»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:03 silug left
tylercurtis sorear++ 00:11
00:14 daniel-s left 00:17 thou left 00:19 Ali_h left, Ali_h joined 00:23 wamba left
ashleydev what happened to the efforts of fglock? 00:25
sorear fglock mostly disappeared
ashleydev btw sorear++ # test passage 00:26
yeah I wonder why he disappeared...
sorear common reasons for people vanishing off the internet: entering a PhD program, getting a job, raising kids
I doubt it's anything worrisome, it's probably something great
00:27 quietfanatic joined
sorear o/ quietfanatic 00:28
quietfanatic hello
I am trying to figure out what the best way to parameterize a grammar is.
sorear details? 00:29
quietfanatic I have a grammar that operates one one line of a file at a time
something like: token TOP { ^ $LINE_BEGIN <directive> $LINE_END $ }
the $LINE_END and $LINE_BEGIN things there
are so that the lines can be embedded into comments in some other file format.
sorear sounds like what you want is subclassing 00:30
token TOP { ^ <.line_begin> <directive> <.line_end> $ }
quietfanatic Hmm.
I see.
sorear token line_begin { <?> }
token line_end { <?> }
quietfanatic subclassing at runtime might be difficult though 00:31
The program is supposed to sniff out the commenting format of the file before it begins parsing.
sorear ...
I'm getting more and more confused. Examples? 00:32
quietfanatic Hm, okay
it needs to see lines like "name: Yadda yadda"
but
tylercurtis quietfanatic: presumably you only have some predictable set of options. In which case you could just dynamically choose one of several subclasses.
quietfanatic in some files, that line will look like "#name: Yadda yadda"
Or it could even look like "#?name: Yadda yadda 00:33
I don't want it locked into a predefined set of formats.
sorear quietfanatic: please un-handwave youir "sniffing out" 00:34
quietfanatic When it reads the file
it looks for an introductory line that says "tcml: v1.0" or something like that
and if that line also has extra stuff around it
tylercurtis rakudo: grammar Foo { has $.a; token TOP { $.a; } }; say Foo.new(a => "a").parse("a")
quietfanatic like " /* tcml: v1.0 */" 00:35
p6eval rakudo 369665: OUTPUT«===SORRY!===␤Malformed regex at line 22, near "TOP { $.a;"␤»
tylercurtis rakudo: grammar Foo { has $.a; token TOP { $!a; } }; say Foo.new(a => "a").parse("a")
quietfanatic then it'll use that format for all of its parsing from then on.
p6eval rakudo 369665: OUTPUT«===SORRY!===␤Malformed regex at line 22, near "TOP { $!a;"␤»
tylercurtis rakudo: grammar Foo { has $.a; token TOP { $.a } }; say Foo.new(a => "a").parse("a")
p6eval rakudo 369665: OUTPUT«␤»
tylercurtis rakudo: grammar Foo { has $.a; token TOP { $.a } }; say so Foo.new(a => "a").parse("a")
p6eval rakudo 369665: OUTPUT«Bool::False␤»
quietfanatic oh, .parse can be used on instantiations of the grammar and not just the class?
tylercurtis rakudo: grammar Foo { has $.a; token TOP { $.a } }; say Foo.new(a => "a").a
p6eval rakudo 369665: OUTPUT«a␤»
sorear quietfanatic: then, you need a variable
quietfanatic ooh, that is just what I wanted.
In fact I should have tried that. 00:36
tylercurtis except that it doesn't appear to work...
sorear tylercurtis' example will not work in nontrivial cases
quietfanatic oh
sorear this, however, will
rakudo: my $x = "a"; grammar Foo { token TOP { $x } }; say Foo.parse("a")
p6eval rakudo 369665: OUTPUT«a␤»
sorear why are you using a grammar anyway? 00:37
tylercurtis sorear: In fact, my example didn't even work in trivial cases. :)
sorear if you only have one regex, just write it directly
quietfanatic Because it's way too complex for a single regex
The grammar is currently about 50 lines long and will eventually be several hundred.
00:38 leprevost joined
quietfanatic and if the grammar is in a seperate module from the routine that's calling it, using a my-scoped variable to control the grammar would be a bit difficult. 00:38
tylercurtis rakudo: grammar Foo { our $x = "a"; token TOP { $x; } }; say Foo.parse("a") # I think this should work. 00:39
p6eval rakudo 369665: OUTPUT«===SORRY!===␤Malformed regex at line 22, near "TOP { $x; "␤»
tylercurtis rakudo: grammar Foo { our $x = "a"; token TOP { $x } }; say Foo.parse("a") # I think this should work.
p6eval rakudo 369665: OUTPUT«a␤»
sorear quietfanatic: then use a dynamically scoped variable 00:40
$*opener, $*closer
tylercurtis rakudo: grammar Foo { our $x = "a"; token TOP { $x } }; say Foo.parse("a"); $Foo::x = "b"; say Foo.parse("b");
p6eval rakudo 369665: OUTPUT«a␤b␤»
quietfanatic rakudo: grammar Foo { our $x = "a"; token TOP { $x } }; $Foo::x = "b"; say Foo.parse("b")
p6eval rakudo 369665: OUTPUT«b␤»
quietfanatic I see
I thought I tried using our-scoped variables, but it errored on me
maybe it's something else that was wrong. 00:41
tylercurtis sorear: To what extent would having a $x in a rule prevent optimizations of a grammar that would otherwise be possible? (I'm guessing "a large extent".) 00:42
quietfanatic well, it works now.
I don't know what was wrong before, but I'm not going to complain. 00:43
00:43 aindilis` left, aindilis` joined 00:45 ymasory joined
pmichaud seems hard to fathom that a recursive quicksort in PIR could be faster than Parrot's builtin sort. 00:49
wrong chan 00:50
sorear tylercurtis: I haven't thought of any significant optimizations that having $x would significantly effect, actually
ARGGGHH
affect
there goes my perfect a/effect record :(
tylercurtis sorear: Would it not impact LTM? 00:51
sorear tylercurtis: it stops LTM outright
but I don't consider LTM an optimization
00:52 perplexa joined
sorear it has such large constant factors that I try to avoid using it 00:52
quietfanatic At the most it'd just be another sequence point, right?
sorear although it really comes in handy for some things
quietfanatic: yes
pmichaud: ping 00:53
tylercurtis realizes that he has forgotten most of what little he knew about Perl 6 LTM.
quietfanatic Though I suppose if there were some more formal way of parameterizing it, it wouldn't have to be a sequence point. 00:54
sorear quietfanatic: there is, but it's NYI in Rakudo
IIRC
pmichaud sorear: pong, but heading out the door 00:55
sorear 16:57 < sorear> pmichaud: I would like to hear your thoughts on OUTER.
pmichaud in rakudo I think it's mostly tied to callframe
sorear perl6: grammar A { token TOP { <.begin> \d+ <.end> } }; role b_e[$b,$e] { token begin { $b }; token end { $e } }; say (A but b_e["x","y"]).parse("x123y") # quietfanatic 00:56
p6eval rakudo 369665: OUTPUT«Attempt to use rebless_subclass where the new class was not a subclass␤ in 'infix:<does>' at line 7658:CORE.setting␤ in 'infix:<but>' at line 634:CORE.setting␤ in main program body at line 22:/tmp/mLclfct6Ev␤»
..pugs: OUTPUT«*** ␤ Unexpected "[$"␤ expecting "::", "-" or trait␤ at /tmp/sbkzwUOlvO line 1, column 58␤»
..niecza v5-85-ga828415: OUTPUT«Unhandled exception: Unable to resolve method at-pos in class b_e␤ at line 0 (ExitRunloop @ 0)␤ at /tmp/EyuEcliQcE line 1 (MAIN mainline @ 2)␤ at /home/p6eval/niecza/lib/CORE.setting line 1421 (CORE C595_ANON @ 2)␤ at /home/p6eval/niecza/lib/CORE.setting line
..1422 (CORE module…
pmichaud in nom we should be able to resolve OUTER at compile time
sorear perl6: grammar A { token TOP { <.begin> \d+ <.end> } }; role b_e[$b,$e] { token begin { $b }; token end { $e } }; say (A but OUR::b_e["x","y"]).parse("x123y") # quietfanatic
pmichaud (that's not exactly easy to do in rakudo master at the moment)
p6eval niecza v5-85-ga828415: OUTPUT«x123y␤»
..rakudo 369665: OUTPUT«Cannot find sub OUR::b_e␤ in main program body at line 1␤»
..pugs: OUTPUT«*** ␤ Unexpected "[$"␤ expecting "::", "-" or trait␤ at /tmp/B_Yja3Bgpf line 1, column 58␤»
sorear quietfanatic: ^^^ more formal way to parameterize, does NOT create a LTM sequence point 00:57
pmichaud anyway, I don't have any profound to say about OUTER at the moment -- it's just another pseudonamespace that the compiler has to deal with
afk, errands 00:58
quietfanatic It uses the parametric role system, eh? 01:00
01:01 keeth left 01:19 am0c joined 01:44 huf left 01:48 Chillance left 01:59 whiteknight left 02:00 lichtkind left 02:10 [particle] left 02:11 [particle] joined 02:15 bbkr_ left, bbkr_ joined 02:22 am0c left 02:36 am0c joined 02:42 silug joined 02:46 keeth joined 02:49 silug left 02:52 am0c left 02:53 GinoMan[A] joined 03:06 am0c joined 03:09 GinoMan[A] left 03:13 zorgnax left 03:23 GinoMan[A] joined 03:36 GinoMan[A] left 03:50 keeth left 04:03 cdarroch left
dalek kudo: 0d3c71c | pmichaud++ | src/builtins/ (10 files):
Add :subid to :vtable entries so they can be profiled via RAKUDO_SUBLOG.
