»ö« | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, alpha:, pugs:, std:, or /msg p6eval perl6: ... | irclog: irc.pugscode.org/ | UTF-8 is our friend!
Set by moritz_ on 25 June 2010.
00:11 hercynium joined 00:17 yinyin joined 00:23 jferrero left, yinyin left 00:29 songmaster joined 00:32 rgrau_ left 00:41 masonkramer joined 01:09 patspam joined 01:10 patspam left 01:14 [Coke] joined
sorear Has anyone tried to translate the YamlReference grammar to Perl 6? 01:32
01:39 PZt joined
sorear TimToady: Does a method call always have to account for the possibility that said method will be overriden with a version that has "is rw" arguments? 01:44
ie class Foo { method bar(int $x) { ... }; }; sub sam(Foo $a) { ...; $a.bar($y); ... }; ...; ...; ...; eval 'class Foo2 is Foo { method bar(int $x is rw) { ... }; }' 01:45
01:59 bgoggin joined, hmmm joined 02:01 bgoggin left 02:09 agentzh joined 02:20 masonkramer left
hmmm About Perls 6, and the choice between loading Parrot or Haskell to try it, what's advised foe Win32? 02:22
foe==for, typo...
[Coke] rakudo works fine on win32. 02:23
SFAIK.
hmmm no Parrot for Rakudo on Win32?
maybe no haskell needed for PUGS also on Win32? 02:24
Though i do like Haskell, and mayy load it anyways... 02:25
02:29 masonkramer joined 02:30 lkk- joined
[Coke] yes, parrot builds on win32. 02:31
if you don't have a compiler, you can just grab strawberry perl. 02:32
I have no idea what the current state of pugs is.
02:36 ash_ joined 02:48 masonkramer left 02:54 hercynium left 02:57 alester joined 02:58 justatheory joined
dalek kudo: 9993ee9 | pmichaud++ | src/builtins/Parcel.pir:
Refactor Parcel assignment to be lazier, use &infix:<=>. Fixes RT #75950.
 03:04
03:04 lkk- left 03:05 tylercurtis joined 03:09 rcsheets joined
tylercurtis Is there anything in Perl 6 like the "[1]" in APL's "+/[1]m"(i.e., something that modifies operators/functions on arrays by causing them to operate on a different axis of a multi-dimensional array than they normally would, in this case, on the first axis(columns, if it's a matrix))? 03:09
[particle] not in the spec, you'd have to build that
03:09 jaldhar joined
tylercurtis I didn't expect there would be. Thanks. 03:11
[particle] p6 stole some things from apl, but not that one 03:12
hmmm I see, parrotWin32+Rakudo is the needed combo for Perls6 on Win32. 03:20
03:20 lkk- joined
hmmm PUGS looks like a version in December... interesting. 03:21
L8R, thxs!
03:25 hmmm left 03:30 Bigshot joined
ash_ tylercurtis: how does that [1]m work in your example from APL? 03:35
pmichaud sub abc(@pos) { say @pos.elems }; abc([1, 2, 3]); # valid, outputs 3?
that should probably say "valid?" 03:36
03:37 TiMBuS joined
ash_ 3 looks right to me... 03:37
rakudo: sub abc(@pos) { say @pos.elems }; abc([1, 2, 3]); # what does rakudo say? 03:38
p6eval rakudo d16a2f: OUTPUT«3␤»
pmichaud sub abc(@pos) { say @pos.elems }; 'a' ~~ /./; abc($/); # valid? outputs 0? 03:40
ash_ $/ has no captures in the current regex, so... sure... 03:41
tylercurtis ash_: normally, "+/m" would sum the last axis(rows) of the matrix, so, if m were "1 2 3\n4 5 6", it would be "5 7 9"(I think, I don't have an actual APL interpreter on hand), but "+/[1]m" would sum the first axis(columns) of the matrix, so, given the same value for m, it would be "6 15"(or maybe "6\n15", again, no APL implementation on hand). I think. 03:42
03:43 tedv left 03:46 Bigshot left
ash_ so your selecting the first column of the matrix, (say ((1, 2, 3), (4, 5, 6), (7, 8, 9))) and you want to add the first column? 03:52
lue pmichaud: how's variable refactor going? I noticed the commit earlier. 03:53
pmichaud lue: it's going a bit slower than I'd like.... there's lots of pieces that have to be fixed. 03:56
lue: but each piece I fix seems to close another ticket or two :)
lue it's sounds like something I would be interested in. Something about modifying a fundamental part of the interpreter I guess. :)
rakudo: say ((1, 2, 3), (4, 5, 6), (7, 8, 9)).perl
p6eval rakudo d16a2f: OUTPUT«((1, 2, 3), (4, 5, 6), (7, 8, 9))␤» 03:57
lue (hrm, I could swear () flattened...)
