perl6-projects.org/ | nopaste: sial.org/pbot/perl6 | evalbot: 'perl6: say 3;' | irclog: irc.pugscode.org/
Set by mncharity on 25 March 2009.
cspencer what does the $?NF variable contain? 00:01
moritz_ the compile-time default Unicode normal form, I think 00:02
though I don't see much sense in having such a variable
cspencer how come? 00:03
00:04 kst left, kst joined
moritz_ why do we need a *compile time* variable for a thing which only affects *run time*? 00:05
cspencer ah, ok :)
00:07 msmatsko left 00:08 Guest34131 joined
Guest34131 hello? 00:09
Diederich howdy
Guest34131 hi, is this the place to ask questions about perl 6?
Diederich aye, but I'm just a lurker here 00:10
Guest34131 i c
Diederich I've seen a lot of Qs ansked and answered
s/ansked/asked/ :-)
brb
Guest34131 ok 00:11
cspencer guest34131: what kind of questions do you have?
i might be able to help, or point you in the right direction, at least :)
Guest34131 thanks, here it is..I've installed Perl 6 on ubuntu 8.10... 00:12
[particle] which perl 6? rakudo?
Guest34131 .. and I got $home/rakudo/perl6 -e 'my Int $x; $x.say' to work
[particle] type perl6 -v
ok 00:13
Guest34131 ... and now I'm trying to install the package "Image::Magick" so I can make it work with perl6
moritz_ that's a Perl 5 module
Guest34131 but I don't know how to go about installing packages into perl6...
moritz_ you can't with Perl 5 packages. 00:14
Guest34131 ahhhhh..........
I spent 2 days on it :)
00:14 kulp joined
moritz_ it'll probably work some day, but we honestly have no idea when that could be 00:14
00:14 kulp left
[particle] ouch! 00:14
moritz_ (at least I don't)
[particle] i wonder what kind of errors you got
so, maybe, we could improve the messages 00:15
Guest34131 I had fun though, installing, and reinstalling perl 6 though...
00:15 kate21de1 left
Guest34131 then if you don't mind me asking another question... this is the reason why I want to install perl 6 in the first place... 00:16
[particle] please, not to run perl 4 code!
moritz_ ;-)
Guest34131 ... I'm trying to use Image::Magick to automatically draw a simple shape, using the Draw command... 00:17
cspencer you should probably use perl5 then :)
Guest34131 ... then entering in numbers (200,100 50,50) as coordinates...
... I got Image::Magick to accept coordinates when I use numbers (200,100 etc) but when I substitute $x after declaring $x = 200; then ($x,100) doesn't work as a coordinate 00:18
I tried "use Integer;" but it didn't work either... 00:19
[particle] yeah, i'm sure that was a move of desperation
Guest34131 then I read about how in perl6 you can actually declare an Int as an Int, which is why I installed it... but now I can't import the Image::Magick package into Perl6... 00:20
[particle] unfortunately, there's no module to use Image::Magick in perl 6 yet 00:21
so, you'll have to use perl 5
Guest34131 is there any way to declare a real integer in Perl 5? I heard it was possible but very hard to do... 00:22
[particle] you don't need to, to get that working
Guest34131 what do you sugges then?
cspencer guest34131: what was your error message? 00:23
if any?
[particle] i suspect you're using single quotes instead of double quotes
my $x = 3; print '$x'; print "$x";
prints: $x3
Guest34131 the code was Draw... '200,100 50,50' but instead I put '$x,100 50,50' 00:24
[particle] use double quotes instead
then $x will be interpolated
Guest34131 ah...
I'll try that, thank you very much 00:25
[particle] for more help, try #perl, since you'll be happily using perl 5 now :)
Guest34131 haha
00:26 __felix__ left
Guest34131 btw, do you know when approximately Perl 6 will have the ability to import packages like Image::Magick? or do you think not until Perl 6 actually gets released in the mainstream... 00:27
[particle] i'm surprised every day by the progress of our volunteers
there is a module installer called 'proto' 00:28
Guest34131 oh?
[particle] but there are very few modules, still. that's changing, and i suspect this summer, after the us/european conferences, there will be more folks writing perl 6 modules
Guest34131 i c... btw, is "GD" in perl 6 yet? or is that a another module that has yet to be written? 00:29
cspencer there's no GD yet :) 00:30
Guest34131 i c... well, I guess I'll wait to see what happens in the summer :-) 00:31
jnthn Eventually we should be able to use and call functions from Perl 5 in Perl 6. But we're a bit off that yet. 00:32
Guest34131 ah...
jnthn (It's non-trivial to do.)
[particle] thanks for your patience. it will be rewarded :)
Guest34131 I appreciate the hard work, where are most of you located? North America or Europe? 00:33
00:34 Alias left
[particle] we're well spread out, actually 00:34
00:35 |jedai| is now known as jedai
Guest34131 i c 00:35
[particle] and jnthn alone covers half of europe and asia in his travels
jnthn ;-)
Hey, travel is *fun*. :-)
Guest34131 hey jnthn, sometimes Japan by any chance?
jnthn Guest34131: Actually, I'd really like to make the next YAPC::Asia. 00:36
I'm just waiting to hear dates, etc. :-)
Guest34131 next week in Tokyo?
or in May?
jnthn Huh? Last I heard of it was in September.
(Erm, that it'd be in September.)
Guest34131 I c, it hasn't been scheduled yet.. 00:37
jnthn Aye. Best info I've seen yet is use.perl.org/~lestrrat/journal/38354
Anyway, I never did visit Japan yet, and I'd really love to, and combining travel and Perl events has worked out wonderfully in the past. :-) 00:38
00:39 frioux is now known as frioux_away
Guest34131 jnthn, do you get paid for your efforts? or is 100% volunteer work.. 00:40
Tene jnthn: purl in #parrot has messages for you
jnthn: I implemented leave() with exceptions, but pmichaud thinks that maube exceptions are wrong there, and wants to discuss with you 00:41
jnthn Guest34131: I am funded, but do some volunteer effort too. 00:43
Guest34131 i c...
jnthn Guest34131: I couldn't give the time I give to Perl 6 without funding, so I've very grateful with it.
*for it
Tene: OK, let me look...
Tene jnthn: leave_exceptions branch in rakudo 00:44
leave.t fails because of &?BLOCK and &?ROUTINE as well as indirect method call syntax
jnthn Tene: Yeah, pmichaud and I originally discussed it in terms of invoking a continuation, but the discussion with you/me/TimToady IIRC in it I think convinced me that exception-based approach maybe was the way...
indirect method call syntax?
Such as? I think we have passing tests on that :-|
Tene oh, maybe I was wrong 00:45
jnthn Tene: Nah, you could easily have found a bug too
00:48 kst left, kst joined
jnthn Tene: Need sleep now; will try and look more tomorrow. Thanks for working on it. 00:50
Tene jnthn: np
00:51 nihiliad joined 00:52 Guest34131 left 00:56 Kisu left 00:57 Kisu joined 01:09 NordQ left 01:11 frooh_away is now known as frooh 01:13 msmatsko joined 01:18 alc joined 01:21 Diederich left 01:24 Diederich joined
cspencer i take it it's not possible to use a "return" in try's CATCH { ... } block? 01:26
pmichaud not yet. 01:27
actually, that should work, though.
std: q :hello:
cspencer rakudo: try { die "exeption"; CATCH { say "Caught: $!"; return fail("Failed") } }
pmichaud one doesn't "return fail"
cspencer oh :)
pmichaud fail() already does a return.
cspencer you just...fail? :)
ah ok
pmichaud but I suspect that CATCH might catch the fail in this case. 01:28
(it probably shouldn't -- but we have to get lexical returns working for that first.)
cspencer alright, i'll work around that for the time being :)
pmichaud std: say q : hello world :
std: say q 'hello' 01:29
std seems lost.
cspencer rakudo: say "foo" 01:30
rakudo seems lost too
pmichaud perl6: die?
p6eval: hello? 01:31
01:35 kulp joined 01:36 kulp left 01:37 FurnaceBoy left
s1n frooh: ping 01:46
frooh s1n: pong
01:53 Diederich left 01:54 Diederich joined
pugs_svn r26003 | pmichaud++ | [t/spec]: Fudge some rx:P5 tests for rakudo. 01:54
r26004 | pmichaud++ | [t/spec] more rakudo fudging for rx:P5. 01:59
02:04 Caelum left 02:05 bacek left 02:06 cotto joined 02:09 Caelum joined, justatheory joined
pugs_svn r26005 | pmichaud++ | [t/spec]: more rakudo :P5 fudging 02:09
dalek kudo: 353aca7 | pmichaud++ | src/parser/ (3 files):
Enable :P5 on m/.../ and rx/.../ patterns.
