»ö« | perl6.org/ | nopaste: paste.lisp.org/new/perl6 | evalbot usage: 'perl6: say 3;' or rakudo:, alpha:, pugs:, std:, or /msg p6eval perl6: ... | irclog: irc.pugscode.org/ | UTF-8 is our friend!
Set by lichtkind on 5 March 2010.
dalek psi: 17f4298 | masak++ | doc/announce/2010.05:
[doc/announce/2010.05] fixed a few typos

typos were found by by wolverian++ before the release announcements were sent out (but after the uploading of the release tarball).		 00:01
00:05 Limbic_Region joined 00:12 am0c joined
lue afk 00:27
sorear rakudo: say BEGIN { rand } 00:38
p6eval rakudo c4857a: OUTPUT«Could not find sub &rand␤current instr.: '_block11' pc 0 (EVAL_1:9)␤»
sorear rakudo: say rand 00:39
p6eval rakudo c4857a: OUTPUT«0.555681770990731␤»
diakopter 3 hours later, I found the bug.
sorear I see that my bug with macros is shared with BEGIN blocks
diakopter not quite a record.
sorear files rakudobug 00:41
rakudo: say raxx 00:42
p6eval rakudo c4857a: OUTPUT«Could not find sub &raxx␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
diakopter yay! 00:43
fixed it
the assemblies perlesque.exe emits are now entirely independent of Sprixel.exe and perlesque.exe, unless of course they use classes from them..
and I can do super-fun stuff like this: 00:44
dalek meta: r237 | diakopter++ | trunk/Sprixel/ (2 files):
[perlesque] more mono fix attempts
diakopter perlesque: my $a = sub int (int $b) { return 1; }; $a = sub int (int $b) { if 0==($b % 100000) { say('reached ' ~ $b) }; if $b == 1000000 { return $b }; return $a($b + 1) }; say($a(0));
p6eval perlesque: OUTPUT«reached 0␤reached 100000␤reached 200000␤reached 300000␤reached 400000␤reached 500000␤reached 600000␤reached 700000␤reached 800000␤reached 900000␤reached 1000000␤1000000␤»
diakopter sorear: a 1m callframe-deep invocation :)
perlesquel: . 00:46
p6eval perlesquel: OUTPUT«reached 0␤reached 100000␤reached 200000␤reached 300000␤reached 400000␤reached 500000␤reached 600000␤reached 700000␤reached 800000␤reached 900000␤reached 1000000␤1000000␤real 1.43␤user 1.38␤sys 0.18␤»
00:48 TiMBuS left
diakopter jnthn: ^^ 00:49
and subs can refer to themselves by name: 00:53
perlesque: sub int foo(int $b) { if 0==($b % 10000) { say('reached ' ~ $b) }; if $b == 100000 { return $b }; return foo($b + 1) }; say(foo(1))
p6eval perlesque: OUTPUT«reached 10000␤reached 20000␤reached 30000␤reached 40000␤reached 50000␤reached 60000␤reached 70000␤reached 80000␤reached 90000␤reached 100000␤100000␤»
diakopter perlesquel: .
p6eval perlesquel: OUTPUT«reached 10000␤reached 20000␤reached 30000␤reached 40000␤reached 50000␤reached 60000␤reached 70000␤reached 80000␤reached 90000␤reached 100000␤100000␤real 0.33␤user 0.14␤sys 0.02␤»
00:57 wolverian left, wolv is now known as wolverian
jnthn sorear: Ah, yes...I suspect some tweaks are needed to run BEGIN blocks within the setting. 00:58
Anyways, sleep for me now. :)
01:00 eternaleye left 01:15 eternaleye joined 01:24 am0c left 01:26 fridim joined 01:35 wknight8111 left 01:36 alester joined 01:50 wolverian left 01:58 alester left 01:59 Psyche^ joined
sorear TimToady: If I have a macro that runs at the 'statement' or 'declarator' category, it seems non-trivial semantically to substitute >1 declarator 01:59
It's always legal to use >1 statement where 1 is needed, but the grammar doesn't handle this regularly
so I'm not sure how to design that feature 02:00
02:03 Patterner left, Psyche^ is now known as Patterner 02:21 BrowserUk joined
diakopter sorear: 02:21
maybe enclose it in a self-executing block? 02:22
sorear diakopter:
diakopter: self-executing blocks create a new lexical scope
diakopter yah
sorear if I want to make a macro which defines stuff, that doesn't work
diakopter hm, yeah.
make a new grammar category that's list_of_statements or something ...? 02:23
then make it stand-in for statement everywhere else
so every "real" statement is a list of 1 02:24
(sorry, I know it's kludgy)
sorear what I'm going to do is far more evil 02:26
replace "statement" with "statementlist" when reparsing string macro results 02:27
diakopter heh 02:28
nice
sorear > macro term:<foo> is parsed( / <?> / ) { Perl6::Grammar.parse("say('foo'); say('bar')", :rule("statementlist"), :actions(Perl6::Actions)).ast } 02:29
> foo
foo
bar
I'd call this "success" 02:30
eternaleye sorear++ 02:34
02:38 kst left 02:39 kst joined 02:40 camenix joined 02:43 BrowserUk left 02:53 xinemac joined
Tene lue: specificaally, I was going for realtime multiplayer, kind of a mud/roguelike mix. 02:56
sorear Tene: Hi! I hear, belatedly, you were doing some stuff with macros? 02:57
03:06 MetaMucil joined 03:26 kst left, k23z__ joined 03:29 kst joined 03:30 MetaMucil left 03:31 Sathiya joined 03:32 Sathiya left 03:52 astrojp left 03:54 gurjeet left, gurjeet joined 04:07 meppl joined, envi^home joined 04:13 Limbic_Region left 04:14 kst left 04:15 kst joined 04:19 redicaps_ joined 04:20 gurjeet left, gurjeet joined 04:33 finanalyst joined 05:01 kaare joined 05:02 kaare is now known as Guest81117 05:05 rv2733 joined 05:06 kst left 05:07 kst joined 05:17 k23z__ left 05:21 rgrau left 05:38 fridim left 05:43 jimmy joined, jimmy is now known as JimmyZ 05:46 kst left 05:47 kst joined 05:55 k23z__ joined 06:15 finanalyst left
snarkyboojum phenny: tell masak, I'd like to pick your brains about Yapsi/Tardis integration. I have something going, but have a bug somewhere :) 06:17
phenny snarkyboojum: I'll pass that on when masak is around.
snarkyboojum: 01 May 17:38Z <masak> tell snarkyboojum that Yapsi is now, finally, ready for Tardis. here's a proof-of-concept: gist.github.com/386512
06:28 camenix left 06:31 xinemac left 06:32 kst left 06:34 plobsing left, kst joined 06:39 camenix joined 06:41 xinemac joined
sorear nqp: sub a(*@b) { say(+@b); }; a(|[], |[]) 06:48
p6eval nqp: OUTPUT«0␤»
06:50 JimmyZ left
sorear nqp: pir::null__P() ?? 1 !! 0 06:52
p6eval nqp: OUTPUT«Null PMC access in get_bool()␤current instr.: '_block11' pc 0 (EVAL_1:5)␤»
06:59 jaldhar joined 07:03 JimmyZ joined 07:16 snarkyboojum left
sorear nqp: say(pir::isa__IPS("foo", "Str")) 07:32
p6eval nqp: OUTPUT«0␤»
sorear nqp: say(pir::isa__IPS("foo", "String"))
p6eval nqp: OUTPUT«1␤»
sorear rakudo: say(pir::isa__IPS("foo", "String"))
p6eval rakudo c4857a: OUTPUT«1␤»
07:33 JimmyZ left, kst left, fda314925 left 07:36 kst joined 07:49 fridim joined, pmurias joined
sorear > macro term:<stub> is parsed(/ <identifier> /) { 'sub ' ~ $<identifier>.Str ~ ' { !!! }' } 08:06
> stub foo; foo; 08:07
Stub code executed
no more PAST!
sorear starts an overnight spectest run 08:08
anyone who thinks this is interesting - I've left the (not yet spectested!) patches on github 08:09
08:10 iblechbot joined 08:20 stepnem left 08:21 snarkyboojum joined 08:28 _jaldhar joined, jaldhar left, kst left 08:29 kst joined 08:31 stepnem joined 08:57 tri1 joined 08:59 tri1 left
eternaleye sorear: Wait. So you've just implemented macros, both hygenic and non, in 1-2 days. I am seriously impressed. 09:10
moritz_ kinda finds that hard to believe 09:11
eternaleye Looking over the patches, it's more 'to a first approximation', but it's still damn cool 09:18
moritz_ enough to give masak something to break :-) 09:19
eternaleye heh, true enough
rakudo: class Foo {}; class Bar {}; sub baz( Foo $a --> :( Foo, Bar ) ) { return( $a, Bar.new ); ); baz( Foo.new ); 09:28
p6eval rakudo c4857a: OUTPUT«Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤current instr.: 'perl6;Regex;Cursor;FAILGOAL' pc 1664 (ext/nqp-rx/src/stage0/Regex-s0.pir:907)␤»
eternaleye rakudo: class Foo {}; class Bar {}; sub baz( Foo $a --> ( Foo, Bar ) ) { return( $a, Bar.new ); ); baz( Foo.new );
p6eval rakudo c4857a: OUTPUT«Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤current instr.: 'perl6;Regex;Cursor;FAILGOAL' pc 1664 (ext/nqp-rx/src/stage0/Regex-s0.pir:907)␤»
eternaleye How would I specify a signature as a return type?
rakudo: class Foo {}; class Bar {}; sub baz( Foo $a ) returns ( Foo, Bar ) { return( $a, Bar.new ); ); baz( Foo.new ); 09:33
p6eval rakudo c4857a: OUTPUT«Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤current instr.: 'perl6;Regex;Cursor;FAILGOAL' pc 1664 (ext/nqp-rx/src/stage0/Regex-s0.pir:907)␤»
moritz_ rakudo: sub f($x --> Int) { 1 }; say f(5) 09:35
p6eval rakudo c4857a: OUTPUT«1␤»
moritz_ rakudo: sub f($x --> Int) { 'foo' }; say f(5)
p6eval rakudo c4857a: OUTPUT«foo␤»
moritz_ seems it's not yet checked
rakudo: my Int sub f() { 'foo' }; say f 09:36
p6eval rakudo c4857a: OUTPUT«foo␤»
moritz_ should work too
09:36 moritz_ sets mode: +v p6eval
eternaleye Yeah, but I'd settle for knowing the right way to phrase it for when it is checked in the future. I know the syntax for single return types, and I could potentially hack around it by specifying a Parchel return type and using unpacking, but I'm not sure that's the best way/ 09:37
s,/$,.,
*Parcel
moritz_ subset ReturnType of Any where { :(Int, Num, Array) } 09:39
without the brackets
09:45 M_o_C joined 09:46 kst left 09:48 kst joined 09:52 iblechbot left 09:53 M_o_C left
moritz_ s1n: (backlog) yes 09:56
s1n: mpl.mpg.de/mpf/php/abteilung1/index...;and=staff 09:57
10:14 wknight8111 joined 10:23 redicaps_ left 10:32 FreeStorm joined 10:33 kst left 10:34 kst joined 10:38 lest_away is now known as lestrrat 10:43 orafu left 10:56 FreeStorm left
jnthn oh morning 11:03
11:03 takadonet1 joined 11:04 takadonet left
moritz_ \o/ 11:11
11:13 orafu joined
colomon o/ 11:18
jnthn colomon: o/
colomon Iceland still cool? :) 11:21
jnthn colomon: Yes :-) 11:23
colomon: I like it here.
