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.
jasho pugs fedora binaries where? 03:23
xinming jasho: hmm, I think you have to checkout the source and compile yourself 03:27
jasho xinming: it seems like perl6 is only for Windows. 03:28
xinming jasho: BTW, the machine with less than 256M, for compiling GHC and Pugs is not recommended. :-)
jasho: No, I am in Linux.
jasho xinming: I mean in terms of community. it seems to be all Windows now. 03:29
there are binaries for Windows, but you are on your own for Linux, etc.
xinming jasho: hmm, I hadn't used Fedora for over months, So I might not be able to help much,
jasho: hmm, I think most linux distro can compile ghc and pugs. 03:30
jasho: hold on.
jasho I can't find anything for Linux which is ready to go. the pugs docs say that by 6.2.0 there will be builds for various common platforms. but so far it seems only Windows. 03:31
xinming jasho: In fact, that's true, I don't know if Fedora has compiled binary, But Mandrake does. :-) 03:33
jasho I looked for Fedora and Suse. couldn't find anything. I didn't check Mandrake. I don't have it. 03:34
xinming jasho: hmm, I think, You don't need to change so much, The easiest way to compile pugs, might be, do a full install for with Fedora, And then, download ghc6.4 and pugs source, compile yourself. 03:36
s/for with/with/
jasho I think perl6 will become a Windows community if there are no builds readily available for Linux. 8( 03:37
xinming: thanks for your help.
xinming is still sleepy, But can't fall asleep, so go on reading the Synopsis.
jasho: well, I am a newbie also, that might be what I can help :-)
revdiablo jasho: I think the prevalence of windows builds are simply an indication of the fact that people using that platform don't tend to build stuff themselves 03:38
jasho: compared to the average unix user, who tends to be comfortable doing their own compiling
jasho revdiablo: but Linux is also trying to become an OS for people who just want to get stuff done. for now, that seems to mean Windows for perl6, or maybe stick with perl5. 03:39
coral whoah 03:40
linux in two windows
revdiablo jasho: pugs is hardly to the Get Stuff Done phase
jasho: it's still a developer's toy.
a mighty nice and fun toy, but a toy nonetheless
jasho revdiablo: I want to test grammars. I understand this was working already with pugs? 03:41
xinming jasho: Most basic grammar is supported. :-) There are lots things to do.
revdiablo the basic framework is there; I don't know how complete it is
jasho does this mean there is nothing even to test? maybe I should forget perl6 for my project and stay with perl5. 03:42
xinming revdiablo: IMHO, If the full OO support for pugs is finished, some( or maybe many ) modules can be started writting in "perl 6", 03:43
spinclad plenty to test. just compile and play.
revdiablo jasho: you can test it by building it
jasho: it builds easily on linux
xinming: *nod* a lot of stuff can be written in perl6 with pugs.
jasho now I think pugs is not worth it to test yet. 03:44
revdiablo don't get me wrong -- playing with pugs helps everyone, I encourage it
xinming revdiablo: But the problem is, There are "many" test exists... :-)
jasho I would play with pugs if there were binaries available for Linux. 03:45
xinming jasho: well, In fact, for Now, You try Pugs, You might get more than learning "only" the grammar. :-)
revdiablo jasho: "to test" and "get things done" are two different things. which one do you really want to do?
jasho revdiablo: I want to install binary pugs and test, to decide if perl6 is ready for my project. but I hear now pugs is not ready and not available for Linux without compile. 03:46
I think I should stay with perl5.
thanks for the help.
revdiablo jasho: I would say that is correct. I would still like you to try pugs and help us make it better 03:47
jasho make it easy for me to try. on Linux.
revdiablo I don't use any of the distros you listed. 03:48
so I can't really make a binary build for you
xinming If the money falls from sky, I would go out. But the fact is, If there is no money fall from sky, We still have to go out, the only thing important is, "when"
:-)
Now is the time to try pugs, It won't make you disappoint. :-) 03:49
jasho xinming: I wanted to try. maybe when pugs is available for my OS. 03:50
revdiablo it is, jasho
xinming jasho: In fact, Even if the pugs finished working on all "perl 6" specification. There might not be enough module for you to use...
jasho revdiablo: you said it is only a toy.
revdiablo jasho: it is available for your distro, just not in the form you want.
spinclad at this point, any binary package will be out of date quickly. parrot has debian packaging, pugs not yet (or vice versa, i forget offhand), but you would want recent packages in any case.
jasho xinming: do I need modules to use grammars?
xinming jasho: well, Linux is a toy also, Why do you use it? :-)
jasho in need modules for perl5.
xinming jasho: No, But for your project, I think There are ( "should" be ) many modules depends on other modules. 03:51
revdiablo I think I'm talking in circles now
jasho xinming: Linux with perl5 works well for me now, but I wanted to test grammars with perl6. but it seems perl6 is for Windows community now.
xinming ...
?eval "painful to go out this circle.".say 03:52
evalbotzy painful to go out this circle. bool::true
xinming jasho: There are still perl 5 modules can't co-work with pugs.
spinclad btw, grammars are not well integrated as yet. there's PGE and there's pugs, but they don't talk in both directions with each other.
xinming jasho: some are working, But some are not.
jasho xinming: yes, I should stay with perl5 for my project. 03:53
revdiablo I think jasho had a set of points he wanted to make when he came in here, and is ignoring any evidence to the contrary.
jasho spinclad: do I need modules for grammars for PGE?
revdiablo I'm done with this
spinclad not sure
jasho revdiablo: I think you helped me to decide to stay with perl5. so I don't have to test pugs now. that helps me. 03:54
I like to try myself. but no binaries, and not sure it is worth my time. you tell me that it is not ready yet, and is still for developers only. so no binaries. I will use perl5. 03:55
spinclad i think they're good points, we could welcome patches to build drop in packages, for debian at least, with each release...
xinming jasho: hmm, If you need the pugs to finish the full "perl 6" specification, you might still have to wait for 4 months, And also, for the other modules, maybe half a year, hmm, But can't assure this exactly.
spinclad (i think that's optimistic...) 03:56
xinming spinclad: pugs can use perl 5 modules. :-)
though It's a kind of pain on "speed". 03:57
spinclad sure. the full specs need a lot of work still though.
xinming spinclad: I just mean the "grammar" specification. 03:58
jasho maybe after this project we can try for pugs again.
spinclad ok, you may be closer there then.
xinming jasho: ?? what do you mean? After pugs, the perl 6 is out.
jasho I mean my project, not the pugs project. my project can't wait for pugs to finish I think. 03:59
spinclad xinming: not quite: plans are to release perl6 written in perl6, which calls for a rewrite after pugs... 04:00
xinming spinclad: hmm, well, the "perl 6"s father. :-) maybe pugs can be grand-pa... :-P 04:01
jql as soon as pugs can be used to write perl6, pugs is perl6 enough for me
xinming jql: So am I, I don't care speed much as I just use it as a toy. :-)
jql it'll be good -- our programs can only get faster 04:02
jasho my project doesn't care about speed, but I want to use perl6 syntax for advanced constructions. but I need correct results, so maybe pugs needs some debugging first before I choose perl6 for projects. 04:04
xinming jasho: you are right. :-) 04:06
pugs still can't be used for production.
spinclad that sounds like a different gate than whether there are binary packages for it. it definitely needs development on many fronts still.
jasho spinclad: I only wanted binaries for quick test. but you give me the answers already. pugs is not ready for me to test. I will use perl5. 04:07
spinclad sure. check again later, things will have gotten further. 04:08
jasho ok. thanks for the help.
luqui jasho, what is your project 04:10
it's always nice to have people using stuff in production 04:11
we get lots of bug reports that way
(and from what I've seen, no bug goes untended for more than 24 hours :-)
xinming luqui: I don't suggest to use pugs on project for now really. :-) 04:13
svnbot6 r6414 | putter++ | PIL-Run - MetaModel is now used for Sub. fglock++ 05:07
r6414 | putter++ | Macro is now the only remaining non-mm Run/Type/.
r6414 | putter++ | PrimX.pm: primitives' argument lists are now written in perl6 form.
r6414 | putter++ | crude_repl.pl: :v verbose is now even more so, including a nicely readable yaml dump of the PIL. iblech++
putter jasho:
jasho: there are binary builds of ghc available for FC from haskell.org. (as long as you dont have x86_64 - that requires 6.4.1, which hadn't been released the last time I checked). 05:08
good night all & 05:10
luqui understands type inferrence... a little bit 05:43
luqui will write a perl module that does type checking on trees to see if he really groks it 05:44
QtPlatypus Type checking? 05:46
What do you plan to do. 05:47
?
stevan luqui: what about Devel::TypeCheck? 05:49
or is it not the same thing you are talking about
luqui That's the new inferrence module for perl 5?
stevan yes sir
luqui that's not what I'm talking about
Perl 6's type inferrence must be much more general than that
stevan then you sir, are talking crazy :P 05:50
can you elaborate?
luqui I'm just going to let the user specify a tree structure and then try to fill it in
stevan type checking on trees?
luqui such as, oh, say, PIL
With typed nodes
stevan ah
luqui and it will check that everything makes sense as you fill it in
stevan so you are not talking about making a "type tree" and checking types with that
luqui I'm not sure what you mean by "type tree" 05:51
stevan testing derived types and such
luqui oh I will be testing derived types and such
It's pretty easy to do -- you don't ever need to look outside the types that were actually mentioned
It's just a little expensive (n^2 in the number of type variables)
stevan no I mean making a tree of types like (Any Node(Scalar), Node(Int)) stuff like that
luqui you mean union types 05:52
?
stevan is going to stop asking and wait for luqui to write some code :)
luqui okay
I have a five hour gap at school tomorrow
that's probably when it will get done
bedtime now 05:53
stevan I think we are probably talking about 2 diff things, but not in the same dialect
luqui probably
good night
luqui sleep &
stevan bah,.. for a college student you are a wussy
it is early still
luqui wake
luqui smacks stevan
stevan no no,.. you need your beauty sleep
luqui tomorrow is the second day of classes
I'd really rather not sleep through that one 05:54
luqui puts in earplugs and sleeps &
stevan :)
nite
stevan cracks open MetaModel 2.0 and gets to work
coral oo, unaffiliated 06:03
geoffb stevan, what is MM 2.0? (Sorry, been away for a while because of RL) 06:22
stevan geoffb: 2.0 is a better tasting metamodel with only half the calories of 1.0 06:31
geoffb: it is based on MiniMetaModel 06:32
geoffb Will it replace MM 1 everywhere?
stevan I am trying to clean things up, and be as minimalist as possible
geoffb: eventually, yes
I am trying to keep the changes to the user-facing API to a minimum 06:33
geoffb nodnod
stevan just re-write the core
geoffb How up to date is STATUS wrt your stuff?
stevan no idea
geoffb ah
stevan I have been totally swamped with $work
geoffb Understood, I've been totally swamped with $life
stevan $life++ 06:34
but $work--
QtPlatypus none($work)--
stevan basically this is my "relaxation" after about 14+ hours of work
geoffb Hmmm, maybe tomorrow or so I need to do another round of asking people for status and updating docs to match
good choice, stevan :-)
stevan :)
geoffb is relaxing by writing notes on optimization strategies 06:35
stevan :D
last night I read papers on implementing subtypes/generics with metamodels
geoffb :-)
stevan and got inspired,.. so I wanted to finish 2.0 enough to try ideas 06:36
I probably actually should be doing $work
geoffb recognizes that behavior
(having been there many times)
bah, must run, thanks for the info, stevan 06:37
stevan geoffb: enjoy :)
06:57 _jql is now known as jql
svnbot6 r6415 | Stevan++ | Perl6::MetaModel 2.0 - 07:45
r6415 | Stevan++ | * $::Object is now defined, but does not yet work
r6415 | Stevan++ | correctly. I need to figure out the best way to
r6415 | Stevan++ | deal with class methods and meta methods still.
r6415 | Stevan++ | * a few more test files, however not many tests
r6415 | Stevan++ | can be written until the class/meta method issue
r6415 | Stevan++ | is resolved.
nothingmuch morning 07:58
QtPlatypus Mornig. 08:05
nothingmuch concludes from the 1.e5 thread that his solution is the best 08:13
anyone want to implement an AUTOMETHOD that does that?
it should be something like 'method AUTOMETHOD (Int $int:) { parse_as_float("$int$AUTOLOAD") }' where parse_as_float is one of haskell's 'read's 08:16
ah, oops 08:17
s/AUTOLOAD/_/; 08:18
as per s10
nothingmuch is off to clean up the house 08:21
xinming sub head(*$head, *@tail) { return $head } 10:01
hmm, I can understand this one, What about this example please?
