: ary=>[] should cause an error because %ary can't take []
xinming: furthermore, the unbound b=> and c=> should cause error, as are the unbound x=> and y=> 18:49
rmr ok master programmers!!!
IGTG
c ya... 18:50
xinming autrijus: Ok, thanks. Now, I recorded it.
rmr hi everyone! 18:53
clkao 6/win28 18:54
rmr can anyone tell me which is the best book to get started in perl?
autrijus rmr: Learning Perl, 4th edition, also known as the Llama book.
rmr right now i'm using o'reilly learning perl 3ed
autrijus 3ed works too, but 4ed is cooler because it talks about Pugs
;)
rmr Pugs?!:-/ 18:55
autrijus rmr: pugscode.org
rmr thanks 4 the info! 18:56
:-)
xinming autrijus: who is the operator of #perlchina?
autrijus xinming: no idea... 18:58
xinming autrijus: hmm, Ok, I will ask myself. :-)
fglock_ is there a description of PIL that I can read? 18:59
iblech fglock_: Pugs.Compile :) And there was a thread about PIL on p6c some weeks ago 19:01
rmr can u tell me about some commonly used programs in linux, written in perl?
iblech rmr: Parts of Debian's dpkg/apt-get are written in Perl 5, IIRC 19:03
rmr thanks iblech
kgftr|konobi iblech: kungfuftr.com/code/dibs-0.01.tar.gz 19:05
if you're intrested
autrijus fglock_: also -CPerl5 output
fglock_: and perl5/PIL2JS/lib/PIL/* 19:06
iblech fglock_: And -CPerl5 | perl -MYAML -we 'print Dump(<>)' 19:07
xinming rmr: I think, You shouldn't ask which programs are written in perl, as most program needs perl, :-)
iblech kgftr|konobi: This is not "dibs -- Distributed Internet Backup System", is it? 19:08
rmr xinming: can u give me some eg: ? 19:09
kgftr|konobi nope... debian (improved/improvised) installer
iblech kgftr|konobi: ah
xinming rmr: I don't think we need much eg. as most Linux distribution "have" to install perl. even in freebsd.
kgftr|konobi came up with it ages ago to provide a better mechanism for debootstrap for one-touch installations 19:10
rmr well, even the cygwin i'm currently using under windows has got perl... so i guess its an important part of every distro, right? 19:11
kgftr|konobi xinming: freebsd 5.0 doesn't have perl 19:12
autrijus ...in its minimal installation
xinming kgftr|konobi: hmm, Well, Sorry for my misleading. :-)
autrijus in practice, though, its default install contains perl.
kgftr|konobi autrijus: true enough
rmr ok
autrijus so it's more like "freebsd5's core doesn't depend on a particular version of perl5". 19:13
rmr so all of u guys r using freebsd? i've never had the opportunity to use freebsd...
autrijus I'd imagine not. :)
integral keeps misreading rmr as rms 8)
autrijus debian-based distros is perhaps more popular here iirc 19:14
kgftr|konobi down with linux! as they say
rmr can i run all the GNU sw i'm using in linux in freebsd as well? 19:16
integral that's what meta^Wautoconf is for
rmr i mean without any modifications to the source code? 19:17
autrijus rmr: freebsd can run linux binaries.
rmr is that because both linux & freebsd r POSIX compliant?
autrijus rmr: also you can install a debian, rh9, suse9, or gentoo userland.
xinming rmr: But only 90%...
autrijus rmr: so unless it's a kernel driver or something, chances are it runs on freebsd. 19:18
rmr: no, it's because freebsd is linux compliant :)
xinming rmr: No, that's because BSD has a "emulator" for Linux binary.
autrijus: hmm, so, why are you choosing bsd?
autrijus: :-)
rmr ok. thanx guys..:-) 19:19
autrijus xinming: clkao forced me to
I had not choice
(translation: freebsd community in .tw more active than linux community.)
that was many years ago though. nowadays it's both got great community.
xinming what does clkao mean ?
rmr i'd like 2 try bsd... do they've a live cd? 19:20
autrijus xinming: clkao is clkao at clkao.org, author of svk.elixus.org, the person introduced me to freebsd, all around crazy guy.
xinming rmr: I suggest you Linux rather than bsd if you have much new hardware.
autrijus rmr: yes, www.freesbie.org/
kgftr|konobi freebsd emulates linux system calls, iirc
xinming rmr: I ever wish to try Freebsd, But I got a Sil 3114 sata card and something more which doesn't well supported in Freebsd, 19:21
rmr what about hacking the device drivers 4 say, linux and using it? 19:22
autrijus sadly that's the one thing that does not carry over :-/ 19:23
kgftr|konobi better than it was though
autrijus yup. thanks to eg unified codebase for wifi with linux 19:24
rmr ok. i've an old PIII system with 64 MB RAM & a 40GB IDE HDD. can i do a minimal install in this system 19:25
autrijus you can do a full install.
but I need to sleep. cioa!
*wave* 19:26
autrijus &
rmr ok ciao
autrijus: Thanks 4 the info...
iblech Night :) 19:27
rmr IGTG now. Good Night Everybody...:) 19:28
xinming hmm, what does mmd mean? 19:31
and also ciao... 19:32
dudley Multi Method Dispatch 19:35
and ciao is goodbye
in italian
dudley slinks back off to Real Life for a while... 19:37
nothingmuch wonders why no one has mentioned the is_gay function... perlmonks.org/?node_id=485453 19:39
Juerd It has an underscore in its name and thus does not at all look like a Perl builtin 19:40
nothingmuch that's because it's part of PDF::Sexuality 19:41
jql does this mean there's a use strict qw(hetero)? 19:44
theorbtw1 looks up the node in question, and says "OY!" loudly. 19:53
xinming by the way, anyone here can tell me why in Synopsis, it will have a big jump. which skip S14 and reached S29?
putter fglock_: ping?
iblech xinming: Only because some topics are more important than others 19:54
xinming: Think of builtin subs vs. threads or so
xinming iblech: So, there are still many syntax of the perl language there? T_T 19:55
iblech: or They just the default functions built-in with perl 6 19:56
kgftr|konobi theorbtw1: NOT LONG NOW SUCKA!
iblech xinming: Sorry, couldn't parse these two sentences, but yes, you can, of course, override all "builtins" 19:57
xinming: There could even be a "no builtins;"
xinming iblech: I mean, If there are still many grammars for perl 6 language there. 19:58
putter Can anyone think of a reason p6-on-p5 could not have a blended type system? eg, p5 str isa P5::Str does String. p6 classes can inherit from p5 cls via AUTOMETH. p5 classes can inherit from p6 cls normally, as long a p6 classes @ISA a UNIVERSAL_Blocker class which throws isa() etc back at the metamodel's AUTOLOAD. use overload helps use p6 objects in p5 code. thoughts?
xinming iblech xinming: There could even be a "no builtins;", hmm, So what does perl 6 do exactly? Just a kind of mind... maybe the spirit of perl 6? :-)
iblech putter: ooh inheriting from native classes 19:59
xinming: I've probably misunderstood you, but yes, there can be as many grammars as wanted/needed 20:00
wolverian Juerd, hello. haven't seen you much lately. (on IRC, that is.) 20:03
Juerd Hello
I've been away for a month, and have been inactive for at least a month before that
I've been on IRC now and then, but it's highly irregular
wolverian ah. I hope it's nothing serious?