 02:10
02:14 dKingston left 02:17 wknight8111 left, justatheory left
p6eval std 26002: OUTPUTĀ«##### PARSE FAILED #####ā¤Unrecognized quote modifier: 1 at /tmp/S4TctYMh3N line 1:ā¤------> q :hello:ā¤FAILED 00:02 34mā¤Ā» 02:32
02:32 p6eval left, p6eval joined
p6eval ..... 02:32
..pugs: OUTPUTĀ«*** ā¤ Unexpected end of inputā¤ at /tmp/xLaLWXX0qr line 2, column 1ā¤Ā»
diakopter std: say q : hello world : 02:42
p6eval std 26005: OUTPUTĀ«ok 00:03 35mā¤Ā»
02:42 rachelBROWN left
diakopter std: q :hello: 02:43
p6eval std 26005: OUTPUTĀ«##### PARSE FAILED #####ā¤Unrecognized quote modifier: 1 at /tmp/t1i1e1d0fu line 1:ā¤------> q :hello:ā¤FAILED 00:02 34mā¤Ā»
03:01 justatheory joined
pugs_svn r26006 | pmichaud++ | [t/spec]: Some updates to rakudo skips for rx syntax. 03:11
r26007 | pmichaud++ | [t/spec] Some #?rakudo fudging for unspace.t .
r26008 | pmichaud++ | [t/spec]: More fudging for unspace.t . 03:12
03:12 Alias joined 03:13 Alias_ joined
dalek kudo: 7c69c7b | pmichaud++ | (2 files):
Add some more t/spec/S05-modifier files to spectest.data .
 03:13
03:24 grwi joined
pugs_svn r26009 | putter++ | [elf] elfish/on_sbcl renamed on_lisp to reflect added ccl support. Also renamed the old elfish/elfX to elfX_smop for greater clarity. 03:29
03:30 Alias left 03:31 grwi left
pugs_svn r26010 | pmichaud++ | [t/spec]: Final bit of rakudo fudging in unspace.t 03:46
dalek kudo: 4929856 | pmichaud++ | t/spectest.data:
Add unspace.t to spectest.data .
 03:48
04:03 cspencer left
pmichaud rakudo: say ('a', 1, 'b', 2, 'c', 3, 'd', 4).sort.perl; 04:19
p6eval rakudo 492985: OUTPUTĀ«["a", "b", "c", 1, 2, 3, 4, "d"]ā¤Ā»
pmichaud say 'a' cmp 1;
rakudo: say 'a' cmp 1;
p6eval rakudo 492985: OUTPUTĀ«1ā¤Ā»
pmichaud rakudo: say 'd' cmp 4; 04:20
p6eval rakudo 492985: OUTPUTĀ«1ā¤Ā»
pmichaud rakudo: say 1 cmp 2;
p6eval rakudo 492985: OUTPUTĀ«-1ā¤Ā»
pmichaud rakudo: say <a 1 b 2 c 3 d 4>.sort.perl;
p6eval rakudo 492985: OUTPUTĀ«["1", "2", "3", "4", "a", "b", "c", "d"]ā¤Ā»
frooh rakudo: <a 1 b 3 c 3 d 4>.sort 04:22
p6eval rakudo 492985: RESULTĀ«["1", "3", "3", "4", "a", "b", "c", "d"]Ā»
frooh rakudo: <a 1 b 3 c 4 d 5>.sort
p6eval rakudo 492985: RESULTĀ«["1", "3", "4", "5", "a", "b", "c", "d"]Ā»
pmichaud rakudo: say 4 cmp 'c'; 04:23
p6eval rakudo 492985: OUTPUTĀ«1ā¤Ā»
pmichaud rakudo: say 'c' cmp 4; 04:24
p6eval rakudo 492985: OUTPUTĀ«1ā¤Ā»
04:31 Util joined 04:35 Util left 04:38 alc left 04:43 kimtaro left 04:47 Util joined
Util rakudo: my %h = ( foo => "bar" ); for %h.pairs -> $one { say $one.perl } 04:47
p6eval rakudo 492985: OUTPUTĀ«"foo" => "bar"ā¤Ā»
Util rakudo: my %h = ( foo => "bar" ); for %h.pairs -> $one { print $one.perl } 04:48
p6eval rakudo 492985: OUTPUTĀ«"foo" => "bar"Ā» 04:49
05:00 kst left, Util left, kst joined 05:05 estrabd_ joined 05:06 estrabd left, Khisanth left 05:07 Khisanth joined 05:14 Woody4286 left, kimtaro joined 05:15 Woody4286 joined 05:16 nihiliad left 05:37 nww joined 05:39 nww left 05:46 Kisu left 05:49 Kisu joined 05:59 masak joined
masak good morning, fellow butterflies. 05:59
Tene goot morning 06:00
good 06:02
06:03 mberends joined
mberends good morning, happy hackers 06:03
06:05 alc joined
mberends masak: top o' the mornin' to you, sir! 06:06
masak mberends: and a very fine morning to you, fellow citizen!
masak tips hat 06:07
mberends bows
masak: may I insert copies of Configure.p6 and Configure.pm into proto? 06:09
masak mberends: for proto itself, or for budding project creators? 06:10
mberends initially for the latter
masak yes. you may. :) 06:11
06:11 DemoFreak joined
mberends will do in next few hours. currently planning a tree-based Pod parser as alternative to current stream-based. 06:12
masak nice. 06:14
mberends it may be nice to apply your SVG class to the task of XHTML emitter :) 06:16
masak aye, been thinking the same. 06:19
both the SVG class and an XHTML emitter would be greatly benefitted by some kind of real-time validation by DTD.
06:19 finanalyst joined
masak but my DTD-fu is still lacking. I'm looking at the stuff, and only almost seeing what it means. 06:20
06:21 charsbar_ left
mberends validation separate from generation is very expensive, so better to it by means of constraints at generate time. DTD is cumbersome, XML Schema is friendlier. I use them in .NET work. But performance is usually a problem when parsing (and also validating) XML. 06:23
masak I'm not saying we necessarily use the DTD qua DTD. 06:24
just that it contains the information I'm interested in validating against. :)
oh, and there should probably be a way to turn validation off in production environments, yes. 06:25
mberends perhaps validate mainly in the test suite
masak hm. 06:26
06:26 parduncia_ joined
masak the group I hope to reach is the one writing one-off scripts of the CGI kind. 06:26
mberends generally systems validate what arrives from the outside, not what is made inside.
masak indeed. 06:27
mberends design your generator classes to make only valid productions
masak that would be ideal, yes.
but... let's take an example to make it concrete. 06:28
a <li> element in HTML can only occur in some other elements.
I'm not sure if it's invalid XHTML 1.0 Strict if it occurs outside, but it's at least doubtful use. 06:29
a warning from the XHTML generator would seem appropriate.
06:30 Alias joined, Alias_ left
mberends a hierarchy of classes would prevent you from creating <li> except where it is valid 06:30
masak mberends: sir, you just set warning bells off in my head with the phrase "a hierarchy of classes" :) 06:32
but that's all right.
I'm just a little allergic to excessive OO.
mberends too Java-ish?
masak might be.
mind you, I've seen wonderful class hierarchies in my day, but they're few and far between. 06:33
most often they're just a waste of ink, or bytes as it were.
my ears perk up when it turns out that the classes use polymorphism in some constructuve way. 06:34
s/uve/ive/
mberends agreed. that reinforces my dislike of frameworks that bloat that way 06:35
06:35 miloux left
mberends Java causes some of that by being too formally strict and correct 06:35
masak Java enforces bloat.
mberends laughs 06:36
the potential number of classes for XHTML is not so bad 06:37
masak I'm just not convinced yet that element types should map to classes. 06:38
mberends SVG is worse, which is probably why your code was so generic. too generic, methinks.
masak maybe they should, if MMD is used in some clever way for each type of valid containment relation.
mberends: aye, too generic. 06:39
mberends you were starting top-down and SVG::Tiny was going bottom-up.
masak mberends: I wish to introduce a convenience into SVG: for some attributes, like style, an array of pairs should be flattened and formatted into a string of semicolon-separated key/value pairs. 06:40
after that, I'll be happy with the syntax. but I'll still want validation.
mberends actually, that's very nice. just make the validation fast, per-item. 06:42
per-item is probably the trick, avoid unwieldy trees 06:43
masak aye.
I just wish that someone gave me the magical ability to understand what the DTD document means. 06:44
www.w3.org/Graphics/SVG/1.1/DTD/svg...030114.dtd
06:46 charsbar joined
mberends what a format. it's a wonder Tim Berners-Lee got something friendly out of it. 06:46
masak I hereby publicly advertise for a person who can speak DTD.
I'm sure I can manage if I get the first three or four WTFs explained to me. 06:47
06:47 bacek_ left
mberends the Microsoft toolchain replaced all DTDs with XML Schemata. 06:48
www.w3.org/XML/Schema 06:49
masak that was probably for the better.
mberends a very good move, indeed.