sub head( *$head, $special, *@tail ) { return $special; ) 10:02
sub head( *$head, $special, *@tail ) { return $special; };
ouch... I think I mislead myself... 10:03
svnbot6 r6416 | rgs++ | * a note on split STRING 10:53
r6416 | rgs++ | * shortly mention =$fh
r6417 | yiyihu++ | Add the test for S06 List parameters, which will test slurpy scalar, slurpy array, and slurpy hash. 11:32
QtPlatypus What is the meaning of a slurpy scalar? 11:33
xinming Slurpy scalar parameters capture what would otherwise be the first elements of the variadic array: 11:34
hmm, QtPlatypus Sorry for my confusion, I think I should say ("slurpy scalar"| "slurpy array"|"slurpy hash") parameters 11:35
s/my/the/
xinming is considering to go to Junior School for learning English. 11:36
QtPlatypus nods. 11:38
xinming seen autrijus 12:50
jabbot xinming: autrijus was seen 1 days 9 hours 16 minutes 18 seconds ago
nothingmuch hola 12:57
Limbic_Region como esta ustede? 13:02
svnbot6 r6418 | fglock++ | * perl5/ Code - TODO update 13:28
itz print @foo.len(); # this returns No compatible subroutine found: "&len" with the SVN version? 13:36
nothingmuch ?eval my @foo = (1, 2, 3); @foo.length
evalbotzy Error: No compatible subroutine found: "&length"
nothingmuch ?eval my @foo = (1, 2, 3); @foo.elems
evalbotzy 3
nothingmuch this is an explicit distinction 13:37
length is ambiguous
otoh the number of elements is unambiguous
for strings too - let me try to remember:
?eval my $foo = "abc"; $foo.chars
evalbotzy 3
nothingmuch ?eval my $foo = "abc"; $foo.graphemes
evalbotzy Error: No compatible subroutine found: "&graphemes"
nothingmuch ?eval my $foo = "abc"; $foo.bytes
evalbotzy 3
nothingmuch ?eval my $foo = "לה לה לה"; $foo.bytes
evalbotzy 3
nothingmuch ?eval my $foo = "לה לה לה"; $foo.chars 13:38
evalbotzy 3
nothingmuch that's bollocks
?eval "fooooo".chars
evalbotzy 6
nothingmuch ?eval "שלום".chars
evalbotzy 1
nothingmuch anyway, nevermind, appearantly evalbot is not grokking the unicode
in my terminal it works correctly 13:39
as for graphemes, Ʊ is one char, but two graphemes, and i don't know how many bytes (encoding dependant) 13:41
?eval 'Ʊ'.chars
evalbotzy 1
nothingmuch ?eval 'Ʊ'.craphs 13:42
evalbotzy Error: No compatible subroutine found: "&craphs"
nothingmuch ?eval 'Ʊ'.graphs
evalbotzy 1
nothingmuch itz: fair enough? 13:44
iblech autrijus: Could you please explain what isNamed indicates? sub foo ($a) -- isNamed is 1, sub foo (+$a) -- isNamed is 0 13:46
nothingmuch iblech: named vs positional params 13:47
iblech nothingmuch: Right, but why is isNamed 1 for a positional?
nothingmuch because it can be named
wait
ughhh, confusion! 13:48
itz nothingmuch: you lost me there :) is @foo.len not valid syntax? I thought it worked in the past?
nothingmuch itz: there is no 'len' method for arrays
iblech nothingmuch: Thought so too. But then I wondered why isNamed is 0 for a name-only param...
nothingmuch and not even 'length' like there should have been
iblech itz: It is valid syntax, though
nothingmuch itz: the error is not a syntax one, it's a "can't find such a method" one 13:49
iblech ?eval "a string can't be .summed".sum
evalbotzy 0.0
itz was it ever in pugs or did I dream it?
iblech d'oh
itz: You must have dreamed it :)
nothingmuch itz: there are no more abbrv func names in p'l 6 13:50
itz hmm I must be confusing p6 with ruby or something
nothingmuch so it was never 'len' but rather 'length', but even that doesn't make sense, because arrays don't have a length
they have a size in a certain dimension, like the number of elements
uh 13:51
itz I want "elems" then?
nothingmuch damnit, shouldn't it be @array.elements ?
i thought they were dropping all the gethostbynm type crap
iblech nothingmuch: Huffmann, etc. chars, graphs, bytes, elems
nothingmuch are there full sized aliases?
iblech itz: If you want the number of elements in an array, yes -- you can use @array.elems or +@array 13:52
nothingmuch: our &elements ::= &elems?
nothingmuch iblech: i mean by default
but yes
?eval my @foo = <1 2 3>; @foo.elements 13:53
evalbotzy Error: No compatible subroutine found: "&elements"
nothingmuch ?eval my @foo = <1 2 3>; @foo.elems
evalbotzy 3
nothingmuch also 'graphs' is not as clear as 'graphemes
and shouldn't the tilde on Ʊ be a different grapheme? 13:54
iblech .graphs is currently the same as .chars 13:55
nothingmuch ah
btw, should we also have phonemes?
iblech doesn't have a strong opinion, as long as he has .bytes and .chars 13:57
nothingmuch en.wikipedia.org/wiki/Terminology_i...aphonomics
well, .chars is sometimes diff
it's the most common usage
and .bytes is totally orthogonal
in hebrew, for example, we have nikud 13:58
we have very few vowels
and most motions are implicit, when there is no nikud present
svnbot6 r6419 | fglock++ | * perl5/ List 14:13
r6419 | fglock++ | - implemented ('a' x 100) style lists
r6419 | fglock++ | this can be used to create sparse arrays like: [ 1, undef x 1_000_000, 2 ]
r6420 | iblech++ | * Usual svn props and EOLs at EOFs.
r6420 | iblech++ | * t/builtins/lists/: map.t, sort.t, grep.t, sum.t, reduce.t:
r6420 | iblech++ | Test that 42.sum, 42.map, etc. do not work (but (42,).sum, etc.) should work.
r6420 | iblech++ | * t/subroutines/slurpy_param.t: Fixed some minor details, changed the expected
r6420 | iblech++ | results of some tests to make more sense, started a p6l thread concerning
r6420 | iblech++ | the test and skipped all tests ("semantics not confirmed by p6l") until p6l
r6420 | iblech++ | replies.
iblech fglock: undef x 1_000_000 should not create a list containing 1_000_000 items, but a string of length 1_000_000: 14:14
?eval "42" x 5
evalbotzy '4242424242'
iblech ?eval "42" xx 5
evalbotzy ('42', '42', '42', '42', '42')
lhooq ?eval my $i = 0; sub i { $i++ }; i() for (1..5); 14:15
evalbotzy undef
lhooq ?eval my $i = 0; sub i { $i++ }; (i, i, i, i, i); 14:16
evalbotzy (0, 1, 2, 3, 4)
lhooq ?eval my $i = 0; sub i { $i++ }; i xx 5;
evalbotzy Error: No compatible subroutine found: "&xx"
lhooq ?eval my $i = 0; sub i { $i++ }; i() xx 5;
evalbotzy (0, 0, 0, 0, 0)
lhooq eager!
iblech lhooq: Right. You need something like (1..5).map:{ i() } 14:17
But IIRC luqui (or was it somebody else?) wants to change that to { i() } XX 5
s/change that to/make the following possible/
lhooq with uppercase XX? 14:18
iblech Yep.
But the proposal hasn't been accepted by @Larry
PerlJam map is easier to type 14:19
nothingmuch iblech: i think it was Juerd
iblech nothingmuch: Ah, thanks
PerlJam I'd take i() xxx 5 before I'd like XX 14:20
nothingmuch &:infix:{'x' x 3} 14:21
nothingmuch thinks it should have been 'x' instead of 'xx' and '~x' instead of 'x'
PerlJam nothingmuch: maybe. We already have that consistency elsewhere, so it makes sense with x I guess. 14:22
nothingmuch PerlJam: did you hear my rant about == ?
PerlJam nope
nothingmuch i think ~== and +== should replace 'eq' and '==' respectively
where '==' is generic equality
PerlJam Hmm. 14:23
nothingmuch so "foo" == "bar";
and "foo" == 1
they work
PerlJam What does "work" mean?
nothingmuch it's stringwise in these cases, because coercion works that way better than numification works
otoh 10 == 10 and 10 == 10.0 is also good
although "10.0" == 10 14:24
that is bad, because it's ~==
for that you need to say "10.0" +== 10, and instead of the right side being stringified, the left side is numified
autrijus hates it
but my argument is that i need generic equality much more often than I need "eq" or "=="
PerlJam nothingmuch: I'll tell you that I don't believe in "generic equality" 14:25
You have to know the types of the things you're comparing at some level.
nothingmuch it's not generic equality in the sense that it's super smart
it's a part of the Eq role
and object must implement it if it wants to
like "use overload '=='" in perl 5 14:26
PerlJam my $a = new Foo; my $b = new Bar; $a == $b # What's this? An error if both don't do the Eq role? 14:27
nothingmuch yes
unless MMD defines it otherwise
perhaps ~== is the default == 14:28
but it lets you have truely meaningful things:
xinming iblech: :-) In fact, I considered the slurpy *%hash carefully, And I think I "might" be right, So I wrote the test. :-P
nothingmuch multi sub &infix:<==> (@l, @r) { @l.elems +== @r.elems && (@l Ā»==Ā« @r) 14:29
xinming iblech: thanks for your modification.
PerlJam trigraphs are little loved though. If you could make them digraphs even they'd be more palatable.
iblech xinming: Lets wait for p6l to respond :)
xinming: But I don't think it makes much sense that you can't pass any parameters by name if there's a *%slurpy_hash 14:30
nothingmuch PerlJam: what bothers me about eq vs == is that in perl 5 the choice was mostly aesthetic 14:31
people overload == more often because it looks more like =
lhooq ?eval my @a = (); @a.push('x') for(1..3); @a; 14:32
evalbotzy ['x', 'x', 'x']
lhooq ?eval my $i = 0; sub i { $i++ }; my @a = (); @a.push(i) for(1..5); @a;
evalbotzy [0, 1, 2, 3, 4]
nothingmuch and for many things, especially structures it makes sense to overload equality (sets, arrays, junctions, etc) for reasons of efficiency as well as accuracy
further uses: 14:33
path handling: the path object's == consolidates 'dir/..' so that Path.new("foo/../bar") == Path.new("bar") 14:34
you can overload == locally for float tolerance:
my &infix:<==> (Float $l, Float $r) { abs($l - $r) < $threshold } 14:35
this doesn't interfere with +== on numbers
PerlJam nothingmuch: careful ... Larry may like your idea but use +eq and ~eq instead ;-) 14:37
nothingmuch does this seem somewhat sane?
lhooq nothingmuch: sorry, but I don't like +== for comparison, it looks too much like += 14:38
iblech nothingmuch: I agree, it's very useful. *But* I want eq to stay string eq and == to stay numeric eq. 14:39
nothingmuch: But now there's eqv :)
nothingmuch: my multi &infix:<eqv> (Float $l, Float $r) {...}
nothingmuch right, i know eqv
and I hate it
lhooq: what do you think about the other ~/+/? distinctions? 14:40
PerlJam There's too much cultural history for == or eq to go away.
nothingmuch what about the consistency? &&, || and ^^ are generalized
?&, ?|, +&, ~^, etc are specialized 14:41
broquaint ?eval class MyArray is Array { sub push { say "pushing things: {@_.perl}" } }my @a = MyArray.new; @a.push("a string"); # What am I missing? 14:42
evalbotzy 2
broquaint No, I don't think there's a 2 missing, evalbotzy. 14:43
But nice guess.
nothingmuch ?eval class MyArray is Array { sub push { method "pushing things: {@_.perl}" } }my @a = MyArray.new; @a.push("a string");
evalbotzy 2
nothingmuch ?eval class MyArray is Array { method push { say "pushing things: {@_.perl}" } }my @a = MyArray.new; @a.push("a string");
evalbotzy 2
iblech broquaint: That wouldn't work, you have to use my @a is MyArray
nothingmuch oops, wrong subs ;-)
iblech broquaint: Otherwise it's simply my @a = (MyArray.new)
nothingmuch ?eval class MyArray is Array { method push { say "pushing things: {@_.perl}" } } my @a is MyArray; @a.push("a string"); 14:44
iblech broquaint: But it's not implemented yet
evalbotzy 1
broquaint Indeed, iblech. Using := instead of = does the trick too.
?eval class MyArray is Array { sub push { "pushing things: {@_.perl}" } }my @a := MyArray.new; @a.push("a string"); # What am I missing?
evalbotzy 'pushing things: [MyArray.new();, \'a string\']'
broquaint Neat. 14:45
And sub Array::push works too, which is very cool.
xinming iblech: where can I find the contents for the `built-in sum`?
iblech xinming: IIRC src/Pugs/Prim/List.hs
xinming iblech: thanks 14:46
iblech xinming: Or perl5/PIL2JS/lib6/Prelude/JS/Array.pm if you want to see a pure-Perl 6 version of it :)
xinming Oops... Isn't there a perl version?
thanks
lhooq nothingmuch: well, I don't like the distinctions too much, especially the ? one
iblech xinming: The sum in the .hs file is in Haskell 14:47
lhooq nothingmuch: but &&, || and ^^ are different, they are always boolean in context
nothingmuch so is ==
wait
no, only == is
&& and || and ^^ are not boolean at all 14:48
or do you mean the context they force on their args?
fglock stevan: ping
lhooq yes, I mean the context
xinming iblech: yes, In fact, I want to know If I can do as the "same" as the built-in ones. :-)
nothingmuch sub &infix:<||> ($l, $r) { return $l if ?$l; return $r }
i always saw them as passthrough 14:49
lhooq there is an explicit ? there
nothingmuch yes
lhooq so they "force" boolean context somehow
nothingmuch if $l in boolean context is true, then return $l, otherwise return $r
$r is never boolean
unless they stack
lhooq well, ok, I think you understood what I'm talking about 14:50
iblech xinming: Not sure I've understood you correctly, but yes, you can override the builtin implementantions. I.e. even sub say (*@args) {...} is fine
lhooq you don't really need something like +||, +&& etc.