The conference folk who made it anyway, despite the cancellation, are hackathoning today.
(Me too.)
colomon p5 hackathon?
jnthn Whatever people want. :-) 11:24
colomon do you have a goal in mind?
jnthn For me, I was planning to spend my time doing Rakudo fixes/get more tests passing, and spend some time on book. 11:25
A group of people here are working on some education software project.
In fact, that's most of them. :_)
colomon noble goals
11:26 kst left
jnthn Well, I already got a fix to get another test file to run so far as it did on alpha (not pushed yet). 11:26
moritz_ works again on match objects
my current idea for the vtable overrides is: use src/cheats/Match.pir which adds a method with :vtable(...) 11:27
11:27 kst joined
jnthn moritz_: Which vtable overrides? 11:27
moritz_ jnthn: set_pmc_keyed
jnthn What's making you need that?
moritz_ jnthn: it's what Regex::Cusors.'!MATCH' calls
jnthn Ah. 11:28
moritz_ I'd prefer to pass all captures as argumenst to .new
but there's a certain overhead associated with that
jnthn Did you try putting a forwarder in Mu.pir like we have many others?
moritz_ what should it forward to? 11:29
jnthn postcircumfix:<{ }> I guess?
It's a confusing of binding and assignment though. :-/
Alternatively, what you suggest for adding a Match.pir with them would also work. 11:30
moritz_ I'll try that next, if my approach doesn't work
jnthn k 11:31
isBEKaml hi guys :) 11:38
jnthn hi isBEKaml
isBEKaml I still don't seem to get a grasp on GatherIterator(). 11:39
rakudo: my $ff = "helloWorld".split(/o/,:all); say $ff.WHAT; say $ff.get() while $ff;
p6eval rakudo c4857a: OUTPUT«GatherIterator()␤o␤o␤␤»
isBEKaml rakudo: my $ff = "helloWorld".split(/o/,:all); say $ff.get() while $ff; say $ff; 11:40
p6eval rakudo c4857a: OUTPUT«o␤o␤␤␤»
jnthn That's internals leaking out more than something you're really mwant to see, I suspect.
moritz_ rakudo: say 'hllowWorld'.split(/o/, :all).join(', ')
p6eval rakudo c4857a: OUTPUT«hll, o, wW, o, rld␤»
isBEKaml jnthn: If I simply say $ff before iterating through it, it prints out the entire thing. why does it just show the split param when I give get() ? 11:41
moritz_: yes. :)
colomon isBEKaml: because .get is the method to get the next element of an iterator 11:42
whereas if we get a naked iterator, we try to do something intelligent with it.
isBEKaml colomon: I get it, but I would expect it to show the next element. not the split element..
colomon: My thinking was this. I split "helloworld" into several elements around "o". Now if I call .get(), I should be seeing the first element, i.e. "hell", "o","w","o","rld". Along these lines. Am I wrong? 11:43
sorry, that should be "hell", and successive calls should yield the next elements.. 11:44
moritz_ isBEKaml: I'd expect the same 11:45
isBEKaml moritz_: is this a bug, then?
moritz_ jnthn: ah, with set_pmc_keyed in Mu.pir I get infinite recursion... but at least it does something; will investigate more 11:46
11:46 nadim_ joined
moritz_ isBEKaml: if the spec says that the return value is soemthing that can be iterated with .get, then yet 11:46
colomon rakudo: my $ff = "helloWorld".split(/o/,:all); say $ff.get; say $ff.get; say $ff.get
p6eval rakudo c4857a: OUTPUT«hell␤o␤W␤» 11:47
colomon rakudo: my $ff = "helloWorld".split(/o/,:all); say $ff.WHAT; say $ff.get; say $ff.get
p6eval rakudo c4857a: OUTPUT«GatherIterator()␤hell␤o␤»
isBEKaml colomon: then I must be doing something wrong in using the while loop. 11:48
moritz_ rakudo: my $ff = "helloWorld".split(/o/,:all); while my $x = $ff.get !~~ EMPTY { say $x }
p6eval rakudo c4857a: OUTPUT«1␤1␤1␤1␤1␤»
moritz_ rakudo: my $ff = "helloWorld".split(/o/,:all); while ( my $x = $ff.get) !~~ EMPTY { say $x }
colomon rakudo: my $ff = "helloWorld".split(/o/,:all); say $ff; say $ff.get; say $ff.get
p6eval rakudo c4857a: OUTPUT«hell␤o␤W␤o␤rld␤»
rakudo c4857a: OUTPUT«hell o W o rld␤␤␤»
isBEKaml rakudo: my $ff="helloworld".split(/o/,:all); say $ff.get while $ff;
p6eval rakudo c4857a: OUTPUT«o␤o␤␤»
moritz_ isBEKaml: the loop with !~~ EMPTY above works
isBEKaml rakudo: my $ff="helloworld".split(/o/,:all); say $ff.get while $ff !~~ EMPTY;
p6eval rakudo c4857a: 11:49
..OUTPUT«hell␤o␤w␤o␤rld␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤
colomon yeah, it's definitely something with the while loop.
jnthn doing $ff !~~ EMPTY won't work, afaik.
moritz_ isBEKaml: you have to check the return value for EMPTY, not the iterator
jnthn You match something that comes back from $ff.get against EMPTY
isBEKaml oh, damn, that stores the fixed block in every iterator initiation..
moritz_: I see.. 11:50
colomon rakudo: my $ff = "helloWorld".split(/o/,:all); my $a; while ($a = $ff.get) !~~ EMPTY { say $a; }
p6eval rakudo c4857a: OUTPUT«hell␤o␤W␤o␤rld␤»
11:50 rgrau joined
moritz_ rakudo: class A is Hash { }; my $h = A.new; $h{'foo'} = 'bar'; say $h{'foo'} 11:52
p6eval rakudo c4857a: OUTPUT«bar␤»
moritz_ rakudo: class A is Array { }; my $a = A.new; $a[0] = 'baz'; say $a[0]
p6eval rakudo c4857a: OUTPUT«Null PMC access in invoke()␤current instr.: 'perl6;Proxy;!STORE' pc 14609 (src/builtins/Routine.pir:151)␤»
isBEKaml colomon: thank you. :) 11:53
moritz_ submits rakudobug
rakudo: class A is Array { }; my $a = A.new; $a[0] = 'baz'
p6eval rakudo c4857a: OUTPUT«Null PMC access in invoke()␤current instr.: 'perl6;Proxy;!STORE' pc 14609 (src/builtins/Routine.pir:151)␤»
11:57 M_o_C joined
isBEKaml I can understand that while checks a boolean and acts accordingly. Doing, while $ff ... , what exactly does an iterator emit to the while loop? 12:01
note, I'm not checking like, say something while iterator is not empty.. 12:03
moritz_ rakudo: my $ff = "helloWorld".split(/o/,:all); $ff.Bool; say $ff.join('|') 12:04
p6eval rakudo c4857a: OUTPUT«o|W|o|rld␤» 12:05
moritz_ isBEKaml: ah, checking for truth also consumes an element
isBEKaml moritz_: yes, but why? 12:06
moritz_ isBEKaml: no idea
isBEKaml rakudo: my $ff="helloworld".split(/o/,:all); $ff.Bool; $ff.Bool; say $ff.join(','); 12:07
p6eval rakudo c4857a: OUTPUT«w,o,rld␤»
12:12 kst left, kst joined
colomon I'll tell you why. 12:13
the only real interface to an iterator is get. 12:14
so the only way to find out if it is empty is to .get an element.
that's why you only got the o's back, too. 12:15
because the while condition was getting every other bit. 12:16
isBEKaml colomon: so whatever we do on an iterator is daisy chained to get() ?
colomon anything you do with an iterator other than .get has to be implemented in terms of .get.
12:16 wolverian joined
isBEKaml colomon: doing, say $ff.get while $ff # dropped everything other than o. I think, if as you say, it should be dropping the "o"s too.. 12:19
but it doesn't..
colomon nope
the sequence is this:
while $ff (does get, gets "hell")
say $ff.get (gets 'o', says it) 12:20
while $$ (does get, gets ' w')
say $ff.get (gets 'o', says it)
etc.
isBEKaml ah,
jnthn grr...seems we get some GC segfault when I fix some bug. :-| 12:21
isBEKaml as moritz_ pointed out, checking for booleans even gets an element. 12:22
jnthn isBEKaml: But a boolean check is the Wrong Way to check if an iterator has more values anyway.
I'm not sure it's a bug.
It's just that you're not meant to use it like that.
moritz_ jnthn: it's surely a bug that .split returns such a fragile object
jnthn moritz_: Right, that is a bug.
But I think that's a bug with gather in general 12:23
Rather than just split.
moritz_ it should return a lazy list that remembers all evaluated values
and .Bool on it should only force evaluation of the first element, if it isn't already
12:25 iblechbot joined
moritz_ jnthn: if a class has $!foo, does it actually install an attribute $!foo in the parrot class? 12:25
12:25 masak joined
masak greetings, perl6beings. 12:26
phenny masak: 06:17Z <snarkyboojum> tell masak I'd like to pick your brains about Yapsi/Tardis integration. I have something going, but have a bug somewhere :)
masak snarkyboojum: cool. I'll be here for a few hours today.
moritz_ jnthn: when I say 'class Match does Associative' instead of 'class Match is Hash' I get the error "No such attribute '$!cursosr'" 12:27
jnthn: which is rather weird, because I still have the 'has $!cursor;' line in there 12:28
colomon But it very definitely *shouldn't* return a lazy list that remembers all evaluated values. 12:29
that's terribly limiting. 12:30
moritz_ ok, but assigning it to a variable should result in something that remembers all values
jnthn moritz_: Yes, it should install it in the Parrot class.
colomon moritz_: maybe.... 12:31
you're essentially saying that when you assign a lazy list to a variable, it becomes a Seq. 12:32
I think that edge between Seq and lazy list is something we still haven't figured out.
moritz_ colomon: I think there would be too much confusion if it continues to be as-is 12:33
colomon for sure we need to bring back List as a type for lazy lists.
but there needs to be a way to treat it as a raw iterator (no memory) if needed, too. 12:34
moritz_ calling .list on it
(no idea if that makes sense) 12:35
colomon I don't think it does, but I could be wrong.