 04:04
04:04 satyavvd joined
sorear at this point I think I could be dealing with a mono bug 04:36
04:36 ymasory left
sorear I just added debugging code to print a hexdump of the serialized blob - and it printed a hexdump of the *recent debug messages* instead 04:36
benabik sorear: Mono's a pretty bad bug. 04:37
sorear hugme: hug benabik
hugme hugs benabik
04:37 birdwindupbird joined
benabik sorear: What? I know people who had to drop out of college because of mono. 04:39
sorear I still think you're trolling. 04:40
04:40 GinoMan[A] joined
benabik I suppose punning is a form of trolling. 04:41
sorear oh, that was actual humor 04:42
sorry, I was getting ready to deploy the ops
I am so sick of RMS fanboys and their secondary targetting schemes 04:43
benabik sorear: I know lots of people who can't tell when I'm being funny IRL too, so don't worry about it. 04:44
I don't mind mono the program... Just never had a use for it.
04:49 mtk left 04:53 d4l3k_ joined 04:54 stepnem left, dalek left
sorear nah, I take that back 04:55
it's OBVIOUS now that I'm looking at a Mono bug
04:55 d4l3k_ is now known as dalek, ChanServ sets mode: +v dalek
sorear there's no way that new byte[1920] should be resulting in an array of bytes filled with what appears to be MonoObject structs 04:58
04:59 mtk joined
sorear I said it was filled with debug messages before. Well, the 3 words before that were 1. the message length 2. 0 3. a data segment pointer (clearly System.String's vtable) 04:59
04:59 stepnem joined
benabik That does sound poor. 05:01
05:03 p6eval left 05:05 p6eval joined, ChanServ sets mode: +v p6eval 05:13 koban` joined
quietfanatic Is it supposed to guarantee zero-initialization? 05:13
sorear quietfanatic: actually I'm using a form with explicit initialization 05:14
quietfanatic ah
05:15 ymasory joined
sorear I'm using the newarr opcode together with the InitializeArray runtime function 05:15
newarr is supposed to zero-initialize 05:16
the Mono JIT has a hack which suppresses zero-initialization if the array is immediately being passed to InitializeArray with a matching size
quietfanatic So it seems that InitializeArray is not doing its job 05:19
05:25 LaVolta joined
sorear also, it only fails with the genrational GC enabled 05:35
dalek ecza: af8a3f2 | sorear++ | / (4 files):
Make my A::B $x work, add lots of debugging code to the blob loader
 05:43
05:43 leprevost left
05:48 pamera joined 05:55 silug joined 05:56 wtw joined 05:59 silug left
sorear and with mono 16fc336 from earlier today... NOT REPRODUCABLE \o/ (?) 06:10
sorear feels slighted out of a chance to debug Mono :P 06:11
dalek ecza: be60cde | sorear++ | test2.pl:
Test :: in parameters
 06:13
06:14 wamba joined
sorear perl6: sub foo(@bar) { say +@bar }; foo <a b c>; 06:18
p6eval pugs, rakudo 0d3c71: OUTPUT«3␤»
..niecza v5-85-ga828415: OUTPUT«Unhandled exception: Excess arguments to MAIN foo, used 1 of 3 positionals␤ at /tmp/M4ZOk45CMw line 0 (MAIN foo @ 0)␤ at /tmp/M4ZOk45CMw line 1 (MAIN mainline @ 1)␤ at /home/p6eval/niecza/lib/CORE.setting line 1421 (CORE C595_ANON @ 2)␤ at
../home/p6eval/niecza/lib/CORE.setting l…
tadzik benabik: how would people be dropped out of college because of mono? 06:20
Souns like RMS being one of the proffesors 06:21
sorear tadzik: benabik was punning. secure.wikimedia.org/wikipedia/en/...onucleosis
tadzik hehe, I see 06:23
06:24 fhelmberger joined 06:27 Mowah joined 06:28 ponbiki left
dalek ecza: fbf986f | sorear++ | / (2 files):
foo <a b c> is only one formal argument
 06:34
sorear rakudo: say index("pie", "i") 06:37
p6eval rakudo 0d3c71: OUTPUT«1␤»
sorear rakudo: say index("i", "pie")
p6eval rakudo 0d3c71: ( no output )
06:50 amkrankruleuen left 06:51 amkrankruleuen joined
dalek ecza: 6c42d59 | sorear++ | lib/CORE.setting:
Add index()
 07:01
ecza: 6bf4ff8 | sorear++ | / (2 files):
Fix build
sorear phenny: tell masak yapsi/niecza 52d74c6 passes all tests on niecza/master 6bf4ff8 07:04
phenny sorear: I'll pass that on when masak is around.
sorear currently at 56.5 changed lines 07:06
std: 1.^.^.^.^foo 07:08
p6eval std 4b1b100: OUTPUT«===SORRY!===␤Confused at /tmp/YRkBAt61YM line 1:␤------> 1.^⏏.^.^.^foo␤ expecting dotty method or postfix␤Parse failed␤FAILED 00:01 113m␤»
07:13 hanekomu joined 07:24 mj41 joined 07:30 amkrankruleuen left 07:33 quietfanatic left
sorear rakudo: $*OUT.^parents(:all).say 07:37
p6eval rakudo 0d3c71: OUTPUT«too many named arguments: 1 passed, 0 used␤ in main program body at line 22:/tmp/fk3RDmbMmF␤»
sorear rakudo: $*OUT.^parents.say
p6eval rakudo 0d3c71: OUTPUT«Cool()Any()Mu()␤»
sorear rakudo: $*OUT.^methods.say
p6eval rakudo 0d3c71:
..OUTPUT«closeeofgetinslinesopenprintprintfsayreadwritegetcslurptdefslzcreatedmodifiedaccessedchangedmovechmodcopylinkautoflushpathstatNumericRealIntRatNumabsconjugateexploglog10sqrtrootsto-radiansfrom-radiansfloorceilingroundtruncatesigncisunpolarchrchrsrandsincostanseccoseccotansinhcoshta…
sorear rakudo: $*OUT.^methods.join(", ").say
p6eval rakudo 0d3c71: OUTPUT«close, eof, get, ins, lines, open, print, printf, say, read, write, getc, slurp, t, d, e, f, s, l, z, created, modified, accessed, changed, move, chmod, copy, link, autoflush, path, stat, Numeric, Real, Int, Rat, Num, abs, conjugate, exp, log, log10, sqrt, roots, to-radians,
..from-r…
sorear rakudo: $*OUT.^methods(:local).join(", ").say
p6eval rakudo 0d3c71: OUTPUT«close, eof, get, ins, lines, open, print, printf, say, read, write, getc, slurp, t, d, e, f, s, l, z, created, modified, accessed, changed, move, chmod, copy, link, autoflush, path, stat␤»
dalek ecza: 3b56aac | sorear++ | / (2 files):
Implement .?