pmichaud it does flatten in item context
rakudo: say ((1, 2, 3), (4, 5, 6), (7, 8, 9)).item.perl 03:58
p6eval rakudo d16a2f: OUTPUT«(1, 2, 3, 4, 5, 6, 7, 8, 9)␤»
ash_ wouldn't @m[0;*] give you a slice with the first column (assuming my @m[3;3])
lue Ah well. For safety, I'll always use [] 03:59
rakudo: my @a = [ [1,2,3], [4,5,6], [7,8,9] ]; say @a[0;*].perl
p6eval rakudo d16a2f: OUTPUT«Method 'Num' not found for invocant of class 'Block'␤ in 'infix:<<>' at line 283:CORE.setting␤ in 'infix:<<>' at line 555:CORE.setting␤ in 'postcircumfix:<[ ]>' at line 1083:CORE.setting␤ in main program body at line 1␤»
ash_ master doesn't do multi-dimensional arrays currently
i forget if alpha did or not... 04:00
pugs: my @a[3;3];
p6eval pugs: OUTPUT«*** ␤ Unexpected ";"␤ expecting "_", fraction, exponent, term postfix, operator or "]"␤ at /tmp/SbgnSqgOkL line 1, column 8␤»
ash_ alpha: my @a[3;3];
p6eval alpha 30e0ed: OUTPUT«Confused at line 10, near "[3;3];"␤in Main (file <unknown>, line <unknown>)␤»
ash_ std: my @a[3;3];
p6eval std 31481: OUTPUT«ok 00:01 109m␤»
ash_ yeah, i think thats just not yet implemented 04:01
pmichaud rakudo has never done multdim arrays
tylercurtis ash_: What "+/[1]m" does is add all of the columns together. with ((1,2,3), (4,5,6), (7,8,9)), the result should be (1+4+7, 2+5+8, 3+6+9).
ash_ oh, like that, got ya
lue
.oO(Then I guess matrix multiplication is out of the question)
ash_ @m.grep: { [+] @_ }; #? 04:02
me shrug just thinking 04:03
ash_ shrug*
does grep flatten?
tylercurtis rakudo: my @m = [1,2,3], [4,5,6], [7,8,9]; @m.map([+] *).perl.say 04:06
p6eval rakudo d16a2f: OUTPUT«(6, 15, 24)␤»
ash_ there ya go, thats not to bad 04:07
pmichaud rakudo: say (1,2,3) Z (4,5,6) Z (7,8,9); 04:09
p6eval rakudo d16a2f: OUTPUT«No applicable candidates found to dispatch to for 'infix:<Z>'. Available candidates are:␤:()␤:(Any $lhs, Any $rhs)␤␤ in main program body at line 11:/tmp/eSDz5bCi4S␤»
pmichaud urgh.
rakudo: say (1,2,3) Z (4,5,6); 04:10
p6eval rakudo d16a2f: OUTPUT«142536␤»
lue gets the crazy idea that Z is supposed to chain 04:12
pmichaud it's supposed to be list associative, iirc
we probably need variadic slice context 04:13
rakudo: say ((1,2,3) Z (4,5,6)).perl; 04:15
p6eval rakudo d16a2f: OUTPUT«(1, 4, 2, 5, 3, 6)␤»
pmichaud that's wrong also, I think.
should be parcels
ciphertext pmichaud: see rt.perl.org/rt3/Ticket/Display.html?id=75818 04:17
pmichaud fixing the two-arg form is pretty easy. 04:18
I'm wondering if I can do the three-arg form.