Juerd And because my hands, wrists and arms have still not yet recovered, it'll stay like this for a while, I fear. 20:04
It's serious alright
stevan putter: slight digression, but where is AUTOMETH and co. documented? which Syn?
wolverian Juerd, damn. I'm sorry. I didn't cause it but hate the world for it.
(really.) 20:05
Juerd I hate the world for it even though I did in part cause it
It would perhaps be easier to live with if it weren't partly my own fault
wolverian it's probably easier to hate the world then. 20:06
I don't consider it a fault, though.
Juerd Stubbornness is a big mistake
I didn't admit it at the time, but I have to ask myself why the hell my largest number of PM posts were done at times I let other things for what they were because of the same very pain. 20:07
I didn't make that mistake this time.
putter stevan! hi. was just thinking of you. checking... 20:08
A12 S1[023] 20:09
wolverian Juerd, at least you've learned. and I'm happy that you're not totally disabled.
Juerd If I were totally disabled, I'd get help from others
Now I am on my own
stevan putter: thanks :)
Juerd Even my parents no longer believe I have to avoid some activities anymore
putter np :)
wolverian Juerd, that sucks. :/ 20:10
Juerd I can still move everything, and I can still carry large boxes
It's hard for the outside world to realise that this all causes great discomfort
putter iblech: on reflection (no pun intended), p6 classes would almost certainly have to do delegation, but with AUTOMETH it could look to clients like inheritance.
xinming @objectĀ».*meth(@args) # calls all methods (0 or more) on each 20:13
what does each mean here?
in parrel dispatch 20:14
putter Juerd: yeah, that's a tough one. It helps to get it named, eg "I have CTS, so I can't do that".
xinming parallel
Juerd putter: Yep. But I don't have CTS --nobody knows what exactly I do have-- and I *can* do it. 20:15
I can just ignore the pain and do things. It'll get worse, but it doesn't disable me.
Hm, I said I didn't make the same mistake this time. That isn't even true.
juerd.nl/vt.txt # I shouldn't have typed all this in, even though I believe in the cause. 20:16
putter sure, but there is a large class of things (lifting vis back or abdom.), typing vis wrists, etc, where yes you can do it, you might even do it without pain, but you know the concequences in pain and incapacity for thus pushing your envelope, and so wisely decline the task. 20:17
Juerd Yes 20:18
I know you people understand it. I just wish the general public understood what it is about.
putter the real hard case is when there isnt even near term feedback, eg truckdrivers and spinal/join wear from lifting, where its "slow job vs abstraction of being crippled for life a decade out". 20:19
Juerd It's so depressing to continually have to answer questions to people who don't understand the answer, and will never do so.
putter at least the universe promptly reminds you you are making a mistake. even if its thus far difficult to explain to others. ;) 20:20
wolverian xinming, it calls @object[0].meth, @object[1].meth, and each .meth that it finds in the inheritance trees of all the @object[n]s 20:21
Juerd By the way - I am looking for people who CAN type (without discomfort) to help me transcribing some videos 20:22
But I reckon that asking this in #perl6 is unsmart. Ignore what I said and get coding :)
putter General advice for this kind of thing - try additional doctors. Often one will eventually say, "oh, of course. I dont know what those other folks were thinking. its obviously foo". They you can say "its foo" and people will be happy. Sometimes even when the cause is still unknown, there are names for bags of symptoms, which doest help you cure wise, but does provide a name.
xinming wolverian: Ok,thanks.
wolverian: I didn't see the @ T_T 20:23
Juerd putter: I've seen too many already, imo
putter: I can say "it's RSI", which at least in .nl is known by everyone. 20:24
putter rms (not rmr;) has had folks providing typing assistance full time. "Ctrl-A Meta-W thank you for your"...
Juerd putter: But because of some very weird media attention, people think that 1/3 of all office workers have RSI, and that it is mostly a psychological and whiney disorder.
putter ah. so "i have rsi. i have to / have been told to avoid x y z"
ooooh. curious. 20:25
Juerd There was a move to rename the spectrum of RSI to CANS, but of course publicizing this decision undid the effect it should have had.
putter well, could always add a modifier. "severe rsi". "tramatic rsi". "impact rsi" 20:26
Juerd oh, severe rsi is the official term for what I have
And that is what I use.
It helps not. :( 20:27
putter :) another hypothesis bites the dust.
err, s/:)/:(/
hmm, well, could use it a chance to do public education. the process of which also helps pursuade. 20:29
Juerd I'm too tired of all this nonsense that surrounds it, that I can't even explain anymore in a friendly tone what is wrong. 20:31
It gets to you after a few years
luqui ?- topic(X).
Juerd luqui: RSI and people not recognising its impact and importance
luqui ahh
Juerd The annoying and awkward situation in which you have a named disorder, but you still have to do everything just because people don't believe the things they ask from you will hurt so much. 20:32
fglock_ putter: is there an overview of how PIL-RUN works? 20:33
putter fglock_: no. let's see...
fglock_ is lost in the void between MetaModel and PIL 20:34
xinming multi sub unimpl (MyFoo $x) { upchuck(); } 20:35
&infix:<+> ::= &unimpl;
hmm, infix needs two arguments, so, the example only gives one, where is the other one please?
fglock_ xinming: that should be an error 20:36
putter crude-repl is the top, MainX loads, EvalX processes PIL to p5, ApiX provides an abstraction on the runtime, PrimP5 has primitives. Type/ contains some object types not yet migrated to Per6-Value etal. That's about it.
luqui xinming, where did you see that? 20:37
putter fglock_: ah, ApiX is most of the interface between mm and pil-run. EvalX handles all PIL processing.
nothingmuch hola luqui
luqui hi nothingmuch
fglock_ putter: I guess it will need Type abstraction? How does it handle multisubs?
xinming luqui: S13
fglock_: S13 show this example. 20:38
putter actually, I was going to talk with you and stevan to see if we cound discard ApiX and have a blended p6-p5 type system.