I don't want to pretend to have all the answers here -- if I did, I'd already have implemented it all myself. :)
isBEKaml colomon: I can't think of a reason we need a raw iterator. AFAIK, every operation that we do, even DB calls, the results are stored in variables. Can you expand? 12:36
the variables may be lists or seqs or even iterators (mostly Iterators for DB calls, I guess) 12:37
colomon isBEKaml: by raw iterator I mean an iterator variable
(more or less)
moritz_ the point is that foo 1..10000 { ... } should never have to store these 10000 elements
colomon rather than a higher level class that wraps the iterator.
isBEKaml containing nothing? but just initalising it?
moritz_ s/foo/for/
colomon what moritz_ said.
moritz_ so we need a data type that lets us iterate without storing the evaluated items 12:38
isBEKaml ah, ok
colomon So there are two ideas, and they are in conflict.
because "raw" iterators are consumed as they are used.
which is what you want to happen when you're doing something like for 1..10000 12:39
but at the same time, for a lot of uses, you prefer they behave like a normal array -- in particular, so you can look at the values without destroying them.
look at the values twice, or N times, or whatever. 12:40
moritz_ and for $stuff uses list($stuff) under the hood
which is why I think that list() should return a raw iterator
colomon moritz_: but the problem is that if you you call list(@normal-array), it can't consume the elements when you iterate them. 12:41
moritz_ colomon: so... list(Range) => Raw iterator 12:42
list(Raw iterator) => raw iterator
list(everything else) => lazy list which remembers values
wait
list(@normal-array) is allowed to return a raw iterator 12:43
as long as @normal-array remembers its values
colomon yes, but if you do that, you can't have gather iterator return something which isn't a raw iterator.
moritz_ that's fine as long as that turns into a List upon assignment
(IMHO)
isBEKaml can we do context here? lists don't appear to be iterators? What's the differentiator?
moritz_ isBEKaml: lists are iterators that remember the evaluated elements 12:44
colomon my little guy just gave me one of his shoes and his sweatshirt.
I think he wants to go out, and doesn't understand it is pouring down rain.
isBEKaml moritz_: How do we say that _remembers_ elements? I may be missing something here, but there should be something that says _remember_ this, don't that. 12:45
colomon moritz_: what you are saying is basically that if you store an iterator, you lose the ability to forget. 12:46
that sounds right for my @a = split
but then how do I write a sub / method which gets the results of split without instantiating those in a variable? 12:47
12:47 k23z__ left
moritz_ colomon: by binding, or by using it without a varialbe 12:47
isBEKaml pass a ref? sounds odd...
moritz_ colomon: $thing.split(..).map
because you can't use it twice unless you assign it 12:48
isBEKaml: the problem is that we want to the remember/don't remeber distinction rather matically
colomon moritz_: binding maybe. 12:49
using it "without a variable" is only using it without a variable at the user's level.
colomon may run out of laptop battery at any minute, alas
moritz_ foo($str.comb) # signature binding 12:50
colomon so, we have a List type which is lazy and doesn't remember it's values.
moritz_ which we've called "raw iterator" so far 12:51
colomon but if you assign it to a variable (other than binding?) it becomes a Seq.
12:51 skangas joined
moritz_ that's my suggestion, right 12:51
colomon moritz_: well, I'm envisioning the List type as a pretty wrapper around the raw iterator.
binding here is something like my \$a = split ? 12:52
moritz_ my $a := split
colomon how do you do that in a sub signature? 12:53
moritz_ signatures already do binding
sub f($x) { }; f(3) # binds 3 to $x
isBEKaml moritz_: I'm looking at the source for the distinction. Or isn't that in yet?
moritz_ isBEKaml: no code yet, all speculative 12:54
isBEKaml moritz_: :( 12:55
12:56 kst left
isBEKaml moritz_: seeing that every list is an iterable, can we do something about new additions that contain only specific methods that allow for iterators? 12:56
12:56 kst joined 12:57 wolverian left
moritz_ isBEKaml: I don't understand 12:57
isBEKaml moritz_: for (@a) -> means a is iterable. Now we have an iterator, doing for (@iterator) should only do it lazily...
or we shouldn't allow for(@iterator) at all..
moritz_ well, we should have a working 'for' construct :-) 12:58
isBEKaml :)
moritz_ and that means that it should iterate stuff
and do it lazily 12:59
I hope we agree so far :-)
12:59 colomon left
isBEKaml or, +iterator = $stuff.split , say $_ while (+iterator); 13:00
13:00 wolverian joined
moritz_ where + is a new sigil? 13:00
isBEKaml maybe...
if that's possible to add, yes.. 13:01
moritz_ well, not quite
13:01 [mark] joined
isBEKaml I just wanted to mark the difference between an iterator and a list. + and @. 13:02
moritz_ I don't think new syntax is a good solution for our problem :( 13:03
isBEKaml then we' d to do some cross-work with what we already have. for eg. in writing a sub that gets the result of an operation returning an iterator, we have to treat the iterator here as something that can be referred to? a ref? or a callable? 13:06
but the problem here is that iterators are not methods, in which case we have to define the default operation on an iterator. 13:07
moritz_ well, there are a few basic operations that you can do to an iterator
binding, assignment, and calling methods on them
so we just have to come up with sensible behaviour for that on an iterator 13:08
isBEKaml we still have to draw the distinction between lists and iterators under the hood. for should do the Right Thing. :) 13:12
13:13 BrowserUk joined, dju left 13:14 dju joined
isBEKaml moritz_: wait, assuming we have a correct for implementation that loops over iterators, what if it exhausts itself? We shouldn't be able to use the iterator variable again, right? 13:18
13:18 dju left
moritz_ isBEKaml: if it's a variable, we should be able to use it again 13:18
my @a = $str.comb; for @a -> $x { }; # we should use @a again here
13:19 dju joined
isBEKaml moritz_: you mean once it's run beyond the last element, it should go back to before the first element? 13:19
moritz_ isBEKaml: it's more robust if 'for @a' takes a new iterator internally 13:20
so that for @a { for @a { ... } } does what you expect
13:21 dual left
isBEKaml wouldn't that make the compiler track the iterator at two different positions? 13:22
in every loop ? 13:23
moritz_ that's why we need two iterators in that case
isBEKaml internally, treat every list as an iterator? 13:25
moritz_ well, we need both 13:26
isBEKaml so we can have for look at them as just iterators and go about its business..
masak I just spent an hour or so explaining to this nilsonsfj guy on my blog why I don't think that the Perl 6 operators are scary or overboard: use.perl.org/comments.pl?sid=44783&...;cid=71948 13:32
moritz_ masak: you should write that as a blog post on its own 13:33
masak I guess I could just copy-paste the good parts, yes.
masak attempts that
moritz_++ # idea
isBEKaml masak: a blog post.. :)
masak title suggestions? 13:36
"Perl 6: zombie operators want your brains"
arnsholt "I Like Operators"?
masak hm, s/want/eat/ 13:37
isBEKaml Perl Ops ain't shadow ops! :)
masak *lol*
funny pun, but no underlying reference :)
unless we by definition consider Perl 6's ops to be ops of the Light. 13:38
isBEKaml I meant they don't scare the s* out of you..
13:38 ruoso joined
pmurias ruoso: hi 13:39
13:39 kst left 13:41 kst joined 13:47 sorear left 13:48 sorear joined 13:51 colomon joined
colomon Sorry for wandering off there, no battery and then needed to finish cooking Sunday dinner. 13:52
I got one of those simple yet maybe really nice ideas.
would it be possible to set things up so that we specify we want access to the iterator (without requiring memorization) simply by declaring that we want an object of type List? 13:53
So my List $list = split 13:54
gets you the List object.
my @list = split
gets you a Seq (or Array, or something like that)
seems like that might require a touch of magic in variable binding, but would be a very clear and simple interface 13:55
13:59 ethel joined
isBEKaml colomon: so you mean to say everything is an iterator except List ? 13:59
colomon: that loops would treat @lists lazily? and we just hold all the values in List without memoization? 14:00
colomon I'm saying that the List object is a wrapper around an iterator that allows you to process the iterator without saving the results. (so that when you say $list.get, $list no longer remembers what the value it just returned is.) 14:05
and I'm proposing that any assignment of the List object to a type other than an explicit List will convert the List to a Seq/Array (which can also contain an iterator, but are required to remember the values the iterator has). 14:06
I'm prosing the exact same thing moritz_ and I were talking about earlier, except that instead of binding being the way to signify that we want to keep the List around as a List, declaring that we want a List type explicitly would be the way to do it. 14:07
jnthn can probably say better than I whether that can be reasonably made to work.
but it seems like it would be very intuitive to use then.
afk 14:08
masak omgiblogged: "Perl 6: the Frankenstein's monster of operators" use.perl.org/~masak/journal/40339 14:09
isBEKaml masak: It's down to Frankenstein! :D 14:10
\o/
masak isBEKaml: as a non-native, I'm confused by that idiom -- "it's down to X". what does it mean? 14:11
isBEKaml masak: I'm a non-native too. I was referring to Frankenstein as a freak thing that happened as was shown in the book. 14:12
masak so you meant "it's because of Frankenstein!"?
isBEKaml and I was also referring to your pseudo vote polling for choicest blog titles! 14:13
you finally picked on one. masak++
masak ah. 14:14
well, Frankenstein's monster emerged all by itself at the last paragraph, and that settled the matter. 14:15
slavik isBEKaml: everyone calls the monster Frankenstein ... even though it's the doctor
masak to the extent that the term 'Frankenstein' is now quite overloaded.
isBEKaml masak: I think you mean overused/abused. 14:16
masak no, I meand polymorphic/polysemic :) 14:17
s/meand/meant/
isBEKaml :)
slavik: Frankenstein's origins are left ambiguous in the book.... So, I think it's common to call the monster Frankenstein.. 14:18
slavik isBEKaml: but wasn't the doc's name Frankenstein? 14:19
and the monster technically doesn't have a name?
slavik never read the book
isBEKaml slavik: I don't know if he's a doc. He was just portrayed to be very much curious about the human frame. There may be allusions to medical practice. It was so long ago that I read it. 14:20
s/frame/body/
masak 'The title of the novel refers to a scientist, Victor Frankenstein, who learns how to create life and creates a being in the likeness of man, but larger than average and more powerful. In popular culture, people have tended incorrectly to refer to the monster as "Frankenstein".' 14:21
from en.wikipedia.org/wiki/Frankenstein
isBEKaml and yes, the monster doesn't have a name..
moritz_ masak: you lost some markup while copying and pasting, namely a link
masak looks
moritz_ I briefly give the rationale for why I like this change in blog post [perl.org] from a while back.
masak ah. yes. 14:22
thanks.