 07:39
ecza: 9e75e8d | sorear++ | lib/CORE.setting:
Add first cut at $*OUT
07:40 wamba left 07:44 wamba joined 07:46 amkrankruleuen joined
sorear down to 37 lines changed 07:48
07:48 wamba left
sorear rakudo: say +[ hash a => 3, b => 5 ] 07:53
p6eval rakudo 0d3c71: OUTPUT«2␤»
08:15 amkrankruleuen left 08:16 amkrankruleuen joined
dalek ecza: b084a1e | sorear++ | / (2 files):
Add &hash, 1-arg .substr, .subst with Str lhs, allow assigning .kv to hashes
 08:19
sorear calls a night 08:23
08:38 am0c left 08:54 daniel-s joined, silug joined
daniel-s is there a text editor that does syntax highlighting for pir? 08:56
08:56 mberends joined
moritz vim does 08:57
there's a pir.vim in the parrot repo somewhere
09:01 daniel-s_ joined 09:03 daniel-s left 09:05 huf joined 09:13 noganex_ joined
jdhore1 is there any language that a syntax hilight script doesn't exist for vim? 09:16
09:16 amkrankruleuen left, noganex left 09:17 amkrankruleuen joined 09:19 awoodland joined
yath jdhore1: RPG. 09:21
jdhore1 Actually, you're right. 09:22
arnsholt jdhore1: I had to write my own syntax file for the Xerox finite state tools =) 09:23
09:27 amkrankruleuen left 09:29 amkrankruleuen joined 09:31 Mowah_ joined 09:35 amkrankruleuen left 09:40 silug left 09:42 snearch joined 09:46 amkrankruleuen joined 10:09 c1sung left 10:16 c1sung joined, daniel-s joined 10:19 daniel-s_ left 10:22 ponbiki joined 10:27 LaVolta left 10:34 _jaldhar left 10:44 gbacon joined 10:45 stepnem left 10:47 stepnem joined 10:48 orafu joined 11:05 _jaldhar joined 11:07 jaldhar_ joined 11:10 _jaldhar left 11:12 wamba joined 11:18 pernatiy left 11:19 ab5tract joined 11:27 jaldhar_ left 11:28 wamba left, jaldhar_ joined 11:29 wamba joined 11:33 og01 left, satyavvd left, donri joined 11:39 wamba left 12:08 wknight8111 joined 12:13 Mowah left 12:14 Mowah_ left 12:15 hanekomu left 12:17 pernatiy joined, wamba joined 12:21 amkrankruleuen left 12:22 amkrankruleuen joined 12:23 silug joined 12:26 ab5tract left 12:28 silug left 12:29 silug joined 12:31 birdwindupbird left 12:33 silug left 12:41 satyavvd joined 12:42 Holy_Cow joined, Holy_Cow left 12:59 birdwindupbird joined 13:00 ymasory left 13:01 ab5tract joined 13:02 pamera left 13:06 donri left, donri joined
daniel-s hello 13:24
anyone around?
takadonet daniel-s: yo
daniel-s hey 13:25
what are you up to?
also, you know grammers and defining regex rules in perl6, is that all new to perl6 compared to 5? 13:26
sbp yep. that and most of the rest of the language... :-)
try perl6advent.wordpress.com/ for an overview of some differences
might also find rosettacode.org/wiki/Category:Perl_6 useful for direct comparisons 13:27
takadonet daniel-s: In p5 regular expression are just string in P6 they are soo much more
13:27 wamba left
sbp oh sorry, I think I misunderstood your question now that I read it back 13:27
most of the grammar and regular expression stuff is changed beyond recognition, yes, if that was your question 13:28
you can still bootstrap perl5 regexps into perl6 though if you so want
13:30 [Sec] joined
[Sec] Maybe I'm just stupid. I want to print all "odd-numbered" elements from an aerray, but neither @a[1,3 ... *] nor @a[1,*+2..+@a] work for me? 13:32
13:33 bacek left 13:34 JimmyZ_ joined
sbp rakudo: say [1, 3 ... 15].join(", ") 13:34
p6eval rakudo 0d3c71: OUTPUT«1, 3, 5, 7, 9, 11, 13, 15␤»
sbp hmm, lemme think
13:36 daxim_ left 13:39 Chillance joined
sbp rakudo: for [1, 3 .. 7] { say <a b c d e f g h i j k>[$_] } 13:40
p6eval rakudo 0d3c71: OUTPUT«bdefgh␤» 13:41
sbp [Sec]: how's that?
(zero based of course)
er, and with ... instead of .. of course too
rakudo: for [1, 3 ... 7] { say <a b c d e f g h i j k>[$_] }
p6eval rakudo 0d3c71: OUTPUT«bdfh␤»
sbp le typo magnifique 13:42
wonder if it works lazily
rakudo: for [1, 3 ... *] { say <a b c d e f g h i j k l m n o p>[$_] }
p6eval rakudo 0d3c71: OUTPUT«(timeout)»
sbp nope
13:42 wamba joined
sbp rakudo: say <a b c d e f g h i j k>[$_] for [1, 3 ... 7] 13:43
p6eval rakudo 0d3c71: OUTPUT«bdfh␤»
[Sec] sbp: it just doesn't return for me 13:44
sbp hmm? 13:45
[Sec] rakudo: my @a=<<eins zwei drei vier fuenf sechs>>;say @a[1,3 ... +@a].perl
p6eval rakudo 0d3c71: OUTPUT«(timeout)»
JimmyZ_ rakudo: say (gather for 1, 3 ... * { take <a b c d e f g h i j k l m n o p>[$_] })[2]
p6eval rakudo 0d3c71: OUTPUT«f␤» 13:46
JimmyZ_ rakudo: say (gather for 1, 3 ... * { take <a b c d e f g h i j k l m n o p>[$_] })[1,3,7]
p6eval rakudo 0d3c71: OUTPUT«dhp␤»
JimmyZ_ rakudo: say (gather for 1, 3 ... * { take <a b c d e f g h i j k l m n o p>[$_] })[1,3,4]
p6eval rakudo 0d3c71: OUTPUT«dhj␤» 13:47
JimmyZ_ thinks it works lazily
sbp thanks JimmyZ_ 13:48
JimmyZ_ rakudo: say (gather for 1, 3 ... * { take <a b c d e f g h i j k l m n o p>[$_] })[100]
p6eval rakudo 0d3c71: OUTPUT«Any()␤»
JimmyZ_ rakudo: my @a; say @a[100]; 13:49
p6eval rakudo 0d3c71: OUTPUT«Any()␤»
JimmyZ_ wonders whether it's a bug or not.
[Sec] rakudo: my @a=<<eins zwei drei vier fuenf>>;say @a[1,3 ... +@a].perl
p6eval rakudo 0d3c71: OUTPUT«("zwei", "vier", Any)␤»
[Sec] with odd number of elements, it works, with even, it hangs. 13:50
13:50 foomator joined
foomator hi 13:50
[Sec] well, works is rleative. I get "Any" appended. But at least it does something.
foomator da sec :) 13:51
sbp rakudo: my @a = <eins zwei drei vier fuenf sechs>; say @a[1, 3 ... 5].perl
p6eval rakudo 0d3c71: OUTPUT«("zwei", "vier", "sechs")␤»
sbp I dunno why it works with 5 but not +@a
rakudo: my @a = <eins zwei drei vier fuenf sechs>; say @a[1, 3 ... 6].perl
p6eval rakudo 0d3c71: OUTPUT«(timeout)»
sbp guess it does screw up the arithmetic progression generator 13:52
13:52 snearch left
sbp rakudo: say [1, 3 ... 5].perl 13:52
p6eval rakudo 0d3c71: OUTPUT«[1, 3, 5]␤»
sbp rakudo: say [1, 3 ... 6].perl
p6eval rakudo 0d3c71: OUTPUT«(timeout)»
sbp there's a reduced testcase 13:53
[Sec] was about to paste exactly that :)
13:53 ponbiki left
colomon 1, 3 ... 6 is the same as 1, 3 ... * 13:55
you have to hit the terminating value exactly to terminate
you probably want something like 1, 3 ... * > 4 13:56
[Sec] so there is currently no nice way to select all odd elements in an array?