fglock_: the mm isnt yet being used for subs. Type/Sub.pm is. 20:39
xinming reading Synopsis is really kind of fun. 20:40
putter that was a next step before something in the mm changed, and crude-repl stopped working.
luqui xinming: fixed
thanks
fglock_ I've been planning to work on the type system - I hoped that being able to undertand PIL would help the implementation 20:41
putter I dont suppose anyone has seen any perl5 magic which makes ->foo method access work on nonreference things?
luqui has been wanting that for a while
fglock_ do you mean on unboxed types?
luqui you could overload all constants :-)
putter, why? 20:42
xinming luqui: What do you mean by "fixed"? do you mean the document is fixed, or just a wrong completion ? :-)
luqui xinming, document has been fixed in svn.perl.org/perl6/doc 20:43
putter perl -we '$x = 3; $x->foo' currently an error, someone out there may well have a hack to make it work...
luqui It would require some C code or a source filter
putter re why, just trying to understand the design space,
luqui or overload::constants
(which doesn't exist, I now remember) 20:44
(it's just part of overload)
putter autrijus's emphasis on FFI has me thinking about the p5backend from the bottom up. "if p6 integration is to appear seamless from the p5 side, then ...".
xinming luqui: I ever thought I was reading the newest doc. But now I know... 20:45
fglock_ putter: if you were using Value.pm, it would be $x=Int->new( '$.unboxed'=>3 ); $x->foo;
luqui xinming, well, you were, until I fixed it five minutes ago
that reminds me.. I have to write my value type manifesto 20:46
putter right. but one could make p5 ints look like say P5::Int (does Integer) on p6 side. I was just looking for a way to close the circle. avoid p5 code having to say make_sure_its_an_object($x)->foo 20:47
fglock_ you mean delay the autoboxing - but that's the optimizer's job (?) 20:48
luqui I think he's talking about user-generated code
(that is, hand written) 20:49
xinming luqui: So, I wish to clarify an example...
sub tmp (+$a, *%hash, *@ary ) { if $a { $a.say } else{ return undef }}; tmp( { a => 100} ); should return undef; 20:50
which ($a %hash @ary ) will get `{ a => 100 }`
luqui @ary
xinming luqui: Ok, then, how about `tmp( a => 100 )` ? 20:51
luqui +$foo and *%foo are both for named arguments
$a would get that
and tmp( b => 100 ); # %hash<b> == 100
putter yes. the ideal FFI is one that doesnt exist. user p5 code says $x = Int->new(3) or just $x = 3 and can use p6 bar($x). and with $x = a_p6_bar(), overloading gets us $x + 1, but $x->foo is still a problem for native p5 non-reference types.
xinming luqui: Ok, the problem now comes out,
fglock_ how about _call_method( $x, 'foo' ) 20:52
xinming tmp( a => 100, 3, 4, 5, 6, );
luqui: If I need to pass a pair which is ( a => 200 ), how can I do that?
pass it to %
luqui what do you mean?
you want $a to be 100, but you want %hash<a> to be 200? 20:53
putter right. I was just wondering whether in the large bag of little used esoteric perl5 hacks, someone had worked out something to permit 3->foo.
xinming luqui: yes.
luqui can't
*% is for overflow named arguments
fglock_ that's the same thing when you do UNIVERSAL::isa()
xinming luqui: So, % will never get a key named 'a' in this example, right?
luqui fglock_, except that _call_method could be smart about it
xinming, I believe so 20:54
fglock_ sure
xinming luqui: In fact, autrijus gave me an example tmp( a => 100, hash => { a => 200 } );
putter isa($x,'C') you mean?
xinming luqui: But from This example, The other problem comes out, which is, other pairs which "should" be in hash will all be ignored.
luqui oh my 20:55
yeah, that's an interesting case
fglock_ sub call_method{ unless(ref($_[0])){ upgrade_to_object(\$_[0]) } $_[0]->$_[1] } or something like that
xinming luqui: So, I think You might need to write more details on this. :-) 20:56
putter right. or the more general but less efficient p6ify($x) which gives you a full p6 proxy object.
(if necessary)
luqui xinming, using hash => { } might be construed as evil, according to some 20:57
xinming luqui: some people who is busy won't care about this, except me... as I am "twindle my thumbs" :-)
luqui thinks he will p6l this issue 20:58
xinming luqui: by the way, In the above example, I didn't mean *%, I mean *%hash, hope you can understand. :-)
luqui yeah
xinming %_ will slurp all the other pairs for methods, But not for sub... odd... 20:59
luqui xinming, the only difference is that in methods %_ is automatically declared 21:00
putter ha! you _can_ create p5 infix operators. just define a receiver class with say + overloaded. so Filter::Simple say --> (that's - - > ) to say + mumble, where mumble's overload + returns the right thing. hmm, have to check precedence... hopefully something higher than +
err, those "say"'s are "perhaps", not say(). ;) 21:01
xinming luqui: I knew. :-) 21:02
luqui putter, that's an interesting idea
the tightest infix you can get is ** 21:03
but it's right associative
putter 3-->ref eq 'P5::Int' ;)
luqui the next one up is *
putter ooo, good.
luqui putter, you will break Damian's code if you do that :-) 21:04
putter which code?
luqui he uses this idiom: 21:05
while ($count --> 0) {...}
oh, there's another issue 21:06
putter ?
luqui 1-->ref eq 'P5::Int' # would be equivalent to 1**ref($_) eq 'P5::Int' 21:07
that is, -> is special in that it quotes its right side
but that's not a huge issue
just turn 3-->ref into 3**_method('ref') 21:08
source filters can do such things :-)
hmmmmmm, I wonder if you could use that technique to use -> 21:09
convert $x->y into $x**_method('y')
and then make _method smart
fglock_ how about p6(1)->foo instead of 1-->foo (no source filter needed) 21:10
luqui fglock_, because that's a lot harder with a source filter
oh
without a source filter
that's awkward though
putter oh drats. I see. you cant overload anything higher than, so has to deal not just with the object, but with any method call. blech. oh well.
luqui you want to minimize pain on the user's part
putter than -> 21:11
fglock_: yes, definitely. also call(3,'foo'). I'm just looking for a third, less efficient but perhaps less invasive, idiom. 21:12
luqui Hmm, is PPI smart enough to be able to transform 1->foo
generally
fglock_ you are using the source filter as a mini-perl6?
in this case, why not use 3.foo ? 21:13
luqui because that's a much more common operator in perl 5
xinming luqui: Why does perl still use pod, why not change it to xml by the way?