06:49 kate21de joined
lucs masak: This looks like a decent tutorial: library.stanford.edu/tools/tutorial...entle.html 06:50
masak lucs: thank you.
lucs np
mberends lucs: were you replying to masak's advert? 06:53
masak seems so.
lucs I was kinda following along, yeah.
masak 2.4 is actually not the bad, thus far.
lucs The example in 2.4.1 is badly formatted as far as I can see in my browser, but the rest looks ok. 06:55
masak aye.
I'm hallucinating line breaks; after that, it looks fine. 06:56
lucs Right :)
mberends view source does not correct the line breaks - a missed opportunity :( 06:58
06:58 meppl joined 07:00 justatheory left, mtve joined
masak oh, the content model has similarities with a regular expression, in fact. 07:01
lucs Maybe more like shell globs, but yeah.
masak and the <ENTITY % foo ... > declarations look a bit like macros to me.
lucs Right.
masak ah, the rush of revelation! 07:02
lucs++
mberends leaves DTD to lucs and masak, shaking his head
masak mberends: I'll get back to you when I've written a Perl 6 grammar that takes the SVG DTD and emits an understandable Perl 6 data structure :) 07:03
lucs I hadn't thought about DTDs for many years (used to work with them before XML became popular), but that intro is quite decent.
mberends yes, the explanation is good. but DTD is obsolete now, superseded. 07:05
lucs Exactly.
masak it seems to me I'm most interested in the ENTITY declarations declaring things ending with '.content'. they contain all the meaty stuff about hierarchy. 07:06
also, I get the feeling that this DTD was written with extensibility in mind.
there's a lot of '.extra.content' defined, but they're all empty.
lucs Good luck with it. 07:08
lucs & # ZZ
mberends masak: the content you will generate is finite in structure, so the entity restrictions can be hard coded. avoid interpreting them from any schema file format at all.
masak of course. :) 07:09
that's why I want to pre-parse it and turn it into a Perl 6 structure.
mberends ah, so you were not joking...
masak :) 07:10
meeting &
07:10 masak left
mberends shower & 07:14
07:14 mberends left 07:15 kimtaro_ joined 07:17 eternaleye left 07:32 kimtaro left 07:35 eternaleye joined 07:46 riffraff joined 07:53 eternaleye left 08:03 bacek joined
pugs_svn r26011 | moritz++ | [t/spec] unfudge unspace.t a bit 08:06
bacek good evening
moritz_ good localtime() ;;-) 08:07
pugs_svn r26012 | moritz++ | [t/spec] unfudge while.t for rakudo 08:08
08:08 mberends joined
bacek moritz_: :) 08:08
rakudo: sub localtime {...}; sub good { say "O RLY?" }; good localtime 08:09
p6eval rakudo 492985: OUTPUTĀ«O RLY?ā¤Ā»
bacek it works :)
08:10 Woody4286 left, Woody4286 joined
moritz_ why doesn't it throw an exception when calliing localtime()? 08:10
mberends only time() is currently implemented. 08:11
bacek moritz_: it's warning, not exception. 08:12
rakudo: sub foo { ... }; foo
p6eval rakudo 492985: RESULTĀ«undefĀ»
bacek hmmm...
moritz_ uhm.
bacek masak's time?
mberends masak went to a meeting about an hour ago 08:13
yesterday masak++ ran a pattern match on a million character document, and Rakudo outperformed Perl 5 in some sense :) 08:15
bacek rakudo: sub foo { ... }; foo; say $!
p6eval rakudo 492985: OUTPUTĀ«Use of uninitialized valueā¤ā¤Ā»
pugs_svn r26013 | moritz++ | [t/spec] unfudge match.t for rakudo 08:16
r26013 | moritz++ |
r26013 | moritz++ | we now support both rx/.../ and lexicals in eval(). YaY!
08:17 amoc joined
bacek moritz_: rakudo crashed later than perl5? :) 08:17
moritz_ bacek: that was masak, not me :-) 08:18
bacek I'm expected it. masak is pretty good in breaking things : 08:19
:)
mberends bacek: irclog.perlgeek.de/perl6/2009-03-26#i_1018129 08:20
moritz_ I usually break things when I want to write articles about them ;-)
bacek mberends: heh :))) 08:21
mberends perl5 segfaulted, but Rakudo++ survived 08:23
pugs_svn r26014 | moritz++ | [t/spec] unfudge comb.t for rakudo 08:25
08:33 Helios left 08:34 Helios joined 08:43 sunnavy left 08:45 sunnavy joined 08:51 pmurias joined 08:54 bacek left 08:58 jedai left 08:59 |jedai| joined 09:01 alc left 09:05 kate21de left 09:09 kimtaro joined 09:11 rblasch joined 09:18 kimtaro_ left 09:33 dduncan joined 09:38 literal left 09:48 jrockway left 09:52 bacek joined 09:54 rblasch left 09:55 Alias left 09:56 Alias joined 10:12 mikehh joined 10:18 dduncan left
jnthn hi all 10:20
moritz_ hi jnthn
jnthn I see pmichaud++ has got us passing more tests! :-) 10:24
10:25 bacek left
moritz_ indeed 10:25
jnthn And moritz++ has been unfudging some too :-)
moritz_ mostly just those found by autounfudge 10:26
10:39 pmurias left, pmurias joined 10:46 zamolxes joined, kst left 10:47 kst joined 11:04 Tene left, diakopter left, buubot left 11:05 Tene joined, diakopter joined, buubot joined 11:15 kimtaro_ joined 11:16 kimtaro_ left 11:17 jrockway joined 11:19 literal_ joined 11:24 kimtaro_ joined 11:25 kimtaro left 11:42 rob___ joined 11:57 ruoso joined, mikehh left
ruoso finanalyst, I've just replied to your mail about junctions 12:00
12:11 kst left, kst joined, pmurias left 12:12 orafu joined 12:15 Trey joined 12:17 ejs joined 12:25 mikehh joined 12:27 bsb left 12:31 dKingston joined 12:32 dKingston left 12:33 dKingston joined
Matt-W mberends: ping 12:33
12:33 dKingston left, dKingston joined
mberends Matt-W: pong 12:33
Matt-W mberends: what did you want last night? 12:34
mberends I was curious if you were working on Form. I'm studying it closely.
Matt-W I was, yes 12:35
and suffering some frustration
mberends sorry to hear that
Matt-W it *looks* like something's trying to call an action method with an insufficient number of parameters 12:36
but since Parrot won't tell me which routine it is that's got the parameter count mismatch, I'm not entirely sure
mberends Matt-W: for diagnostics, replace your 'say' trace statements with 'warn', then they still appear during 'make test' 12:38
in Actions.pm
Matt-W ah 12:39
hadn't thought of that
12:39 skids left
Matt-W they work nicely when I run the test file directly though :) 12:40
jnthn Matt-W: Akshually the routine that is normally at the top of the backtrace is the one that didn't get enough parameters passed.
12:40 rodi joined
Matt-W jnthn: it's still an action method then, it's just a different one 12:40
hmm
except it manages to run say statements from within that on
jnthn Matt-W: The way you normally get these, is having an "#= foo"
But not $key parameter to receive it.
Matt-W that would be the other way round wouldn't it? too many parameters 12:41
jnthn Though maybe not in this case. But it's the first thing I check when i'm dealing with actions/grammars.
Ah, yes, so you try to take a $key when none is passed.
mberends therefore "#= foo" missing in Grammar.pm 12:42
Matt-W It really looks a lot like that 12:43
but I can't see that there is any way they could be missing
mberends Matt-W: the version last pushed passes all tests, care to push the current files?
Matt-W sure
done 12:44
mberends pulled. making test...
Matt-W it *looks* like it's the action method for aligned_field, but it's got all the #= it could possibly want... 12:45
jnthn URL?
Matt-W maybe I just need somebody else's eyes to look at it without expecting to know what's there
github.com/mattw/form 12:46
lib/Form/Actions.pm and lib/Form/Grammar.pm being of particular interest
no doubt it will be something simple and embarrassing
mberends yep, left_justified_field 12:47
Matt-W oooh 12:48
I've seen it
I was right wasn't I
very simple and embarrassing 12:49
mberends++
now I just need to make the grammar work
but that's a different problem
mberends me doesn't see it yet 12:50
Matt-W extra {*} in the grammar
with no #=
mberends Grammar.pm:54
Matt-W yup 12:51
pushed the fix
now back to making the new bits of grammar work
mberends pulled, now fails test 11 12:52
Matt-W yes I know about that
that's th ehalf-finished work that got derailed by finding the other bug
I just fixed that, actually, and am proceeding to the next five 12:53
12:53 gfldex left 12:54 bacek joined
Matt-W and that's done too 12:54
I was closer than I thought
12:54 finanalyst left
Matt-W it can now handle a string like "{[[[} some text {>>>>} some other text" 12:55
mberends jnthn, is the "#= foo" a short term kludge or a permanent language feature? 12:56
jnthn mberends: Better to ask pmichaud, but I'm not aware of it being endangered. 12:57
mberends has little faith in comments-as-syntax 12:58
jnthn mberends: In general, the need for it in Rakudo's grammar will mostly disappear as we get proto-regexen.
mberends good. riddance.
jnthn Though whether that's mostly disappear or completely go away, I'm less sure.