QtPlatypus lhooq: You need it to distingiush between +|| and ~|| 14:51
lhooq QtPlatypus: you _don't_ _need_ 14:52
14:52 autark is now known as jp-autark
lhooq the distinction is +(A || B) vs. ~(A || B), maybe 14:52
QtPlatypus lhooq: So using context to deside which one? 14:53
broquaint Let's just drop boolean operators and be done with it.
lhooq QtPlatypus: no, because the || is always the same
nothingmuch how is operator precedence specified for 'sub &infix:<blah>' ?
lhooq the _external_ context may be different, but || doesn't care
nothingmuch or which synopsis is that in?
QtPlatypus Hold on || is bitwise or isn't it? 14:54
lhooq (doesn't care in the sense that it is "passthrough", as nothingmuch noted)
nothingmuch or (~A || ~B)
xinming iblech: what I mean, is, I want to know, If I catch the "right" idea of writing perl 6 program. :-) the built-in certainly would be the best example for me to follow.
iblech: by the way, did you read the Synopsis?
nothingmuch the passthrough-ness of &&, and && should apply to ==, IMHO
xinming the S06
nothingmuch ^^ returns Bool anyway
so == is not in a different ballpark 14:55
xinming iblech: It was updated this "morning" (here now is evening, maybe midnight)
lhooq nothingmuch: I disagree
nothingmuch lhooq: i know you don
t
it's just my opinion =) 14:56
lhooq I don't see _how_ == is supposed to passthrough
xinming away to go out to take something to eat.
bbl
iblech xinming: Of course, I have read all AES several times :)
xinming: Right, but the update didn't effect the section about slurpies, right?
lhooq variables in Perl are not strong-typed, so unless you use == or eq, you can't possibly know what to numify and what to stringify 14:57
iblech xinming: I've thought of slurpy hashes similar as of slurpy arrays -- they slurp all *remaining* arguments (positionals or pairs)
nothingmuch lhooq: it's not passthrough, but it's generic 14:58
lhooq define "generic"
nothingmuch does anybody remember how you say what an &infix's precedence is? i can't find it
lhooq: multi sub &infix:<==> (@l, @r --> Bool) { @l.elems +== @r.elems and @l.elems }
uh, crap, that's 50%
lhooq: multi sub &infix:<==> (@l, @r --> Bool) { @l.elems +== @r.elems and @l Ā»==Ā« @r } 14:59
lhooq nothingmuch: is tighter(...) or something like that
nothingmuch lhooq: thanks
lhooq nothingmuch: this can't be what you want 15:01
nothingmuch it is
lhooq nothingmuch: if I want a == to compare each element in two arrays, I prefer to explicitly use >>...<<
nothingmuch what if it's a generic collection? like a tree? 15:02
lhooq eg. when I do something that could bog down my CPU like a slug, I don't want it to be _so_ easy to do
nothingmuch do you need to define multi &circumfix:{'>>', '<<'} (Code $op, Tree $tree) { .... } ? 15:03
lhooq: why bog down?
stringification of an array takes just as long
and @array == @other is very confusing
that is really (@array.elems == @array.elems)
newbies will weep 15:04
lhooq but their CPUs will rejoice :-)
nothingmuch (@array +== @other) is much more "clear cut"
but it'll be wrong code
and infix == is more "optimized" than >>==<< because it checks @l.elems +== @r.elems first, too =)
and in fact, i think that >>==<< can be misleading if the shape of two arrays isn't the same
are undefs substituted in the shorter list? 15:05
lhooq then it is (@array == @other && @array >>==<< @other)
nothingmuch i know, but i think that's aweful
what it is now is much better: 15:06
if (@array eqv @other)
but that still kinda sucks
IMHO at least
iblech nothingmuch: That won't do what you think it does, I think.
nothingmuch: eqv expects two scalars, so that really is \@array eqv \@other
nothingmuch ugh
lhooq shouldn't that be @array ~~ @other?
nothingmuch is it multi? 15:07
lhooq: ~~'s conveyed meaning is completely different, IMHO
iblech nothingmuch: Then these refs (in the absense of some rebinding) don't point to the same container, so it'll always be false
lhooq: Right, @array ~~ @other should work, I think
nothingmuch my @array = <a b c>, my @other = (rx/a/, rx/[abc]/, rx/\w+/); @array ~~ @other;
*BOOM*
it's way too smart 15:08
which is good for most things
but not for some things
lhooq nothingmuch: anyway, your example still doesn't answer this question: is each element in @l and @r compared numerically or stringish?
nothingmuch lhooq: that depends on each pair of elements, and the MMD that matches
=)
iblech nothingmuch: Hm, to the problem is that it recurses too much... @array ~~ @other :recurse(0)
nothingmuch iblech: i beg to differ 15:09
~~ is just as recursive, even if it's implicit
Array Array arrays are identical match if $_ Ā»~~Ā« $x
from s04
iblech nothingmuch: Right, I meant that by using an adverbial modifier you could change that behaviour 15:10
nothingmuch in a sense I think that &infix:<==> should have the same "recursive" semantics as &infix:<~~> in the prelude
oh
it's just that == is applied instead of ~~ to the elements
so why not have @array == @other :recurse(0);
iblech Because == means number, and I doubt that this changed 15:11
(I don't want it to change, too :)) 15:12
But that's what ~~, paired with modifiers, is there for, I think
nothingmuch ick 15:13
i'm mailing to p6l
please make sure to voice your protest
iblech I will :)
nothingmuch &infix:<~~>.add_variant(sub (Code &op, Array @l, Array @r) { @l >>&op<< @r }.assuming(&infix:<~~>)) 15:15
iblech ooh
but
how do you specify your own comparator, given that ~~ is not ternary?
nothingmuch rephrase please
iblech I mean, ~~ takes only two arguments, LHS and RHS 15:16
nothingmuch right
iblech But your snippet adds a third param
nothingmuch so does ==
iblech Right.
nothingmuch assuming =)
iblech Right.
But:
If I don't want to use .assuming 15:17
But if I want to override it only for one single call
nothingmuch then it's just a weird variant
that takes an entirely different set of params
iblech I.e. something along the lines of @foo ~~:{ $^a eq $^b } @bar
(But this look very ugly)
nothingmuch huh? 15:18
iblech Hm, @foo ~~ @bar :comparator{ $^a eq $^b }
nothingmuch ah
you just take the generic comparator generator from the prelude, and apply it to your own op
in my email it will be cleared 15:19
iblech ok, /me waits :)
nothingmuch sent
iblech (I like that general idea, as far as I've understood it, BTW)
nothingmuch yes, i like it too =) 15:20
btw, i've changed my mind slightly
=='s affinity is towards multi sub &infix:<==> (Num $l, Num $l) { ... }
lhooq nothingmuch: anyway, I think that "generic equality" a) causes more problems than it tries to solve and b) is too CPU-intensive to implement
nothingmuch so "10.0" == 10 is backwards compatible
lhooq: most of perl 6 is too CPU intensive already 15:21
lhooq so you fight fire with fire? :-)
nothingmuch as for a) we're trying to clean this up... MMD is helping, but I definately don't want 'use overload' all over again
lhooq: no.. See also Psyco, type inferencing, and what not to make it "just work" 15:22
as well as the fact that despite this we still have >><< and [] which are *very* dangerous in that respect, but so useful we can't give uup
the caveat in my proposal is that (@a == @b) cannot be translated as is
and makes things slower for that case
but when I want to compare the number of elements we have all sorts of disambiguations anyway.. .for example, it's not @a.length like in other programming languages, it's @a.elems 15:23
so in that sense I think Perl 6 is progressing to another plane of usefulness
and correctnes
s
and in that sense we should strive to give bazookas to shoot your foot with
lhooq but you're giving a bazooka that looks like a lollipop 15:24
nothingmuch because in the long run, as people become adept in perl this will pay off, as a programmer time optimization, and that's what really matters at the end of the day
i agree
but i think perl 6 is far enough from history to allow us to make it look like a lollipop
lhooq well
nothingmuch because it's a very sweet bazooka 15:25
lhooq I remember the first time I saw a regexp
nothingmuch as in totally cool
lhooq I thought "bazooka!"
nothingmuch (ref: www.realultimatepower.net/)
lhooq now I master it, and can shoot with it
nothingmuch iblech: extend_comparators on the bottom 15:26
These guys are cool; and by cool, I mean totally sweet 15:27
i got it backwards =(
anyway, perl 6 is all about real ultimate power... think about it:
1990s: regexes... perl 6: full blown parser engine
1990s: for loops... perl 6: vectorizations
1990s: overloading... perl 6: higher order infix subs, and MMD dispatch make it all just functions 15:28
1990s: boolean logic... perl 6: junctive types encapsulate logic in data
(slow! autohtreading! dangerous! but still in there!) 15:29
anyway, i have to get back to house work 15:30
i have to finish putting the books back on the shelf
and vaccum the cement dust from my room
nothingmuch thinks about restarting fryTunes in perl 6 15:35
the metamodel for music data is much easier to do with such a tough object system
nothingmuch will think about it while cleaning... Ciao!
iblech nothingmuch: Could you elaborate a bit on extend_comparators later?
nothingmuch iblech: still here, was reading mail 15:36
iblech: are you asking? 15:37
iblech nothingmuch: Yep :)
nothingmuch okay =)
btw, is control over whether --> is used for inferencing or not in the lexical scope? 15:38
(strong inferrencing)
iblech Huh? Can it be used for something different that inferencing?
nothingmuch (we should always have [partial] static inferencing)
well, if &infix:<==> (.... --> Bool) (regardless of whether it's generic or numerical) 15:39
then can I use it and ignore the fact that it wants to return a Bool ?
or do I have to use it in a type safe way? (BAD BAD BAD)
iblech Ah, I understand
nothingmuch i'd like to say "use types 'inferred'"
and that way all functions can have (-->) 15:40
or maybe that's just moot
anyway, i'd like to be able to make use of any library that says what it returns in a type inferred way, and in an unsafe way
and i'd like types to be inferred in both cases, and only when there is a conflict for the distinction to shine: safe - compile time error, unsafe - coercion compiled in 15:41
iblech I think that will definitely work, given that we're talking about Perl (and not Haskell) :)
You always have coercion, I think
I.e. "1" + 1 will always work
nothingmuch should it always work? 15:42
iblech Hm, I think so
Because
nothingmuch even if i "use types 'nazi'"?
iblech There exists a coerce:<as>(Str $str, Num ::to)
I think you need to talk with autrijus about that :)
nothingmuch "use types 'inferred' :forbid<coercion>"
there, problem solved =) 15:43
iblech Right, something along this line
nothingmuch i think everyone agrees that it's time for a language that is as type safe as the programmer wants it to be =)
so what is your question about extend_comparators?
btw, is the sigil & just a macro for 'Code $' ? 15:44
i think that is pretty cool, except for 'my $var; my &code'; 15:45
uh, i meant 'my $foo; my &foo;
QtPlatypus nothingmuch: Not in all cases.
nothingmuch QtPlatypus: morally =)
QtPlatypus sub foo {...}; &foo; 15:46
nothingmuch iblech: faster!
think of sub $name as just a macro for BEGIN { &symbol = $parsed }
if it's not preceeded by 'my' or 'our' it gives our
iblech nothingmuch: Sorry, was afk for a moment
nothingmuch: Right, autrijus and I want to change it that way
nothingmuch: But the sigil is till & 15:47
s/till/still/
QtPlatypus nothingmuch: I like that.
nothingmuch sorry
iblech So there's no conflict between $var and &var
nothingmuch there is the macro 'sub'
and the function 'sub'
which is just like 'id'
but also takes parameters
(a prototype)
beh, it's a bit more difficult than that, obviously
but in a sense the declaration 'sub' is a macro just like 'use' is a macro 15:48
iblech nothingmuch: WRT extend_comparators, could you give me some code lines to illustrate your idea? :)
nothingmuch iblech: think of that as the part of the prelude that defined ~~
iblech ok
nothingmuch there are already &infix:<~~> ($_, $code)
and so on and so forth
and then more generic comparisons regarding collections are generated 15:49
like $item ~~ %hash;
or @array ~~ @array
the rules for how ~~ matches WRT to collections can be applied to == too
iblech ok
nothingmuch btw, can I define aggregate functions like this? 15:50
&foo := for.assuming(:block -> $x, $y { ... });
iblech &foo := &statement_control:<for>.assuming(block => -> $x, $y {...}) should work 15:51
nothingmuch woot!
iblech: is the extend_comparators issue cleared up?