luqui xinming, because xml sucks 21:14
I should clarify
humans hate writing xml
and you can't read it easily
but its good for generated markup
putter fglock_: in a very limited sense. the idea being -> would work as expected on p6 objects. but if p6 objects arent forced to do outgoing border checks, a p6 sub could return a p5 native int. and it would be nice not to have to saturate your p5 code with "an make sure it's not a native int before I call a method" code.
luqui but pod is written and read by humans
putter luqui: PPI?
wolverian luqui, I assume PPI is good enough, as it (AFAIK) parses 99% of CPAN correctly. 21:15
luqui wolverian, it *tokenizes* 99% of CPAN correctly
xinming luqui: well, In fact, For perl, There isn't much markup, So, with autocomplete feature. I didn't find xml is hard to written really. :-)
luqui I was wondering if it was smart enough such that you could put a precedence hack around it
wolverian luqui, ah, right.
putter hmm... an alternative would be P5::Int does _not_ does Integer, but does coerce, so everything which hits p6 ceases to be native. but then one has to deal with proxy objects and such. which it would be nice to avoid. 21:16
luqui putter, but is not as nice to avoid as source filters are 21:17
xinming, well, you could write your documentation in xml
because we're already supporting both pod and kwid
so an =xml plugin shouldn't be so hard
putter luqui: --> would be an optional import. trades filterness for being able to "think objects" even in p5. 21:18
luqui fair enough 21:19
however, as long as you're adding syntax
we could do as fglock_ says and treat /\.\w+/ as a method call
instead of the heaviweight /-->\w+/
putter hmmmmmmmmm 21:20
luqui oh shit
putter lol :) 21:21
loudly
luqui how would you transform this: 3.foo(24)
putter "what HE said!"
3**mumble->foo(24)
luqui which means 3**(mumble->foo(24))
however... 21:22
as long as you have control over mumble
you could make that work
putter mumble still has to play "capture the chain of ->" game.
fglock_ 3**unboxed**->foo(24) :) 21:23
self-documenting
putter **-> syntax error?
luqui s:2nd/\*\*// 21:24
xinming luqui: Ok, by the way, hmm, will be perl 6 default assume the codes are perl 6 in the future? I know now, It is assume the codes are perl 5 codes.
putter oh, so using "unboxed" as the name of mumble? i still havent thought to hard about what mumble needs to be. 21:25
luqui xinming, you mean when you run perl script.pl
it is assumed to be perl 5?
putter users will never see it anyway.
luqui putter, I think he was suggesting that you make the users write that
putter eeep.
fglock_ unboxed{3}->foo(24) -- using a tied hash 21:26
luqui fglock_, how is that better than p6(3)->foo(24)
xinming, well, the question is, when do we switch?
because as soon as we do, we break things
xinming luqui: hmm. :-)
luqui so it will probably stay perl 5 forever, until we give you the option not to build perl 5 at all 21:27
and then it's the administrator's choice
the other way around it is to use a different extension for perl 6 scripts
xinming luqui: why not ask people to change all the perl extension into .p5 or .pl5 ? :-)
luqui how about the other way
xinming luqui: and the default .pl will always stand for the newest version of perl. 21:28
luqui ask people to change from .pl to .p6 when it's perl 6
putter $x**ip6->foo() does give you infixness wo filters, which p6($x)->foo doesnt. so I guess it could make sense to expose mumble so you can do that.
luqui xinming, because then you're still breaking everything
it doesn't matter what you ask people to do
they won't read your request
fglock_ luqui: it allows more explicit declarations, such as unboxed{3:int}->foo(24)
luqui fglock_, that's a syntax error
p6(3, 'int')->foo(24)
xinming luqui: hmm, Do something rude... :-) 21:29
putter $x**as6->foo()**as6->bar() vs p6(p6($x)->foo())->bar() vs $x.foo().bar()
xinming luqui: just do something like M$. :-)
luqui ?
oh
putter ok, can anyone think of any showstoppers for the blended type scheme? 21:30
xinming luqui: Ok, so, how about extension .pl6? will it be used for perl ?
luqui you mean "HI THERE, I SAW YOU HAVE A FILE THAT HAS A .PL EXTENSION. YOU PROBABLY MEANT PERL 5, BUT WE WANT YOU TO MEAN PERL 6. SO CHANGE IT, OR ELSE I'LL CRASH"
putter everyone believe we can create a HIDEUNIVERSAL class which bounces UNIVERSAL methods back to the p6 object's mm AUTOLOAD?
luqui just rebind UNIVERSAL? 21:31
xinming luqui: Well, No... You should rather say, "Perl 5 is outdated... We need a wonderful technoledgy, which will make your life easier... your program wrong faster... and safer... "
luqui *{'UNIVERSAL::'} = \%HIDEUNIVERSAL::'
xinming s/wrong/run/
luqui xinming, though the original wasn't too far off :-) 21:32
putter p5 calling $p6o->isa() on a p6 object should get p6's version of is, rather than UNIVERSAL's. I think.
luqui so why don't you just define "sub isa" for the perl 6 Object class
putter $p5o->isa() on non-p6 objects should get the usual UNIVERSAL isa. 21:33
fglock_ putter: I think it already works like this in the MetaModel
luqui putter, right. isa() is nothing special, you can override it just like any method
xinming luqui: because perl is a really different language, It will keep changing... and in the future... I believe perl 7 will be even harder to "pharse", So, why not make these scheme default? .p6 or pl6,
putter oh, duh. by simply defining an isa method.
xinming luqui: And also, It's really ugly to write use v6; as the first statement.
luqui xinming, agreed 21:34
putter so that just leaves the puntable issue of folks adding methods to UNIVERSAL and p6 having to avoid them. something for version 2.
luqui I believe if you give perl a file that ends with .pl6 or .p6, it should do the right thing
xinming luqui: or even more, you can have symbolic link, which links to /usr/bin/perl and the link names are /usr/bin/perl5 /usr/bin/perl6
luqui and we can just bind .pl (and .plx, and .pm) to mean perl 5 by default 21:35
xinming, yeah, that was planned already
xinming luqui: will the perl 6 have a different extension for "module"?
putter Ok, so which version of the metamodel should we be using. stevan? mm2.0? doesnt matter?
luqui no, but perl 6 modules start with "module", not "package"
so we can tell
xinming I rather prefer going with the newest perl. 21:36
luqui again, that breaks stuff. we don't want to break stuff
one of the reasons perl has been so successful is because it never breaks backwards compatibility
putter If -Bperl5 is to be integrated with pugs, we should probably move a metamodel, along with its tests, out of pugs/perl5/ and into the main distribution.
luqui we're gutting the language, so we have to have safety features to assume backwards compatibility
putter, maybe so 21:37
fglock_ luqui: does isa() test for subtypes? like $list->isa(Lazy)
xinming luqui: hmm, I think, by the time goes by, we can make .p5 .pl5 as perl 5, and px plx for the specified version. and .pl for the newest version of perl.