colomon The final "Any" above is actually a Rakudo bug, I think
sbp rakudo: my @a = <eins zwei drei vier fuenf sechs>; say @a[1, 3 ... * > +@a].perl
p6eval rakudo 0d3c71: OUTPUT«("zwei", "vier", "sechs", Any)␤»
colomon rakudo: my @a=<<eins zwei drei vier fuenf>>;say (1,3 ... +@a).perl
p6eval rakudo 0d3c71: OUTPUT«(1, 3, 5)␤»
13:57 jaldhar_ left
JimmyZ_ rakudo: my @a=<<eins zwei drei vier fuenf>>;say +@a; 13:57
p6eval rakudo 0d3c71: OUTPUT«5␤»
colomon @a[5] = Any, but I think it's supposed to be automatically ignored if it's @a[1, 3, 5]
13:58 daniel-s left
colomon I think what you're looking for is probably more like 13:58
rakudo: my @a = <eins zwei drei vier fuenf sechs>; say @a[1, 3 ...^ * >= +@a].perl
p6eval rakudo 0d3c71: OUTPUT«("zwei", "vier", "sechs")␤»
colomon rakudo: my @a = <eins zwei drei vier fuenf sechs>; say @a[0, 2 ...^ * >= +@a].perl
p6eval rakudo 0d3c71: OUTPUT«("eins", "drei", "fuenf")␤»
[Sec] colomon: yay! 13:59
@a[1,3...*] would've been nice, though. :)
JimmyZ_ rakudo: my @a=<<eins zwei drei vier fuenf>>;say @a[1,3 ... (+@a-1)].perl 14:00
p6eval rakudo 0d3c71: OUTPUT«(timeout)»
colomon [Sec]: I'm not sure @a[1, 3 ... *] will ever work, because the series there is infinite.
sbp is this step stuff taken from haskell? 14:01
foomator why shouldn't it work?
colomon even if we were smart enough to ignore values over +@a, there's still an infinite number of those values to ignore
[Sec] colomon: but the array isn't, and the series should be lazy?
colomon the sequence is lazy, but we're still going to ask for an infinite number of terms from it. 14:02
foomator hm, according to what i heard yesterday it should work afair
colomon I mean, it's obvious to us that once the sequence is past +@a, it's never going to be less than +@a again. 14:03
but that's not obvious to the compiler
foomator the array implementation should know how to handle the range, shouldn't it?
JimmyZ_ sbp: I dunno why it works with 5 but not +@a #because the last elems is +@a - 1
colomon It's not a range, it's a sequence.
If it were a range, it should work, yes.
foomator gotcha :)
colomon @a[1..*] should work 14:04
In fact...
rakudo: my @a = <eins zwei drei vier fuenf sechs>; say @a[2..*].perl
p6eval rakudo 0d3c71: OUTPUT«(timeout)»
colomon urk. well, that definitely should work
but @a[2...*] shouldn't. Or at least, will require a much more sophisticated framework to handle 14:05
[Sec] colomon: i might be spoiled from playing with functional languages which "just manage" this correctly %) 14:06
colomon are there functional languages that have array slicing? 14:08
[Sec] colomon: no. but stopping on lazy list evaluation when no more elements will match :) 14:10
sbp perl6: my @a = <eins zwei drei vier fuenf sechs>; say @a[2..*].perl 14:11
JimmyZ_ rakudo: my @a=<<eins zwei drei vier fuenf>>; say @a[1,3 ... ((+@a)-1)].perl # why time out?
p6eval rakudo 0d3c71, niecza v5-93-gb084a1e: OUTPUT«(timeout)» 14:12
..pugs: OUTPUT«("drei", "vier", "fuenf", "sechs")␤»
rakudo 0d3c71: OUTPUT«(timeout)»
14:12 MayDaniel joined
colomon JimmyZ_: +@a - 1 == 4. 1, 3 ... 4 is an infinite sequence 14:12
JimmyZ_ oh 14:13
colomon [Sec]: errr... that's not even theoretically possible in the general case, is it? it's the halting problem.
obviously it's possible in simpler cases... 14:14
14:16 zamolxes left
JimmyZ_ rakudo: my @a=<<eins zwei drei vier fuenf a b c>>; say @a[1,3 ... @a - 1 ].perl 14:17
p6eval rakudo 0d3c71: OUTPUT«("zwei", "vier", "a", "c")␤»
14:22 wtw left 14:23 buubot_backup left 14:33 mberends left 14:36 kaare_ joined, dual left 14:42 kjeldahl joined 14:45 ymasory joined 14:46 buubot_backup joined 14:51 koban` left 14:52 MayDaniel left 14:55 kjeldahl left 14:57 hercynium joined 15:03 mkramer1 joined 15:05 mkramer1 left, mkramer1 joined 15:07 silug joined 15:08 ab5tract left 15:12 silug left, silug joined 15:18 mkramer1 left 15:24 silug left 15:25 silug joined 15:31 silug left, silug joined 15:33 JimmyZ_ left, JimmyZ__ joined 15:35 molaf joined 15:37 envi joined 15:45 thou joined 15:46 silug left, bluescreen10 joined 15:47 MayDaniel joined, silug joined 15:48 JimmyZ__ left 15:49 mberends joined 15:51 ymasory left, silug left, silug joined 15:52 mj41 left
sorear colomon: it ought to work 16:05
colomon sorear: what ought to work?
sorear S09:218
@array[1,3...*]
the operative word here is "truncated" 16:06
as opposed to "filtered"
this means that after the first invalid subscript, it does a last;
colomon ah
16:06 tomaw left
sorear does anyone else think masak has been out an uncharacteristically long time? 16:07
colomon agreed on masak
16:07 silug left
colomon sorear: where do you something suggesting "truncated" is allowed to work the way you suggest? (I mean, what you suggest makes sense, but I'm not sure I understand where it actually says that in the spec.) 16:08
16:09 molaf left
colomon sorear: specifically, it looks to me like that section was written for ranges (which always increment, and are thus trivial to truncate), and then someone added "sequence" to it without clarifying any of the sequence-specific things. 16:11
or putting it a different way, it seems to me we should be modifying the spec so that it (more?) explicitly says what you are claiming it does say. :) 16:12
and not just for sequences -- for any list argument to [ ] 16:13
sorear probably I got it from truncate(2)
which removes bytes from the end of files, never the middle 16:14
it might even be in a real dictionary, I dunno
16:17 GinoMan[A] left 16:19 tomaw joined 16:31 MayDaniel left 16:32 kjeldahl joined
dalek ecza: 18545a8 | sorear++ | src/niecza:
Fix /< foo bar >/
 16:32
16:33 kaare_ left
mberends It is interesting that the software interpretations of "truncate" all drift around minimizing, fixing or setting some variable length, which has drifted some way from the original "eliminate branches" (keep only trunk) concept. 16:35
en.wiktionary.org/wiki/truncate 16:36
sorear mberends: consider the truncated icosahedron
mberends yep, that's the classic meaning
PerlJam mberends: consider perldoc -f truncate # :-) 16:38
mberends PerlJam: that one at least reduces, the Unix one is emulates will even lengthen a file :/ 16:39
PerlJam perhaps if it had been called "entrunken" or something, the meaning wouldn't have been diluted.
thou sorear: does your announcement up there mean that niecza is feature complete (as defined by the roast test suite)? 16:40
sorear thou: no 16:41
thou: yapsi is a Perl 6 app with its own tests (about 100 of them) 16:42
yapsi was developed for Rakudo alpha; I just got it to work in niecza yesterday
mberends
.oO( don't use SQL data types in C. There is no 'unsigned small' ) :P
 16:44
16:45 pernatiy left
thou sorear: ah, OK, got it. thanks. 16:46
16:48 fhelmberger left
sorear mberends: did you do that? 16:51
16:51 dual joined
sorear my main problem seems to be writing Vector<int> in C# and std::list<int> in C++ 16:52
mberends sorear: just finished $work on SQL server and then switched to developing a bigint library for Perlduino.
sorear can't just use bigint.pm? 16:53
mberends nope, I have severe NIH syndrome
add and multiply are done and tested, next will be subtract and divide 16:55
sorear how fancy is it?
mberends small^Wshort int arrays, very streamlined, pure C, max capacity +/- 2^524272. 16:57
and prettier source code than that C# one you just pasted into niecza ;) 16:58
sorear I mean more like, does it know about subquadratic division? 16:59
mberends no, long multiplication and long division algorithms (this is destined to cross compile to tiny 8 bit CPU's). 17:00
17:03 satyavvd left
mberends I hope implementing a bigrat library on top of bigint will not be too hard. 17:04
17:04 cdarroch joined, cdarroch left, cdarroch joined 17:05 birdwindupbird left 17:10 masak joined
masak hola, zebry. 17:11
phenny masak: 07:04Z <sorear> tell masak yapsi/niecza 52d74c6 passes all tests on niecza/master 6bf4ff8
mberends o/ masak! We were missing you!
colomon masak!
masak is... *drum roll* ...on a train
mberends !!
masak sorear: wow!
well, guys. I've missed you too ;) 17:12
the backside of the coin of a terribly successful life is that one's time gets more limited :P 17:13
dalek ecza: 7745f0d | sorear++ | / (2 files):
repeat, until, while, unless take pointy blocks
masak <-- happy bevause a course he taught went well
17:13 ymasory joined
sorear It's MASAK! 17:13
I was just beginning to wonder/worry about you :D
masak sorear: congratulations on making yapsi/niecza work! 17:14
17:14 Moukeddar joined
Moukeddar Hello Perl6 \􏿽xB0 17:15
masak Moukeddar, sir!
sorear HI Moukeddar!