because I really need to become useful now
holy shit! leaf.dragonflybsd.org/mailarchive/s...00028.html 15:52
iblech nothingmuch: No, but I have to think a bit longer about that and I don't want to keep you from becoming useful :)
nothingmuch iblech: then reply to the p6l post please =)
iblech nothingmuch: Yep, will write it in a minute :) 15:53
nothingmuch ciao!
iblech bye nothingmuch :)
xinming iblech xinming: I've thought of slurpy hashes similar as of slurpy arrays -- they slurp all *remaining* arguments (positionals or pairs) 16:00
iblech: then, This would cause Error.
the positional argument is normally required, right?
slurpy *%hash slurped it, Then, You didn't specify the $positional, It would be a compile time error. 16:01
iblech: hmm, But what I do not agree with is if there is a ( pair ) need to pass to the hash, but not the keyword is the same as positional's name, then, It's an error. just like 16:05
sub fun( $a, *%h ) { ... }; In my opinion, If there is a slurpy hash, pass the arguments with "pair-form" should be disabled. 16:06
iblech: Sorry, I went out for the "supper" :-)
iblech: In fact, I even thought that if @Larry hesitates to choose in these two choices. 16:08
mambotech hi, can anyone here tell me what is missing please Can't locate Config/IniFiles.pm in @INC (@INC contains: /usr/local/lib/perl5/site_perl/5.8.7/mach /usr/local/lib/perl5/site_perl/5.8.7 /usr/local/lib/perl5/site_perl /usr/local/lib/perl5/5.8.7/BSDPAN /usr/local/lib/perl5/5.8.7/mach /usr/local/lib/perl5/5.8.7 .) at /usr/local/sbin/vuser line 9. 16:09
BEGIN failed--compilation aborted at /usr/local/sbin/vuser line 9.
xinming mambotech: How do you run pugs? 16:10
do you use make install as root?
mambotech ?
xinming mambotech: command make just compile the sources. It seems that your installation isn't complete 16:11
iblech xinming: luqui replied to my question on p6l 16:16
mambotech: I think you want #perl, not #perl6, #perl6 is only about Perl 6
xinming: The point is, that slurpy hashes only slurp remaining pairs, i.e. only pairs which were *not* already consumed by naming other positionals 16:18
I.e. sub foo (*%hash, *@rest) -- %hash always gets all pairs
sub foo ($x, *%hash, *@rest) -- foo(bar => "baz") -- %hash gets (bar => "baz") 16:19
sub foo ($x, *%hash, *@rest) -- foo(x => "baz") -- %hash gets (), $x gets "baz"
xinming iblech: hmm, Ok, I think your right. So, will there be a "shadow-copy" of x => "baz" in %hash? 16:24
iblech xinming: luqui wasn't sure 16:25
xinming: But currently, not
xinming iblech: :-) I think If we need to explicit specify the pair to hash, we could use a keyword to explicitly specify it. 16:26
iblech: I would change the test later. :-)
jql goes even farther off the deep end... foo("baz") should imply %hash<x> eq 'baz'
positionals should just default the key-name, IMO 16:27
iblech xinming: Or we could simply use sub foo (*%hash), which will unconditionally slurp all pairs
xinming iblech: hmm, But that lack the flexibily of foo "bar", "this", ( x => 'y' ) 16:28
why @Larry made spcify the *%hash might because of, Some of the arguments might not be needed, And the optional 'named-param' can't handle all the situation. 16:30
hmm, That's just my opinion. :-)
iblech Right, then you'd have to use sub foo (*%hash, *@rest)
xinming iblech: The problem also exist, *@rest can't do "type-checking" :-) 16:31
s/exist/exists/
eg: sub foo ( Str $str, *%h ) { ... }, If you use @rest, Then, The first param can be every thing. 16:32
iblech Right. In this case, simply don't use *@rest :)
I.e., sub foo ( Str $str, *%h ) { ... } is perfectly fine 16:33
xinming hmm, I think, sub foo( [Type $var, ... *@rest ], *%hash ); might help.
iblech oh, right! :)
xinming iblech: well, these example doesn't cover all the cases. And @Larry wish to cover them all, to obey m0tto, "TMTOWTDIT" :-) 16:34
seen autrijus
jabbot xinming: autrijus was seen 1 days 13 hours 20 seconds ago
xinming hmm, Where did autrijus go?
xinming is missing him... 16:35
iblech xinming: Right, TIMTOWTDI++ :) 16:36
He's deep into $work, I think
deadline, etc.
xinming hopes he can be more helpful in the future.
fglock iblech: I'm reading PIL2JS - trying to see if we can unify some code 16:40
iblech fglock: ooh (unifying code)++ 16:43
fglock: Especially lib6/Prelude/JS/ControlFlow.pm could probably share some code with PIL-Run
E.g. map, for, while, until, etc. all use loop 16:44
fglock why is it written in p5 (not p6)?
iblech It is written in P6 :)
With some JS code, of course 16:45
fglock I mean, the main program, not the library 16:46
iblech ah. 16:47
Speed, I think
fglock goes away for a while 16:50
xinming ?eval ( a => 'abc', b => 'xyz' ).key 16:55
evalbotzy ('a', 'abc')
xinming ?eval ( a => 'abc', b => 'xyz' ).keys 16:56
evalbotzy (0, 1)
xinming hmm... :-)
iblech m19s28.vlinux.de/iblech/stuff/not_p...smoke.html # New PIL2JS smoke, 65.59% ok, ~~40min
Hm, seems to be a bug 16:57
xinming ?eval ( a => 'abc', b => 'xyz', c => 'ddd' ).keys
evalbotzy (0, 1, 2)
iblech (the first one)
?eval (a => "abc").key # should give "a"
evalbotzy 'a'
iblech ?eval (a => "abc", bar => "baz").key # should be an error
evalbotzy ('a', 'abc')
iblech as (a => "abc", bar => "baz") is not a Pair
food &
xinming ?eval { a => 'abc', b => 'xyz', c => 'ddd' }.values 16:58
evalbotzy ('abc', 'xyz', 'ddd')
xinming ?eval { a => 'abc', b => 'xyz', c => 'ddd' }.keys
evalbotzy ('a', 'b', 'c')
ods15 hi nothingmuch 17:06
nothingmuch: do you use your laptop as your main box?
nothingmuch i have two "main boxes" in this sense 17:07
the laptop does everything PCish: text editing, movie viewing, browsing
ods15 ircing?
nothingmuch and I also have pasta, which is headless... it does apache, postfix, dhcp, dns, routing, etc
irc is on the laptop
i tried irc in screen on pasta for a while 17:08
but then I realized that I don't really need to be available 24/7 through backlogs
I do read my email on pasta, via mutt inside screen
muttng actually
but that is the only regular use for pasta that isn't implicit
ods15 muttng? 17:09
nothingmuch mutt + many patches (maildir header caching, side bar, nntp, pgp password forgetter, and more and more and more)
iblechbot: :comparator is trivial, read mail 17:10
wolverian I use muttng just for the caching :)
nothingmuch it's wonderful
wolverian yes.
nothingmuch i switched to reiserfs when i made the move towards LVM + software raid
so that I could have online resizing of partitions
wolverian online? 17:11
nothingmuch so old mutt became unbearably slow (maildirs + reiser + nocaching)
wolverian: no unmount
you can only grow though
wolverian hm. I thought rfs is supposed to be fast in a scenario like that?
nothingmuch i thought so too, but it's not fast, it's just size efficient, i guess
i have notail now
maybe I should copy my maildir and swap 17:12
anyway, muttng and screen were very good optimizations
ods15 caching?
as in, when you open a really big maildir?
nothingmuch ods15: instead of reading each maildir file it keeps an index of the headers
17:12 Maddingue__ is now known as Maddingue
nothingmuch each message file in a maildir, that is 17:12
ods15 doesn't mutt already support this? 17:13
nothingmuch yes, but with a patch that isn't in mainline
ods15 header_cache Type: path Default: ""
The header_cache variable points to the header cache database. If header_cache points to a directory it will contain a header cache database per folder. If header_cache points to a file that file will be a single global header cache. By default it is unset and so no header caching will be used.
thats from man muttrc, and afaik i'm using regular mutt
nothingmuch ods15: it may be your distro being nice to you
ods15 20:13) ods15@linux15:~ $ mutt -v 17:14
Mutt 1.5.9i (2005-03-13)
no sidebar that i know of btw
wolverian hmm. I have that too. odd. 17:15
nothingmuch the sidebar is actually quite nice, b ut I dont really use it
wolverian: that's in muttng
wolverian it might have come with 1.5.9i, I guess.
nothingmuch, I have that in man muttrc
nothingmuch oh
wolverian (debian sid)
nothingmuch the headercache, not the sidebar
wolverian right.
nothingmuch i used mutt + headercache for a long time though, and i think muttng felt faster 17:16
ods15 whats the sidebar
wolverian I'll try it.
nothingmuch so I think it caches on another level
ods15 folder list?
nothingmuch ods15: a mailbox list on the side
that shows you counts
ods15 ah, one of the things i'm MOST missing in mutt
but tbh doesn't sound "quite" right in console
nothingmuch tbh?
ods15 even with high res framebuffer, console isn't THAT big that you can stuff a sidebar too :/
to be honest
nothingmuch ods15: you toggle it 17:17
default binding to toggle is 'b'
ods15 hmm
wolverian ah. mutt doesn't have header_cache_compress
ods15 btw, is there any SIMPLE mime command line tool?
wolverian not that that should affect speed, at least negatively...
ods15 and btw, how d you deal with hebrew in mutt?
nothingmuch ods15: i don't deal with hebrew too well 17:18
the terminal is utf-8, which is Good Enoughā„¢
but i don't write mail
in theory it's possible, with vim supporting it
ods15 i don't deal with it at all.. if it's something i REALLY want to read, i open it with kmail
justathe1ry removes it
I'm an Emacs guy, after all.
17:18 justathe1ry is now known as justatheory
ods15 i find the emacs/vim war extremely funny, because i personally think both suck :P 17:19
i REALLY like nano
xinming justatheory: So am I, :-) I love Emacs really...
ods15 seriously
anyway, nothingmuch , you can READ hebrew mails in mutt though?
justatheory I don't love Emacs, but I do love cperl-mode. :-)
nothingmuch ods15: nano is decent because it's simple, and all the commands are displayed on the screen, so you don't even need to find out what command brings up help
but for anything more it's... eh... 17:20
ods15 nothingmuch: i use for C coding, perl coding, bash scripting, mail editor, etc. etc.
nothingmuch no quantification, no pattern matching, no complex regular expressions, no pipes...
xinming ods15: hmm, But I don't think Vi is a poor Editor, Why I don't like Vim is because, For long time programming, If I am in Vim, The left hand (arm) won't feel so good.
nothingmuch I can read most hebrew mails
ods15 i have syntax highlighting, copy paste of entire lines or selected region, etc. etc.
nothingmuch xinming: tried ctrl + bracket?
ods15 nothingmuch: you lost me with all you said regarding pipes etc. 17:21
if i need to do some kind of 'generic' editing to a file, i do in bash, no an editor.. that's insane
nothingmuch ods15: how easy is it to pipe your buffer through 'expand' for example, if you're editing ingy's code?
ods15 expand? ingy?
oh, ingy is a person
nothingmuch ingy doesn't use tabs, I do
ods15 expand is "indent"?
nothingmuch expand is s/\t/ /; 17:22
buut smarter
xinming nothingmuch: How do you konw my problem? -)
ods15 i've never seen a need for that...
nothingmuch ofcourse, this is permanently fixable by saying '# vim:set expandtab:' or whatever the syntax is
ods15: do you edit other people's code?
ods15 if i do need such a thing, i do: c bla.pl P 's/\t/ /' > bla2.pl nano bla2.pl
nothingmuch WTF?! 17:23
use expand -4
ods15 i've never even friggin heard of expand
i've never expanded tabs
nothingmuch x\ty is not x y it's x y
xinming nothingmuch: In fact, I would prefer Emacs to reindent the code. C-M-h C-M-/ :-)
nothingmuch ods15: that's not my fault
ods15 that was just an example
i've never done such a thing or wanted to do such a thing
i wasn't complaining about not knowing expand :) because i've never had a use for it 17:24
nothingmuch and btw, ods15 if yhou're doing bla.pl and bla2.pl
ods15 in MPlayer, tab expansion is illegal anyway
nothingmuch you might as well do `perl -pi -e's/\t/ /' bla.pl'
ods15 (you can't send patches which change indentation)
nothingmuch: yes i know, but i already have the aliases..
nothingmuch ods15: in ingy code tabs are illegal
(you can't send patches which change indentation)
ods15 i really should make a 'perl -pi.bak -e' alias...
need a name for it.. 'P' is taken.. 'p' maybe? 17:25
nothingmuch pi
ods15 hmm, yes, 'p' good enough
nothingmuch added a short alias today
i have a sogudi alias 'pd' that is 'search.cpan.org/perldoc?@@@' 17:26
ods15 heh
nothingmuch so when I started typing 'pd' in the command line for already installed modules I realized I need an alias
obra sogudi?