Matt-W proto-regexen? 12:59
because I'm really relying on its functionality in Form right now
and while proper embedded closures would fix it, it wouldn't for the case of having multiple sets of actions for the same grammar
jnthn *nod* 13:00
I think best is to ask pmichaud
Matt-W I think we can manage that
mberends Matt-W: while I like your structure, obviously I'll make a big effort to use alternatives to "#= foo". 13:01
Matt-W I was wanting to just use { make whatever } but that doesn't work yet 13:02
since I don't need multiple sets of actions
I just want to build a nice pile of meaningful objects 13:03
but I agree with you on the syntactically meaningful comments thing
I know regex is a different language, but still...
13:04 Alias left
mikehh rakudo (4929856) builds on parrot r37753 - make test/make spectest PASS - Ubuntu Intrepid Amd64 13:05
13:08 kimtaro joined, kimtaro left 13:09 szabgab left 13:21 mj41 left 13:22 Alias_ joined, donaldh joined 13:24 mj41 joined, kimtaro_ left
PerlJam greetings 13:32
13:35 skids joined
diakopter hi 13:36
moritz_ hi 13:37
PerlJam anything interesting happen in the last 20 hours ? :)
moritz_ yes 13:38
Alias_ Although the next 20 hours should be more interesting
Since they're going to be completed in only 19 hours
moritz_ rakudo: "abc" ~~ m:Perl5/(?:..)/ && say 'alive'
p6eval rakudo 492985: OUTPUTĀ«aliveā¤Ā»
moritz_ rakudo: "abc" ~~ m:Perl5/(?:..)/ && say $1 13:39
p6eval rakudo 492985: OUTPUTĀ«Use of uninitialized valueā¤ā¤Ā»
moritz_ rakudo: "abc" ~~ m:Perl5/(..)/ && say $0
p6eval rakudo 492985: OUTPUTĀ«abā¤Ā»
moritz_ perl 5 regexes are new in rakudo :-)
diakopter Alias_: how's that
PerlJam cool
Alias_ diakopter: Daylight savings switchover
PerlJam I suppose they don't support \1, \2, etc. though :)
diakopter Alias_: oh; not in usa; that happened a couple of weeks ago
PerlJam Alias_: maybe for you, but we already did that
Alias_ uk 13:40
moritz_ doesn't it happen on sunday morning?
Alias_ Not according to the news...
Saturday morning, early
Maybe I wasn't listening well enough
PerlJam moritz_: given that DST is of political determination, there's no telling when it could happen ;) 13:41
moritz_ :-)
here (Germany) it's Sunday 2am
PerlJam same in the USA
Alias_ Not a bad last 20 hours for releases though 13:42
New Module::Install, new Padre, new Catalyst, and new DBD::SQLite
PerlJam There is a female cardinal determined to peck a hole through my window. Every day for the last week she comes at random times during the day and pecks at my window for a few minutes.
moritz_
.oO( new TAP::Parser? )
PerlJam meanwhile, her mate just hangs out and eats the birdseed that I put out and drinks from the water in the bird bath that I set up. 13:43
new Catalyst?
13:44 bsb joined
PerlJam needs to pay more attention perhaps 13:44
Alias_ search.cpan.org/~mramberg/Catalyst-...e-5.71001/
only minor release though :/
PerlJam 5.8 will get here eventually :) 13:45
13:50 kimtaro joined
dalek kudo: 9fa0fca | pmichaud++ | docs/spectest-progress.csv:
spectest-progress.csv update: 339 files, 7795 passing, 0 failing
 13:51
moritz_ updates the chart
jnthn wow, that chart has a nice ending 13:52
moritz_ that's the p5 regex spike ;-) 13:53
13:54 exodist joined
pmichaud if I can get (?m) and (?i) working in p5 regexes we might get another spike. 13:55
13:58 [particle] left
pmichaud svn.pugscode.org/pugs/t/spec/S05-ma...operties.t # test file that would get us a huge spike in passing tests 13:58
14:03 kanru joined
moritz_ pmichaud: what's so hard about (?i) and (?m)? weird scoping? 14:06
pmichaud moritz_: weird scoping is part of it -- mostly it's just tuits to implement it 14:09
I'm a little surprised that PGE::P5Regex handled as many as it did :-)
14:14 ejs left 14:19 justatheory joined 14:21 alester joined 14:26 Alias_ left 14:34 justatheory left, kst left 14:35 kst joined
pmichaud std: sub infxi:<abc>($x, $y) { ... } 14:35
p6eval std 26014: OUTPUTĀ«ok 00:02 36mā¤Ā»
PerlJam "infxi:"? 14:36
pmichaud that's what I was checking.
PerlJam I figured, but why? Was there a typo that continued to work surprisingly? :) 14:37
amoc std: sub anyany:<abc>($x, $y) { ... }
p6eval std 26014: OUTPUTĀ«ok 00:02 36mā¤Ā»
pmichaud I'm curious how STD parses names of the form category:<...> so I can see about doing the same in rakudo.
Matt-W pmichaud: A question was raised earlier about whether the {*} #= stuff in rules is permanent
pmichaud Matt-W: it's permanent for at least as long as STD.pm has them.
14:38 orafu left
PerlJam I think it would be crazy for them to NOT be permanent (at least as an option) 14:38
14:38 OuLouFu joined
diakopter (I thought I read in STD.pm they were deprecated, eventually) 14:38
14:38 OuLouFu is now known as orafu
Matt-W PerlJam: we were wondering if there was something coming along that didn't use comments as meaningful syntax 14:38
pmichaud Matt-W: that's only because you insist on thinking of #= as being a comment :-) 14:39
PerlJam That they're comments is a feature!
14:39 [particle] joined
pmichaud I haven't seen or heard anything to suggest that #= is going away. 14:39
Matt-W Well it does start with a #
PerlJam you could write your parser in perl 6 and your "actions" in some other language if that were desirable. 14:40
Matt-W just feels icky to me
but
the functionality is definitely needed
so I'm going to not complain too much, especially as I don't have a better option at the moment :)
14:41 rob___ left
PerlJam Matt-W: just keep using it until it doesn't feel icky any more :) 14:41
Matt-W Or alternatively I could grow a couple of extra brains and use them to implement embedded closures 14:42
mberends so, is S02:128 wrong about comment, because the {*} exception is not specified?
pmichaud that's a slightly different question :-) 14:44
if you're asking "is #= part of the spec", the answer is "no", or at least "not yet".
diakopter jnthn: ping
(sigh)
jnthn diakopter: pong
[particle] #? isn't part of the spec yet, either
mberends relatedly, where is the action form of 'make' specified?
pmichaud if you're asking "is #= going away anytime soon", the answer thus far is "not likely" or "not until TimToady comes up with an alternative"
'make' is specified in S05 14:45
mberends ok, thanks
pmichaud even {*} isn't specified in the synopses yet.
(where "yet" includes "possibly never will be")
[particle] sure, but it's valid perl 6 anyway
pmichaud well, #= is valid perl6 also :-) 14:46
[particle] it's semantics aren't specced
pmichaud actually, it semantics would currently be the same as { * }
[particle] i can imagine #= and even #? becoming some single-line pod directive
pmichaud the "special" semantics that PGE currently gives to {*} in regexes isn't specced. 14:47
[particle] right
and it's not only pge, larry is using that for viv
moritz_ maybe we should just spec it :-) 14:48
ruoso wonders if āŽØ1,2,3āŽ¬ could be used as a synonim for set(1,2,3)
[particle] wonders what those utf8 chars were, and why they didn't display here
pmichaud or maybe Ā«{ 1, 2, 3 }Ā»
ruoso U+23AC RIGHT CURLY BRACKET MIDDLE PIECE
U+23A8 LEFT CURLY BRACKET MIDDLE PIECE
pmichaud with <<{ 1, 2, 3 }>> as the Texas set :-) 14:49
Matt-W well, if you implement custom operators... why not?
or maybe a macro is more suitable
hurrah for playing with the grammar
ruoso considering sets will end up in core
ruoso points at a very recent thread on p6-l about this issue 14:50
mberends Matt-W: see what trouble you've got us into!
moritz_ :-)
pmichaud as I see it, the problem most people have with junctions isn't their collapsing behavior, but that they then want to know "what particular set of collapsing behavior led to this outcome"? 14:51
ruoso pmichaud, that's the easy problem...
pmichaud i.e., they want to look inside the collapsing wave function to figure out how we arrived in the universe we're in.
ruoso the harder problem is "how do I get a subset of a junction?"