Moukeddar it's been a while
how are you all doing?
sorear awesome 17:16
finals ended yesterday
PerlJam masak: hey! how's it going? (besides happy :)
Moukeddar me getting started in learning UML clarified a lot of things :)
masak PerlJam: tired ;) 17:17
I miss Perl 6. and #perl6. :)
PerlJam Moukeddar: was one of those things "don't use UML"?
Moukeddar i wasn't , i just didn't get to really understand how important it is
masak Moukeddar: don't pay too much attention to UML, IMO. just learn what you need, and move on to actual stuff.
masak is liberal with advice :P
colomon sorear: assuming you're correct on what is meant by truncate, what do you do if you are using @a[1..*] as an lvalue? 17:18
masak Moukeddar: ...not terribly important. :P
Moukeddar masak, that's exactly what i'm doing , learning what'll be usefull to me
sorear colomon: replaces the array's contents but doesn't change its length
colomon: unassigned values become Any, excess values are discarded
masak I haven't been this behind with backlogging in I don't know how long.
17:19 jaldhar_ joined
Moukeddar masak, any book on how to alanylse real world problems ? 17:20
colomon sorear: "(When used as an lvalue, any non-existent subscripts generate WHENCE proxies that can receive new values and autovivify anything that needs it.)" That sounds to me like changing the length is allowed? 17:21
Moukeddar and apply design patterns while designing the solution
analyse*
colomon rakudo: my @a = 1..10; @a[2,3,4] = <a b c>; say @a.perl
p6eval rakudo 0d3c71: OUTPUT«[1, 2, "a", "b", "c", 6, 7, 8, 9, 10]␤»
sorear colomon: that line is impossible
colomon sorear: why? 17:22
colomon has a deer in his backyard
sorear because 1. it would require expressions to know if they are lvalues 2. it would require creating an infinitely large Parcel with an infinite number of WHENCes 17:23
Moukeddar prepares for some Deer Hunt
colomon oh, that line of the spec, you mean?
sorear yes 17:24
17:25 masak left
colomon okay, then I'm going to try to implement truncating in Rakudo and see what I break. :) 17:25
oh, ack, what a minute.
PerlJam sorear: I don't think so ... it could generate a WHENCE generator that lazily fills the parcel as needed 17:26
colomon *wait
PerlJam :-)
sorear PerlJam: parcels are not lazy.
s/^/Nice try, but /
colomon sorear: does it ever say @a[blah] returns a parcel? 17:27
rakudo: my @a = 1..*; @a[2,3,4] = <a b c>; say @a[^10].perl 17:28
p6eval rakudo 0d3c71: OUTPUT«(1, 2, "a", "b", "c", 6, 7, 8, 9, 10)␤»
sorear it's the only way to make list assignment work.
colomon yeah, that's the catch with the sorear++ solution
(the catch: either the array or the slice might be infinite. if I naively check subscript members against the length of the array, I might trigger an infinite loop getting that length) 17:31
sorear IIRC you can do something like !pir::isnull($array!fill(1000)) which will return true iff $array has >=1000 elements 17:32
been a while since I hacked on the Rakudo list guts though 17:33
17:33 masak joined
colomon unfortunately, postcircumfix:<[ ]>(@pos) is on Any, not Array 17:33
17:34 stephanepayrard_ left
colomon hmmm... maybe that's not a stopping point, though 17:34
masak trains. they move in and out of the Internet.
mberends Moukeddar: before you practically program a computer, you will only know what others write about programming and you will learn very little that is usable. After you regularly program a computer, you will see which theoretical information is useful to you, but your practical experience will have a far higher value. 17:35
colomon rakudo: my $a = 10; say $a[0]
p6eval rakudo 0d3c71: OUTPUT«10␤»
masak Moukeddar: what mberends said.
colomon rakudo: my $a = 10; say $a[100]
p6eval rakudo 0d3c71: ( no output )
colomon rakudo: my $a = 10; say $a[100].perl
p6eval rakudo 0d3c71: OUTPUT«undef␤»
masak Moukeddar: I reached for something similar, but mberends got there before me and said it better, too. :)
colomon rakudo: my $a = 10; say $a[100].WHAT
p6eval rakudo 0d3c71: OUTPUT«Failure()␤» 17:36
masak would be hard pressed to recommend a Patterns book
mberends is not in a train
Moukeddar let me see if i get your point , i have to get into the business ?
colomon is inclined to agree with the folks who argue that Patterns are basically attempts to work around major holes in some popular programming languages. 17:37
Moukeddar colomon, your idea is quite familial
masak I'm more of a "classics" guy: TAoCP, SICP, Little Schemer...
Moukeddar so design patterns are useless or what ? 17:38
17:38 masak left
PerlJam colomon: yes and no. Patterns aren't meant to be some kind of "universal answer", they're just "stuff that works when you it" 17:38
17:38 masak joined
PerlJam sometimes you need it because your programming environment has some shortcoming or other. 17:38
Moukeddar i'm very aware that design patterns can't be used anywhere and anytime 17:39
PerlJam It's the *idea* of patterns that is really important IMHO
Moukeddar but they're good to learn , right?
PerlJam Moukeddar: common solutions to common problems? Absolutely! 17:40
Moukeddar good, so i'm on the right way ,still , i think i didn't quite get what Mr mberends meant , can someone rephrase it ? 17:41
mberends Moukeddar: they might help you only later in your learning curve, because you should understand quite a lot of the concrete implementations that the patterns describe. (more)
17:41 masak left
sorear Moukeddar: (familial) Did you mean "familiar"? 17:41
17:41 masak joined
Moukeddar mberends, so the patterns are the tree that hides the forest right ? 17:42
PerlJam Moukeddar: no
colomon hackerne.ws/item?id=2291417
Moukeddar sorear, yes familiar , French syndrom :p
sorear needs to learn that soonish
mberends Moukeddar: for example, there are many ways of sorting data. You will appreciate the theory of sorting much, much better if you try to write a few sort routines yourself.
Moukeddar sorear, no rush , learn it for romance :p 17:43
mberends, i got me a book called
wait a sec
Introduction to Algorithms
it has all this stuff
PerlJam and if you study big-O notation you'll appreciate that you can sometimes use a bubble sort (because it's the only algorithm you can remember for sorting) in an environment where there is no built-in sorting routines
mberends +1 17:44
masak Moukeddar: I like both reading theory/principles, and trying out things in practice. I think mberends++' point is something like doing the latter with make you appreciate the former more fully.
if that's his point, I agree fully. :)
toying around with Perl 6 has taught me loads about languages and compilers.
Moukeddar: is it red, white and green on the outside?
Moukeddar masak, it's CHM , books are like Aids cure here :p 17:45
pmichaud good afternoon, #perl6 17:46
Moukeddar so ,what you say is learn what's under the hood right ?
colomon pmichaud: good afternoon!
masak Moukeddar: I should ask: is it this one? www.amazon.com/Introduction-Algorit...262033844/
pmichaud: \o
Moukeddar like all the algorithms and data structures
colomon pmichaud: and I was just poking through your code....
PerlJam Moukeddar: as much as you can, yes.
Moukeddar understood , that was my plan for this summer :)
masak, that's the one :) 17:47
is it good ?
masak Moukeddar: I've never learned an algorithm or data structure and then thought "OH NOES! why did I learn *that*?". (barring, perhaps, AVL trees.)