nothingmuch obra: a safari hack
you type into the URL bar 'pd Some::Module' 17:27
and it takes the string on the right and substitutes for the '@@@'
ods15 heh i've never used url aliases
obra ah. *nod*
nothingmuch very convenient, except you can no longer type spaces in URLs and have safari replace them with %20
obra: I have wikipedia, regular cpan search, cpan perldoc, c2.com
obra *nod* I tend to work from the cli, not a browser 17:28
ods15 btw what's your browser nothingmuch
nothingmuch oh, and recently 's 06' etc for synopses
obra I need to get a quicksilverish microterminal.
nothingmuch ods15: safari which isa konqueror
ods15 mine is 50:50 between elinks and firefox
nothingmuch obra: me too, but for things that aren't installed I use the browser
as for quicksilver, I think it complements the terminal 17:29
www.atamadison.com/w/kitzkikz.php?page=Sogudi
ods15 nothingmuch: anyway, nano is surprisngly more powerful than most people think
nothingmuch ods15: i know nano is not like pico
obra nothingmuch: I'm not on a mac ;)
nothingmuch but it's not a full blown editor either
obra: for some reason I thought you were... hmm
ods15 wtf is a "full blown editor"
nothingmuch tries to remember
ods15: one with at least one embedded language 17:30
ods15: which can autoindent smartly
ods15 heh?
nothingmuch and nest syntax highlighting modes
ods15 autindent is crap, all you need is keeping previous line's indent
nothingmuch ods15: not when you're writing bullets in email instead of code
ods15 nano has syntax highlights, and i wrote pretty nice regexes for it for C and Perl
nothingmuch regexes for perl? EEK 17:31
xinming hmm, for a newbie, I really don't know why don't you choose Emacs, :-P for it is slowly?
nothingmuch you can't really write a *GRAMMAR* for perl
ods15 yes.. they are no failsafe (mostly with # inside strings), but they work pretty damn well
obra nothingmuch: I used to use a mac. I got better :)
or rather, I got a laptop that suits me better. my desktop is still a mac
nothingmuch obra: congrats.. I'm still too lazy and happy with macs =)
did you see the mac I bought at YAPC?
obra The thing that sold me was the Toshiba R100. 17:32
nothingmuch: no. what is?
nothingmuch PB 17"
ods15 my only 2 pet peeves with nano is bad pasting (if i forget auto-indent is goes wild), and it has some weird mem leak or something that causes it to slow to a crawl if you use it for a long while with syntax highlighting
obra The R100 weighs 2 pounds. but is a real laptop.
ods15 other than that i love it
nothingmuch ods15: ah, the opposite of what I was searching for =)
i needed a desktop computer I could take to work
ods15 nothingmuch: btw, know the hebrew song 'kawaka'? 17:33
nothingmuch: ?
desktop what huh
nothingmuch uh, haven't heard of it
ods15 hold on
obra nothingmuch: I travel too much ;)
ods15 www.supersite.co.il/manzur/kawakafliz.mp3
friggin hilarious song, dl it :)
nothingmuch ods15: i don't understand it 17:34
only a few words 17:35
oh wait, the verse isn't as bad
xinming ?eval sub fun { ... }; fun();
evalbotzy *** ... - not yet implemented at <eval> line 1, column 11-15
xinming ?eval sub fun { ; }; fun();
evalbotzy undef
xinming hmm, So what are the differences please?
nothingmuch xinming: ... is an error under 'use fatal' 17:36
and it also says "not yet implemented"
xinming nothingmuch: Ok, thanks, just for "visual" effect. :-)
nothingmuch xinming: not only that, it
bah
it really means 'undef but error "not yet implemented" 17:37
so if it propagates upwards, i think it should be collected in the error stack for any errors happenning because of it
ods15 nothingmuch: listen to the words, the "punchlines" are always completely clear... :)
nothingmuch ods15: it was... OK, i guess
xinming :-)
ods15 nothingmuch: mp3music.gpg.nrg.co.il/lyrics/9881.html
nothingmuch not really hilarious
ods15 bleh :) 17:38
nothingmuch oh wait, if I remove the 'default' sogudi thing then I can type spaces again... Woohoo! 17:40
ods15 nothingmuch: so could you explain to me the tab thing? 17:43
nothingmuch ods15: tab thing?
ods15 why on earth would you want such a thing
nothingmuch i use tabs
ingy uses spaces
ods15 oh, i just remembered what else i syntax highlight, .patch :) and mail ofcourse (starts with '> ') 17:44
nothingmuch if I edited code and forgot to :set expandtab
then I pipe to 'expand -4'
ods15 :set expandtab does what 17:45
nothingmuch it means that every time I type a tab, or it autoindents, it'll be with spaces 17:46
but other tabs in the file are not replace
d
ods15 wait, if you forgot to set it, other tabs are replaced?
nothingmuch ods15: no, if I set it after a while, then other tabs are still tabs 17:47
and it's only new tabs
also, another feature I bet nano doesn't have is hints to the editor placed in comments
ods15 explain? 17:48
nothingmuch for haskell I do use spaces instead of tabs
it just works out bettetr
so I have {- vim:set expandtab: -}
in the top of the file
{- -} is haskell for "block comments" 17:49
ods15 i don't think i've really ever piped code through a program.. i have pipes data through scripts to make em generate code though, but rarely anything else
ok
ah, i got what you meant
no, it doesn't have that
there aren't that many options so you really don't need such a thing 17:50
nothingmuch ods15: also, sometimes when I'm editing xSV files I want to stay in my editor to see the tabulated data properly, but use "real men" commands to edit them, like awk or perl
ods15 nano's "auto-indent" is straightforward - whatever whitespace was in the begginning of the line, is repeated when you press enter...
nothingmuch anyway, I've got to get back to house work 17:51
ods15 i've never really used awk :)
ok bye
nothingmuch ciao!
ods15 heh
geoffb Limbic_Region, you tried to reach me a couple days ago, and I've been just missing you ever since (like sitting down at keyboard a minute after you quit, sheesh). What did you need me for? 18:00
xinming ?eval sub foo is rw { state $var }; foo = 1; 18:02
evalbotzy Error: Can't use readline() or getc() in safemode.
xinming ?eval sub foo is rw { state $var }; foo() = 1;
evalbotzy \1
xinming ?eval sub foo is rw { state $var }; foo() = 1; foo.say;
evalbotzy \1
xinming ?eval sub foo is rw { state $var }; foo() = 1; foo().say; 18:05
evalbotzy 1 bool::true
nothingmuch heh, larry appearantly forgot that I'm "nothingmuch", and that I lobbied for lazy { } in the first place ;-) 18:09
geoffb Hey iblech, how is the JS backend coming? 18:10
iblech nothingmuch: Seems your lazy proposal is accepted :) 18:11
nothingmuch yep =D
hurrah, dancing, whatnot
iblech geoffb: Great, 65%, latest smoke is at m19s28.vlinux.de/iblech/stuff/not_p...smoke.html
geoffb iblech: do you have a blog I can just follow?
iblech, schweet
iblech No, no blog
geoffb What's the best way to follow PIL2JS work, other than bugging you? Do diffs on the README? 18:12
iblech Bugging me is fine :) 18:13
geoffb OK then: how is MM integration coming? 18:14
iblech And yes, I try to keep README in sync with the code
geoffb iblech++ # Keeping docs updated
iblech Everything I could integrate is integrated -- but that's only methods.
geoffb Why not other stuff?
iblech Waiting for -PIL2 so PIL2JS gets class declarations etc.
geoffb AH!
OK, that answers the next question about the status of PIL2 -- I take it autrijus's $work has delayed that? 18:15
iblech (Currently, even small things like class Foo { has $.x } do not compile to PIL, so PIL2JS can't do anything.)
Yep, he has deadlines etc. as far as I know
geoffb bleah, suckage 18:16
.oO( What that boy needs is a MacArthur (sp?) award)
xinming iblech: could you please reveal some information about the deadline? 18:18
iblech xinming: I dunno, all I know is from autrijus' journal (use.perl.org/~autrijus/journal): "It is tomorrow now and I'm still neck deep in this $work deadline thing" 18:19
geoffb Who's working on the P5 backend these days? 18:20
nothingmuch geoffb: fglock stevan and putter IIRC 18:21
i pretend to do it ;-)
fglock what kind of information 'want' returns - is it a type or class?
geoffb iblech, is the failing 35% mostly because of lack of PIL2 and therefore full P6 OO?
nothingmuch fglock: i think 'want' returns an object which yuou can ask questions
and the object is a Context, i presume
and maybe you can say Context::Scalar.new.enforce(&code); 18:22
that could be fun
ods15 bleh
nothingmuch but either way, it stringifies
and has the notation of arity
ods15 firefox is taking 200mb of ram :(
nothingmuch (how many items you want from a list)
iblech fglock: nothingmuch's correct, and there're test for it in t/builtins/want.t
nothingmuch and laziness, too, i think
ods15 total used free shared buffers cached
Mem: 515500 511812 3688 0 5552 125320
-/+ buffers/cache: 380940 134560
Swap: 787148 393572 393576
i have a total of 750mb of "ram" used...
heh more actually 18:23
nothingmuch ods15: want to do some porting? there was a patch to linux 2.2 that scanned processes after they forked
it remerges identical but spoiled pages
ods15 scanned how
nothingmuch so that they are read only again
geoffb any(< nothingmuch fglock stevan putter ): what is the current status of the P5 backend? Pointer to appropriate place is OK, I'm updating STATUS right now.
nothingmuch and shared
iblech geoffb: The main missing parts are OO and packages, P5 and P6 regexes, temp/state/let vars, and some missing builtins
nothingmuch geoffb: I only pretend to know ;-)
ods15 what are you talking about
and i know nothing of kernel programming
nothingmuch ods15: say you launched two firefox processes 18:24
it would scan them (eventually, i think it's an idle time loop)
and compare identical memory pages
ods15 firefox is stupid and doesn't allow multiple instances, but ok, go on
nothingmuch when it finds pages that are the same, it will merge them
so that they are read only, and shared
ods15 that's somewhat dangerous and mostly not worth it
nothingmuch dangerous?
iblech geoffb: OO and packages needs PIL2 support, P5 would be relatively simple (because JavaScript has a Regex class), ask putter for P6 regexes :), temp/state/let vars needs a little bit of work on -CPIL/-CPerl5 and then some work on the PIL2JS side 18:25
ods15 unless it does it safely.. copy on write...
it's not worth it though
nothingmuch ods15: ofcourse it does copy on write
fglock geoffb - stevan is working on the metamodel-2; putter is improving PIL-RUN; I'm working on the implementation of base classes
nothingmuch and it's very worth it if similar processes are starting up the same way
geoffb (nothingmuch: interestingly, at least one virtualization product (VMware ESX) does this across VMs, which I find quite cool. I wonder if the Xen guys do it)
ods15 brk hardly ever has common data, stack is tiny, and code is shared anyway by linux kernel
nothingmuch geoffb: i like virtualizations... they have many cool features
like a cache of translated code pages, and whatnot 18:26
geoffb virtualization++
nothingmuch ods15: if brk has common data due to malloccing of the same stuff
like common initialization
especially if these are aligned to arenas
(a new window has a new javascript interpreter, for example) 18:27
ods15 firefox is mem leaking anyway, it should not be taking 200mb of ram
(and 320mb of swap...)
nothingmuch ods15: another option is to poke in the "heavy" config params
geoffb fglock, happen to how is PIL-RUN coming, since putter appears offline? Passing any of test suite?
nothingmuch it might cache history pages in memory
safari does that so that the back button is almost instantenous
the prev page is completely parsed, all it has to do is draw 18:28
ods15 yes, i did find some option.. "browser.cache.memory.enable"
nothingmuch well, back to chores
ods15 but i disblaed it now.. only way to check if it helped it restart firefox and wait again..
iblech geoffb: IIRC, 01-sanity.t passes
ods15 still, 200 mb is extreme for it
iblech geoffb: But there's no "make p5smoke" or so yet 18:29
geoffb iblech, the whole dir? Cool.
ods15 heh, that's funny, X is using 400mb of virtual :P
iblech geoffb: No, only the first test.
geoffb oh, 01-sanity/01-tap.t
iblech geoffb: Right, meant that 18:30
geoffb: But let me look so we know for sure
geoffb: oooh even 03-equal.t passes! :)
ods15 anyway, i can probably free quite a bit of ram by just closing kdevelop and qt designer, not using em right now anyway 18:31
putter hey guys. hi geoffb.
fglock hi putter
ods15 but come to think of it, why should i :) the whole point of having a lot of ram and swap is to use it.. (althought not ABUSE it, like firefox is doing...)
geoffb iblech, heh: 18:32
er, heh.
putter, hey there
putter hi fglock. putter was rather hoping fglock was working on PIL-Run, and putter was just dabbling. ;)
ingy BURN ALL TABS! 18:33
geoffb iblech, when you say -CPIL/-CPerl5 are needed for temp/state/let -- are these PIL2 or PIL1?
iblech geoffb: PIL1 is sufficient 18:34
geoffb: (Currently, let $foo = 3 is emitted as my $foo = 3)
geoffb: So that's a relatively simple change
putter as I consider creating pirsmoke and perl5test and perl5smoke and ... how about make test-pir test-js test-per5 smoke-pir smoke-js etc test-all smoke-all. thoughts? I expect we will have additional backends, and things are already getting crufty... no?