14:51 frooh left
pmichaud iiuc, junctions aren't sets any longer in Perl 6. 14:52
Matt-W pmichaud: as long as they don't want a career programming quantum computers...
ruoso it actually never was
but people kept using it as if it were
so TimToady removed the hability for it to work as a set
pmichaud Synopsis 2 once indicated that Junctions were sets
ruoso ah... 14:53
I should have missed that
[particle] my Set|Junction $x;
pmichaud S02 used to give the following for Junction: 14:54
Junction Set with additional behaviors
ruoso right... but it's a wrong definition even then... 14:55
pmichaud agreed, which is why it was changed :-)
ruoso unless you believe in anti-sets
i.e. none(1,2,3)
what I was really wondering... is at which point Sets can DWIM 14:56
I mean... I think "set(1,2,3) + 3 eqv set(4,5,6)"
but to which extent that can be applied to other operators/ 14:57
?
specially... "set(1,2,3) > 3"
probably it should be used in conjunction with junctions at that case... 14:58
"any(set(1,2,3)) >= 3"
pmichaud aren't these just lists, though?
ruoso pmichaud, not really
moritz_ autothreading lists?
ruoso if we want to DWIM, we need an actual type
pmichaud (with the exception of the uniqueness)
ruoso moritz_, I don't think autothreading applies to sets 14:59
pmichaud set(1,2,3) + 3 is really just (1,2,3) >>+>> 3
ruoso pmichaud, exactly... that's how "multi infix:<+>(Set, Num) {...}" implements it
pmichaud so, you're just wanting to have a Set type that hyperizes its operators without using the >> << 15:00
ruoso so it DWIM on regular operators
just as jucntion does
(which is why people keep wanting to use Junction as Set)
pmichaud but I don't think this gets at the true issue 15:01
i.e., someone wants to be able to ask "are any of the elements greater than X" -- which means it's a junction again.
or if someone wants to ask "which of the elements are greater than X", it's a map
or if someone wants to say "give me all of the elements greater than X", it's a grep
diakopter perhaps more contextualizers are needed, then, for Junction context and such 15:02
15:02 parduncia_ left
moritz_ if we still had .eigenstates it wouldn't be all that hard :) 15:03
pmichaud istr that TimToady has recently declared that junctions are meaningful only in boolean context.
moritz_: even .eigenstates doesn't help out much for "give me all of the elements greater than X"
PerlJam junction -> list -> grep seems fine to me :) 15:04
pmichaud PerlJam: sure, but if you do something like any(...) > 3 then the resulting junction is a junction of booleans
moritz_ what PerlJam said :-)
pmichaud i.e., we've lost the original values
moritz_ right
ruoso the problem is that using junctions as sets will lead to unexpected behavior at some point
moritz_ but for that question you use (...).grep({$_>0}) 15:05
PerlJam pm: so, don't do that? It seems like we have enough operators to convert lists into whatever we nee
pmichaud PerlJam: yes, that's kinda my point.
ruoso I see your point...
moritz_ currently Junctions are kind of like a singularity
ruoso but it's all about DWIMmery
moritz_ you can throw items in, but you'll only ever get a boolean out 15:06
heh, but I know I can cheat to get eigenstates back
rakudo: my @x; sub e($x) {@x.push: $e }; e(1|2|3); @x.perl.say 15:07
ruoso I currently think of Set having all the dwimmery junction has, but no collapsing and no autothreading...
p6eval rakudo 9fa0fc: OUTPUTĀ«Scope not found for PAST::Var '$e' in eā¤current instr.: 'parrot;PCT;HLLCompiler;panic' pc 146 (src/PCT/HLLCompiler.pir:102)ā¤Ā»
moritz_ rakudo: my @x; sub e($x) {@x.push: $x }; e(1|2|3); @x.perl.say
p6eval rakudo 9fa0fc: OUTPUTĀ«[1, 2, 3]ā¤Ā»
moritz_ ruoso: but the junction dwimmery comes exactly from autothreading
PerlJam ruoso: it's different dwimmery isn't it?
moritz_ ruoso: if you don't autothread, you won't get any dwimmery with user defined subs 15:08
ruoso moritz_, well... it does in terms of implementation... not in terms of expected behavior...
moritz_, autothreading is not necessary to call the same sub several times
15:08 literal_ is now known as literal
ruoso that's just an hyper invocation... 15:09
moritz_ ruoso: ... and we call it autothreading.
not that "normal" hyper invocation doesn't work with user defined subs with two parameters 15:10
that's where you need autothreading to dwim
ruoso indeed... that's the point I'm talking about
Set is Any
15:10 nihiliad joined
ruoso it's not autothread.. it's simply overrided operators 15:10
moritz_ so you want a type that doesn't DWIM on user defined functions, but autothreads on built-in functions/operators? 15:11
ruoso not autothread... simply follows regular multi dispatch
there's nothing "automatic" there
moritz_ but from a user's point of view it looks like autothreading? 15:12
pmichaud you want a type that has overloaded operators that dwim the hyperization for builtins
ruoso pmichaud, exactly
pmichaud but wouldn't dwim the hyperization for non-builtins
i.e., if I define my own custom operator, I'd have to define its Set version also.
moritz_ ... which would introduce a distinction between builtins and non-builtins that's undesirable IMHO
ruoso exactly...
it's not really a distinction... 15:13
pmichaud I think it is.
ruoso it's like Whatever
moritz_ but it makes it more work to define new operators that DWIM
that's the big advantage of junctions
ruoso just like with Whatever
moritz_ 90% of all code just works with them
ruoso moritz_, and the danger in being used as if it were a simple Set
PerlJam 90%?
ruoso my point is... Set is as special as Whatever 15:14
not as special as junctions
moritz_ ruoso: yes, but if you add more such concepts the complexity for defining user defined operators increases quadratically
pmichaud I can understanding having Whatever as a special case. I can't understand using it to justify other special cases :-)
moritz_ no, actually exponentially
right, one special case is bad enough already 15:15
ruoso ok... what would happen in the case of new operators?
it would fall into the Any,Any signature
which should deal with unknown types
moritz_ but doesn't add any meaningful behaviour
ruoso so, Set has a very well defined behavior... 15:16
diakopter pmichaud: why would you need a Set version of the custom operator as well? isn't that just like saying you can write an optionally parallel implementation of an operator if you want to? Why wouldn't we want to allow folks to write parallelized and nonparallelized implementations of operators that work in the proper contexts
ruoso that's my point
pmichaud diakopter: suppose I have set(1,2,3) + 3 .... according to ruoso, that gives me set(4,5,6)
(more
diakopter: now then, if I define postfix:<!> to do factorial 15:17
and write
(set(1,2,3) + 3)!
my generic expectation is that I end up with (4!, 5!, 6!)
but unless I write a special Set-form of postfix:<!> what I will really end up with is 3!
moritz_ ruoso: it's well defined, but much less useful than autothreading
[particle] (because 3 is the number of elements in the set, for those wondering) 15:18
moritz_ have a nice weekend all 15:20
15:20 donaldh left
moritz_ ciao 15:20
ruoso hmm...
pmichaud it gets a little worse if I can't even tell that something is a Set
ruoso pmichaud, that you could...
pmichaud because $a + 3 and $a customOp 3 would end up doing different things.
ruoso ~~ Set
pmichaud, but that's just MMD
15:20 donaldh joined
ruoso but point taken... I don't think I have a strong argument... 15:21
I think it shouldn't be a problem to have Set in the same level as junctions
but without the collapsing
pmichaud ruoso: I'd suggest completing the example you gave in the thread, though. 15:22
ruoso: i.e., you proposed the idea of Set, but then didn't illustrate how it could be used to determine that @p beats @d
diakopter correct me if I'm wrong... there's already a metaop syntax for hyper.. but ruoso thinks it can be inferred, sometimes (but pmichaud disagrees).
pmichaud diakopter: ruoso proposes that Set DWIMs hyper for builtins 15:23
ruoso pmichaud proposes it DWIMs for everything ;)
[particle] it auto-hyperizes, that's the dwim here
does that apply to bag, too?
but not array and list? 15:24
Matt-W what happens if you do something that expects a set to a set of sets?
pmichaud I'm not proposing it DWIMs for everything -- I'm illustrating why DWIM for somethings and non-DWIM for others may be problematic
[particle] is Set.keys defined?
pmichaud if anything, I'm saying that we already have a hyperizing notation, we should use it.
ruoso Matt-W, it doesn't collapse...
[particle], I'm not sure it should be named "keys"
Matt-W pmichaud: that seems sensible to me 15:25
[particle] i'm basically asking how array-like a Set is.
does it Enumerable?
ruoso hmm,... Enumerable?
you mean... Positional?
PerlJam It seems like ruoso's idea could work if there were some way to specify that operations "pass through" the Set to its members. (i.e., if there were a generic mechanism to specify the pass-through) But I'm not sure there's much utility beyond the specific example of Set
ruoso PerlJam, the way for doing that would be the same of the one for junctions... 15:26
PerlJam ruoso: right.
ruoso for those who missed it, I agreed it should be in the same level of Junctions...