14:22 fridim left
masak fixing. 14:22
there. also fixed one un-unicodified smiley. 14:24
14:25 kst left 14:27 kst joined
isBEKaml masak: Ah, yes. I remember the story now. slavik, sorry if I sounded confused. It was quite a long time ago I read it, really.. 14:33
slavik isBEKaml: I'm alive ... :-\
moritz_: what other languages have a half decent parrot compiler?
moritz_: I want to try compiling a Perl6 class to pir and then using it in another language :) 14:34
isBEKaml slavik: I was referring to the book. :D
pmurias slavik: tcl?
slavik hmm ... 14:35
anything else?
pmurias: my tcl knowledge is non-existant
moritz_ lua 14:36
14:36 M_o_C left
moritz_ mostly feature complete 14:36
isBEKaml slavik: parrot.org says python, ruby, lua,scheme and other dynamic languages..
I don't know how far they are done, though..
14:36 dual joined
slavik isBEKaml: it will prolly be a while before I try 14:36
google 14:37
google's app engine would be most beneficial if it ran on top of parrot
moritz_ it would use far too much memory :( 14:38
slavik moritz_: Java uses too much memory ...
;)
moritz_: point is, that if there were like a j2ee app server but for other languages, java would die in a year 14:39
or two
moritz_ just like cobol died immediately, as soon as alternatives were available?
and fortran? 14:40
slavik fortran is awesome
isBEKaml slavik: java already has a light weight appserver in jetty, purely coded in java. But it hasn't died yet.
slavik is jetty an actual app server? or is it more like a library?
isBEKaml it's more like a library, but can be used as an appserver. AFAIK, people use it only for personal projects. I am yet to come across it being used in production. 14:41
slavik isBEKaml: looks like it's a library, similar to resin 14:42
but can be used to make an app server
actually, jetty seems to be more like tomcat ... since it's a container.
isBEKaml you can write a standalone app in java and deploy it into jetty with a few lines of code. It'd still be a container - that's what I meant as an appserver. More of an embeddable appserver. 14:43
masak the trouble with irony is that I don't know if the resulting congratulations are sincere :P twitter.com/lamech/status/13217278101 14:44
moritz_ mentions yapsi in his talk 14:46
arnsholt "It's complete in the sense that all of the things it implements, it does." =D
isBEKaml "It's official because we have little labels with the word 'official' on them" :D 14:47
arnsholt masak: l. 51, "a server" perhaps?
Also, your style is occasionally reminds me of the prof I'm working for 14:48
slavik isBEKaml: the other thing is that Java is very verbose :(
masak: so yapsi is Perl6 in Perl6?
arnsholt He puts 'celebrate approriately' in his assignment texts from time to time ^^
masak arnsholt: typo already fixed in a later commit. thanks. :) 14:49
isBEKaml slavik: I tend to agree with this.
slavik isBEKaml: Perl/6 (IMO) would allow faster/better development of stuffs
arnsholt Oh. Never mind me then =) 14:51
isBEKaml slavik: I have a little under 3 years of experience with Java. I have encountered several issues with it - one of them being that it takes too long to deploy/undeploy apps without running into heap space issues. Another thing is that any development/deployment quickly turns bloated with too many third party libraries - jar hell. :(
slavik isBEKaml: I support app servers for a living ... I know :( 14:52
isBEKaml: the best one is when developers code against a specific version of a library belonging to an app server
and when you upgrade the app server their stuff breaks
isBEKaml slavik: yeah, I have seen that happening too. the heart breaking thing here is, some third party apps don't support backward compatiblity. We had to include two such apps with different versions in deployments. :( 14:53
slavik yeap 14:54
14:54 fda314925 joined
isBEKaml I hope P6 provides a better way or a different way.. :) 14:55
slavik easy, precompile to pbc and deploy, done 14:56
isBEKaml including hot swapping?
slavik you mean undeploy/redeploy? 14:57
isBEKaml no,just deploy the changes without asking for a full build. appserver should only look for changed files and build deps lazily.
s/deps/dependencies/
slavik isBEKaml: compile on the fly?
isBEKaml slavik: yeah. 14:58
slavik isBEKaml: hot deploy?
isBEKaml: most/all app servers do that with JSP
isBEKaml slavik: yes, again. I called that hot swapping.
slavik ahh, ok
14:58 fda314925 left 14:59 fda314925 joined
slavik isBEKaml: we,, jboss has a scanner thread that can invoke the deployer ... although it should be turned off for prod 14:59
isBEKaml: for weblogic, it has 2 parameters which make it check every so often for changed JSP files
isBEKaml slavik: I know JSP does it. but we precompile JSPs before deploying them into webserver.
slavik pagecheckseconds and something else
isBEKaml: grats, you are doing the better thing :)
and websphere has a specific checkbox to pre-compile JSP
when you deploy 15:00
isBEKaml slavik: We use a
slavik a?
isBEKaml slavik: we use glassfish - it only loads up on memory usage..
sorry, hit enter too soon..
slavik heh
isBEKaml: you mean high heap utilization? 15:01
15:01 kst left
slavik isBEKaml: there is def some heap tunning that can be done ... 15:01
isBEKaml slavik: yeah, that's on the high side.. causes webserver to go down several times. We had to manually bring that back.
slavik isBEKaml: make sure there are no memory leaks (if you are keeping objects around) and then you might need a bigger heap ... 15:02
isBEKaml slavik: we were all very confused as it was our first experience with GF. people were throwing around ideas like upgrading the server's RAM and so on.. We simply backed out of it.
s1n moritz_: that's great! is that you in the middle? what made you interested in optics (and how did you go from physics to perl6)? 15:03
slavik isBEKaml: who suggested the RAM idea?
15:03 kst joined
moritz_ s1n: I'm not on the picture 15:03
s1n moritz_: oh, i was just guessing :/ 15:04
isBEKaml slavik: sys admins, most of them. We weren't all that confident about it...
moritz_ s1n: and optics is just a nice topic, it's taught in some introductory courses... and it's stuff you can *see* (at least some of it :) 15:05
isBEKaml slavik: I think it's better we take this offline. We can just discuss perl6. :)
s1n moritz_: did you get a BS in physics? 15:06
isBEKaml slavik: even as I am only a little over a week into perl6. :)
moritz_ s1n: master with honours, and "Diplom" (a German thing)
15:06 rv2733 left
s1n moritz_: ahh, that's excellent 15:07
moritz_: i myself am finishing my masters :)
moritz_ s1n: if you're interested in a PHD position in physics here in Germany... we're still looking for fresh meat... erm, PHD students :-) 15:08
s1n moritz_: honestly, i am terrible with physics, the one subject that never stuck with me 15:09
i'm more of a math / stats / CS guy
moritz_ actually I came to Perl 6 because while I was in Scotland for one year, I was bored
and read about pugs, and joined this channel to report a broken link
I stayed.
s1n heh 15:10
slavik isBEKaml: but the point of our conversation is that a parrot app server would be friggin' awesome
dalek p-rx: f8e8a77 | Coke++ | build/Makefile.in:
Don't check for missing parrot config item.
 15:11
isBEKaml slavik: yes, that'd be great... with faster deployments, low downtime, not running into memory/heap issues.
s1n i was studying PEGs when i learned perl6 had it built-in, came to see what it was all about
slavik isBEKaml: IMO parrot should have Xmx and Xms ...
s1n only i've been mostly on the outside looking in
isBEKaml slavik: But do we really need it? I mean, _really_ need it?
slavik isBEKaml: maybe not Xms ... but def Xmx ...
isBEKaml: my thinking behind it is that you don't want your applciation code to bring entire box down for OOM killer to come in 15:12
isBEKaml: better an instance of the server die than the entire box
isBEKaml slavik: that was the idea behind introducing clustering in app servers. Did it succeed? People only complained more.. 15:13
slavik isBEKaml: well ... in java world, your objects have to be serializable ...
isBEKaml: but imagine have 2 isntances of parrot running diff apps 15:14
would you want 1 app to kill the second one?
in worst case scenario of course
s1n moritz_: out of curiosity, how'd you find me on linkedin?
masak std: for my $a (1, 2, 3) { say "OH HAI" } 15:15
p6eval std 30537: OUTPUT«===SORRY!===␤This appears to be Perl 5 code at /tmp/Mh3YJ59pV6 line 1:␤------> for ⏏my $a (1, 2, 3) { say "OH HAI" }␤Parse failed␤FAILED 00:01 108m␤»
masak STD++
slavik lol
nice!
isBEKaml slavik: ewww... I'd hate that.. If we can instantiate multiple instances of parrot, don't we want to set a top limit on heap usage. Not instance wise, but at the top-level. 15:16
masak std: for (my $a = 0; $a < 10; ++$a) { ... }
p6eval std 30537: OUTPUT«===SORRY!===␤Unsupported use of C-style "for (;;)" loop; in Perl 6 please use "loop (;;)" at /tmp/xuDGiUw02M line 1:␤------> for ⏏(my $a = 0; $a < 10; ++$a) { ... }␤Parse failed␤FAILED 00:01 109m␤»
slavik isBEKaml: you mean for all parrot instances put together?
masak std: when True { ... }
p6eval std 30537: OUTPUT«Potential difficulties:␤ Smartmatch against True always matches; if you mean to test the topic for␤ truthiness, please use :so or *.so or ?* instead at /tmp/N4RuhZiMoQ line 1:␤------> when True⏏ { ... }␤ok 00:01 107m␤»
slavik isBEKaml: for Java, it is generally accepted that you don't want a very high heap size ... 2-3GB seems to be where most people stop 15:17
isBEKaml slavik: My thinking is, instances can consume as much memory as they want, they can even eat into the memory used by other instances (provided, it's freely available). But don't consume more than this to bring the appserver down. Just throw some other exception or redirect to a temporarily unavailabl epage.
jnthn rakudo: for my $a (1, 2, 3) { say "OH HAI" }
p6eval rakudo c4857a: OUTPUT«This appears to be Perl 5 code at line 11, near "my $a (1, "␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
slavik isBEKaml: so you mean to tell 2 instances of parrot that together they are not allowed more than X memory and to share it between themselves?
isBEKaml slavik: yes, that's the basic idea..
slavik isBEKaml: I think in that case ... there might be some shell/system foo that could accomplish this ... 15:18
isBEKaml slavik: the thinking is to allow dynamic usage of memory..
slavik isBEKaml: the exception would be either out of memory or no extra memory
isBEKaml: set max memory on user account specified for parrot ... or something 15:19
this should be a #parrot discussion methings
thinks*
isBEKaml slavik: OOM sounds scary even in perl world, I'd go with no extra memory. But clueless people would just increase heap and get on with it. I'm worried about this..
slavik isBEKaml: that's why clueless people are clueless
isBEKaml: apps should also be able to handle OOM
moritz_ most do. If you consider "dying" a form of "handling" :-) 15:20
pugssvn r30538 | jnthn++ | [t/spec] Fudge a test we were passing bogusly.
slavik moritz_: handling gracefully ...
isBEKaml slavik: that's why I suggested a redirect - if apps don't handle OOM or other memory issues..
slavik isBEKaml: you mean handle on server level instead of app level?