xinming Just read the comment of autrijus' Journal, I sometimes wonder why will people like wish perl to "become" php. :-)
s/just/just now/ 18:35
putter pugs -BJS -e 'use php; ....php code...'
iblech putter: I like that :) (But I don't mind them staying jssmoke, pirsmoke, etc.) 18:36
putter iblech: any thoughts on jssmoke and jstest becoming somewhat depreciated?
ok. awsome. will do.
iblech putter++
putter wonders if there is a word for irc messages that cross in the writing...
mjl69 cross? 18:38
putter two msgs written at the same, sent with seconds of each other. each of which obsoletes the other. if you could see the other person's keystrokes, it wouldnt happen... 18:39
or a weaker form, one of which obsoletes the other. 18:40
mjl69 I remember years ago, we used to chat online and you would see keystrokes in real time. backspacing was a talent. 18:41
geoffb mmm, ytalk . . . .
fglock would it take much coding to replace pugs 'List' implementation with a perl6 version? 18:43
putter pugs ability to use p6 code is currently a bit limited. Prelude.pm has issues, as does the type checker, and ... . rules development is currently stopped waiting for something to improve just enough for regexs (Prelude rx_common_) to return a Rul. 18:47
s/(currently stopped)/$1 (modulo pm's continuing work on pge etal)/ 18:48
iblech: jssmoke uses PUGS_RUNTIME rather than -BJS, any particular reason, or just historical artifact? 18:49
geoffb putter, you were working on P6 rules implemented in P6, yes?
iblech putter: PUGS_RUNTIME is a hack, -BJS is the way to go
putter iblech: ok. thanks. 18:50
iblech putter: PUGS_RUNTIME is only used to tell run-smoke to skip the ext/ tests and use a perl5/PIL2JS/pugs-smokejs.pl instead of ./pugs
putter geoffb: I was working on all sorts of rules stuff. on p5, on p6/pcre, on p6/native, etc. except with p5, spent most of the time dealing with pugsbugs. was too ambitious a project for the then/now state of pugs. finally, switched to frontend-backward development approach. get something in, improve it, allowing others to see and contribute. after a lot of trying, failed to get something in. well, have warnings about modifiers now. 18:54
but cant return a Rul, means everthing else is stuck. :(
geoffb putter, bleah 18:56
putter, do you mean Rule, or is Rul a magic type? 18:58
putter Rule is currently (against/un-spec) a role, with Rul intended to be the usual class doesing it. 18:59
geoffb ah 19:00
OK, pugs' memory requirements to build have gotten completely insane 19:01
putter Rul is just a couple of lines. infix:<~~> calls a sub. oh, and will carry around a copy of the pattern and modifiers.
optimized? 19:02
putter still feels guilty about the rx rules hook in Prelude. its a non-trivial fraction of Prelude, and Prelude is a bit painful to compile, and the hook isnt buying much at the moment. I have a "rip out hook" patch just incase we need it for a release... 19:03
Juerd iblech: No, there was no discussion about *changing* map to repetition 19:04
iblech: Map on a range, and then using none of the numbers in the range, is misuse of a range and of map.
iblech Juerd: Right, I s// my wording (unless I forgot it) 19:05
putter iblech: does PIL2JS/pugs-smokejs.pl suggest we need a pugs wrapper script, say p5, which deals with these kinds of things, to insulate pugs? or is that not worth it...
iblech Juerd: I agree with you completely
Juerd iblech: This also means that, contrary to what some people have been requesting with the XX, there should be no easily accessible standard counter variable
For when you do need the number, you should use map. 19:06
geoffb putter: *un* optimized (sorry for delay, this box is thrashing its brains out)
iblech putter: Hm. runjs.pl (which is a thin wrapper around pugs -CPerl5 and pil2js.pl) is theoretically sufficient. The only problem is that runjs.pl doesn't stop pugs from compiling Test.pm.
putter: Therefore pugs-smokejs.pl applies this evil hack (s/use Test//) 19:07
putter: If we had a cleaner way to say pugs "trust me, I've precompiled that module to JS, don't load it", we wouldn't need pugs-smokejs.pl
Juerd: Right.
putter -m (the case inverse of -M;)? 19:08
iblech putter: I pondered a TrustMeTestPMIsAlreadyLoaded.pm, which'd only contain "%*INC<Test.pm>++", and then using pugs -MTrustMeTest... -CPerl5 ..., but unfortunately, pugs' option parser doesn't like that 19:09
geoffb Why not just a flag that specifies which modules (generically) have been preloaded? 19:10
Given the thrashing on this system, I can't look for a free flag letter ATM :-) 19:11
iblech This would work too, and is cleaner, I think, so I like it :) 19:12
fglock Juerd: isn't lazyness going to fix this? (sorry maybe I don't know what the problem is exactly)
Juerd fglock: Oh, Perl 6 will have this operator one way or another, I'm sure.
If not in the standard language, then in the standard dialect.
PerlJam Juerd: and if not there, then in your own personal dialect :) 19:13
Juerd PerlJam: That is the standard dialect.
I plan perlego, that has stuff that didn't make it to the base language, but adds value without breaking existing code. 19:14
This can be read as the esperanto word perlego (perl-eg-o, the noun meaning: a greater pearl), or as per-lego (adding building blocks) or as perl-ego (MINE!)
geoffb nice, Juerd 19:15
bah, have to boot yet more systems 19:16
Juerd So far, my perlego document that I have been maintaining has had all its original items removed because they got implemented somehow in the base design.
This strengthens my belief that my ideas aren't that stupid :)
geoffb feels the temperature in his office climb a couple more degrees (again)
Juerd If only I used a versioning system for it, then I'd know exactly which they are
PerlJam Juerd: Those weren't your ideas. You just imagine that you had some really great ideas that were then implemented only it's that they were implemented and then you thought of them. (hint: get a versioning system :) 19:19
Juerd Oh, that's perfectly possible :) 19:21
PerlJam I'm still quite ... shocked? surprised? in disbelief? over all the ./ hoopla 19:22
geoffb sheesh, is that *still* going on? 19:23
svnbot6 r6421 | iblech++ | * t/subroutines/slurpy_param.t: unSKIP as luqui confirmed my thoughts.
r6421 | iblech++ | * t/unspecced/lazy.t: Added three more tests to it and added a pointer to
r6421 | iblech++ | Larry's p6l post which implicitly accepted nothingmuch++'s lazy {...}
r6421 | iblech++ | proposal.
r6422 | iblech++ | Moved t/unspecced/lazy.t to t/var/lazy.t, as nothingmuch's proposal has been accepted.
Juerd I kind of stopped following all Perl 6 developments because of that.
I usually abandon ships with drunken captains
PerlJam geoffb: no, no one has mentioned it on-list in a while, but I bet if you did, it would start all over again. 19:24
xinming luqui: are you there please?
Juerd It'll get better
xinming oops, hmm,
Juerd But I think all Larry's done is disappoint people and make them less ethusiastic
geoffb oh holy crap, this newly-booted box hasn't been updated in a while -- 233 packages to update, according to apt
putter notes Juerd is always gloomy about @Larry and p6 design decisions... 19:25
Juerd Oh, not always
Let it be and stay clear that the majority of the design decisions together make a really great language that I'm looking forward to use 19:26
putter ah 19:27
is there a pattern to the kinds of things you disagree with? 19:28
Juerd Yes, they are all syntax.
My primary concern is with syntax, in any case.
Some very nice features can be implemented with operators that are new, but don't require any learning 19:29
The best example I can give for that is <->. When used in the position where we now use ->, there can be no doubt in any intelligent lifeform's mind that it is closely related to -> itself. 19:30
It doesn't live in one's brain as "another way to create a sub". -> and <-> are the same operator, except that <-> implies "is rw" for all parameters
And there are really many occasions where "is rw" on multpile parameters is the best solution 19:31
xinming Juerd: I think <-> will confuse newbie with <=> ... 19:32
Juerd It'd suck to ever have to write "for @foo -> $foo is rw, $bar is rw, $baz is rw, $quux is rw { ... }" if we could have "for @foo <-> $foo, $bar, $baz, $quux { ... }" at absolutely no cost.
xinming Juerd: They are too smimilar.
Juerd xinming: I think newbie should commit suicide if that's a problem.
We already have -> and => which are radically different and in no way related.
putter (that was putter's first thought, what is <->? a variant on <=>?)
(no doubt clearer in context) 19:33
PerlJam Juerd: for @foo -> ($foo,$bar,$baz,$quux) is rw { ... } # :-)
Juerd <-> by itself is unclear. But when used in context, I have yet to encounter one species that has no clue what it means while knowing Perl 6's -> operator.
xinming Juerd: well, In fact, => is looks like a key point the "value"
Juerd: and -> should make people think for a little while,
putter what portion of the syntax concerns are foward compatible? (eg, define op foo) not, but parse? (eg, redefine ==) and require fiddling with the parser rules? (??) 19:35
xinming Juerd: maybe suggest @Larry consider my ($a, $b, $c, $d) is rw...
Juerd xinming: I think you're missing the point
PerlJam xinming: I think he'd rather have $Larry consider <-> instead
Juerd xinming: The point is not that there should be a way to do it. The point is that there should be the *obvious* way to do it. 19:36
This particular example has an appendix
xinming for @foo -> ($foo, $bar, $baz, $quux) is rw { ... }
Juerd: I don't, I think This might be clear to me... :-)
Juerd The default is now "-> $_ is rw", which is a pattern seen nowhere else: a trait added by default, but no longer when you specify a variable.
PerlJam Juerd: yeah, that is a bit odd. 19:37
Juerd This asymmetry can easily be broken by simply saying that the default operator is <->, instead of ->. The effect and hopefully bytecode of which are equal, but this being a much easier way to live with the situation consistency wise.
PerlJam But I figure that Larry is trying to unhobble the hobgoblin of foolish consistency where ever he can. Sure he makes mistakes, but in the long run, he gets most things right. 19:38
Juerd Most things, yes.
wolverian is there a unicode version of ->?
Juerd Foolish consistency would be adding <- as well.
wolverian (just curious...)
PerlJam Besides, I recall when perl5 was first released there were tons of things that got changed right away as Larry's intuition diverged from the perl populace 19:39
Juerd Because write-only is hardly ever used, as opposed to read-write and read-only, and <- introduces a parsing nightmare (qw<-...) that -> and <-> do not.
fglock { say } <- for 1..10
Juerd fglock: Have you any idea what the -> operator does? 19:40
fglock something with binding a variable to the block scope, I think
Juerd fglock: That's only half, if not less, of what it does. 19:41
xinming fglock: In my understanding, It stands for "sub"
Juerd fglock: "-> $foo { bar }" is a way of writing "sub ($foo) { bar }" (except the class of the block is different)
$foo = -> { ... }; is valid syntax. 19:42
<-, if it existed, would be in the same position:
for @foo <- $elem { ... }
xinming Juerd: then, what does for @foo <- $elem { } mean?
Juerd xinming: @foo <- $elem { $elem = 5 } would assign 5 to each of @foo 19:43
s/: /: for/
While for @foo <- $elem { say $elem } would raise an exception because $elem is write-only.
BUT AGAIN: "<-" would be incredibly silly 19:44
svnbot6 r6423 | putter++ | Makefile.PL: added targets {test,smoke}-{pir,js,perl5,all}. their old names still work. some of the targets, eg *-perl5, dont work yet.
Juerd The correct way to write this would be: @foo = { ... } XX +@foo
For now assuming an XX op
Otherwise: @foo = map { ... } 1..@foo 19:45
Or any variant on that (consider pipes)
1..@foo ==> map { ... } ==> @foo
wolverian @foo>>= 5 19:46
Juerd Well, yes, in the specific case of assigning a single value, that would be a better way to go.
I don't think that'll ever happen in code written by any of us, though.
putter realizes what has been bugging him about Juerd's descriptions - "_Larry_ this and that", personalities, rather than just discussing positions and their merits. feels un-academic. ;-) (err, at least in engineering! ;) some fields it seems like sop) 19:47
PerlJam putter: It's personal. Larry has the ultimate say for better or worse. Juerd just focuses on the "worse" aspects sometimes (for those issues where it's bothersome) 19:48
iblech putter: Does it affect PIL-Run if PSub gets a boolean pSubLValue indicating "sub foo is rw {...}" vs. "sub foo {...}"? 19:49
putter: (I.e. just like pSubName) 19:50
putter iblech: dont let any aspect of PIL-Run slow you down. just go for it. PIL2JS is critical path on improving PIL. any PIL-Run fallout can be cleaned up in its wake. that said, no affect, no problem. ;) 19:53
iblech putter: ok then :)
putter :)
Juerd PerlJam: I hope the reason for the focus is obvious, though 19:57
PerlJam Juerd: it is to me.