PerlJam ruoso: right now, that's part of the "special case" for junctions. They have this behaviour, but other objects can't get at it easily.
[particle] hrmm, yeah, i guess i mean Positional. perhaps there's room for an Enumerable role, that doesn't implement .keys
ruoso I'd suppose Set doesn't support .] 15:27
.[] nor .{}
because a set is not ordered...
I guess
[particle] right, nor Bag
Matt-W .[] would be pretty meaningless
and .{} has no keys to work with
diakopter well, it has an ordering on *some* time dimension 15:28
ruoso I guess it implements .exists
and .values
Matt-W I can imagine doing a for loop ove ra set though - just in no particular order
PerlJam diakopter: yeah, but you don't want to imply anythign about the order at the interface.
diakopter heh :)
[particle] Set.values is Set.pick(*)
ruoso fair analogy 15:29
[particle] so set(1,2,3).values!
no, that probably still gives 3! 15:30
ruoso just set(1,2,3)!
and it will autothread, returning the same as set(1!, 2!, 3!)
15:30 kst left
ruoso just like junctions... 15:30
but without any type of collapsing
15:31 kst joined
[particle] yes, that's your proposal. 15:31
ruoso yes
[particle] a set is a junction with list context
a junction is a set with boolean context
something like that
ruoso not really
because none(1,2,3) doesn't map to any set concept 15:32
[particle] set(Nil)
ruoso the collapsing of junctions is what makes it "Not A Set"
[particle] i wonder about set(Nil).elements
ruoso if I remember first-grade school, it is 1 15:33
skids I always viewed a set as keys without values, not values without keys.
15:33 Cybera joined
ruoso [particle], if I see what you mean... Nil.elements will be 0 15:34
[particle] yep
PerlJam If we had AutoThread as a role and Junctions does AutoThread and Set does AutoThread ... :)
ruoso PerlJam, that isn't necessary...
[particle] so you have the whole container/values method problem, like junction
ruoso Any vs Object already does that distinction 15:35
diakopter so, a junction is a membership predicate of a [perhaps lazily enumerated] set?
(sorry for continually translating the Perl terminology...)
PerlJam doesn't remember the hierarchy
ruoso diakopter, I don't really see a reason to stablish a hierarchy between Junction and Set... they are both Object and both not Any 15:36
diakopter I'm not tryign to do that
ruoso [particle], ah... I guess that's the main difference...
diakopter they can have logical relationships in both directions
ruoso (1,2,3).foo doesn't autothread...
it simply calls the method foo on the set
diakopter set(1,2,3).foo ? 15:37
[particle] set(1,2,3).foo, you mean
ruoso yeah... sorry
[particle] so, set(1,2,3)! is 3!
ruoso why?
[particle] set(1,2,3).postfix<!>
ruoso [particle], ops are not method calls 15:38
they are multi sub calls
[particle] ah, right, different dispatcher
skids Do folks really want sets, or Bags? Bag is a hash of integer counts where count=0 keys cannot exist.
ruoso exactly
[particle] there are three dispatchers, right?
skids: perl 6 specifies both Set and Bag
skids Yes, but speaking of what the demand is for. 15:39
ruoso [particle], multi sub dispatch, object dispatch and private dispatch.... yes... I think that's all...
[particle] a Set is a Bag with uniqueness constraints
skids What are these junction abusers really after? :-)
ruoso skids, implement blackjack programs
[particle] :)
[particle] wonders if blackjack will be a one-liner in perl6 15:40
diakopter .21
[particle] well, the problem and the proposal seem well-defined to me. 15:41
diakopter re-reads osdir.com/ml/lang.perl.perl6.langua...00196.html
pmichaud I'd still like to see the complete blackjack solution done with sets 15:42
I'm not convinced it solves the problem.
[particle] i'm not sure it solves it, either, but it's clear to me.
ruoso pmichaud, basically... replace 1|11 by set(1,11)
pmichaud I understand that part. I don't see how to do the rest of the problem. 15:43
[particle] right, the trick is in the op and method calls involving sets
ruoso my @hand = set(1,11), 3, set(1,11);
my $sum = [+] @hand;
pmichaud what does $sum become after that step? 15:44
ruoso say $sum.perl; # set(5,15,25)
pmichaud okay, so keep going
in reality (from the original) we would have 15:45
ruoso if any($sum) < 21 { #there's a game here
pmichaud my $p = [+] @p;
my $d = [+] @d;
where $p is set(4, 14, 24) and $d is set(5, 15, 25) 15:46
ruoso ok...
pmichaud how do we determine that $d beats $p ?
diakopter mangles any( $sum, (x) => x < 21 ) in my head
ruoso pmichaud, that's where the set part really enters... since we need the subset of valid games... 15:47
pmichaud and we do that with ... ?
ruoso my $valid_p = $p.subset: { * < 21 };
literal Hm, ok, I'm definitely gonna apply for GSoC this year and I have Perl 6 in mind. However, I not sure which aspect I should work on. What would be of most benefit to Perl 6? Spec tests? Docs (u4x)? Something else? 15:48
diakopter <=
ruoso right...
my $valid_p = $p.subset: { * <= 21 };
[particle] literal: both are good. i suggest you take that to the list (tpf-gsoc-students@googlegroups.com) 15:49
PerlJam literal: pick the one you want to work on.
[particle] the mentors not here can reply there
literal PerlJam: a lot of them look good to me, so I'd like to pick one that's more likely to be accepted :)
pmichaud what's the difference between $p.subset and $p.grep ? 15:50
ruoso $p.subset returns another set...
pmichaud $p.grep couldn't do the same?
ruoso maybe...
but even then...
my $valid_p = $p.grep: { * <= 21 }; 15:51
[particle] does grep change the context of its elements?
ruoso which is what people is trying to do with junctions
[particle] er, the elements of the object over which it's operating.
ruoso [particle], what do you mean by "change the context"? 15:52
[particle] i guess i mean, does grep return List?
or are you suggesting a multi for grep that returns Set?
ruoso [particle], well... that was why I didn't suggest $p.grep at the first example
but I think there isn't really a problem in that case...
pmichaud so far it looks as though the only advantage of Set over List is in the [+] step. 15:53
15:53 Tene_ joined
ruoso in the blackjack case... maybe 15:53
but... 15:54
if you really think as sets...
you could extend the dwimmery to...
my $valid_p = $p <= 21;
pmichaud wouldn't that be a set of booleans, then?
ruoso no... a set of the values that passed the test 15:55
where an empty set is, obviously, false
pmichaud I would think an empty set would remain empty 15:56
ruoso I mean... an empty set in boolean context is false
[particle] i'd like to see set operators that create sets, like union, intersection, etc
pmichaud (but you can see why this gets weird, because we're taking what was a boolean operator and turning it into a filter)
there's no boolean context in my $valid_p = $p <= 21; 15:57
ruoso yes... I only said that as an additional comment
15:57 |jedai| left
pmichaud does prefix:<+> DWIM on sets? 15:58
ruoso does it DWIM on junctions?
pmichaud yes.
ruoso so yse
since Set is not Any
[particle] maybe it should be spelled 'set', not 'Set' 15:59
15:59 |jedai| joined
[particle] as now we have junction 15:59
pmichaud how would I get a list of the elements out of a set?
ruoso that usually differentiates native vs non-native types
pmichaud, $set.values?
or just @$set
pmichaud so methods don't autothread
ruoso no... they don't...
[particle] right, ops autothread, and not methods, on sets. 16:00
that's ruoso's idea.
pmichaud how would I truncate all of the elements in a set?
ruoso what do you mean by truncate?
[particle] make it the empty set.
pmichaud rakudo: 3.4.truncate
p6eval rakudo 9fa0fc: RESULTĀ«3Ā»
[particle] ah, like floor.
pmichaud yes, or floor.
diakopter truncate in-place or make a new set 16:01
pmichaud make a new one
[particle] set(1,2,3).<<truncate
pmichaud particle: a set in list context becomes a list of its elements?
ruoso that's my plan
diakopter (flattens) 16:02
pmichaud can I have lists of sets?
or sets of lists?
[particle] (set(1,2,3).<<truncate).WHAT() # set, or List?
ruoso hmmm
why are you using the hyper operator there>
?
[particle] i want to call that method on the elements of the set 16:03
pmichaud he's treating the set as a list.
(the hyper is pointing the wrong way, I think)
ruoso I guess you need an explicit cast, in that case
pmichaud so it would be set(1,2,3).values.truncate ?