luqui: backwards compatibility will confuse many "newbie" 21:38
They might not be able to do so much comparision with perl. That's what I think, might be wrong anyway. 21:39
luqui: just do a suggestion. ;-) and most people will change by the time goes by. :-)
luqui fglock_, in perl 6, no
xinming, you know how perl tutorials say "always write #!/usr/bin/perl at the top of every script" 21:40
xinming this is a bit like w3c recommendation.
luqui I don't see how perl six is any different: "always write #!/usr/bin/perl6 at the top of every script"
fglock_, in fact, it would really be nice to abolish isa() in perl 6
or at least hide it in is_subclass_of or inherits_from
xinming luqui: hmm, So, Some people will use #!/usr/bin/perl6 at the top line, and use perl 5 syntax. That's what I can image. :-) 21:41
luqui because most of the time, what people are asking is whether it does()
fglock_ or move it to .meta?
luqui xinming, they have more to learn :-)
putter iblech, etal: reality check: with this kind of FFI-free scheme, one could still cps transform p6 code, passing the continuation in a local() var. is anything (aside from p5-compiled-from-p6 code size) lost?
luqui fglock_, yeah, perhaps
fglock_ how to test for subtypes? .does(Lazy) ? 21:42
luqui fglock_, yeah
though I think you're misusing "subtype"
fglock_ I've read that Lazy is a subtype ? 21:43
luqui fglock_, of what, with what constraint?
xinming luqui: hmm, So, If a script extension name is .p5 or .pl5, will this be pharsed as perl 5? 21:44
luqui xinming, sure
xinming, just like everything else that isn't .p6 or .pl6
if it ends with .p5 and starts with "v6" or "use v6", then it's perl 6
perhaps with a warning on that case
putter mutters "pity there isnt a UNIVERSAL_UNBLESSED"... ints and strings could be validly copied into p6 objects, and arrays and hashes could have methods out of U_UB. 21:46
xinming luqui: Ok, then, I'd suggest to write a simple line with, "we suggest use .px .plx (x = perl version ) name extension instead of .pl(Note: It's not necessary but recommend)."
luqui xinming, agreed 21:47
.p5 is clearer than .pl
xinming luqui: then, you can watch how long will this come to reality. :-)
and also, for .pl the autodetection way is needed, but It will assume it's the newest perl version. 21:48
luqui uh, no. .pl assumes perl 5 21:49
because that's when .pl was in use
xinming luqui: hmm, what's extension for perl 4?
cgi? :-)
luqui perl 5 is backwards compatible with perl 4, so it doesn't matter
putter looks like having arrays use methods in ARRAY has been talked about for a decade, but never done. :(
luqui haha
xinming or no extension, but use `cat filename | perl`
luqui: perl is amazing, But I think, the extension scheme flexibilty is really needed. as perl isn't like any other language in the world. 21:50
luqui: That's just my Humble Opinion. 21:51
wolverian extensions are a hack anyway.
leo__ what I'm doing wrong: lt@feather:~$ pugs -e'use Set;say "hallo"'
(the plain say "hello" works fine) 21:52
luqui hmm 21:53
well, this ought to work, but doesn't:
fglock_ luqui: so Lazy and Eager are roles - I was trying to find out this for days :)
nothingmuch wow... /me just implemented the forth looping constructs in 10 lines of forth
putter ARRAY autoload could even give the same "unblessed reference" error. ah well.
luqui pugs -Mblib6/lib -euse Set'
er, s/-e/-e'
nothingmuch well, 2/3rds of them
putter is craving ruby just about now... none of this idiocy.
luqui nothingmuch, cool
xinming luqui: So, In fact, we can say, How many people there are still using perl 4? In fact, most people will use perl 5, hmm, except for older working scripts.
luqui putter, what idiocy? 21:54
nothingmuch i love the way forth is like that over and over again
i should give it more than an hour a month =(
luqui oh, the ARRAY stuff and whatnot
fglock_ programmed forth for years
before perl
nothingmuch fglock_: feather.perl6.org/~nothingmuch/harrorth 21:55
uh, s/org/nl
specifically feather.perl6.nl/~nothingmuch/harro.../Interp.pm
look for ': BEGIN'
putter if native p5 arrays behaved as if they were blessed as ARRAY, its not clear anything would be lost (the arguments against I've seen so far have all be flawed), and we wouldnt now have to be designing hoops and struggling to jump through them. the right idea was had by the mid 90's, and yet here we are. :( sigh.
nothingmuch fglock_: i'm learning forth by implementing it with perl, so that I can implement it better in haskell and learn haskell that way 21:56
the only problem is that I haven't done enough of this over the past 2 months 21:57
luqui nothingmuch, do you grok monad transforms?
nothingmuch luqui: nope
luqui oh
nothingmuch i know how to use ReaderT
luqui well, that's a start
nothingmuch i understand what it's purpose is, and what it does conceptually 21:58
luqui I'm writing Monad.pm (for both perl 5 and perl 6)
nothingmuch but I have no idea how the implementation actually works
cutesy
luqui autrijus suggested that I have Monad::Transform
leo__ luqui: nope doesn't (-I neither) 21:59
luqui leo__, I know. I just said that it *should*
you could try installing pugs
so it will look in the right place
leo__ oh, well compiling it
luqui aha! 22:00
leo__, I've got it
export PERL6LIB=/home/whatever/pugs/blib6/lib 22:01
putter oh, drat. cant use dot ofcourse. not with &-free subs. sub bar{"bar"} "foo".bar() Back to -->. and wishing p5 had been properly patched.
luqui putter, I was suggesting that you treat /\.\w+/ as a method 22:02
and if they don't put space around the ., that's their fault
putter sigh. I suppose. I hadnt appreciated the incompatibility before. But I guess use Dot; would be ok. 22:03
nothingmuch sleepy time 22:04
putter use Dot; could also, instead of being p6-specific, fix the unblessed- and non- reference problem, treating them as ARRAY, etc, and then the p6 side of things could twiddle ARRAY.
nothingmuch all employee meeting at 8 am
xinming nothingmuch: good night.
nothingmuch hates time zones
putter good nigh.
luqui aginite
sleep wellish
putter surely someone has already done a use Dot before... ??
luqui (or welsh, whichever you prefer) 22:05
xinming luqui: how the way, How long did @Larry write the "Apocalypse" for perl 6?
nothingmuch if i sleep in welsh i won't understand anything
xinming how long had @larry*
luqui xinming, "the" apocalypse?
xinming luqui: yes, I mean, the language design should be slowly gathered. 22:06
luqui there are eight of them
xinming luqui: only for perl 6?