(er, n-arg form)
04:29 eternaleye joined
tylercurtis rakudo: ([>>+<<] ([1, 2, 3], [4, 5, 6])).perl.say 04:39
p6eval rakudo d16a2f: OUTPUT«[[5, 7, 9]]␤»
tylercurtis rakudo: ([>>+<<] ([1, 2, 3], [4, 5, 6], [7, 8, 9])).perl.say
p6eval rakudo d16a2f: OUTPUT«Sorry, sides are of uneven length and not dwimmy.␤ in 'hyper' at line 113:CORE.setting␤ in 'hyper' at line 136:CORE.setting␤ in 'reducewith' at line 1␤ in main program body at line 1␤»
tylercurtis rakudo: ([>>+<<] \[1,2, 3], \[4, 5, 6], \[7, 8, 9]).perl.say 04:42
p6eval rakudo d16a2f: OUTPUT«[[[12, 15, 18]]]␤»
tylercurtis rakudo: my @m = [1,2,3], [4,5,6], [7,8,9]; ([>>+<<] @m).perl.say 04:43
p6eval rakudo d16a2f: OUTPUT«Sorry, sides are of uneven length and not dwimmy.␤ in 'hyper' at line 113:CORE.setting␤ in 'hyper' at line 136:CORE.setting␤ in 'reducewith' at line 1␤ in main program body at line 1␤»
mberends [>>+<<], now that's a good looking operator (maybe there's a single Unicode character for that ;)
songmaster rakudo: say [+] 1, 2, 3 04:48
p6eval rakudo 9993ee: OUTPUT«6␤»
songmaster Strange, my perl6 installation responds with "Could not find sub &prefix:<[+]>" 04:49
ciphertext songmaster: i'm getting that too: it seems to be a new REPL bug
mberends commute -> $work 04:51
songmaster ciphertext: Thanks, I though I was going crazy...
tylercurtis songmaster, ciphertext: I've gotten that sometimes, too. It usually stops happening if I close my REPL and start it again, though.
tylercurtis has no idea why...
ciphertext trying that
> [+] 1,2,3; 04:52
6
o.O
weird
songmaster Ditto; strange.
ciphertext although, i actually tried that once earlier, and it didn't work that time... 04:53
ingy pmichaud: when you terminate a ec2 running instance, does it blow away all your work? 04:55
pmichaud ingy: yes. There's a way to use a different storage type to keep your work, I think (i.e., mounted volumes) 04:56
ingy: I haven't tried it yet.
ingy pmichaud: so I can't turn this thing off without nuking my whole setup, eh? 04:57
pmichaud uec-images.ubuntu.com/releases/10.04/release/ has a list of available ubuntu images -- I think the ones that say 'ebs' have some sort of persistant storage available. 04:58
04:58 synth joined
songmaster rakudo: say [||] 0,0; say [||] 0,1; say [||] 1,0; say [||] 1,1 04:58
p6eval rakudo 9993ee: OUTPUT«0␤1␤1␤1␤»
04:58 Mowah joined
songmaster rakudo: say [&&] 0,0; say [&&] 0,1; say [&&] 1,0; say [&&] 1,1 04:59
p6eval rakudo 9993ee: OUTPUT«===SORRY!===␤Could not find sub &infix:<&&>␤»
hejki rakudo: say 0 && 0
p6eval rakudo 9993ee: OUTPUT«0␤»
songmaster Bingo! I have not managed to get [&&] to work yet.
hejki rakudo: say [+] 5,1,3 05:00
p6eval rakudo 9993ee: OUTPUT«9␤»
tylercurtis songmaster: That's because infix:&& and infix:and aren't actually defined. &&, ||, and, and or are implemented specially in the grammar. I'm about to submit a patch on RT to define sub forms of && and and, though. 05:02
05:03 synth left
songmaster tylercurtis: Thanks, I'll do a work-around then. 05:03
ciphertext tylercurtis: looks like someone beat you to it: rt.perl.org/rt3/Ticket/Display.html?id=76206 05:05
tylercurtis ciphertext: yay! I don't have to learn to use RT yet. 05:08
05:10 synth joined 05:15 synth left
hejki rakudo: say [and] 0,1 05:19
p6eval rakudo 9993ee: OUTPUT«===SORRY!===␤Could not find sub &infix:<and>␤»
hejki ahh.. what tylercurtis said about those :P 05:20
05:22 jhuni joined 05:23 agentzh left
dalek kudo: aea3bbc | pmichaud++ | src/Perl6/ (2 files):
Add statement_mod_cond:sym<when>. Resolves RT #75916.