[particle] yes, it is, and i copy/pasted the bad code again :(
ruoso or does the hyperop implies list context? 16:04
pmichaud er, >>.truncate
[particle] hyperop implies list
ruoso well... if it implies list, you'll get it called in the elements of the set
16:05 Tene left
ruoso pmichaud, but why can't set implement truncate as well? 16:05
(although I don't really see what truncate would do in a set)
[particle] it would operate on its elements
pmichaud ruoso: this gets back to the question of "how do I know which methods are defined for Sets and which aren't?" (more)
[particle] basically, most anything Any does, set should do.
pmichaud In the case of Junctions, methods autothread over the Junction, so we don't have to think about whether or not it's implemented for Junction
diakopter so, a set has its own dispatcher that iterates, optionally parallel. 16:06
ruoso pmichaud, that problem doesn't arise in Set, because it never delegates a method call to its elements
pmichaud it's just inconsistent then (more)
for Junctions, we have autothreading of methods and operators
[particle] so sets only produce sets with operators. with methods, they produce lists. 16:07
pmichaud for Lists, we have hyperizing of methods and operators
for Sets, we hyperize operators but not methods
(or DWIM operators but not methods)
ruoso right... I see the point...
but even if we have to face the problem of "which method does Set implement" 16:09
there's still a major difference from Junctions...
there's no collapsing
pmichaud I'm asking if there's a major difference from List
16:10 cognominal left
ruoso it DWIMs in operators and methods... 16:10
pmichaud no, you just said it DWIMs operators only
ruoso pmichaud, I accepted your point five lines ago ;) 16:11
pmichaud oh.
I didn't know that "see the point" meant "accepts it"
ruoso sorry...
pmichaud so, the purpose of a set is to get rid of the hyper angles on lists.
skids Well either way I think we would want to specify that sets/bags only act like hashes when treated that way, and when naked they explode to the keys, not the values.
ruoso pmichaud, the purpose of set is provide all the DWIMmery people love in Junctions, but without collapsing, which makes "getting the values" a reasonable idea... 16:12
pmichaud but the effect is simply to have a type that doesn't require hyper angles
ruoso not really...
pmichaud because I could do the same thing with List
ruoso the effect is having a type that can be used as a singular value, even if it's a plural value 16:13
which means that you can intermix singular and plural
i.e.:
my @hand = set(1,11), 3, set(1,11)
my $sum = [+] @hand;
the code works the same if you have no sets there 16:14
16:15 |jedai| left
pmichaud what if we have ... 16:15
16:15 |jedai| joined
pmichaud my $x = set(1,11); say ($x < 3).perl; 16:15
16:16 mberends left
pmichaud do I get set(3) or set(True, False) ? 16:16
sorry, set(1) or set(True, False)
and that leads to 16:17
my $ordered = [<=] @hand;
ruoso the question is "what does comparing a set to a number means"? 16:18
hmm... actually... that's not the question... 16:19
pmichaud in junctions, $x < 3 would produce a junction of true/false values
(assuming $x is a junction)
ruoso rakudo: my @hand = 1|11, 3, 1|11; my $ordered = [<=] @hand; say $ordered.perl; 16:21
p6eval rakudo 9fa0fc: OUTPUTĀ«get_integer() not implemented in class 'Junction'ā¤current instr.: '!MAKE_JUNCTION' pc 9241 (src/classes/Junction.pir:222)ā¤Ā»
16:21 kst left
pmichaud we still have some autothreading issues in rakudo. 16:22
ruoso but what would it return?
16:22 kst joined
pmichaud well, it ought to be the same as 1|11 <= 3 <= 1|11 16:22
which is going to be a junction of true/false values of some sort 16:23
ruoso rakudo: my $ordered = 1|11, 3, 1|11 ; say $ordered.perl;
p6eval rakudo 9fa0fc: OUTPUTĀ«[any(1, 11), 3, any(1, 11)]ā¤Ā»
ruoso rakudo: my $ordered = 1|11 <= 3 <= 1|11 ; say $ordered.perl;
p6eval rakudo 9fa0fc: OUTPUTĀ«any(Bool::False, Bool::True)ā¤Ā»
ruoso thinking 16:24
skids That's logical. It could be true, or could be false, depending on what value collapsed out of each junction.
pmichaud anyway, I need to get to work on some other tasks. Sets are a good idea, but I don't think we have quite the right magic yet.
ruoso I think "set(1,11) > 3" should somehow return "set(1)" 16:25
but I still don't know how it could get there
skids I think that's more like the functionality a user that goes looking for "sets" would want. 16:26
But as far as "the purpose of sets" it was my reading that the original intent was just that set(1,2){1} == 1, but set(1,2).>>say said "1 2" or "2 1" 16:27
And nothing fancier.
Not that fancier might not be good. 16:28
ruoso skids, the problem is, if you want to be able to use a set where you use a plain value... it needs to be much fancier
and if that's not possible, there's no difference from a regular list 16:29
skids Not if all you expect is a hash that you don't have to call .keys on all the time.
ruoso what's the purpose of that?
I mean
set(1,2){1}
what's the meaning of that?
you're trying to extract a value that you already have
skids A lot of uses of hashes is in the keys, not the values, either counting(bag) or presence(set). So I read it as just a way to ensure you could have very efficiently stored keys that were easier to work with syntactically. 16:30
ruoso skids, right... but .{} is not the api to that 16:31
probably .exists
skids $a = something that generates a set; if ($a{"foo"}) do something.
16:31 DietCoke joined, DietCoke is now known as Coke
skids ruoso: think bags for .{}, then ask why not sets too. 16:31
Coke is there a way to collapse nested anys?
e.g. any(any(1,2),any(2,3)) == any(1,2,3) ? 16:32
skids bag is the more important of the two.
Coke .
16:33 awarefish joined
ruoso Coke, you collapse it when you use it 16:33
rakudo: say any(any(1,2),any(2,3)) > 2;
p6eval rakudo 9fa0fc: OUTPUTĀ«Junction<0xb5b45928>ā¤Ā»
ruoso rakudo: say (any(any(1,2),any(2,3)) > 2).true; 16:34
p6eval rakudo 9fa0fc: OUTPUTĀ«1ā¤Ā»
ruoso rakudo: say (any(any(1,2),any(2,3)) > 2).perl;
p6eval rakudo 9fa0fc: OUTPUTĀ«any(any(Bool::False), any(Bool::False, Bool::True))ā¤Ā» 16:35
skids ruoso: and also pay attention that anything a Set/Bag can do a KeySet/KeyBag should also do, but they are mutable. 16:37
ruoso skids, I'd prefer a syntax like: if "foo" āˆˆ $a {...} instead
skids I think .{} falls out from the overarching "if you want it to behave like a hash, treat it as a hash" edict. 16:39
ruoso skids, not really... KeySet implements the behavior you want... not Set
it's not just immutable/mutable conterparts
Set doesn't behave like a Hash 16:40
nor does Bagh
just KeySet and KeyBag
skids Is there something in the standard that says or implies so?
ruoso yes... S02 16:41
KeySet KeyHash of Bool (does Set in list/array context)
KeyHash Perl hash that autodeletes values matching default
Set Unordered collection of values that allows no duplicates
16:41 Cybera left
skids That doesn't say anything about {} 16:42
16:42 awarefish left 16:43 orafu left
ruoso skids, Associative implements .{} 16:43
Set is not Associative
16:43 FurnaceBoy joined, OuLouFu joined, OuLouFu is now known as orafu
jnthn Heading off for the weekend; probably offline most of it; back Tuesday. Have fun. :-) 16:43
pmichaud have fun, jnthn! 16:44
ruoso pmichaud, do you see any way to make "set(1,2,3) > 2" to return "set(3)" ? 16:45
pmichaud not without introducing some other confusion, no.
ruoso unless there are specific signatures for that operators 16:46
since we want them to return a different thing
16:46 Psyche^ joined
pmichaud rakudo: say 'c' cmp 3; 16:47
p6eval rakudo 9fa0fc: OUTPUTĀ«1ā¤Ā»
pmichaud rakudo: say 3 cmp 'c';
p6eval rakudo 9fa0fc: OUTPUTĀ«1ā¤Ā»
pmichaud (...sigh...)
ruoso rakudo: say +'c'
p6eval rakudo 9fa0fc: OUTPUTĀ«0ā¤Ā»
ruoso lunch & 16:49
pmichaud yes, I should lunch here also.
skids same here 16:50
[particle] you people are crazy, it's too early for lunch! 16:55
literal initial draft: nix.is/u4x-gsoc.pod ... comments? 16:56
16:59 hercynium joined
diakopter ruoso: what about another meta operator... meaning apply (really *permute*) the operator's argument(s) (if/where they are sets), and return a set of the [union of] the results. set(1,2,3) S> 2 gives set(3), and set(1,2,3) S> set(1,2) gives set(2,3) 16:59
ruoso diakopter, the point is about being able to use a set as if it were a singular value... 17:00
diakopter I know; I'm suggesting a way to do it, assuming there's no reliable way to infer it. 17:01
ruoso the point is if that can't be done, sets are not more usefull than plain lists 17:02
17:03 Patterner left, Psyche^ is now known as Patterner
diakopter but they are, if they have a meta-op that enforces uniqueness when hyperizing. 17:03
17:05 zamolxes left
diakopter ? bags are eagerized lists that disregard ordering; sets are bags that enforce uniquenes ? 17:07
I see your point, they're not more useful as language intrinsics if operations on them don't have possibly special behavior... but of course they're still useful as logical constructs 17:11
(I think) 17:12
17:12 kst left, kst joined
ruoso I think I'll get back to the idea of Set being Any 17:12
the consequences of putting it in the same level as junction are enormous 17:13
and in the end, you always need to use it as a Set
so "set(1,2,3) + 3" works by the existance of "infix:<+>(Set, Num)" 17:14
it provides limited DWIMmery... but it's still better than no DWIMmery.... 17:15
diakopter .... idea of Set being the result of an Any [applied to a predicate producing a bool that's used for membership]?