Moukeddar: yes, it's quite alright.
Moukeddar: I haz it. I come back to the section on graphs quite a lot, because I like its explanations about graph algorithms.
Moukeddar lol masak
i have a long way to go :)
masak Moukeddar: have you seen ufo? it uses a topological sort from that book ;) 17:48
colomon pmichaud: in particular, trying to figure out how to make @a[1, 3 ... *] work
Moukeddar masak, i haven't started reading yet , i'm just collecting resources and preparing for the exams 17:49
my internet subscription will end this month
this sucks 17:51
pmichaud colomon: I've been round-and-round with TimToady many times about truncation of subscript indexes 17:52
it's still not resolved, afaict
17:52 ymasory left
masak PerlJam: I explained patterns yesterday and this morning. I gave the example of, instead of giving 75 detailed steps of how I moved through town, just being able to say "I went to work". that's a pattern, focusing on some things and clearing away innessentials. 17:52
colomon pmichaud: sorear and I were discussing it, and I'm reasonably happy with his suggestion... though maybe I'm missing something
pmichaud I backlogged, but might've missed it, so give me the latest version of his suggestion :-) 17:53
PerlJam masak: excellent!
masak PerlJam: also, good pattern descriptions tend to contain a "Forces" section, describing what'd make you want to use the pattern, and what'd make you avoid using it.
humble, self-deprecating patterns :)
Moukeddar lol 17:54
PerlJam masak: it's that crazy "designs don't live in a vacuum" thing :)
Moukeddar self what?
colomon pmichaud: his notion (as I understand it, anyway) is that as you process @pos (in postcircumfix:<[ ]>(@pos)), once you hit an index which is outside the current range of the array, you stop.
masak Moukeddar: the patterns contain instructions on when not to use them. that's very self-deprecating.
pmichaud doesn't work for lvalues 17:55
my @a; @a[5] = 'hello';
masak PerlJam: yes, that's what makes every new project... new and interesting :)
Moukeddar lol , things like that exist?
when to no use?
masak Moukeddar: you said that you understood that patterns are not to be used all the time. this is just an example of the patterns containing that information. 17:56
colomon pmichaud: understood, and I don't know how to resolve that.
pmichaud I know that TimToady already decided once before that truncating on first non-existing wouldn't work in the lvalue case.
sorear colomon: you only stop when the current index came from a range or sequence
pmichaud (Looking fr the reference now)
sorear: you don't always know that it comes from a range or sequence 17:57
Moukeddar i meant ,"with only not to use cases"?
what about when to use :p
PerlJam I'm (still) of the opinion that software systems need to be built with the idea of some sort of periodic-ish review process to make sure that, say, 5 years after its been in production, the conditions under which is was designed are still valid and useful.
17:57 lumi_ left
colomon sorear: I'd argue that should be anything @-ish, not just ranges and sequences 17:57
masak Moukeddar: no, it's generally both. :)
colomon anything iterable with more than one element? I'm not sure the best way to say that. 17:58
masak PerlJam: what you just said was basically on slide 2 of yesterday's class :P
pmichaud note that a parcel can be iterable and have more than one element.
Moukeddar interesting talk 17:59
pmichaud TimToday has said that ranges are supposed to autotrim. He's never explained how.
dalek ecza: 673862f | sorear++ | src/ (4 files):
Mergeback, start on is copy
ecza: 5967651 | sorear++ | / (5 files):
Implement 'is copy'
colomon pmichaud: ranges and sequences, as I read the current spec 18:00
masak PerlJam: or rather, that it's an architect's role to look not just at the situation now, but to anticipate concerns 1, 5, or 10 years into the future.
pmichaud colomon: yes, but I've come up with examples using ranges and sequences that obviously can't autotrim
I think the latest version was that the subscript autotrims if @pos is detectably infinite
masak pmichaud: do you have such an example handy? 18:01
colomon pmichaud: using ranges? really? other than the case where the @array[ ] is infinite?
pmichaud colomon: yes, using ranges.
PerlJam masak: yep. (and since we can't actually predict the future with 100% accuracy, when 5 or 10 years in the future becomes "now", we should revisit our designs)
pmichaud colomon: because it's pretty clear that my @a; @a[0..5] = $IN.lines; should grab six lines.
masak PerlJam: full ACK. 18:02
18:02 pernatiy joined
pmichaud people would be very surprised if @a[0..5] doesn't work the same as @a[0,1,2,3,4,5] 18:02
masak rakudo: say "hello pernaity!"
p6eval rakudo 0d3c71: OUTPUT«hello pernaity!␤»
colomon pmichaud: considering how easy it is to say $IN.lines[^6], I'd certainly be willing to throw that one under the bus.
tadzik masak o/
masak tadzik: \o!
PerlJam colomon: waterbed 18:03
colomon pmichaud: yeah, I'm arguing @[0, 1, 2, 3, 4, 5] shouldn't work either. :)
pmichaud colomon: my @a; @a[0..5] = ($IN.lines, $IN2.lines);
masak PerlJam: you can't throw things under a waterbed :P
PerlJam colomon: (you've just moved the problem around)
pmichaud and $IN.lines[^6] doesn't solve the case I just gave.
PerlJam masak: clearer? :)
colomon pmichaud: does ($IN.lines, $IN2.lines)[^6] not work?
pmichaud namely, that of assigning to the first six elements of @a.
colomon: doesn't matter. The subscript is on @a, not on $IN. 18:04
masak PerlJam: I'm overjoyed with explanation. :)
pmichaud my @a; ... ; @a = $IN.lines[^6] is certainly not the same as @a[^6] = $IN.lines;
(the first truncates @a if it already had more than six elements)
PerlJam yeah, I'd be surprised if the rest of @ disappeared 18:05
er, @a
colomon ah, true
masak me too.
I think we've established once before that we have a set of irreconcilable wishes in this area.
pmichaud TimToady: there's supposed to be some way of asking an iterator if it's *ish, but rakudo doesn't do that yet, afaik 18:06
sorry
that's supposed to be a quote, not a request
colomon masak++
pmichaud (cut/paste fail)
irclog.perlgeek.de/perl6/2011-02-08#i_3267485
PerlJam masak: That's why we have a language designer make the hard decisions for us :)
masak PerlJam: I wish him all the best in this case ;)
pmichaud the 2011-02-08 repeats the notion that we have to be able to ask an iterator for *-ishness in order to determine when we truncate 18:07
masak someone should write the irreconcilable wishes down, so that we can look at them and take a pick. 18:08
Moukeddar anyone here had a experience with the one-man Army case?
pmichaud which means that iterators have to figure it out somewhat transitively or lazily, since we can't know from @a[foo(), bar(), baz()] which if any of the things returned are *-ish
18:08 mkramer joined
Moukeddar where you have to do everything 18:08
colomon Moukeddar: you mean, to put together a program? 18:09
masak Moukeddar: I'm not sure why you're asking that here. :)
18:09 pmurias joined
pmichaud anyway, the idea that subscripts stop upon encountering the first non-existent element has been nixed already. 18:09
masak Moukeddar: sounds more like hiring issue than a programming issue.
Moukeddar hiring?
nah
masak Moukeddar: right. you agree to some terns, and then work according to those. 18:10
Moukeddar how does it feel when you're doing all the work ? from graphics to code to design
masak if the terms say "do everything", you do that :P
Moukeddar: it can feel quite nice.
Moukeddar: gives a lot of personal control, for sure.
pmichaud I can work on autotrimming a bit 18:11
need to dive into the List stuff anyway
Moukeddar true
masak hm, maybe it's called "artistic control", I dunno. 18:12
PerlJam Moukeddar: we are always a one-man-army ... it's just the enemy that changes :)
pmichaud wants to be a one-man air force.