 05:29
05:32 alester left 05:33 kaare joined, kaare is now known as Guest94095 05:41 eternaleye left 05:46 synth joined
tylercurtis Is "[@a, @b]" the idiomatic way to concatenate two arrays together in Perl 6? 05:59
06:06 justatheory left
pmichaud tylercurtis: (@a, @b) 06:08
tylercurtis: [@a, @b] also works if you want an array instead of a list. But most people would consider (@a, @b) to be a bi tmore canonical, I think.
tylercurtis rakudo: my @a = 1, 2, 3; my @b = 4, 5, 6; (@a, @b).perl.say 06:09
p6eval rakudo 9993ee: OUTPUT«([1, 2, 3], [4, 5, 6])␤»
06:10 xinming joined
tylercurtis pmichaud: (@a, @b) appears to produce a nested list, at least in Rakudo. Should it not do that? Or is there something I'm failing to understand(most likely option)? 06:12
06:15 synth left 06:17 synth joined 06:20 xinming left
pmichaud it's not a nested list 06:20
it only looks that way because of the way that .perl works (which I disagree with atm)
rakudo: my @a = 1, 2, 3; my @b = 4, 5, 6; my @c = (@a, @b); say @c.perl; 06:21
p6eval rakudo 9993ee: OUTPUT«[1, 2, 3, 4, 5, 6]␤»
pmichaud currently .perl forces arrays to be display with [ ]
*displayed
note the difference:
my @a = 1, 2, 3; my @b = 4, 5, 6; my @c = (@a, b); my @d = [@a, @b]; say @c.elems; say @d.elems; 06:22
rakudo: my @a = 1, 2, 3; my @b = 4, 5, 6; my @c = (@a, b); my @d = [@a, @b]; say @c.elems; say @d.elems;
p6eval rakudo 9993ee: OUTPUT«Could not find sub &b␤ in main program body at line 11:/tmp/z7SGd8jhbx␤»
pmichaud rakudo: my @a = 1, 2, 3; my @b = 4, 5, 6; my @c = (@a, @b); my @d = [@a, @b]; say @c.elems; say @d.elems;
p6eval rakudo 9993ee: OUTPUT«6␤1␤»
pmichaud there.
dalek kudo: 05684c0 | pmichaud++ | src/Perl6/Actions.pm:
Change handling of return value in empty statementlist. Partially resolves RT
06:26
06:31 xinming joined 06:34 foodoo joined 06:40 synth left 06:41 synth joined 06:47 synth left 06:50 synth joined, pmichaud joined 06:55 synth left, synth_ joined, Helios left 06:58 colomon joined, eiro joined, thepler joined, tomaw joined, Gothmog_ joined, opx^away joined, card.freenode.net sets mode: +o colomon 06:59 synth_ left, synth joined
cosimo_ how do I convert a Match object to a Str ? 07:05
moritz_ ~$/ 07:08
or $/.Str
hejki Match() has method .Str now? 07:09
moritz_ rakudo: say ('abc' ~~ /../).Str
p6eval rakudo 05684c: OUTPUT«ab␤»
hejki nice
cosimo_ $0.Str worked, thanks
moritz_ hejki: pmichaud and I fixed Match objects for the last release
cosimo_ it's fantastic! LWP::Simple now uses MIME::Base64 for HTTP basic auth support
hejki last time i updated my anydice.pl Match() didn't have that method :>
moritz_: ahh nice :)
07:10 Helios joined
cosimo_ \o/ 07:10
moritz_ \o/ 07:11
cosimo_++ # LWP::Simple, and blogging about it
cosimo_ now I need chunked transfers support, and it'll be even usable 07:15
07:18 songmaster left 07:26 Su-Shee joined 07:28 synth left, synth joined 07:41 Guest94095 is now known as kaare_ 07:53 eternaleye joined 08:03 Ross joined
sorear phenny: tell hmmm pugs has been effectively dead for >1 year. There are more recent releases, but only to track incompatible changes in GHC Haskell 08:03
phenny sorear: I'll pass that on when hmmm is around.