15:21 nihiliad joined, _TrueFX3 joined
slavik either way, the thread handling the request needs to be killed ... 15:21
so I guess server level
isBEKaml slavik: if there's no handling in the app level, do it at the sever level.
slavik unless the thread can handle it and not ask for as much objects ...
isBEKaml: the other thing that I see in java is that apps do try catch and then just print the stacktrace in catch ... without any output of what they are trying to do 15:22
s1n downside of never having to call malloc in perl is never having to think of OOM :/
isBEKaml slavik: that's just clueless devs coding mindlessly..
slavik: I have seen worse happening. catch and silently pass over..
dalek kudo: 925629d | jonathan++ | src/core/operators.pm:
Make $x does (R1, R2) work.
 15:23
kudo: b2476f7 | jonathan++ | src/Perl6/Grammar.pm:
Add a missing check so our scoped packages work as they should; allows us to run submethods.t again.
kudo: 72ef184 | jonathan++ | src/metamodel/Role (2 files):
More fixing of multis composed from multiple roles.
kudo: 7d700b4 | jonathan++ | t/spectest.data:
Turn on S12-methods/submethods.t.
slavik isBEKaml: I have seen that as well
isBEKaml slavik: I have even seen NPEs and NFEs passed over. :(
slavik s1n: you mean not even in the java style new?
isBEKaml: NPE/NFE?
isBEKaml NullPointerException and NumberFormatException 15:24
data errors! :(
s1n slavik: java style new is actually the c++ style new
which is just a wrapper for malloc()
slavik s1n: didn't know about the first ... ewwwwwwwwww
isBEKaml: yes ... :( 15:25
15:25 camenix left
slavik isBEKaml: it is a reason I dislike java ... 15:25
s1n slavik: nothing gross about having to know more about memory
slavik s1n: not that ... but that java's new is also a wrapper around malloc ...
15:25 xinemac left
slavik s1n: but what java developer knows about memory? 15:26
isBEKaml slavik: I'm just trying to dissociate myself with java and do something else. This was the primary reason I was looking at perl6..
s1n slavik: well, not so much, but what's wrong with malloc?
slavik s1n: nothing wrong with malloc
s1n: everything wrong with people writing java code
s1n slavik: a java dev _has_ to know about memory, otherwise you get 900M GF apps
slavik s1n: huh? the last 3 words confuse me
s1n nothing wrong with java itself, it's a pretty/verbose version of c++
slavik s1n: I don't like the verbosity :( 15:27
isBEKaml s1n: java devs just gloss over app memory issues. They tend to think it's the job of java's GC..
s1n slavik: lots of apps ignore memory and end up consuming all your memory for no reason
isBEKaml: agreed, which is a problem i don't dispute
slavik s1n: isBEKaml: we all agree there :)
s1n: can you explain what you mean by "900M GF apps"
isBEKaml slavik: I think he means 900 million GlassFish apps.. 15:28
s1n slavik: write a hello world j2ee app, then it'll make sense :)
no
900MB sorry
slavik s1n: I like perl because it lets me think in the messed up functional way that I tend to think in ...
masak rakudo: class A {}; class B { has A $.foo .= new }
p6eval rakudo c4857a: OUTPUT«too few positional arguments: 1 passed, 2 (or more) expected␤current instr.: '!dispatch_.=' pc 440 (src/glue/dispatch.pir:118)␤»
slavik s1n: it will take 1 month to do that ...
masak submits rakudobug
alpha: class A {}; class B { has A $.foo .= new }
p6eval alpha 30e0ed: OUTPUT«error:imcc:syntax error, unexpected '\n'␤ in file 'EVAL_25' line 186␤Null PMC access in find_method('new')␤in Main (file <unknown>, line <unknown>)␤»
s1n slavik: well, it's still turing complete, so it's all the same to me
slavik s1n: ... :-\
s1n: that's why we need to care about amortized time :P 15:29
15:29 christine left
s1n i dunno, the fanboyism for a language is pointless to me as long as the language is turing complete, you can express all the same ideas but one language may need more work than others 15:30
either way, that work is either on the app dev, the library dev, or the compiler dev
the perl app dev has much less work because the library and compiler devs did many things for them 15:31
java is the opposite
isBEKaml s1n: I don't dispute your point. But we are humans too and we try to solve problems as humanely as possible. if it's just too much work in one language, we can take another language. But if it's the same tiring job over and over again, we just jump boats..
slavik s1n: that is exactly my point ... I would factor in language dev into that :) 15:32
TimToady++ :)
moritz_ torment the compiler writer on behalf of the user
s1n yeah, that's fine, having a language take care of things the app/lib dev doesn't want to do is nice, i don't argue
15:32 kst left
s1n still work someone has to do :) 15:32
java is fine if you want to do that work yourself :/ 15:33
15:33 kst joined
s1n or you don't mind the vm bloat 15:33
isBEKaml s1n: I tend to think of java world as divided into library and app devs - compiler devs are almost always never disturbed with issues... library devs don't care about app devs and app devs are cursing lib devs for not providing more API functionality.. 15:34
s1n well, lib devs provide plenty, just _way_ too complex to be quickly/easily usable
isBEKaml and app devs either go searching for more third party apps or hand-code their functionality themselves (most don't do this, though)
s1n j2ee is a giant mess
but it can still make webapps (poorly though) 15:35
slavik CPAN ftw
isBEKaml s1n: well, part of that is due to the long and protracted java community process.. most of them just drop out at the elongated affair...
slavik I like searching CPAN for "thingIneed Simple" ... usually gives me a module that take like 2 lines to return some kind of a structure or whatever I need to work with data
s1n slavik: i recall an article somewhere comparing the modern developer to a child building with legos, how depressing that seems 15:36
isBEKaml s1n: it's too much work to even ask for a simple functionality..
slavik s1n: have a link? I'd be interested in reading
15:36 _TrueFX3 left
s1n isBEKaml: i agree, i had to make a j2ee app and i almost wanted to cry 15:37
slavik: i'd have to look, i don't remember where i read it
moritz_ kinda still likes lego
slavik I needed to parse XML the other day ... XML::SImple ... give and XML file or string to XML::Simple::XMLin() and it returns a hash ... done ... easy ... 2 lines of code ...
looking at java libraries made me cry and forget what I wanted to do in the first place 15:38
same with Config::Simple
or maybe just Config
15:39 BrowserUk left
s1n slavik: but nothing about perl makes that library particularly easy or great, you could do that in C if you were so inclined 15:39
isBEKaml slavik: JAXP hurt so much I never did XML again in java..
s1n (_could_, though i wouldn't) 15:40
slavik s1n: I think it has to do with the culture behind the language
moritz_ could you?
first you'd have to write a struct that can either be an array or a hash or a string
s1n slavik: true, too many people have their hands in java that it usually turns out a mess
CPAN is usually one or a small group of people on a module
moritz_ and then accessors for that
s1n easier for the module to stay pure to the design
slavik s1n: I think we agree that many things are easier to do in Perl vs. Java
s1n well some things 15:41
slavik Easy things easier, difficult things not impossible.
:)
isBEKaml :)
15:41 christine joined
slavik which is why I can't wait for feature complete version of rakudo 15:41
s1n again, i heavily disagree with language favoritism, it all comes down to what you're doing
i like perl6 because of the built-in PEG, that's just cool to me 15:42
slavik PEG?
s1n but many other languages can be used to express the idea similarly
parsomg expression grammar 15:43
slavik lol
s1n oops
parsing
slavik s/om/in/ ?
s1n hands were offset there
slavik brb 15:44
15:45 BrowserUk joined 15:46 M_o_C joined, BrowserUk left 15:59 [mark] left
masak I was just about to turn to the poker hand example for the book, when I realised that this bug is still not looked at: rt.perl.org/rt3/Ticket/Display.html?id=74276 16:00
it's the one where defininf a class starting with the letter 'Q' and then using it doesn't work.
seems it's still open. has someone had a chance to look at it yet? 16:01
otherwise, I might give it a shot tonight.
16:01 kst left
arnsholt has a peek 16:01
I've just spent the better part of half an hour fighting with LaTeX and layout
16:02 kst joined
arnsholt Time for something completely different 16:02
isBEKaml afk &
masak is it just me, or has the Rakudo build actually gotten faster lately? 16:03
moritz_ faster? you've got to be kidding 16:04
masak maybe I misremember.
src/gen/core.pir is still the big bottleneck, it seems. but it got through Actions.pm quite quickly. 16:05
moritz_ masak: are you tracking rakudo HEAD right now?
masak yes. 16:06
but I still build Parrot independently.
moritz_ but with a recommended (or newer) parrot version?
16:06 plobsing joined
masak yes. right now the recommended one, r46148. 16:06
moritz_ with a *very* evil hack, I've got non-quantified, named captures working on Match objects 16:08
and with *very* evil I mean introducing a method in Mu 16:10
masak somehow the fact that the class is called 'Mu' makes it feel less evil :)
moritz_ I was about to say that I introduced a global method 16:11
masak now that does have an evil ring to it.
arnsholt nqp: class Foo {}; my $a := Foo.new;
p6eval nqp: ( no output )
arnsholt nqp: class Quad {}; my $a := Quad.new; 16:12
p6eval nqp: OUTPUT«Confused at line 1, near "my $a := Q"␤current instr.: 'parrot;HLL;Grammar;panic' pc 500 (src/cheats/hll-grammar.pir:197)␤»
arnsholt masak: ^^
It's actually an NQP bug
slavik := ???
moritz_ binding
masak arnsholt: good to know. thanks.
arnsholt (Because Rakudo uses NQP's quoting code)
I'm gonna have a look when I get home
slavik moritz_: but = is assignment?
masak ++arnsholt
arnsholt Right now I'm waiting for the bus =)
16:12 tri1 joined
arnsholt slavik: := is binding, which is subtly different form assignment 16:13
slavik is there a short explanation on the difference?