Juerd putter: I've referred to Larry once only. Did this really bother you, or did you mistake the many references by others for mine? 19:58
Oh, and please do note that the things I mention were (and some still are) widely supported ideas that didn't make it for one man's veto. 20:00
There may be more opposers, but they are silent and thus ignored. 20:01
xinming Juerd: well, In fact, @Larry design Perl 6, with mutable features, You can write rule your own. ;-) 20:02
Juerd xinming: I know that
But I'm a very strong believer in sane defaults.
putter Juerd: no bother. perhaps my mistake.
xinming Juerd: well, So am I, I use Emacs, And I never erased any default key-bindings. 20:03
Juerd putter: I found your remark surprising, as I constantly try to keep it technical.
xinming Juerd: But the problem is, Perl 6 now is "a bit" like Emacs, Some one would use Gnus to read email, some may not, And some others like w3m better than w3, 20:05
s/better/more/
Juerd I don't see how that relates to recent discussion, xinming
xinming Juerd: So, default might not be the best, But might be most suitable to many people to others. 20:06
putter re "constantly try to keep it technical", err, "I usually abandon ships with drunken captains" and "I think all Larry's done is disappoint people and make them less ethusiastic"...
;) 20:07
Juerd We were discussing default existing of syntax
xinming Juerd: hmm, In my opinion, You can define any "op" right? But Your perfect might not be perfect to other.
Juerd Not a default of options.
xinming: I'm not even referring to perfection
It's leaving things out of the language that fit in so well that bothers me most.
xinming Juerd: hmm, Well, the "perfect" there mean "suitable", sorry for the confusion,
putter bbiab 20:08
xinming what does bbiab mean? T_T
Juerd putter: I can't help but be sad about some recent decisions. This is, by the way, about the ./ decision, which upset almost everyone.
putter: And thus goes much further than syntax. 20:09
xinming It's a pain to chat with "hacker". Too many Jargon...
PerlJam putter: When Larry makes a poor choice, I expect to see "Larry screwed up"
anything else is just putting "political correctness" before truth 20:10
xerox xinming: Be Back In A Bit
iblech Hm, does t/01-sanity/06-use.t fails for others, too? (Working with Pugs.Parser...)
xinming xerox: Thanks. 20:11
xerox xinming: yw.
fglock xinming: see www.computerhope.com/jargon.htm 20:13
putter xinming: google bbiab. you can even just use the first hit;) 20:14
xinming fglock: bookmarked.
putter: hmm, Sometimes, I am lazy to open the google to search.. :-) 20:15
xerox Yeah, that's why you have `wtf` :-)
xinming The three virtues of a perl programmer: laziness, impatience, hubris 20:16
I wish to be a programmer, So I have to learn how to become "Lazy". :-P
and you seee, "Laziness" comes to the first... 20:17
putter "open" google? putter looks buffudled. ooohhh, you dont roll out of bed in the morning thinking, "hmm, lets google what to have for breakfast"! ;-)
svnbot6 r6424 | iblech++ | * Pugs.PIL1: Added pSubLValue/pLValue to PSub and PCode, indicating the 20:18
r6424 | iblech++ | difference between "sub foo {...}" and "sub foo is rw {...}".
r6424 | iblech++ | * Pugs.Compile: Extract the (subLValue vcode) and put in in the appropriate
r6424 | iblech++ | field of PSub/PCode.
r6424 | iblech++ | * Pugs.CodeGen.PIR: Accomodate for the above changes.
r6424 | iblech++ | * Pugs.Parser: Parse traits for anonymous subs, too (i.e. "sub () is rw
r6424 | iblech++ | {...}" parses and works now).
putter I've been a great believer in the "laziness" standard. autrijus has forced me to... reconsider how broadly to apply it. 20:20
xinming m:w/ \( <expr> [ , <expr> ]* : \) / 20:22
is this the same as `m:w/ \( <expr> \)+ /` ?
svnbot6 r6425 | iblech++ | PIL2JS: 20:23
r6425 | iblech++ | * PIL, PIL::Subs: Previously, all subs were implicitly "is rw", because
r6425 | iblech++ | Pugs.Compile didn't include pSubLValue. As this is fixed now, PIL2JS
r6425 | iblech++ | distinguishes "sub foo is rw {...}" from "sub foo {...}" now.
r6425 | iblech++ | * PIL2JS.js: Minor cosmetic fix regarding the error message you get if you try
r6425 | iblech++ | to assign to a readonly box.
xinming I can't understand what the : mean here really,
anyone would tell me? 20:27
fglock they mean different things - (a,b,c:) and (a)(b)(c) 20:30
putter Juerd/PerlJam: Despite solutions always being social+technical, its of course best to separate policy from mechanism in design. and to, at least initially, to discuss issues independently of the people involved. so when I see "my-position-technical-argument vs another-position-people-argument", my alarm bells go off. but no doubt there has been more strictly technical arguments elsewhere. and, err, I vaguely recally someone named pu 20:31
tter flaming at length a day or two ago about the repeated decision not to allow p5 ->methodcalls on unblessed- and non- reference values. ;) anyway, back to hacking.
fglock putter: may I help you? (maybe I need a break with my modules) 20:32
Juerd xinming: read S5 20:34
putter oh, with PIL-Run? have at it. or with integrating js and p5 backends with pugs? or... ;)
PerlJam xinming: colon is the "cut" operator inside of a p6 rule.
xinming: :, ::, :::, <cut> and <commit> are the various flavors of cut operators.
xinming PerlJam: Yes, I know, But What is it does for one : 20:35
PerlJam a single : means don't backtrack into the previous atom
xinming PerlJam: I can feel that :: and ::: and even <commit> ( maybe like :::: ) will do a break( return ) in a level
putter re PIL-Run, one next step might be to get the next sanity test working. needs if(). which means if() bodies, Thunks and their Code Applications, need to be handled better by EvalX (which currently ignores both Thunks and Code). 20:36
PerlJam xinming: for the rule you posted, once we've matched to the :, if the RE doesn't match a ) then the whole rule will fail (because it can't backtrack) 20:37
iblech putter, fglock: FYI, implementing thunks were easy in PIL2JS -- they're simply subs which never take params and which only have a single expression as their body 20:38
xinming PerlJam: Thanks, Might catch the main idea... 20:39
svnbot6 r6426 | fglock++ | * perl5/ List - fixed "ambiguous call to int"
putter statement_control:<if> is in PrimX. takes three p5 strings, and returns a combined string of code. perhaps "if (p6_to_b($xx1)) { $xx2 } else { $xx3 }". or without the {}, depending on what thunks get converted to. 20:40
iblech: thanks! :)
fglock putter: crude_repl.pl is refusing to run, with errors in PrimP5.pm
putter bleep! checking... 20:41
iblech putter: Oooh, so PIL-Run inlines &statement_control:<if>?
putter: (Instead of really calling &statement_control:<if>)
fglock iblech: the P5 runtime is actually a mix of P6/P5 structures 20:42
putter fribble. there is no such thing as an ok hand-written change after a pre-commit test. missing semicolon on the end of the <if> line. 20:43
fixed. r6427 20:44
svnbot6 r6427 | putter++ | PrimP5.pm: typo fix. fglock++
putter iblech: yes. so far at least, things are being agressively converted to p5 code. but no interesting control flow is done yet, so we'll see if that lasts. macros got put in to do && . 20:46
iblech putter: Ah! This is probably a very good optimization
iblech ponders this for PIL2JS... 20:47
fglock putter: are you using Value::List? I'd like to test lazy lists
putter could be. I'm not sure how it will interact with (potential?) cps...
fglock and implement functions in Prelude
cps?
putter fglock: I dont think lists are in yet. "say 3" "say 'hi'" and I havent checked "say 3 + 4" in a while...;) 20:48
iblech Could somebody check with latest Pugs whether our still works? I might have broken it...
fglock: Continuation Passing Style 20:49
fglock: Instead of foo(); bar(), you compile foo({ bar() }, basically
putter fglock, what was the question again? (sorry, I've become confused:/) 20:50
fglock putter: can I use lazy lists?
putter iblech: I was thinking one might pass the continuation using local(), to permit calls from plain p5 code. 20:51
fglock: in PIL-Run?
or in pugs and Prelude.pm?
iblech putter: Ah! That should work, too
fglock yes - the object is implemented in Value::List, and the functions would be in Prelude 20:52
it needs callbacks - but it can use P5 callbacks
putter call, bbiab
back 20:53
fglock p5ugs> (1,2,3) -- Method not found: .do in Scalar.pm 224 20:54
this actually means that the scalar is undef 20:55
xinming hmm, where to find the `built-in` method for default Class? 20:57
putter fglock: assuming we are talking PIL-Run, Value::List can be required in MainX.pm. There is a lib6/P5Runtime/PrimP6.pm for extra stuff, but I dont think the file is used at all yet. (thats probably a matter of changing the pugs line (-I etc) in EvalX - I could do that). EvalX would need a PIL entry for however PIL represents the lists. which uses either a new p6_{to,from}_list in ApiX, or uses Value::List directly and we'll worry abo
ut ApiX later. I think that's it...
:v will show the PIL and generated p5 code. 20:58
p5ugs> :v
xinming Just like, Array will get a sum method by default, I still didn't know there is a built-in sum method until iblech told me.
putter p5ugs> ...whatever... 20:59
iblech xinming: Look at src/Pugs/Prim.hs, there's a table at the end of the file
xinming thanks.. 21:00
svnbot6 r6428 | putter++ | PrimP5.pm: remove debugging output from previous too-rapidly-applied patch. 21:02
r6429 | fglock++ | * perl5/ Container - more descriptive error message in Scalar
putter fglock: ok, it looks like that error message is the current way of saying "the function you tried to apply isnt defined". 21:03
there is a prefix:<,>, but no infix:<,>
is infix:<,> what defines Lists?
fglock don't know - but it looks like it is 21:05
iblech putter: Right, &infix:<,>(1,2,3) is (1,2,3)
putter (With :v , I evaluated (3,4), which gave the p5 code p6_apply(p6_var('&infix:,'),p6_new('Int','3'),p6_new('Int','4')); but checking Prim, infix:<,> didnt exist...)
fglock and infix:<..> implements ranges 21:06
iblech &infix:<,> can be implemented extremely easy: sub infix:<,> (*@elems) { @elems }
(Presuming, * works, of course)
putter ok, so MULTI SUB infix:<,> (*@a) { p6_to_list(@a) } and ApiX defines p6_to_list using Value::List...
fglock so I can hack ApiX p6_to_list? 21:07
putter err, thats p6_from_list
err, ok, that naming scheme is fraying around the edges.
yes, ApiX, whateveryouwanttocallit(@a) { Value::List->new(@a) } 21:08
in ApiX
fglock I think the function is not there 21:09
putter and the reverse too, getting an array from a List, for the primitives which need to deal with it...
(I assume some primitives need to deal with it...)
fglock putter: you can't always map lazy lists to P5 array 21:10
putter p6_from_list? no, not there yet. no lists yet at all.
fglock but lazy list to P6 Array is fine
putter ah well, do p5 array -> List now, worry about using List later, as the need arises. 21:11
iblech notes that PIL2JS doesn't differentiate between Array and List, without causing any (apparent) problems 21:12
fglock so it would be p6_to_l / p6_from_l ?
iblech: I'm not sure what the difference is, actually - maybe just because Array are containers (mutable)
the p5 implementation of Array is much more complex than List 21:13
putter p6_from_l for now. we can revisit how they are named at some point.
iblech fglock: I don't know either. Probably something along that line
putter or p6_list() 21:14
whatever. there are only three files. change is easy.
iblech: prefix:<,> is just (*@a){@a} as you describe. say() should take a List rather than a p5 list, that might need to change. 21:16
s/say/but if say/
iblech Hm, does a plain @a evaluate to a P5 list, instead of a P6 List/Array? 21:17
(If not, I don't understand your sentence.) 21:18
putter fglock: bottom line - there are only three smallish files of code (well, Prim is big, but only the first few lines are hand-written). so follow your wimsy. do anything. its easy to frob it back and forth. And I kind of suspect noone be me has ever run it. And I sure dont mind it breaking. ;) 21:19
xerox Hey. 21:20
Why do I have a pypy supporter saying "perl6 people copied the object spaces idea" ?