08:30 felipe joined 08:31 clintongormley joined, eternaleye left 08:40 dakkar joined 08:42 Ross left 08:44 eternaleye joined 08:49 thebird joined
_sri trading my firstborn for non blocking sockets! 08:52
imarcusthis what if it's a girl? 08:58
_sri hmm 08:59
then i have to raise the offer by a cow or so? 09:00
moritz_ copy on write FTW! 09:01
09:02 lkk- left
cosimo_ moritz_: is proto.perl6.org updated regularly from proto's master tree? 09:09
moritz_ cosimo_: yes
cosimo_ made a change there, hope it's useful
sorear _sri: as it happens, I adopted blizkost specifically to be able to use non blocking sockets from Rakudo, and succeeded.
09:11 proller joined
moritz_ cosimo_: I've now triggered an update there manually 09:11
cosimo_++ # nice update. It bothered me too at one point, but not enoough to fix it :-) 09:12
cosimo_ moritz_: what about ufo's make install ? 09:16
that's next in my list 09:17
is that done already?
moritz_ cosimo_: ufo && make install should work
cosimo_: with a sufficiently recent ufo
cosimo_ i need to update ufo then
yes :)
moritz_ github.com/masak/ufo/commit/293547c...a97dc0111f
github.com/masak/ufo/commit/f5ddb07...331d70593a
09:18 plobsing left
arnsholt I like yapsi's icon ^^ 09:18
09:20 Guest37381 left 09:21 jhuni left
cognominal pmichaud++ # fixing bugs 09:24
cosimo_ moritz_: mmh, do i need 'make install-src'? 09:25
09:26 tylercurtis left
kaare_ Anyone with some parrot-fu who can decipher this parrot/t/library/pg.t into something rakudo? 09:29
09:29 azert0x joined
kaare_ The central section seems to be load_bytecode 'postgres.pir' + load_bytecode 'Pg.pir' and then by Parrot magic there is a cl = new 'Pg' 09:29
I know there is a load_bytecode in rakudo. But does that help me? 09:30
arnsholt You can call parrot opcodes from Perl 6 in rakudo with pir::opcode 09:31
kaare_ Please pretend I know nothing about Parrot ;-) 09:32
arnsholt For example, you could do pir::load_bytecode("postgres.pir")
kaare_ OK, so far so good. How do I create a new Pg object? 09:33
afaics that is what's going on in the parrot test file with cl = new 'Pg' 09:34
pir::new('Pg') ? 09:35
... nope 09:41
09:53 pragma_ left, pragma_ joined, pragma_ is now known as Guest48110 09:54 Guest48110 left, Guest48110 joined
sorear cognominal: I just got what appears to be a reply to a bug report autopost in my inbox 09:54
cognominal: did you mean to send it as an actual bug report addition?
09:55 Guest48110 is now known as pragma_
sorear #76068 09:55
09:57 masak joined
masak oh hai, #perl6! 09:57
sorear hello masak! 09:58
09:59 gbacon joined 10:00 shade\ left
arnsholt kaare_: You might have to specify the signature of the opcode (there can be several variants) 10:01
10:01 cygx joined
arnsholt Try pir::new__ps 10:01
10:02 cygx left
masak wow, I noticed from the commit log that pmichaud++ was active yesterday. now I see it in the RT messages! :) 10:05
10:05 cygx joined
jnthn oh hai #perl6 10:05
jnthn back from his ickle trip
masak what's an ickle? 10:06
welcome back, jnthn!