...
luqui let's see, Apocalypse 1 was written 2 Apr 2001 22:07
And Apocalypse 12 was written 13 Apr 2004
so it seems about 3 years
nothingmuch written or updated?
luqui written
nothingmuch oh right, we didn't start till feb of *this* year 22:08
xinming luqui: I mean start writing the Apocalypse.
nothingmuch anyway, mail done... *poof*!
luqui probably around March 2001
where'd nothingmuch go?
I wish I could disappear
xinming no wonder perl 6 succeed in the future. :-)
someone who learnt perl 5 for only 2 weeks, he said, Hi, You see, It's really convient to do something I like computer do. ;-) 22:09
luqui: hmm, I believe perl 6 will be easier to program, But harder to grasp the internal. 22:10
luqui xinming, I hope the internals are easier to grasp, being written in perl 6 and all
xinming, but I think the underlying concepts will be a lot harder than perl 5
but we're trying to make it so that you don't have to learn those
putter simon wrote an Acme::Dot, but it is "a class can provide ." rather a client can use it on everything. 22:11
xinming luqui: hmm, I like to learn the internal rather than use it. :-)
luqui: If I don't wish to know the internal, I think Windows is really good except the blue screen. 22:12
luqui: But I don't know when I will know that. :'(
luqui putter, interesting, and it seems he did it without source filters
putter sure, just overload '""'... 22:13
(at least that would be my guess... checking...)
oooo, there is a use overload '.' ! 22:15
xinming luqui: by the way, theoretically, will perl 6 ( not compiled to byte code ) run faster than or equeal with perl 5?
gantrixx OK, I have a question on a print statement. Here is the print statement
print "3rd card is " ~ $deck.sequence[2].display() ~ " \n";
why does this result in the following output
4c3rd card is 1
luqui xinming, you can't really help but compile it to byte code 22:16
we're of course trying to make it run faster than perl 5
but if we can't, then we're giving you mechanisms (like the unboxed types) that you can use to write locally fast code
which is really all you need most of the time
gantrixx, woah, wtf 22:17
xinming luqui: just like perl 6 a rule engine, pharse rules.
pdcawley_ gantrixx: Probably because display does it's own printing and returns true.
xinming and compile the rule into byte code. :-)
luqui pdcawley, ahh yes, that makes sense 22:18
gantrixx why would it display the results of the display method before the "3rd card is" string?
luqui because your statement is equvalent to:
Khisanth because display executes first
pdcawley_ Because, obviously, it's evaluated before the print .
luqui my $str = $deck.sequence[2].display; print "3rd card is $str\n"
gantrixx Oh, that makes sense 22:19
and where does the 1 come from?
luqui the return value of print
gantrixx display returns nothing
pdcawley_ Wrong.
Khisanth subs always return something :)
pdcawley_ It returns whatever the last executed statement returns.
luqui the return value of the print in display()
gantrixx ok, well I don't have a return in the method
luqui that doesn't mean it doesn't return anything :-) 22:20
sub foo() { "hello" } say foo; # hello
pdcawley_ should stop doing this. I'm in a foul mood, which is never a good mood to be in when doing tech support.
gantrixx well it is annoying that it returns it to stdout
Khisanth it doesn't
luqui gantrixx, you should have display() return a string
not print it
Khisanth you printed out the return value
gantrixx Oh OK, I know then, I should have it return the string
putter pdcawley_: do you have a feel for why calling methods on unblessed- and non- references continues to be prohibited?
xinming my @ary; my @ary2 = ( 1, 2, 3, 4, 5, ); @ary = @ary2; 22:21
pdcawley_ It is? So I can't write 10.as_roman?
putter not in p5.
pdcawley_ Who cares?
xinming in this example, will the values pass to @ary is a copy?
gantrixx hmmm, maybe I can just say "return" and have it return nothing 22:22
luqui xinming, if you change @ary, @ary2 will not be affected
xinming luqui: So It's a copy. :-)
luqui yeah
(just being clear)
QtPlatypus But only to one leval. 22:23
pdcawley_ Yeah, shallow copy.
putter if the p6-on-p5 backend emphasizes lightweight FFI, the lack causes pain and suffering yet again...
xinming pdcawley: Do you mean "Copy on write" ? :-)
putter for people using p5 that is. perhaps that's correctly a "who cares?".
ah well. 22:24
pdcawley_ @ary := @ary2 works if you want to alias 'em.
FFI?
luqui foreign function interface
as in libffi, but not for C
pdcawley_ ISTR writing something that would pretty much ensure that all unblessed reference types would autobless into ARRAY/HASH/WHATEVERs, but I can't, for the life of me, remember how I did it. 22:26
I think evil was involved.
with assitance from some of the hairier parts of overload.pm 22:28
putter a p5 implemtation?!? there have been patches floating around for years, but I dont recall seeing a pure p5 one...
pdcawley_ I may be going mad. Or it may simply be that I worked out something that I thought should work, but didn't do it. 22:29
putter ah, yes. as long as you can source filter, you can use overload and a helper class to fake it.
theorbtw1 Doesn't autobox do that?
dduncan quite by accident, my last commit was a nice round number, by computer or comp history terms
pdcawley_ I don't think it used a source filter.
putter autobox?
putter goes to search for autobox... 22:30
leo__ ?eval use Set; "ok".say 22:31
evalbotzy pugs: *** No compatible subroutine found: "&use" at tmp-Prelude.pm line 38, column 30-59
theorbtw1 search.cpan.org/~chocolate/autobox-1.03/
pdcawley_ leo: That's not going to work on evalbot -- IO's generally frowned on.
svnbot6 r6399 | fglock++ | * perl5/ Value - added 'Lazy' and 'Eager' List roles
r6400 | Darren_Duncan++ | updated /LICENSE/GPL-2 so that it reflects the current address of the Free Software Foundation, as seen at www.fsf.org/ ; its whitespace was also cleaned up
putter oooooo, autobox... search.cpan.org/~chocolate/autobox-...autobox.pm 22:32
luqui theorbtw1++
22:32 theorbtw1 is now known as theorbtwo, theorbtwo is now known as theorbtw1
theorbtw1 theorbtwo++ 22:32
pdcawley_ Bah. It still resorts to XS.
leo__ pdcawley_: the evalbotzy can't use any modules?
theorbtw1 theorbtw1--
22:33 theorbtw1 is now known as theorbtwo
pdcawley_ leo: That does tend to follow. However, I'm not entirely sure of this. Ask whoever wrote it. 22:33
putter theorbtw1++ theorbtw1++ theorbtw1++ Yay! 22:34
pdcawley_: yes, but its all PASS on CPAN testers...