16:13 tri1 left, lue left
moritz_ binding replaces the container 16:13
it's like aliasing
slavik whereas assigment makes a copy? 16:14
moritz_ yes-ish 16:15
at least of the reference
masak hm, nilsonsfj has replied yet again: use.perl.org/comments.pl?sid=44783&cid=71949 16:17
this is the first comment of his that I'm not tempted to reply to.
he's still very cogent, but I just can't relate to what he's saying because I don't have those experiences myself.
colomon how can you reply when you try to explain things clearly, and his response is batshit fucking insane? 16:19
where else is there to go?
masak I'm not sure I agree that he's insane. he's just caught up in his perspective. 16:20
moritz_ some things just can't be changed easily
colomon sorry, his response is "batshit fucking insane". 16:21
isBEKaml he doesn't appear to be insane. He just appears to be intimidated by the enormity of changes in P6. ;) 16:22
colomon (ie those are his words.)
slavik I quite like twidgils ... IMO the whole sigil/twidgil thing is better in Perl6
ie: $*Blah means that Blah is a global scalar 16:23
colomon he's not making a reasoned response, he's just utterly dismissive.
isBEKaml Just as I was when I first saw it on perl6.org, but quickly got over it by the friendliness seen on this channel. I'm still motivated to learn more here. :)
masak isBEKaml: and annoyed that the only effects of Perl 6 on his life are people's negative reactions.
moritz_ isBEKaml: which part of perl6.org scared you?
isBEKaml masak: sadly, people tend to go by others' judgements.. :( 16:24
moritz_: I didn't mean perl6.org, I meant the fact that perl6 is an entirely new language and there are lots of new unicode ops. I was thinking, well, as long as they provide ascii ops alonside unicode ones, I can go with it. 16:25
:)
moritz_ isBEKaml: ok. I just thought there might be potential to reduce scariness from perl6.org 16:26
isBEKaml moritz_: I think I said this before, perl6.org doesn't exactly say what parrot, pugs and rakudo are and how they stack up on the perl6 ecosystem. Can that be fixed? I was misled into thinking all perl6 implementations are on VMs..
that includes thinking pugs and rakudo were VMs, when they are not.. 16:28
moritz_ isBEKaml: usually we try not to special-case rakudo or parrot on perl6.org ... but maybe I can come up with a clean explanation on /compilers/
isBEKaml moritz_: no, not showcasing them. Most people, I think, come to perl6.org with a fairly informed knowledge about parrot, pugs but just don't get the "big picture", if you know what I mean.. 16:29
eww, s/informed/rudimentary/ 16:30
moritz_ oh, perl6.org got stuck in an svn lock :( 16:31
isBEKaml oh, does that happen? I just find it odd to see svn locks unless the svn account was disabled/locked out or something.. 16:35
16:35 pmurias left
slavik moritz_: IMO, the periodic table of operations should be part of the docs :) 16:35
moritz_ svn locks its .svn
isBEKaml oh, that.. 16:36
moritz_ slavik: it is
isBEKaml moritz_: does svn cleanup help?
moritz_ isBEKaml: yes 16:37
it's udpated again
isBEKaml moritz_: this would revert all your changes and release locks. you'd have to do your changes again..
moritz_ there should be no changes on that server
16:38 kst left, justatheory joined 16:39 kst joined
moritz_ for some reasons not quite understandble to me, quantified named captures refuse to work in my branch 16:40
masak in the end, I did reply anyway: use.perl.org/comments.pl?sid=44783&cid=71951
time to go nommin & 16:41
16:41 masak left
isBEKaml nomming? 16:41
moritz_ eating 16:42
isBEKaml oh. :)
masak's was a great reply. Not a hint of frustration/disappointment... :) masak++ 16:43
16:45 slavik left 16:51 BrowserUk joined 16:58 envi^home left 17:04 takadonet1 left 17:05 rgrau left
arnsholt What's the proper way to add info to an RT ticket? Reply or comment? 17:07
17:16 slavik joined 17:22 slavik left
moritz_ yes 17:22
doesn't really make a difference 17:23
I usually use 'reply'
17:24 patrickas joined 17:25 kst left
arnsholt Cool 17:25
17:25 rv2733 joined, kst joined
Juerd arnsholt: If you're repyling to a specific item, use reply. If you're adding information but not in response to a previous part, use comment. 17:26
arnsholt Oh, right. That makes sense
Juerd IIRC, replies go to the bug reporter, comments do not. I'm not sure.
17:27 slavik joined 17:33 molaf joined 17:37 BrowserUk left
isBEKaml how do I invoke a sub of a class? I'm looking at the Date module, I find that I can just invoke methods but not subs. Are they private? 17:39
17:39 opx^away left
isBEKaml rakudo: use Date; say Date.assert-valid-date(2010,04,31); 17:40
p6eval rakudo c4857a: OUTPUT«Unable to find module 'Date'.␤current instr.: 'perl6;Perl6;Module;Loader;need' pc 35817 (src/gen/role_pm.pir:932)␤»
isBEKaml here assert-valid-date is a sub. 17:41
17:41 k23z__ joined 17:54 renormalist left
arnsholt I'm starting to suspect that this is an issue with LTM 17:57
Which means I need to understand that
arnsholt heads for S05
17:59 rv2733 left 18:01 renormalist joined, renormalist left 18:03 kst left 18:04 kst joined 18:05 Freakezoid joined
lisppaste3 Panzen pasted "Perl Regexp find the n word and replace it." at paste.lisp.org/display/98684 18:10
Freakezoid can anyone tell me how i can replace the n word in a text? my approach is at but when a word occourrs more than one time, all occurrences will be replaced :( 18:13
source is at: paste.lisp.org/display/98684 18:14
moritz_ Freakezoid: this channel is about Perl 6 development
Freakezoid oh sorry
moritz_ isBEKaml: subs are my-scoped by default
isBEKaml: so yes, private
18:15 Freakezoid left
isBEKaml moritz_: thanks. I got that impression as I was reading the spec after posing it here. :) (though it's bit a vague, but lot clearer than the java spec docs ;) ) 18:16
moritz_ rakudo: Date.new('2010-05-00') 18:17
p6eval rakudo c4857a: OUTPUT«Invalid date: day < 1␤current instr.: '&die' pc 17391 (src/builtins/Junction.pir:410)␤»
moritz_ rakudo: Date.new('2010-04-31')
p6eval rakudo c4857a: OUTPUT«Invalid date: day > 30␤current instr.: '&die' pc 17391 (src/builtins/Junction.pir:410)␤»
moritz_ that's how you call assert-valid-date :-)
isBEKaml moritz_: :)
arnsholt moritz_: Should `Q' being a more specific prefix for quote than name somehow block backtracking when quote fails? 18:18
moritz_ arnsholt: not sure... 18:19
arnsholt: I'm trying to add a >> after Q
arnsholt A >>?
18:20 dju left
moritz_ >> is right word boundary 18:20
arnsholt: seems to help 18:21
isBEKaml moritz_: There are some methods that are chained to subs in the Date module. Can't we have something for assert-valid-date too?
arnsholt moritz_: Good, good 18:22
isBEKaml rakudo: my $ff= Date.new(2010,05,30); say $ff.leap-year();
p6eval rakudo c4857a: OUTPUT«0␤»
moritz_ isBEKaml: why? just call .new(), and if it fails, you know it's not a valid date
isBEKaml rakudo: my $ff= Date.new(2010,05,30); say "leap" unless $ff.leap-year();
p6eval rakudo c4857a: OUTPUT«leap␤»
isBEKaml moritz_: sure, that's right. Talking about consistency, when checks are there for leap year, why not dates too? 18:23
moritz_ arnsholt: will push soon, after more testing 18:24
arnsholt Cool. NQP can probably be fixed in the same manner
moritz_ I am talking about NQP
18:24 renormalist joined
arnsholt Oh, of course ^^ 18:25
*facepalm*
colomon 31806 passing tests, woo-hoo!
slavik colomon: out of? 18:26
moritz_ approx. 40k
colomon 38898, according to tools/test_summary.pl
isBEKaml cool.. 18:27
moritz_ std: mfoof
p6eval std 30538: OUTPUT«===SORRY!===␤Undeclared routine:␤ 'mfoof' used at line 1␤Check failed␤FAILED 00:01 109m␤»
colomon not sure how real that number is, however... does it count tests that are commented out in spectest.data?
moritz_ std: m:g foof
p6eval std 30538: OUTPUT«===SORRY!===␤Alphanumeric character is not allowed as delimiter at /tmp/9EvPIOw2X7 line 1:␤------> m:g ⏏foof␤Parse failed␤FAILED 00:01 107m␤»
moritz_ colomon: it should count all
std: Qbarb
dalek p-rx: ec48321 | moritz++ | (2 files):
allow sub names to start with Q
p6eval std 30538: OUTPUT«===SORRY!===␤Undeclared name:␤ 'Qbarb' used at line 1␤Check failed␤FAILED 00:01 107m␤»
p-rx: d6815bf | moritz++ | src/stage0/ (3 files):
update stage0 files
moritz_ rakudo: Qbarb 18:28
p6eval rakudo c4857a: OUTPUT«Confused at line 11, near "Qbarb"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
18:31 Transformer joined 18:34 Transformer left
arnsholt nqp: say(pir::typeof__sp("{1}")) 18:37
p6eval nqp: OUTPUT«Null PMC access in set_string_native()␤current instr.: '_block11' pc 0 (EVAL_1:6)␤»
18:39 lichtkind joined
lichtkind hai people 18:40
phenny lichtkind: 30 Apr 15:33Z <TimToady> tell lichtkind in best-of tablet, for 1..15:by(-2) -> $i {...}
lichtkind: 30 Apr 15:35Z <TimToady> tell lichtkind ...is wrong two ways, no ':by' anymore, and range doesn't autoreverse. use something like for 15,13...1 -> $i {...}
lichtkind TimToady: damian already wrote me and noted that, its fixed since yesterday :)
moritz_ lichtkind: github.com/moritz/perltalk might interest you 18:41
lichtkind moritz_: ive seen your p-c.de post and have link on your slides
moritz_ lichtkind: ok 18:42
lichtkind moritz_: a possible intro for the book im writing on is also still in the pipe
moritz_ you're writing a book?