Moukeddar who's your enemy now?
masak PerlJam: against a sea of troubles? to die, to sleep? 18:13
PerlJam pmichaud: not a one-man space force?
pmichaud PerlJam: air force is sufficient for now. :-)
colomon suspects he is a one-man air farce
PerlJam pmichaud: is that humility I sense? ;)
masak pmichaud: I've heard it requires replacing one's heart with a nuclear reactor...
pmichaud PerlJam: unlikely. :)
right now I could use a new throat and lungs (have a really nasty cold) :-( 18:14
PerlJam colomon: as long as it really is air and not methane that you're generating
masak suddenly wishes he had a photo montage of Iron Man pmichaud
Moukeddar lol @ farce
good pun
pmichaud okay, so others can help me prioritize tasks
1. Make a new Star release (nobody seems terribly excited about this one) 18:15
masak 1. buy nuclear reactor
PerlJam pmichaud: make rakudo faster. :)
pmichaud 2. Work on rakudo speed
3. Improve nom/nqp build system
4. Work on lists/iterators/subscripts
5. Work on nom/nqp
PerlJam pmichaud: if you wait long enough on #4 the design will change anyway ;)
pmichaud PerlJam: on #4 I'm currently in charge of the design
Moukeddar masak, i heard QUADAFI has an unused one , i cal talk him to sell it cheap :p
pmichaud at least for lists and iterators 18:16
masak Moukeddar: I forgot to mention it has to be really small. like, really small.
PerlJam oh that's right ... just wait until you've changed your mind at least once, then do something.
pmichaud all of the above are -Ofun for me, so I'm interesting in working on whatever people think is best for progress
*interested
Moukeddar masak, he carries is in his costumes , ever wondered why they're so vast
PerlJam at least you'd cut out one round of rewrites
pmichaud I'm a little hesitant to work too much on Rakudo speed if we're going to throw a bunch of stuff out in nom anyway
masak +1 18:17
pmichaud (unless I work on it *in nom*, but there are other more pressing tasks there before we can get down to optimization)
sorear well, there's always Niecza speed :D
PerlJam heh 18:18
pmichaud oh, 6. Plagiarize niecza's LTM engine
(that one is less -Ofun for me at the moment, for some reason)
PerlJam I guess I'm implicitly assuming that nom will land soonish. (whatever "land" means)
sorear please, call it "research"
pmichaud hums a Tom Lehrer tune 18:19
PerlJam pmichaud: if I were you, I'd probably go with #5. But, that's just me.
masak Lobachevsky!
colomon pmichaud: IMO, your priorities should be help with nom strictly as needed and sorting out #4 otherwise 18:20
pmichaud PerlJam: wfm. Although I might do #3 first, to simplify the install/build path for others wanting to play with nom
right now it's fairly convoluted (not surprising, given the history of the projects) 18:21
PerlJam pmichaud: yeah, if I were me, I'd want to you do #3. Wait ... I am me! do #3 ;)
But I have no plans on hacking on nom, so it doesn't much matter to me. 18:22
pmichaud speaking of nom, time for lunch 18:23
PerlJam pmichaud: isn't #4 tied to a grant ?
pmichaud I'll focus on #3 and #4
PerlJam: yes, and I should finish that too.
so #3 and #4 it is.
masak \o/ 18:24
sorear tries to figure out what hyperoperators are supposed to do.
pmichaud they make everyone say "oh! Perl 6 has just reinvented APL!" 18:25
"Including the impenetrable syntax!" 18:26
masak or "This is why Perl 6 will never be anything but an *academic* language." 18:27
pmichaud news.google.com is coming up blank for me.
maybe Paypal froze Google's accounts. 18:28
masak there were no news today. ;)
pmichaud pmichaud.com/sandbox/nonews.jpg 18:29
masak maybe you accidentally turned of JavaScript, or something. 18:31
pmichaud no, it's working now
well, with only a limited set of news
masak well, you know what they say about no news. 18:32
pmichaud afk, nom
sorear masak: I'd be more than happy to ignore them. I don't like them either. But... yapsi uses them. 18:33
masak sorear: oh! sorry about that. 18:35
masak isn't very sorry. not really. :)
sorear down to 29 lines changed, most of them involving hyperops, junctions, or the "use package bug" 18:38
masak sorear: nice. that's better than I'd have thought. 18:39
sorear: also happy you're focusing on convergence. that's good for everyone.
someone(TM) should continue to enumerate/explain the results from running the Contest entries on Niecza, too. 18:40
probably me.
sorear hmm 18:42
was there ever a yapsi 2011.04?
or 05 for that matter 18:43
18:43 MayDaniel joined
masak no .04, yes .05 18:44
but I might've forgot to tag it :/
18:44 masak left, masak joined
masak sorear: there. now Yapsi 2011.05 has a tag. 18:46
sorear tosses masak feather.perl6.nl/~sorear/Yapsi-2011...niecza.zip 18:47
masak thanks.
masak investigates
sorear: hmm? Windows executable? 18:48
sorear CLR executable 18:49
masak ah.
sorear will run with mono
masak mono MAIN.exe 18:50
>>> say 42
42
\o/
that is so cool!
and *fast*!
sorear++
18:50 ymasory joined
masak now anticipates making the 2011.06 announcement quite a bit more 18:50
"Yapsi -- the only Perl 6 implementation with cross-Perl6-implementation support!" 18:51
pmurias how fast is yapsi on niecza compared to rakudo? 18:55
masak I just timed the niecza variant for -e 'say 42'. it took .2 seconds on the Ubuntu VM of my laptop. 18:56
the rakudo variant will take... longer. hold on. 18:57
18:58 lumi_ joined, Moukeddar left
masak not sure any comparison I can make will be very fair. 18:59
the niecza version is all precompiled; the reakudo version has a bin/yapsi script.
18:59 mkramer left
masak but, apart from that... running the rakudo variant takes 4 seconds, quite exactly. 19:00
so Niecza's Yapsi is about an order of magnitude faster. 19:01
sorear "reakudo" eh 19:05
sorear is very dubious about the claims in S03 that hyperops will be the fastest way to operate on lists 19:09
pmurias sorear: what would be faster? 19:13
sorear &map, for one 19:14
pmurias why should that be faster? 19:15
19:15 wallberg left, cognominal joined
sorear it has much less magic 19:15
19:16 wallberg joined, silug joined
Tene isn't map specced as strictly ordered? 19:16
pmurias sorear: hyper ops have magic? 19:17
19:23 birdwindupbird joined 19:27 Holy_Cow joined, Holy_Cow left
sorear Tene: yes 19:28
19:28 envi left
sorear Tene: but automatic threading is a pipe dream 19:28
What stops (0, @array) »+« (@array, 0) from being a shape error? 19:30
19:31 masak` joined
sorear my current interpretation of S04 tells me that (0, @array) «+» (@array, 0) ==> (@array, @array) 19:31
19:31 silug left 19:32 masak left
sorear pmichaud: how confident do you feel about the correctness of Rakudo's current hyper-op implementation? 19:32
19:32 pernatiy left 19:34 masak` left, PacoLinux left 19:35 masak` joined
pmurias sorear: what's unrealistic about automatic threading? 19:35
sorear pmurias: there's no way to tell the difference between -«[1,2] and [&expensive1,&expensive2]»() 19:36
pmurias: creating threads in the first case would destroy performance. as would setting a timer 19:37
19:37 snearch joined
sorear I've figured out how to make an efficient hyperop, but I still think that parallelization requires user input 19:37
maybe a use hyper :threaded; pragma
mberends pmurias: if an array has, for example, 32 elements, is it performant to create 32 threads to process it? That will depend on the hardware. 19:38
pmurias my impression is that the intent is to parallelize for loads of elements 19:39
if the array has 32000000 elements the choise is more obvious
mberends sure, but it's hard to work out where to draw the lines 19:40
pmurias doing it manually doesn't seem much easier 19:41
19:42 masak`` joined, mkramer joined
pmurias and people seem to have some success in doing it in haskell (and moving computation to the GPU) 19:43
19:44 masak` left
mberends we could probably learn a lot from their experiences, because we are all n00bs in that area. 19:45
19:46 masak`` left
sorear I was semi-involved in the Data Parallel Haskell design work 19:46
if that's what you're talking about
19:46 masak`` joined
sorear although most of Roman's work still blows my mind 19:46
S03-metaops/hyper.t:687 makes no sense 19:48
why should eval accept a Buf, *ever*?