pdcawley_ That's gloriously evil... a chap could rather like that... 22:35
putter ok. so that takes care of that. "when writing p5 code, remember p6 code may return an unblessed- or non- reference value. and you dont have the p6 compiler hiding this from you. so if you want to safely use -> everywhere, you need to use autobox." 22:39
hows that? 22:40
svnbot6 r6401 | iblech++ | PIL2JS:
r6401 | iblech++ | * PIL::Subs: Slurpy hashes, thanks to xinming++ for reminding me that
r6401 | iblech++ | *%slurpy_hashes exist :)
r6401 | iblech++ | It uses the semantics of
r6401 | iblech++ | www.nntp.perl.org/group/perl.perl6....age/22860, i.e.
r6401 | iblech++ | sub foo (*%hash) {...}, foo(hash => {...}); # works
r6401 | iblech++ | sub foo (*%hash) {...}, foo(hash => {...}, foo => bar); # dies
r6401 | iblech++ | * PIL::Types, PIL::Params: Restored junction autothreading (was broken
r6401 | iblech++ | in certain cases).
theorbtwo Putter: Why would people expect the perl5 bit of perl5/6 code to act like perl6 code instead of perl5 code? 22:41
Now, if perl5 code can pass an unblessed reference to perl6, and you can't call methods on it in perl6, that'd be a problem. 22:44
pdcawley_ Because being able to do $arbitrary_thing->isa(...) is just gleefully useful, whatever language you think you're in.
Used to seriously piss me off, having check the blessedness of a reference before doing that. 22:45
theorbtwo Eh, UNIVERSAL::isa($thing, 'class').
The thing that always annoys me is that it's UNIVERSAL::isa, and not UNIVERSAL::ISA.
pdcawley_ Which is *so* bloody clunky in mostly OO code.
And sucks satans balls if you need to override isa somewhere. 22:46
theorbtwo I don't use isa much; I generally prefer to just die on errors.
You're 100% right about needing to override isa, though.
pdcawley_ I had to use it in Pixie, at least until I worked out how to force objects to inherit behaviour from classes associated with their base types. 22:47
Which is a whole other level of evil involving redefining bless and...
theorbtwo Wow.
That's some pretty hardcore evil. 22:48
pdcawley_ Well, you just look at the type of the object you're blessing, and, if the class you're blessing it into doesn't inherit from the implementation type, you stick the implementation type on said class's @ISA.
theorbtwo ...and I've had occasion to want to JMP $scalar.
pdcawley_ Easy.
The really cunning bit was when we added magic to allow you to store out of band information on an arbitrary scalar. 22:49
theorbtwo OK, really, to want to $functionpointer->(@arguments), where $functionpointer is a /C/ function pointer.
pdcawley_ Most importantly, any package needed to be able to do it to any scalar.
theorbtwo Isn't that what attributes are for?
svnbot6 r6402 | Darren_Duncan++ | /ext/Locale-KeyedText : converted all of my indenting from tab characters to space characters, at 4 spaces per indent level; updated version to 1.6.0 22:50
r6403 | iblech++ | * PIL2JS: libjs/PIL2JS.js, pil2js.jspl: Fixed exit() not terminating the
r6403 | iblech++ | program flow if it was used in a sub using PIL2JS.cps2normal.
r6403 | iblech++ | * New t/builtins/io/exit-in-if.t: wrote a test for that.
r6403 | iblech++ | * PIL2JS: PIL::PNil: Very minor cosmetic fix.
pdcawley_ Theorbtwo: At the time we wrote pixie, we didn't *have* attributes. Not that you could hang off *values* anyway.
And you couldn't use hash keys, because it's not safe. And we wanted to be able to hang data off objects implemented using arbitrary references. 22:51
xinming It's just plain cool, that's why. Oh, and Perl 6 then magically
runs inside every vaguely recent web browser.
so, What does this exactly mean?
pdcawley_ Pixie does some evil, evil, evil stuff in order to look lovely to its users.
xinming It's from docs/note/plan
pdcawley_ Catch is, the locking is broken as designed.
And a complete bastard to fix.
dduncan pdcawley_, a quick question ... 22:56
pdcawley_ Fire away.
dduncan is pixie and SQL::Translator the same thing or different things?
pdcawley_ I don't know what SQL::Translator is.
dduncan just a minute ... 22:57
pdcawley_ Definitely not the same thing.
dduncan oops, pardon me, confused 'pixie' with 'sqlfairy' 22:58
pdcawley_ Pixie does object persistence in a single database table. There were plans afoot to allow for dedicated class tables with indexy things, but we never got that far.
dduncan bad memory, bad@
!
putter theorbtwo: (long pause) I'm not sure. The background is considering creating a very-low-overhead FFI version of a p5backend. Which suggests allowing native types, rather than mutating or proxying them at border crossings. Which lead to "what is the cost?". And on the p5 site, one cost was losing the ability to do -> on everything. Unfortunately, within the runtime implementation
pdcawley_ I hate writing schemas. We wanted a thing you could chuck pretty much any object at and it would Just Store It.
theorbtwo Well, that seems like a benifit, not a cost: In perl5, stuff acts like perl5.
putter autobox doesnt really help, because being XS, we will likely need to avoid it for portability.
pdcawley_ It didn't have queries, it didn't have all sorts of stuff, but it pretty much worked with everything, and the things it didn't work with you could write hooks for. 22:59
theorbtwo Sounds lovely, pdc.
putter yeah. point taken. though it would be nice in the p5 code which implements the runtime, if we could forget that unboxed types are a posibility.
pdcawley_ But it really, really sucks at concurrency. 23:00
theorbtwo Databases annoy me -- my perl code ends being warped to be nice to the database. I'd much rather have the database be warped to be nice to me.