ruoso no...
the idea of Set ~~ Any
diakopter oh.
ruoso meaning it wouldn't autothread at all
and all DWIMmery would depend on signatures in each operators that we want to DWIM 17:16
just like Whatever
ruoso really lunch & 17:17
diakopter speaking of operators... can an operation (like infix:<+>(Set, Num)) be declared as commutative, so that infix:<+>(Num, Set) is also applicable? 17:18
[particle] no 17:20
there's no commutative adverb
diakopter no, as in "not yet", or no, as in "I don't like that idea"? :)
[particle] right now that's R[]
17:20 frioux_away left, frioux joined
[particle] not yet. 17:20
we have 'is assoc' but not 'is commutative' 17:22
s/adverb/trait/
diakopter I suppose 'is commut' ('is commu'?) could also apply to n-ary operations (not just binary) 17:23
17:23 cognominal joined
diakopter er, not n-ary operations :) I mean, operations of all arities. 17:24
17:25 pmurias joined
diakopter pmurias: hi 17:25
pmurias diakopter: hi
diakopter pmurias: lots to backlog... 17:26
:D
pmurias diakopter: yes, someone has to start this week/day in #perl6 17:31
;)
17:34 Coke left 17:37 donaldh left
ruoso hi pmurias 17:37
pmurias ruoso: hi
ruoso pmurias, I was thinking the LOwlevel STate machine can still be helpfull, specially when we support signatures in the RI DSL 17:41
because if the capture is native and the values are not wrapped into a scalar, it will be able to have all the values avaialble before returning the first time... it could even use some trick to avoid getting back to the runloop... 17:42
while it would still do all the necessary SMOP_DISPATCH (one at a time) if that's not the case... 17:43
pmurias, what do you think?
pmurias isn't that a bit of premature optimalisation 17:44
?
btw we should add a wrapper around SMOP_DISPATCH which calls SMOP_RELEASE on the it's return value
ruoso pmurias, you know you can simply use SMOP_RELEASE around it, don't you? 17:45
pmurias yes
ruoso btw... I'm not sure it's premature optimization... because generating the mold code for the signature is not going to be much fun... specially if the code that will use the values is not a mold block 17:46
generating a specialized code might be easier 17:47
pmurias what i'm wondering is if it would be possible to write s1p in Perl 6 17:49
ruoso I'm not sure it's worth the trouble...
pmurias as most of the things s1p classes do atm is calling mold blocks and keeping data in a struct
ruoso some of them, yes 17:50
S1P__Scalar, for instance...
is not one of them
nor is S1P__Array
or Hash
pmurias, what certainly would help would be a way to write a method with a signature in mold 17:54
like...
%moldmethod foo($bar, $baz)
$mold."code"("here"); 17:55
pmurias that would eliminate a fair bit of the boiler plate 17:56
ruoso but not many methods are implemented just in mold
they usually access some internal attribute of the object...
which is set in the given mold frame 17:57
which gets execute
d
pmurias, maybe %moldmethod provides an additional signature, for getting things from this object's structure... 17:59
like...
%moldmethod foo ($bar, $baz) ($bla = invocant->member, $ble = invocant->othermember) 18:00
18:01 schmalbe joined
ruoso where $bar, $baz, $bla and $ble are made available as registers in the mold 18:02
pmurias that would be helpfull 18:03
pugs_svn r26015 | putter++ | [elfparse] IRx1_FromAST2_create.pl: Kludge around /:my $x; a | b/ parsing as /[:my $x; a] | b/. 18:04
pmurias ruoso: i'll have to think over and compare it to have a set of primitives and write s1p in Perl 6 (without lexical scopes) approach 18:05
skids how about a parameter trait "will explode" or something, lets MMD know if it gets a Set to feed to a scalar, to thread.
(as long as the Set is "of" the appropriate type, of course) 18:06
pmurias ruoso: what license to we use for smop? 18:07
skids Though I guess for foo(set(1,2),$bar) that doesn't get around the set list context explosion before the MMD. 18:08
ruoso same as perl
pmurias shouldn't we use Artistic 2? 18:19
amoc
.oO( it is just a checking for myself that failed in my local test )
 18:20
rakudo: is (<a ab>, <bc ad ba>).comb(m:Perl5/\S*a\S*/), <a ab ad ba>, 'comb a list';
p6eval rakudo 9fa0fc: OUTPUTĀ«Could not find non-existent sub isā¤current instr.: '_block14' pc 190 (EVAL_17:67)ā¤Ā»
18:21 M_o_C joined
amoc rakudo: (<a ab>, <bc ad ba>).comb(m:Perl5/\S*a\S*/), <a ab ad ba>; 18:21
p6eval rakudo 9fa0fc: RESULTĀ«["a", "ab", "ad", "ba", "a", "ab", "ad", "ba"]Ā» 18:22
amoc oh?
pmurias ruoso: native types should support arbitary captures too?
18:26 kst left, kst joined
ruoso pmurias, Perl uses Artistic 2 + GPL 18:38
pmurias, yes, they need...
pmurias, I think the only type we can accept requiring only native capture is the interpreter 18:40
18:46 Tene_ is now known as Tene 18:49 eternaleye joined 18:50 japhb left 18:59 nihiliad left, nihiliad joined 19:01 riffraff left 19:06 |jedai| is now known as jedai 19:24 rodi left 19:26 cognominal left, icwiener joined
pmurias ruoso: Artistics 2 is gpl comaptible so there is no need to make a disjunction 19:29
ruoso it's not a disjunction
19:29 cognominal joined
ruoso it's a dual licensing... 19:29
you can choose whichever you want
pmurias Artistic 2 can be changed to GPL 19:31
and i think perl5 is Artistic 1
s/changed/sublicensed 19:32
19:33 dKingston left 19:44 REPLeffect joined
pugs_svn r26016 | pmichaud++ | [S05]: Correct typo "by use if" ==> "by use of" 19:45
ruoso pmurias, maybe let it just Artistic 2 19:52
diakopter yeah, al2 software may be relicensed as any certified by opensource.org 19:54
19:55 icwiener left 20:03 Cybera joined, Cybera left, Cybera joined 20:07 nww joined 20:27 eternaleye left
Matt-W whee 20:32
new passing tests for Form.pm
20:34 frioux left 20:35 skids left 20:47 orafu left, OuLouFu joined, OuLouFu is now known as orafu 20:49 eternaleye joined 20:50 mikehh left 20:51 ZuLuuuuuu joined, rblasch joined, Grrrr left 20:52 Grrrr joined 20:54 sri_kraih_ joined 20:56 ruoso left 20:58 justatheory joined 20:59 M_o_C left 21:08 sri_kraih left, orafu left, OuLouFu joined, OuLouFu is now known as orafu 21:20 donaldh joined 21:24 mberends joined 21:34 kate21de joined 21:36 skids joined 21:39 ron_ joined 21:40 ron_ left 21:43 Whiteknight joined 21:44 NoirSoldats left 21:45 ron_ joined 21:49 NoirSoldats joined 21:52 ZuLuuuuuu left, Cybera1 joined, Cybera left 21:53 Cybera1 is now known as Cybera 21:54 dKingston joined 21:55 nihiliad left, FurnaceBoy left 21:59 justatheory left 22:01 ron_ left 22:02 pmurias left 22:07 mberends left 22:12 donaldh left 22:17 hercynium left 22:19 Whiteknight left 22:25 nww left 22:28 exodist left 22:31 alester left 22:33 dKingston left 22:44 Whiteknight joined 22:46 rblasch left 22:48 orafu left 22:52 orafu joined 22:54 nacho joined, dKingston joined 23:00 nacho left 23:01 schmalbe left 23:02 nacho joined
skids how do you bind in PIR. e.g. $P0 = find_lex('$a') $S0 = $P0 but I want $S0 to BE the string in $P0 not a copy of it, 23:10
Tene skids: look in docs/ops/set.pod 23:24
skids not finding it. 23:25
23:27 jan_ left 23:28 kate21de1 joined 23:29 jan_ joined 23:44 kate21de left 23:47 orafu left, OuLouFu joined 23:48 OuLouFu is now known as orafu 23:54 eternaleye left

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!