I mean, is it like this or the other way around?
wolverian object spaces idea?
xerox So he says O_o 21:21
wolverian what does he mean with it?
putter yes, right now, say(3,4) compiles as p6_apply(p6_var('&say'),p6_new('Int','3'),p6_new('Int','4')); and... (hmm... the 4 doesnt get printed at all!?!?) ... anyway, so a regular p5 list of arguments is being used. once we have List, perhaps something needs to change there.
iblech putter: Ah, ok.
xerox wolverian: I'm asking... 21:22
wolverian xerox, in any case, it doesn't much matter. everything was done with lisp in the 60s or 70s, anyway. we're just reinventing the wheel - with nicer syntax. :)
putter xerox: perhaps the same reason some people think "you used my software patent!"? ;) 21:23
integral I got the impression that "object spaces" was just that you have objects which have an interface to the code, but they're implemented by something in your runtime engine, but you can implement them in different ways :-/
xerox wolverian: indeed.
putter: haha, good point.
iblech geoffb: I just implemented temp in PIL2JS :)
geoffb: (As you were asking earlier today) 21:24
xerox k thanks :)
putter pugs is a bit unusual, in that autrijus is a paper monster. a lot of projects putter along with only a fuzzy feel for the state of the art, and consider the solutions the work out to be novel.
s/the/they/
wolverian xerox, so, did he mean what integral said? 21:25
putter oooo, temp! putter considers doing rules development on PIL2JS...
xerox wolverian: he meant nothing, he seem to do not know what he was talking about - it should be something like "having different backends for code generation"
wolverian xerox, right.
xerox Sorry for bugging you on this. 21:26
iblech putter: let will follow in a minute :)
fglock putter: done ApiX
putter: List contains a Perl6::Value::List object, so that there is no problem if you 'unbox' it 21:28
svnbot6 r6430 | fglock++ | * implemented to/from p6 List functions in ApiX.pm
putter iblech: :) 21:30
fglock: neat! so... should say(3) be, instead of p6_apply(p6_var('&say'),p6_new('Int','3')), something like p6_apply(p6_var('&say'),p6_from_l(p6_new('Int','3'))) ? 21:31
fglock I think so
did you find what is the node type for List? 21:32
putter sorry, I havent looked...
fglock I'm trying to find out
putter if you do :v and then (3,4) you get a nice yaml dump of the pil tree. 21:33
fglock yep. I'm trying to understand it
putter looks like nothing more than a call to infix:<,>
svnbot6 r6431 | putter++ | PrimP5.pm: say() now prints all its arguments, not just the first one.
iblech nothingmuch: As I'm going to commit a change to Pugs.AST.Scope (which could possibly break everything), I'd appreciate if you restart your smoke script, thanks very much :)
fglock I tried (1,2).shift 21:34
putter which compiles to an excessively brief p6_apply(p6_var('&shift')); , because infix:<,> doesnt exist yet... 21:35
the key things to look for in the pil dump is the PApp's, pVarName's, and simple literals. 21:36
fglock: did you define an infix:<,> in Prim? 21:38
fglock no - I'll open it
putter something like MULTI SUB infix:<,> (*@a) { p6_from_l(@a) }; should do it. 21:40
fglock there is one already - I'll try to modify it
what is prefix:<,> used for? 21:42
iblech fglock: To construct a list/array 21:43
putter I was wondering that a moment ago. must be something, or I wouldnt have defined it. maybe leave it, write the infix:<,> we need now, and worry later.
svnbot6 r6432 | iblech++ | * Pugs.AST.Scope: Reorder the declarations of the various scope types -- SLet
r6432 | iblech++ | and STemp do something to *existing* variables. Accidentally, this causes
r6432 | iblech++ | temp.t to pass on normal Pugs!
r6432 | iblech++ | * Pugs.Lexer, Pugs.Compile: Acommodate for that change.
r6432 | iblech++ | * PIL2JS: temp! let!
r6432 | iblech++ | * PIL: All JS functions have a block_leave_hooks array now, containing native
r6432 | iblech++ | JS functions to be called at block exit.
r6432 | iblech++ | * PIL::PPad uses this variable to implement the (possible) variable
r6432 | iblech++ | restoration needed for temp and let.
r6432 | iblech++ | * t/var/let.t, t/var/temp.t: Should "temp @array[$index]" work? I think so, but
r6432 | iblech++ | Pugs doesn't even parse that. Added appropriate tests in a =pod block.
putter iblech: so when infix:<,> and when prefix:<,>?
svnbot6 r6432 | iblech++ | BTW, let.t passes 7/7, temp.t passes 22/22 (with TODO tests).
iblech putter: Never prefix:<,> 21:44
putter hmm.... 21:45
iblech (At least, I've never seen that one.)
svnbot6 r6433 | putter++ | ApiX.pm: moved Value::List->from_single() call to p6_new 21:46
r6434 | putter++ | ApiX.pm: oops, added missing return.
putter guess I imagined it... 21:47
iblech, does temp() unwind when exceptions are thrown? 21:51
and does temp.t passing on pugs mean it really works (!?!), or that temp.t needs more tests...? 21:52
svnbot6 r6435 | putter++ | PrimP5.pm: deleted mythical prefix:<,> primitive.
iblech putter: Oh, no, haven't thought of that. Writing test and thinking about implementation
putter ok
iblech ?eval $?PUGS_VERSION
evalbotzy \'Perl6 User\'s Golfing System, version 6.2.9, August 3, 2005 (r6378)'
putter my fuzzy recollection is doing unwind-protect (aka temp) in the face of full continuations is a "long studied but regretably inherently hard problem". 21:53
fglock I'll commit what I have so far and go after a problem with 'List' 21:54
iblech putter: oh. hm. 21:56
putter k
hmm... temp $obj.attribute = 5; ;) 21:57
svnbot6 r6436 | fglock++ | * small fix in ApiX
iblech That's actually extremely easy to implement in PIL2JS, presuming that this parses :)
putter !:) 21:58
nothingmuch parrot is headed in the right direction... ;-) 21:59
www.sidhe.org/~dan/blog/archives/000425.html
dan's parrot, at least
putter iblech: oops. I'm wrong (vis continuations vs unwind-protect) bc.tech.coop/blog/050731.html (bottom) and perhaps www.nhplace.com/kent/PFAQ/unwind-pr...tions.html 22:02
dan's parrot would have been an interesting thing to play with... 22:03
fglock putter: p6_from_l works, but there seems to be a problem with the object somewhere else 22:04
putter hmm... just in case it was too hidden, note that in crude you can :5 <perl5 expression>, which helps with debugging... 22:05
fglock It might be a problem with the object - I'll check this first
putter k
nothingmuch: one thing I take away from having seen both parrot and pugs, is that if the domain is amenable to prototyping (it was in both cases), at least in an open source setting, it really helps if your system architect can spew code, instead of just spec. 22:09
svnbot6 r6437 | fglock++ | * updated ApiX.pm
nothingmuch putter: fairy nuff, but MMD as an implementation detail (even when it's exposed to the high level language) is a smart move, since it vastly simplifies optimization when static analysis and type inferrence becomes involved 22:10
and optimization is incremental, and modular
for example the assignment operation on integers as given in dan's blog post makes a lot of sense 22:11
especially when static analysis lets you just emit machine code that has absolutely no respect to an encapsulated object's guts, since the data type is shared 22:12
putter goes back to read article more closely...
nothingmuch this can be bolted on later, after you've got "safe" assignment implemented
which is very good, because it can be profile driven
if your integer operations are too slow, you meter them, and figure out that assignment is taking too long, so you special case assignment on integers, and you save a little here 22:13
then you find that too much time is wasted on MMD dispatch, so you improve static assignment, and inline the MMD candidates as machine code where they can be known in advance
this approach to making things fast is known to work, and is easy to design well, and is good for many-hacker few full timer projects 22:14
where someone who cares can whip up a specialization of the system for the small part they cares about
s/cares/care/; 22:15
uh, damnit
the second cares
fglock weid error - calling a closure as a method doesn't pass $_[0] (in Perl5) 22:16
/weid/weird/
nothingmuch fglock: really? i've relied on it a million times
if (my $sub = $obj->can("method")){ $obj->$sub(@args) }; 22:17
for example
iblech putter: Thanks much for the links :) I'll (have to) read them carefully, tomorrow
svnbot6 r6438 | iblech++ | t/var/: let.t, temp.t: Test that let and temp restore variables even when the
r6438 | iblech++ | block they're in has been left via an exception (and not via an (possibly
r6438 | iblech++ | implicit) call to &return).
fglock I just had to fix this in List.pm
putter np :)
they were just a quick google, so there may well be better stuff out there.
iblech Maybe I dream of the solution, similar to autrijus :)
fglock sub str { $_[0]->{cstringify}( $_[0] ) } -- doesn't work without ($_[0]) 22:18
iblech Which reminds me, I should sleep now. Night all! :)
putter Good night iblech! :)
wolverian fglock, {cstringify} is not a method 22:20
putter fglock: but the () is sub-call (not method call) to the sub ref in the hash (index cstringify) pointed to by $_[0].
nothingmuch wtf is this hoodwink.d thing?
putter ?
fglock mm right - I thought it was a method
putter: it works now - the list prints weird, but that's because stringification is very simple 22:21
putter neat!
fglock 1,2 prints Int=HASH(0x88f6894)....Int=HASH(0x88f6270)
putter and actually, the answer to the earler "does say get called with a list" is.... only if PIL says so! ;) which it doesnt at the moment. 22:22
putter svn up's to look...
svnbot6 r6439 | fglock++ | * perl5/ List - fixed stringification
fglock now trying to find out how to do 1..2 22:23
actually 'shift' is easier 22:24
putter err, did you define infix:<,>...? I'm getting an error on (3,4) 22:25
fglock ah ok, I didn't commit PrimP5
putter k
fglock done 22:26
putter iblech: prefix:<,> is in Pugs/Prim.hs (after DESTROYALL) 22:27
wolverian prefix ,? that's weird
putter fglock: I'm still seeing 6439 22:28
fglock I'm getting a svn conflict - resolving... 22:29
putter takes a moment to marvel at working with someone in another part of the planet, and thinking a 3minute latency in tossing a file back and forth, via taiwan, is like *taking forever*. ;) 22:30
fglock: oops, sorry. probably my dropping prefix:<,>.
fglock done - 6440 22:31
(1,2) works, but (1,2).shift doesn't 22:32
putter prefix:<,> came back, and that's fine. I'll add a note to it saying we're puzzled.
wolverian (1,2).shift is invalid perl6, if that's the problem :)
(I don't know what you're doing, so just ignore me if you know that.)
svnbot6 r6440 | fglock++ | * defined (1,2).shift
wolverian well, I thought it would be. oh well. (lists are mutable now?) 22:33
fglock strange - how do you take a value from a list?
putter hmm, (3,4) gives me Can't use string ("") as a subroutine ref while "strict refs" in use at /home/net1/perl6/pugsxpl2/perl5/PIL-Run/../Perl6-Value/lib/Perl6/Value/List.pm
wolverian my question exactly, fglock. :)
fglock putter - you have to update Value::List
putter hits his head, "doh!". svn up in wrong (two low) dir. 22:34
too low, even
fglock lists aren't exactly mutable, but you have to traverse them (?)
wolverian [] traverses, doesn't it? I don't understand what (1,2).shift would do.
fglock it returns 1, and keeps a list (1). the original list doesn't exist anymore 22:35
oops, returns 1 and keeps 2
wolverian but the list isn't "kept". that's the definition of a list. only arrays are kept. 22:36
putter leaves fglock to do interesting things with List's and goes to get a snack...
wolverian (well, the definition of a list is that it's immutable, I guess. :)
fglock anyway, shift it is not an error at this level of implementation - I'm getting a message saying that I'm trying to shift from an 'undef' 22:38
wolverian right. (the error should be that you're trying to shift from a List) 22:40
putter feeping creatures, snack -> dinner. later & 22:42
fglock bye - I'm getting home too
putter err, fglock, hmm... if you do :v, (3,4).shift for me is compiling as p6_apply(p6_var('&shift')); so somethings not quite right... 22:43
fglock I'll try again later - maybe storing the list in a Scalar first 22:44
putter say(say 3) works...
ok, later! :) &
fglock my $a=(1,2); $a.shift doesn't compile too
putter $a = (3,4) 22:46
:5 p6_var('$a')
Scalar=HASH(0x1164710)
which perhaps isnt what you intended?
oh, duh. yes, that's right. 22:47
5 p6_var('$a')->fetch
List=HASH(0x116c0c0)
i'm late. this was fun, thanks! :) & 22:48
luqui ?eval 42 22:53
evalbotzy 42
luqui good
coral ?eval 42.pick
evalbotzy Error: pick not defined: VInt 42
luqui ?eval (42|5).pick
evalbotzy 42
luqui ?eval (42|5).pick
evalbotzy 5
coral whoah! neat!
someone++
luqui I think that's been there awhile 22:54
buu Holy wtf.
?eval (42|5)
evalbotzy 5
buu ?eval (42|5)
evalbotzy 5
luqui I think that's a parsing error
buu Really?
luqui there is no automatic collapse of junction
wolverian I think evalbot doesn't output junctions right
buu shux.
wolverian ?eval 1 & 2
evalbotzy 1 22:55
wolverian ?eval "a" | 42 | "c"
luqui ?eval (42|5).states
evalbotzy 42
Error: No compatible subroutine found: "&states"
luqui ?eval (42|5).values
evalbotzy (5, 42)
luqui there we go
wolverian ?eval "a" | 42 | "c" | "d"
evalbotzy 42
wolverian that's still not right. :)
luqui it seems to give you the first one when sorted
wolverian right.
or it's trying to tell us something important..
?eval "a".."b".pick 22:57
evalbotzy Error: pick not defined: VStr "b"
wolverian ?eval ("a".."z").pick 22:58
evalbotzy 'x'
wolverian just curious about precedence. now, good night. :)
nothingmuch *yawn*... time for shower & bed
so much dust.. yuck!
luqui eew dust
nothingmuch the worst kind too - cement dust 22:59
luqui sticky
nothingmuch it's corrosive, and it dries up the skin
luqui hmmm.. that's worse than sticky 23:00
I've never been around it too long
nothingmuch we're rennovating, so there's lots of it

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!