Hm, you could always box upon calling into the p6, but that'd be inefficent. 23:01
pdcawley_ Take a look at pixie. I should really take the time to write a design doc for a concurrency handling system that might actually *work*.
putter pdcawley_: have you seen www.rubygarden.org/index.cgi/Librar...orial.rdoc
pdcawley_ Not yet. 23:02
fglock_ putter: does PIL makes the unboxed/boxed conversion explicitly, or is it left to the runtime to do?
pdcawley_ Ruby really wins over perl if you're trying to write an object persistence thing because there's only one way of implementing an object, and there's all sorts of notification features built in. 23:03
putter it would have to be a runtime thing. so p5 code could use a module compiled by pugs and call methods without knowing/caring that they were p6-based.
theorbtwo Well, there's already the argument unpacking code that you generate, no?
pdcawley_ And having Object be extensible really doesn't hurt in the slightest :) 23:04
putter pdcawley_: true. sometimes lots of ways to do something just means "I have to handle _how_ many cases?!"
pdcawley_ Indeed.
putter Hopefully PIL as kernel language will help mitigate that. 23:05
pdcawley_ I just wrote a rather neat thing in Ruby for generating a composite from an XML tree. Most of the time, you're happy with a basic node type, but some things want to have custom behaviour. Initially I was writing an explicit thing for mapping from XML tags to classes, but they just looked like: 'TagName' => 'Node::TagName', which was tedious. But Ruby lets you notify a class when you subclass it, so I arranged things so that simply making 23:07
fglock_ putter: sorry to insist on this, but how about: Int::add( 3, 4 ) - p5 code for saying 3.add(4)
leo__ pdcawley_: what is the $cookie from $pixie->insert
pdcawley_ leo__ it's the object id you use to get the object back. 23:08
Opaque, so it's usually a good idea to insert named 'root' objects. It's been a while, so I can't remember all the methods.
leo__ valid in that very program only?
putter fglock_, theorbtwo: I guess the argument goes, if you want to accept natives, then you either tolerate them internally or proxy them as they come in. the main downside being proxies are notably slow. but perhaps that's acceptable.
pdcawley_ leo: No, persistent.
But generally based to use the 'name_object' (if memory serves) method for that sort of thing. 23:09
leo__ kind of a hash?
value
pdcawley_ In a web app, you'd save your session data say, and stick the OID in your session cookie.
leo__ ah ok
putter fglock_: no sorry needed. i'm thinking fuzzily. it's almost like having a conversation in front of a whiteboard! (wistful sigh) 23:10
pdcawley_ I think we just used UUIDs, guaranteed unique, opaque(ish) strings.
putter fglock_: err, where? I guess it doesnt matter. yes, that should work. Oh, you mean internally... and yes, that should work... 23:12
putter tries to figure out WHAT he has been thinking...
leo__ parrot's builtin '$S0 = freeze foo' should simplify these apps vastly 23:13
theorbtwo Donno. Will be helpful, but not much more then Storable. 23:14
The problem is getting things into the database in a mildly sane manner.
putter Oh, that call is fine if you know you have an Int. But dispatching if there are natives is a pain. (blessed($o) ?? $o->add(4) :: is_scalar_and_number($o) ?? Number::add($o,4) :: ...)
err, s/::/:/ s/??/?/ 23:15
leo__ theorbtwo: true
fglock_ but that would be a special case, and then you would use the slower dispatch( 'add', 3, 4 ) 23:17
theorbtwo fglock: But who is the "you"?
The compiler can't know, and you don't want to force the user to write code like that.
gantrixx wouldn't as a rule "best practices" practices sort of thing, be better to always reference your attributes, even if not inherited, as $self.attribute?
leo__ putter: parrot MMD handles that ;-) 23:18
putter But regardless. The next steps are the same regardless. Get a metamodel out of pugs/perl5/ and its tests into t/ . Probably Int->new(3) rather than Int->(unboxed=>3) to reduce FFI cost. Then we can add overloads to the mm again reduce FFI cost. And while
this is going on, attempt to write the PIL compiler/evaluator/whatever which uses these objects. 23:19
Oh, and using the mm, write the assorted runtime library primitives.
putter just which he didnt have a sneaking suspicion that the flexibility sacrificed by going minimal-FFI wasnt going to bite us later... ?) 23:20
leo__: :)
fglock_: all sounds plausible. let's code. :) 23:21
leo__ this means that all optimizations at he 'upper' level are contraproductive
putter steven: ping?
fglock_: does the minimal-FFI thing of p5: my $i = Int->new(3); print $i+2; # 5 sound plausible? 23:22
fglock_ It might break in MM2, I think it will not export classes to p5 by default. But it can be fixed. 23:24
I'd make the optimizations later.
putter fglock_: ok. 23:25
fglock_ about $i+2, yes, we can overload Int to do that
but overloading objects is even slower than accessing object methods, I think 23:26
putter leo__: yes. but the current puzzling was over the perl backend, rather than the parrot backend... so were stuck with what perl can/cant do. 23:27
leo__ ah ok - sorry for the interruption
putter fglock_: yes, it is. oh well. it hopefully wont harm the p6 side, and it will allow folks to easily use p6 in their p5 programs, so... 23:28
leo__: oh, no problem. just took me a while to think "hmm, maybe it _wasnt_ humor..." ;-) 23:30
fglock_ putter: actually people would use $i = Scalar->new(3) - that's a p6 scalar holding an unboxed value 23:31
putter what's $i->ref ? 23:33
fglock_ that's the class $i belongs to
like ref($i), but in p6 you get the Class object
putter so in perl5, print Scalar->new(3)->ref(); will say... 'Int'? 23:34
'int'? 23:35
fglock_ no, it says 'Scalar' when the content is untyped (but I'm not sure if this is correct)
putter can one still say Int->new(3) when one really wants an Int? 23:36
fglock_ sure
putter k
fglock_ $i = Scalar->new( Int->new(3) ) - a Scalar holding an Int 23:37
putter Scalar->new(Int->new(3))->ref() # 'Int' yes?
fglock_ yes
putter k
sounds like a plan 23:38
I wonder how we handle the p5 and p6 namespaces. Here it feels like they want to be unified, but that causes collisions. 23:40
Maybe P6:: on the p5 side and P5:: on the p6 side?
fglock_ I should be using Perl6::Value::Int, but that's a lot to type
putter (to get to the other one) so class P5::Foo {} declares a p5 package Foo. 23:41
Oh, yeah. Can p5 use Frotz qw(Int Str); ? 23:42
use Perl6::Easily qw(Int Str); 23:43
theorbtwo If you're going to do that, make it Int(3) instead of Int->new(3).
fglock_ Perl6::Value::Int could export $Int so that you call $Int->new(3)
theorbtwo (Though I suppose there's some int/Int confusion for some people.)
Or it could just export &Int. 23:44
Function call overhead is a lot lower then method call overhead.
fglock_ theorbtwo: looks like a good idea
theorbtwo I'd still much rather the argument unpacker did it itself, though. 23:45
fglock_ theorbtwo: I think the idea is to not need an unpacker 23:46
theorbtwo Um, how can you not need an unpacker? 23:47
What else lets you call sub foo(Int $a, Int $b) as foo({b=>12, a=>3}) ? 23:48
theorbtwo shrugs... it's bedtime. 23:50
fglock_ (just thinking aloud) the actual foo is a p6 Sub object. when you call p5 foo() you are calling a wrapper
putter: do you think this make sense? 23:52
leo__ thinks that with current p6 design most of HLL interop doesn't work out of the box 23:56
fglock_ dinner time 23:58

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!