lichtkind moritz_: no the p6 book your writing on 18:43
moritz_ ah, I was misparsing your statement :-)
lichtkind i have just alternative ideas for the beginning and post that
18:46 M_o_C left 18:47 dukeleto joined
arnsholt nqp: say(pir::typeof__pp("{1}")) 18:50
p6eval nqp: OUTPUT«Null PMC access in set_string_native()␤current instr.: '_block11' pc 0 (EVAL_1:6)␤»
moritz_ nqp: say(pir::typeof__pp("foo{1}"))
p6eval nqp: OUTPUT«String␤»
18:52 rgrau joined
jnthn will be, like, totally offline for next 3 days. 18:56
Going to the countryside, camping. :-)
arnsholt How will you ever survive? =)
jnthn ;-) 18:57
arnsholt moritz_: In a nutshell, it's the string interpolation issue I prodded during the CPH hackathon
moritz_ arnsholt: thought so 18:58
18:58 lookieloop joined
arnsholt I figured I'd add a test to the test suite for it 18:58
18:58 plobsing left
moritz_ that would be nice 18:58
arnsholt But pir::typeof blows up, which is inconvenient =) 19:00
moritz_ isa?
nqp: say(pir::isa_pp("foo{1}", 'String'))
p6eval nqp: OUTPUT«error:imcc:syntax error, unexpected PREG, expecting '(' ('$P15')␤ in file 'EVAL_1' line 11␤Could not find non-existent sub isa_pp␤current instr.: '_block11' pc 0 (EVAL_1:6)␤»
dalek kudo: fad910a | moritz++ | src/Perl6/Grammar.pm:
Attemp to allow names starting with q, qq, Q etc.
moritz_ nqp: say(pir::isa_pp("foo{1}", String)) 19:01
p6eval nqp: OUTPUT«error:imcc:syntax error, unexpected PREG, expecting '(' ('$P15')␤ in file 'EVAL_1' line 10␤Could not find non-existent sub isa_pp␤current instr.: '_block11' pc 0 (EVAL_1:6)␤»
arnsholt nqp: say(pir::isa__ips("{1}", "String"))
p6eval nqp: OUTPUT«Null PMC access in set_string_native()␤current instr.: '_block11' pc 0 (EVAL_1:6)␤»
arnsholt nqp: say(pir::isa__ipp("{1}", String)) 19:03
p6eval nqp: OUTPUT«Null PMC access in set_string_native()␤current instr.: '_block11' pc 0 (EVAL_1:6)␤»
arnsholt nqp: say(pir::box__pi(pir::isa__ipp("{1}", String))) 19:04
p6eval nqp: OUTPUT«Null PMC access in set_string_native()␤current instr.: '_block11' pc 0 (EVAL_1:6)␤»
19:04 synth left
arnsholt Yeah. It doesn't like the assignment of the string register result of "{1}" to a PMC register 19:04
moritz_ so probably the result of the block should be coereced to String first 19:06
19:08 nadim_ left
moritz_ rakudo: say < >.WHAT 19:12
p6eval rakudo c4857a: OUTPUT«Str()␤»
19:15 molaf left 19:21 vegard joined
dalek kudo: f82af0b | jonathan++ | src/ (2 files):
Somewhat unbreak candidate introspection on multis.
 19:30
19:30 synth joined 19:31 synth left
pugssvn r30539 | jnthn++ | [t/spec] Unfudge some spectests. 19:31
19:31 synth joined 19:34 lue joined
pugssvn r30540 | moritz++ | [t/spec] unfudge tests for regexe in boolean context 19:36
moritz_ one weird behaviour down, 616 to go :/ 19:37
19:40 kst left, kst joined
dalek kudo: 8abe0a6 | moritz++ | src/core/Regex.pm:
Regexes in boolean context match against $_
 19:41
19:41 nadim_ joined
moritz_ std: class A { has $!a; method foo { :$!a } } 19:43
p6eval std 30538: OUTPUT«ok 00:01 110m␤»
moritz_ rakudo: class A { has $!a; method foo { :$!a } }
p6eval rakudo c4857a: OUTPUT«Symbol '$!a' not predeclared in foo␤current instr.: 'perl6;PCT;HLLCompiler;panic' pc 152 (compilers/pct/src/PCT/HLLCompiler.pir:108)␤»
19:44 chitragupt left 19:47 gurjeet left
sorear eternaleye: No hygenic macros. 19:49
Hygenic macros are much, much harder.
Mainly because Rakudo doesn't have a true parser - we do a lot of symbol hacking during what should be the parse phase
consider quasi { has $x }
colomon TimToady: ping?
sorear if we just parse the quasi, the 'has' will go into the macro itself! not good 19:50
TimToady wants to defer symbol injection using something almost, but not entirely unlike, a role
moritz_ sorear: and there's a good reason for that... otherwise we couldn't do most of the parsing in syntax-check-only mode
sorear that'll take some time to work out the details of.
jnthn moritz_: I was just about to say that. :-) 19:51
STD does the same kinda thing. But there's an $*IN_QUASI (IIRC) flag that can be checked. 19:52
I would guess that suppresses the name stuff.
Or do the right kind of thing.
19:52 BrowserUk joined
moritz_ rakudo: my $x = 3; say (:$x).perl 19:54
p6eval rakudo c4857a: OUTPUT«"x" => 3␤»
lichtkind moritz_: what is the right term for www.perlfoundation.org/perl6/index...._sequences ? 19:59
moritz_ escape sequences. Or backslash escapes. Or so 20:00
lichtkind moritz_: so escape sequences sounds good enough to mention it under that titel in th index 20:01
moritz_ yes
lichtkind thanks
20:02 nadim_ left, nadim_ joined
vegard hi. something went wrong with the perl6 book pdf for april. 20:05
moritz_ please elaborate
vegard starting from page 14, the code samples are messed up
either that or perl6 code looks truly horrible ;-) 20:06
20:06 plobsing joined
moritz_ nope, it's really the pdf 20:06
vegard: try moritz.faui2k3.org/tmp/book.pdf instead 20:07
vegard looks better, thanks
20:07 snarkyboojum left
colomon rakudo: say 31 !% 3 20:10
p6eval rakudo c4857a: OUTPUT«0␤»
colomon rakudo: say 30 !% 3
p6eval rakudo c4857a: OUTPUT«1␤»
colomon rakudo: say 32 !% 3
p6eval rakudo c4857a: OUTPUT«0␤»
20:15 isBEKaml left
moritz_ rakudo: my $x ||= 3; say $x 20:15
p6eval rakudo c4857a: OUTPUT«3␤»
moritz_ rakudo: my $x &&= 3; say $x
p6eval rakudo c4857a: OUTPUT«Null PMC access in invoke()␤current instr.: '!assign_metaop' pc 17195 (src/builtins/Junction.pir:315)␤»
moritz_ std: my($x) 20:17
p6eval std 30540: OUTPUT«===SORRY!===␤Variable $x is not predeclared at /tmp/pQngSFVGBQ line 1:␤------> my($x⏏)␤Undeclared routine:␤ 'my' used at line 1␤Check failed␤FAILED 00:01 109m␤»
moritz_ rakudo: class X {method x (*@_) {say @_} }; X.new.x(5) 20:19
p6eval rakudo c4857a: OUTPUT«5␤»
colomon turned up a Rakudobug locally just in time to be off to "Folk Music Sunday" 20:20
20:24 chitragupt joined 20:25 kst left 20:26 kst joined, lichtkind left
moritz_ rakudo: role Str { } 20:29
p6eval rakudo c4857a: ( no output )
moritz_ rakudo: say [^^] 1, 1, 1 20:31
p6eval rakudo c4857a: OUTPUT«Could not find sub &infix:<^^>␤current instr.: '' pc -1 ((unknown file):-1)␤»
moritz_ rakudo: say (+'23').WHAT 20:33
p6eval rakudo c4857a: OUTPUT«Num()␤»
sorear my spectest failed with unexpected todo passing... 20:34
moritz_ hopes that was not the reason for the fail
sorear S05-mass/properties-general, several # TODO Uninvestigated nqp-rx regression
moritz_ that's platform dependent :(
20:35 chitragupt left, chitragupt joined
moritz_ rakudo: if 5 > my $a = 4 { say $a } 20:35
pugssvn r30541 | moritz++ | [t/spec] rakudo unfudges
p6eval rakudo c4857a: OUTPUT«Use of uninitalized value in numeric context␤Cannot assign to readonly value␤current instr.: '&die' pc 17391 (src/builtins/Junction.pir:410)␤»
moritz_ rakudo: my $x = :a<5>; say $x.map({.key => .value + 1}).perl 20:37
p6eval rakudo c4857a: OUTPUT«Method 'key' not found for invocant of class ''␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
moritz_ rakudo: my $x = :a<5>; say $x.map({.WHAT}).perl
p6eval rakudo c4857a: OUTPUT«(Pair)␤»
20:42 chitragupt left 20:43 pmurias joined 20:44 meppl left, chitragupt joined
sorear oh? hmm 20:58
std: quasi { has $x }
p6eval std 30540: OUTPUT«ok 00:03 114m␤»
20:58 chitragupt left
sorear std: quasi { my $x; $y } 20:58
p6eval std 30540: OUTPUT«===SORRY!===␤Variable $y is not predeclared at /tmp/jYitZNhroH line 1:␤------> quasi { my $x; $y⏏ }␤Check failed␤FAILED 00:01 109m␤»
sorear I had no idea STD implemented quasi 20:59
20:59 chitragupt joined
sorear I guess I need to learn how STD.pm works now 20:59
20:59 BrowserUk left
sorear (does anybody have a clone of my rakudo repo? I'd like to do some branch rearranging) 21:00
21:01 orafu left, orafu joined
jnthn gets a crappy first cut at enums that may make The Chese work some more together. 21:05
sorear Chese? 21:07
jnthn Cheese 21:11
The app cheese.
:-)
21:11 Guest81117 left
jnthn Might make masak++ happy. :-) 21:11
Well, until he realizes how little of them works so far. :-) 21:12
But it should cover all the basic cases.
cognominal what is the purpose of YOU_ARE_HERE? 21:18
21:19 lookieloop left 21:20 pmurias left 21:21 camenix joined 21:25 camenix left
sorear cognominal: it's a placeholder for user code 21:29
Perl 6 has a circumlude, not a prelude
the initial definitions are regarded as an outer lexical scope 21:30
jnthn: But what *is* app cheese?
jnthn sorear: It's the existing Perl 6 code that runs/has run on Rakudo.
cognominal sorear: I still don't get it :( 21:33
21:34 camenix joined
lue o hai! 21:34
21:36 xinemac joined
jnthn cognominal: Programs run with the setting as its outer lexical scope. 21:37
YOU_ARE_HERE represents a point in the lexical scope where the program should be "inserted". 21:38
cognominal thx
dalek kudo: a4112c2 | jonathan++ | src/ (2 files):
First hacky cut of named enums. Mostly just works like a bunch of constant declarations, kinda. It's certainly not what we want in the long run, but it removes a blocker for much of the app cheese.