cygx hi
jnthn masak: slang for "little" :-)
masak ah. figures :)
sounds like something a mother would say to a toddler :)
sorear wonders what to make of the trend where people use their native writing systems for From: on English mailing lists
jnthn Yes, well, I did visit my mum. :-) 10:07
sorear on the one hand, Unicode FTW
jnthn managed to be in six different countries in the last 5 days
sorear on the other hand, it's hard to keep characters straight when they are all identified as <hanzi><hanzi><hanzi>
arnsholt sorear: OTOH, "what are these people called?"
masak sorear: maybe those foreigners will finally learn to read hanzi. :P 10:08
arnsholt But there are so many of them! =) 10:09
masak arnsholt++ # ambiguity 10:10
10:10 mberends joined
arnsholt Oh, heh. Pun not intended, actually ^^ 10:10
sorear Right. Yes. Hanzi. And Cyrillic. And Icelandic. And Greek. And Arabic. And ...
masak lolitsmberends!
moritz_ oh hai 10:11
mberends oh hai from $work in .nl!
jnthn Wow, pmichaud++ had quite a commitfest yesterday!
hi mberends :-) 10:12
masak sorear: I think it's a positive trend. even when you transcribe Chinese names, it requires effort to tell them apart. so the latin alphabet doesn't fully solve the problem either.
jnthn: yes! whatever he ingested, we need to send him more of that. :P
jnthn
.oO( pivo )
More likely a few gallons of Dr Pepper. :-) 10:13
moritz_ masak: sub pmichaud(@tuits --> Code) { $magic } :-)
arnsholt sorear: Icelandic is ok. Just ð and þ
masak :)
make $magic # :)
arnsholt Arabic and Hanzi is probably the hardest for me
masak arnsholt: I'd like to learn en.wikipedia.org/wiki/Devanagari 10:15
sorear arnsholt: icelandic orthography is so different from english orthography that the similarity of the underlying character sets is lost to me
I mean, sure, I can read it
but it has no mnemonic value 10:16
masak this symbol frightens and fascinates me: en.wikipedia.org/wiki/File:JanaSans...ddhrya.svg
sorear (to this native english speaker, &c)
10:16 sftp joined
arnsholt Oh, right. The names are so similar to Norwegian I don't have that problem 10:16
masak talk about a ligature!
arnsholt Yeah, but that's an intentionally perverted example
sorear NFG is fun stuff 10:17
arnsholt Devanagari is pretty straightforward. And the ligatures aren't really used in Hindi
It's only in Sanskrit ligatures are mandatory
masak oh!
arnsholt Of course, Sanskrit is an awesome language. But I would say that, I suppose =) 10:18
masak while we're on the topic of complex scripts... en.wikipedia.org/wiki/File:Bi%C3%A1...cript).svg
I believe that character has been discussed here before.
sorear 2010.06.22.17.37.44 <@darbelo> P<LATIN CAPITAL LETTER O WITH FIVE UMLAUTS AND A HAT>ST
masak at least I recall TimToady dissecting it. 10:19
sorear I wonder if I should submit that to ParrotQuotes
masak sorear: please do!
moritz_ please do :-)
arnsholt Whoa. Not that's complicated =)
sorear well, it was a conversation that involved me
moritz_ so what? 10:20
sorear so I might not be authorized? 10:21
moritz_ you are.
masak it's not Wikipedia :)
10:22 mmcleric joined
bbkr can I achieve in pure P6 something that P5 Moose offers: "has 'x' => ('lazy' => 1, 'default' => \&some_sub_that_sets_x )" ? 10:23
sorear no.
you probably will be, eventually
masak why does all of JimmyZ's bug reports involve augmenting Str and then looping on a range of strings with numbers in them? :)
sorear but Moose is (understandably) a lot more mature on the finer points of classes
masak s/does/do/
bbkr sorear: thanks 10:24
masak I believe rt.perl.org/rt3/Public/Bug/Display.html?id=76202 is invalid. am I right?