sorear pokes masak`` in eir capacity as Buf person 19:49
19:51 masak`` left, masak`` joined
pmurias sorear: wouldn't it cast it to uft8? 19:51
19:52 mkramer left 19:53 mkramer joined
pmurias sorear: the intent of the test is to check that latin1 operators in source file don't work without predeclaration 19:54
19:54 wamba left 20:02 masak`` left 20:04 snearch left, mkramer left
sorear Why are 'isa' and 'does' separate? 20:06
mberends someone once said 'isa' = class = instance management, 'does' = role = behaviour management. You may not need classes because if you create an object that does a role, the compiler or runtime is supposed to create a private anonymous class that does that role, exclusively for use by that object. 20:15
oh, classes have an inheritance hierarchy, roles do not. 20:17
20:21 MayDaniel left, ymasory left
sorear In niecza I shall make isa/does and is/does synonymous in every case 20:28
20:28 wknight8111 left 20:33 PacoLinux joined 20:34 orafu left
pmurias perl6: role Foo {method a {}};role Bar {method a {}};class Baz is Foo is Bar {} 20:34
p6eval pugs, rakudo 0d3c71: ( no output )
..niecza v5-97-g5967651: OUTPUT«Unhandled exception: Unable to resolve method linearized_mro in class Role␤ at /home/p6eval/niecza/src/Metamodel.pm6 line 339 (Metamodel Class.close @ 27)␤ at /home/p6eval/niecza/src/NieczaPassBegin.pm6 line 412 (NieczaPassBegin C175_ANON @ 14)␤ at
../home/p6eval/niecza/src/Niecza…
20:34 mberends left
pmurias rakudo: role Foo {method a {}};role Bar {method a {}};class Baz does Foo does Bar {} 20:34
p6eval rakudo 0d3c71: OUTPUT«===SORRY!===␤Method 'a' collides and a resolution must be provided by the class␤»
20:34 orafu joined
pmurias sorear: that's one difference 20:34
20:35 mberends joined 20:45 dukeleto left, dukeleto joined 20:50 birdwindupbird left 20:52 bluescreen10 left 20:56 wooden joined 20:57 benabik left 20:59 meteorjay left, meteorjay joined 21:00 mberends left 21:06 bluescreen10 joined
Util New RC solution: rosettacode.org/wiki/Anagrams/Deran...ams#Perl_6 21:11
21:17 Gruber joined 21:18 frank joined, Grrrr left 21:19 BinGOs left, aesop left, BinGOs joined, nrr left, nrr joined 21:21 Gruber is now known as Grrrr 21:24 MayDaniel joined 21:32 bluescreen10 left 21:46 noganex_ left 21:49 Patterner left 21:51 noganex joined 21:53 hercynium left 21:54 Psyche^ joined, Psyche^ is now known as Patterner
sorear bah, niecza's STD chokes on <<+>> 21:55
the "LTM" component is adamnt that there's some bit shifting going on
22:00 silug joined 22:09 donri left 22:10 dukeleto left, dukeleto joined 22:18 breatharian joined
breatharian Howdy 22:19
sorear Hello.
breatharian It appears that Perl 6 allows "block expressions". By block expression I mean: gist.github.com/996274
My question is, does Perl 5?
sorear yes 22:20
breatharian sorear, Yes to 5 or 6? :-)
sorear buubot_backup: say 1 + (do { my $x = 5; $x })
buubot_backup sorear: Couldn't match input.
sorear buubot_backup: eval: say 1 + (do { my $x = 5; $x })
buubot_backup sorear: ERROR: syntax error at (eval 28) line 1, near "say 1" syntax error at (eval 28) line 1, near "$x }"
sorear buubot_backup: eval: print 1 + (do { my $x = 5; $x })
buubot_backup sorear: 61
sorear buubot_backup: eval: 1 + (do { my $x = 5; $x })
buubot_backup sorear: 6
sorear breatharian: also, if you wrap the C example in parens so that it becomes ({ ... }), it will become legal[1] 22:21
[1] actually a gcc extension, but icc, clang and I think also msvc have added it also 22:22
breatharian sorear, Aha! So the example above is legal Perl 5. Is it also legal Perl 6? 22:24
sorear yes 22:26
but mostly by accident
the meaning of "do" has changed fairly significantly
a Perl 5/Perl 6 polyglot is about as hard as a Perl 5/Ruby polyglot; don't try to write them, except as a game 22:27
22:27 MayDaniel left
sorear welcome, btw 22:28
breatharian sorear, I updated the gist with the Perl example. 22:31
sorear, By the ({...}), are you referring to this? en.wikipedia.org/wiki/Blocks_%28C_l...tension%29 22:33
22:34 silug left
breatharian sorear, The page says it's an Objective-C only extension. 22:34
sorear, Er... I mean an Apply only extension.
sorear breatharian: gcc.gnu.org/onlinedocs/gcc-4.6.0/gc...Exprs.html
it's a gcc extension, it's been around since 2.x if not before 22:35
breatharian sorear, Aha!!!
sorear I know for a fact icc and clang have it; glibc and linux use it massively
jlaire those blocks you linked to are more like lambda functions
and very non-portable 22:36
breatharian sorear, The C example in the gist wasn't quite legal with the ( and ). I needed to remove the 'return'.
sorear in perl 6 it's really easy 22:37
niecza: say (my $x = 5; $x)
p6eval niecza v5-97-g5967651: OUTPUT«5␤»
tadzik niecza: say { my $a = 5 } 22:38
p6eval niecza v5-97-g5967651: OUTPUT«Potential difficulties:␤ $a is declared but not used at /tmp/5pRMrXdnfU line 1:␤------> say { my $a⏏ = 5 }␤␤Sub()<instance>␤»
tadzik perl6: say { my $a = 5 }
p6eval rakudo 0d3c71: OUTPUT«_block131␤»
..niecza v5-97-g5967651: OUTPUT«Potential difficulties:␤ $a is declared but not used at /tmp/yr1Phjsi6F line 1:␤------> say { my $a⏏ = 5 }␤␤Sub()<instance>␤»
..pugs: OUTPUT«<SubBlock(<anon>)>␤»
tadzik hrm
sorear perl6: say { my $a = 5; $a }()
p6eval rakudo 0d3c71, niecza v5-97-g5967651: OUTPUT«5␤»
..pugs: OUTPUT«<SubBlock(<anon>)>␤*** Cannot cast from VBool True to Pugs.AST.Types.VCode (VCode)␤ at /tmp/DEQxBTYidI line 1, column 1 - line 2, column 1␤»
22:39 wallberg left
breatharian Does Python support block expressions? 22:39
sorear doubt it
that's not Guido's style
breatharian What about Ruby? 22:42
sorear nice, I'm already finding bugs in the hyperop tests 22:47
no clue
22:47 am0c joined
sorear also, I've lost the ability to type "nice" quickly 22:47
I send n+i+c+e to my fingers and it comes out "niec..." 22:48
sometimes "niecza"
breatharian sorear, Thanks alot for the help. Dinner time.
22:48 breatharian left
pmurias sorear: what bugs did you find? 22:52
sorear pmurias: my @a = 1,4,9; @a»++; is_deeply @a, (2,5,9) # NO 22:53
pmurias perl6: my @a = 1,4,8;@a>>++; say @a; 22:54
p6eval niecza v5-97-g5967651: OUTPUT«Unhandled exception: System.Exception: Unable to find lexical &hyperunary in mainline␤␤Server stack trace: ␤ at Niecza.CLRBackend.NamProcessor.ResolveLex (System.String name, Boolean upf, System.Int32& uplevel, Boolean core) [0x00000] in <filename unknown>:0 ␤ at
..Niecza.CLRBacken…
..rakudo 0d3c71: OUTPUT«259␤»
..pugs: OUTPUT«148␤»
pmurias perl6: my @a = 1,4,9;@a>>++; say @a;
p6eval niecza v5-97-g5967651: OUTPUT«Unhandled exception: System.Exception: Unable to find lexical &hyperunary in mainline␤␤Server stack trace: ␤ at Niecza.CLRBackend.NamProcessor.ResolveLex (System.String name, Boolean upf, System.Int32& uplevel, Boolean core) [0x00000] in <filename unknown>:0 ␤ at
..Niecza.CLRBacken…
..pugs: OUTPUT«149␤»
..rakudo 0d3c71: OUTPUT«2510␤»
pmurias sorear: any idea why they weren't detected?
sorear: i think the rationale behind eval(Buf) is to allow something like eval('use latin1;...'); 22:55
does perl6 have a way to specify buf as a literal? 22:56
sorear yes 23:03
:256{0,1,234,92,32...}
they were not detected because they were skipped for Rakudo 23:04
rakudo apparently has issues with »++
(niecza doesn't :D)
23:08 kid51 joined
sorear o/ kid51 23:08
23:10 whiteknight joined 23:15 dukeleto left 23:16 dukeleto joined 23:17 pmurias left 23:35 cdarroch left 23:54 benabik joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!