 21:42
kudo: 7ec16e1 | jonathan++ | src/core/Regex.pm:
Merge branch 'master' of github.com:rakudo/rakudo
sorear jnthn: Can you tell me more about the etymology of app cheese? 21:45
jnthn sorear: At a conference in Oslo a little while back, masak gave a talk in which he depicted Rakudo as being kinda like swiss cheese (the kind of holes in it), where the cheese represents what is implemented feature wise and the holes relate to the bugs. 21:46
That's where this all derives from. :-) 21:47
sorear aha.
jnthn Yes, it's not an obvious one. :-)
sorear goes back off to study STD's implementation of hygiene
lue swiss cheese is very holy. It's not a putdown :)
jnthn tries to fix sorearbug
sorear the crazy BEGIN one? 21:48
jnthn It's Swiss, it must be awesome. :-)
jnthn likes Switzerland lots.
sorear macro work has been moved to topic/macros 21:50
jnthn :-) 21:51
sorear so my 'master' is now consistent with rakudo master 21:52
21:56 snarkyboojum joined 21:59 justatheory left
jnthn > say BEGIN { rand } 21:59
0.466257142692001
Whee.
lue Just so I know, what's with all the capitalization? BEGIN, PAST, and so on. 22:00
I know they're the deep magic, but that's it.
jnthn lue: Well, BEGIN is uppercase because all of the phasers are. 22:01
Phasers are things that run at different times in your program. 22:02
22:02 nadim_ left, iblechbot left
jnthn Rather than just the normal flow of execution. 22:02
sorear: Think I have a fix for the BEGIN issue you filed in RT. :-) 22:03
22:04 rv2733 joined
lue rakudo: say $a; BEGIN { my $a = 4 } 22:05
p6eval rakudo c4857a: OUTPUT«Symbol '$a' not predeclared in <anonymous>␤current instr.: 'perl6;PCT;HLLCompiler;panic' pc 152 (compilers/pct/src/PCT/HLLCompiler.pir:108)␤»
lue guess it doesn't work like that :)
22:05 PacoLinux left
jnthn No 22:08
Plus BEGIN + outer lexicals = BOOM in Rakudo at the moment.
sorear jnthn: I'll have to study your fix so I can port it to macros! 22:09
jnthn ;-) 22:10
sorear++ # hacking on scary hard fun stuff
sorear have you seen what I've got already? 22:14
tribble, sibble? I think I need some explanations here 22:19
jnthn s/// 22:21
and tr///
They're related to those.
But actually I didn't really like that formulation so much so I did it in a simpler way in Rakudo.
22:21 xinming left
jnthn We may need to end up doing what STD does at some point. 22:21
sorear: I haven't. I'm sorta preparing for a camping trip amongst writing easy-ish patches :) 22:22
lue Isn't STD(-esque) our goal in the first place :)
jnthn I will certainly look at it when I get back though.
22:22 xinming joined
jnthn lue: Well, it's a process of convergence rather than STD being unchangable gospel. :-) 22:22
STD-esque though for sure.
We may find there's a reason to do it the way STD has gone yet. 22:23
sorear lue: In a few years, there will be one Perl 6 grammar, and it will be called STD
jnthn And our formulation is pretty close.
dalek kudo: 2acc11d | jonathan++ | src/Perl6/Actions.pm:
Fix problem with calling things in the setting from BEGIN blocks.
22:23 takadonet1 joined
sorear it will have elements of current STD, current Rakudo, current Yapsi, and many things yet to be invented 22:23
22:26 patrickas left
sorear perl6.vim has ... trouble with STD 22:27
22:27 kst left
lue as <dalek> would say, "The Fiiinal ImplementaTION!" 22:27
22:30 kst joined
sorear jnthn: Can you please explain how your last patch works? 22:31
jnthn Well, I drank beer and then the patch just happend...
;-) 22:32
Basically though...there's a YOU_ARE_HERE term.
We compile it down to some "magic" that installs a sub that knows how to install things within th elexical scope of the setting.
sorear So, with that patch, BEGIN cannot be used inside the setting? 22:33
jnthn So all the patch does is not run the BEGIN block, but instead call the "magic thing" and passes the block.
sorear I see
jnthn sorear: Erm. That..may well be the case.
Welcome to the delightful world of bootstrapping. 22:34
sorear :)
jnthn Anyway, it makes BEGIN blocks rather more useful now. :-)
22:35 patrickas joined
sorear What does .blocktype('declaration') do? 22:37
jnthn PAST supports two types of block
Those that we run immediately 22:38
And those that we don't
So in
say 42; { say 69; } say 100;
The block is immediate.
In
my $x = { say 99 }; 22:39
The block type is declaration.
Just controls the code gen.
I guess in a sense, immediate is sugar for a declaration and a call to it.
But it's kinda useful to have.
sorear really ought to learn more PAST for this deep compiler hacking stuff 22:40
jnthn I found it fairly easy to pick up. 22:41
Mostly through using it. :-)
22:41 chitragupt left 22:42 stepnem left
sorear Yeah, me too 22:44
22:44 chitragupt joined
sorear but I have a feeling I'm generating some excessive past sequences 22:45
jnthn That's sometimes a sign that things need some kind of factoring out.
It is generally quite verbose compared to the code it's describing, though. 22:46
22:49 stepnem joined
lue has a flashback to snowman/comet whilst browsing unicode 22:52
colomon snowman! 22:54
lue that was such a wacky bug, it was almost worth letting in :) 22:55
jnthn There will be modules. 22:57
sorear Method 'set_outer' not found for invocant of class 'Perl6Sub' 22:59
I wonder what I did wrong
jnthn Gave it a wrapped sub.
Pass the .do of that to it.
23:00 stepnem left
sorear I'm currently passing it create_code_object($past, 'Sub', 0, $sig_setup_block) 23:01
What should I change that too?
sorear doesn't currently grok most of the $sig stuff
jnthn Does the block have a signature?
lue rakudo: my @a=(3,2); @a[2] := @a[0]; @a[2]=4; say @a[0] 23:02
p6eval rakudo c4857a: OUTPUT«:= binding of variables not yet implemented␤current instr.: '&die' pc 17391 (src/builtins/Junction.pir:410)␤»
lue
.oO(grr...grr...)
jnthn sorear: It's hard to give you a good answer without a little more context.
sorear jnthn: macro foo { ... } causes anon sub ($/) { ... } to be generated at parse tme
this sub is then injected into Perl6::Actions
the parse time generation thing involved stealing a bit of code from BEGIN 23:03
jnthn ah, ok
try passing it
PAST::Block.new( :hll($?RAKUDO_HLL), PAST::Op.new( :pasttype('call'), create_code(....) )) 23:04
sorear How does that work? 23:05
(The code object takes a parameter)
jnthn oh, the ... is meta
I meant, but the call you already have there. 23:06
Oh
wait
Hmm
23:06 stepnem joined
jnthn Yeah, that's a bit tricky. 23:06
23:06 _jaldhar left
sorear generally speaking, I think all the code I've written for generating a sub is completely wrong and only works by accident 23:06
jnthn thinks a bit
How does it get injected into Perl6::Actions? 23:07
add_method ?
Maybe though 23:08
PAST::Block.new( :hll($?RAKUDO_HLL), create_code(....) )
If you pass that to the PAST::Compiler
Then run what you get back 23:09
You should in theory get a code object back that has been put into the correct lexical scope (that of the setting)
And can install that
Though it feels...weird. :-)
But it may well work.
sorear That's what I'm currently doing, but it's not landing in the right lexical scope. 23:10
What I have: $<blockoid>.ast 23:11
jnthn Which is the "right one"?
sorear The enclosing one.
23:12 kst left
jnthn Oh, enclosing in the program sense? 23:12
sorear yes
in the simplest case, this will be SETTING
23:12 kst joined
jnthn Yeah, the problem is that at compile time, the runtime case of that scope doesn't yet exist. 23:12
One of the limitations we have in present-day Rakudo.
The setting is a good approximation
And as good as we get for BEGIN blocks. 23:13
23:13 rgrau left
dalek kudo: 7b1a064 | jonathan++ | src/Perl6/Compiler/Signature.pm:
Fix placeholder paramters bug when the same placeholder parameter was mentioned twice in a block (was treated as two different parameters; now correctly as one).
 23:15
sorear jnthn: That approximation would still be better than what my current code does. 23:16
Macros are run with no outer lexical scope at all
jnthn nod
I think the suggestion I gave above (compile code that returns the block in the correct scope) should probably do it. 23:17
As in, in the scope of the setting.
sorear Your example seemed like it was going to call the block with no arguments? 23:18
jnthn no, the later one
sorear Which is correct for BEGIN
jnthn rihgt, the first was wrong for you
the second I gave after was better. 23:19
PAST::Block.new( :hll($?RAKUDO_HLL), create_code(....) )
sorear Your later one didn't mention !YOU_ARE_HERE, and was in fact identical to my existing version
jnthn Oh, sorry
I'm not being very clear.
jnthn tris again
*tries
23:20 patrickas left
sorear Neither am I. 23:20
github.com/sorear/rakudo/blob/17130...pm#LID1126
jnthn PAST::Block.new( :hll($?RAKUDO_HLL), PAST::Op.new( :pasttype('call'), :name('!YOU_ARE_HERE'), PAST::Block.new( :blocktype('declaration'), create_code(....) ))) 23:21
sorear *nod*
jnthn Essentially I'm suggesting that you compile code that calls !YOU_ARE_HERE with something that will return your code object (what create_code produces)
Kinda closure-like
And that closure-ness hopefully will get the setting stuff right. 23:22
lue is very confused :) 23:27
jnthn hands lue a beer...it'll make it all clear :-) 23:28
sorear syntax error at ../STD.pm line 1, near "but not " 23:33
hmm... I think this is stale. :)
23:36 japhb left, justatheory joined
jnthn gets the $x.SomeRole::methodname(...) syntax working 23:44
(worked for only SomeClass before) 23:45
Finally I can stop giving talks saying "one day you'll be able to..." :-)
jnthn has shown that syntax a few times before and then had to add "but NYI"
sorear use macros? 23:46
ah
what does SomeRole::methodname do?
jnthn sorear: The usual use case is that you have a collision of methods from two roles and you want to write a method that calls the method from a particular role
Then you can put in your class something like
method foo() { self.TheRoleThatIPrefer::foo() } 23:47
sorear mm, not at all like the Moose way of handling that
jnthn Well, Moose <-> Perl 6 ain't a one-to-one mapping :-) 23:48
sorear no 23:49
Moose is my gateway drug though
> macro term:<foo> { ~rand }
> sub moo { say foo; }; moo; moo;
1.39539001596827
jnthn :-)
\o/ 23:50
snarkyboojum sorear++ 23:52
jnthn++
:)
programming-machines++
23:54 kst left 23:55 kst joined 23:57 ethel left
sorear tries to figure out what's in lex/ 23:58

Powered by Cro and the Raku® Programming Language.

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

Thank you!