sorear in current Perl 6, the standard way to do that is to tie $x to a cusrom container that lazily self-initializes
moritz_ bbkr: it will be a simple matter of has $.x = lazy { foo() };
10:24 liuchong joined
sorear wonders how to make that work with clearers 10:24
and predicates. especially and predicates. 10:25
bbkr moritz_: awesome syntax :)
sorear The circularity saw strikes again. I need to allow non-containerized returns from FETCH. 10:26
liuchong ls
sorear Password:
moritz_ no such file or directory.
sorear Input/output error
I wonder if containers should be a native type. Implementing FETCH as a completely ordinary method is raising many circularity saw issues, and the spec doesn't actually talk about custom container *types* at all 10:27
moritz_ how do I check the presense of a contextual variable in NQP without dying if it doesn't exist? 10:28
arnsholt moritz_: I suspect pir::somethingorother 10:30
sorear trac.parrot.org/parrot/wiki/ParrotQuotes last entry 10:31
moritz_ sorear: can't method calls on the container type use a more primitive, non-overridable container? 10:32
sorear moritz_: I use containers as part of the calling convention. Containers which need to be FETCHed 10:33
also, TimToady wants me to make absolutely everything augmentable
moritz_ he learned his p5 lesson well :-) 10:34
10:34 eternaleye left
kaare_ arnsholt: At least that got me one step further :-) 10:39
10:41 frmadem joined
frmadem Hey how can I help? 10:41
arnsholt kaare_: Good, good
masak frmadem: in many ways! welcome.
10:41 shade_ joined
moritz_ frmadem: the question is more, what do you want to do? 10:42
frmadem: write Perl 6 modules? documentation? blogging?
kaare_ I kinda naively hoped to have a perl6 object, but my $dbh = $pg.connectdb(''); returns an error,
frmadem I'm a Spanish perl developer
masak frmadem: most of the ways involve getting a copy of Rakudo and becoming gamiliar with it.
s/gamiliar/familiar/
kaare_ Null PMC access in invoke()
moritz_ kaare_: try say $pg.connectdb('').WHAT 10:43
kaare_: and tell me what kind of error it produces :-)
frmadem Iwould like to test some already developed modules in Perl6
masak frmadem: here's a list: proto.perl6.org/ 10:44
moritz_ frmadem: that's a good start
frmadem I've already got a rakudo copy from the git repository
10:44 pmurias joined
kaare_ moritz_: The same :-( 10:44
Null PMC access in invoke()
frmadem I've tested some of the features, running well in my Mac 10:45
masak frmadem: some projects on that list work in Rakudo today. others are old and not updated after the last refactor. so you're likely to find things that need fixing.
cognominal rakudo: "{}" # :)
p6eval rakudo 05684c: OUTPUT«Null PMC access in get_string()␤ in main program body at line 11:/tmp/OBctzGOoQZ␤»
masak submits rakudobug
cognominal++
10:45 liuchong left
moritz_ frmadem: of my modules that there some which could use more tests... 10:46
frmadem problem is i'm not familiar with parrot vm
moritz_ there's the json module
you don't need to be, for writing tests
masak frmadem: intimate knowledge with Parrot is not needed.
I know fairly little about Parrot, but I can still read and write Perl 6 code :)
frmadem Right I can try to port my Json module to perl 6
? 10:47
It's okay
?
masak frmadem: there is a JSON::Tiny already...
moritz_ but of course diversity is encouraged
github.com/masak/svg/ needs tests
masak oh, certainly. 10:48
moritz_ github.com/moritz/CGI-Application needs tests
masak frmadem: give it a shot. writing grammars is fun, except when it's frustrating. :P
frmadem All right take a look... mucas gracias
mucas/muchas
10:48 felliott left
masak :) 10:49
masak shudders to think what a "mucas gracias" looks like...
frmadem ps. I found grammars an explendid idea...
hasta luego!!
10:50 frmadem left
jnthn masak: I was trying not to think that. :P 10:50
masak jnthn: sorry :P
jnthn: sometimes a little knowledge is a dangerous thing... :)
moritz_ grammars are just like programming: you need quite some experience with them to be able to write more than 3 lines at once without it all blowing up 10:51
now, how much grammar experience do you have, compared to your normal programming experience?
kaare_ how do I see the class of an object? I don't even know that :-(
kaare_ hugs good ole perl5 ;-) 10:52
moritz_ kaare_: say $thing.WHAT
kaare_: if you have some code to share, I can take a look and try to help you a bit
kaare_ moritz_: oh. So I don't have an object from my $pg = pir::new__ps("Pg"); 10:53
moritz_ kaare_: that might be because you're in the wrong HLL namespace 10:54
kaare_: Perl 6 code runs in the 'perl6' HLL, Pg is in the 'parrot' HLL
masak is rt.perl.org/rt3/Ticket/Display.html?id=76142 a dupe of rt.perl.org/rt3/Ticket/Display.html?id=63786 ?
sorear out 10:56
jnthn masak: ano
masak merges 10:57
moritz_ kaare_: github.com/cosimo/perl6-digest-md5/...est/MD5.pm contains an example of how to call some parrot stuff out of the parrot HLL
kaare_ moritz_: nopaste.snit.ch/21628
jnthn If you're looking at Postgres from Perl 6, you may also be interested in github.com/jnthn/zavolaj/blob/maste...lclient.p6
kaare_ jnthn: I am and I am 10:58
jnthn Ah, OK 10:59
moritz_ kaare_: looking at it... but first I have to compile a new rakudo, my current one is broken :)
kaare_ moritz_: All OK. Just note that it fails in line 5.
jnthn: thanx
dalek meta: r351 | stefa...@cox.net++ | trunk/vicil/Kernel.cs:
[vicil] As a concession to circularity, FETCH and STORE become responder
 11:00
bbkr is there working P6 phaser that does the same functionality as Moose BEGIN{} block, or do I have to overload new() method?
kaare_ back to $$$ work. & 11:01
11:01 xinming_ joined
moritz_ bbkr: what do you want to do with it? (I'm not too familiar with Moose) 11:01
arnsholt Not sure what Moose's BEGIN blocks do, but if you want to run code when an object is constructed, implement method BUILD
moritz_ s/method/submethod/
arnsholt Right you are =) 11:02
I should probably fix that in Algorithm::Viterbi 11:03
11:03 xinming left
masak lunch & 11:04
11:04 masak left
bbkr arnsholt: yes, I want to run some code when object is constructed. but it found something that looks like a bug; 11:12
rakudo: class ConditionBlock { has Int $.id; method BUILD { $.id = 666; }; }; ConditionBlock.new(); # why read-only ?
p6eval rakudo 05684c: OUTPUT«Cannot assign to readonly value␤ in '&infix:<=>' at line 1␤ in 'ConditionBlock::BUILD' at line 11:/tmp/LWTYBgAyhj␤ in main program body at line 11:/tmp/LWTYBgAyhj␤»
jnthn has Int $.id is rw; 11:14
or use $!id = 666;
But make it a submethod 11:15
moritz_ or 11:16
bbkr rakudo: class X { has Int $.id; method BUILD { $.id = 666} } X.new # jnthn - but this one works, while previous one doesn't ?
p6eval rakudo 05684c: ( no output )
moritz_ has Int $.id = 666;
bbkr $.id is rw by default, isn't it? 11:18
moritz_ nope
bbkr rakudo: class X { has Int $.id; method BUILD { $.id = 666; }; }; X.new
p6eval rakudo 05684c: OUTPUT«Cannot assign to readonly value␤ in '&infix:<=>' at line 1␤ in 'X::BUILD' at line 11:/tmp/cp70pg547y␤ in main program body at line 11:/tmp/cp70pg547y␤»
bbkr ?! I just added few semicolons :)
rakudo: class X { has Int $.id; method BUILD { $.id = 666; } }; X.new 11:19
p6eval rakudo 05684c: OUTPUT«Cannot assign to readonly value␤ in '&infix:<=>' at line 1␤ in 'X::BUILD' at line 11:/tmp/Flmay942pR␤ in main program body at line 11:/tmp/Flmay942pR␤»
bbkr rakudo: class X { has Int $.id; method BUILD { $.id = 666 } }; X.new
p6eval rakudo 05684c: OUTPUT«Cannot assign to readonly value␤ in '&infix:<=>' at line 1␤ in 'X::BUILD' at line 11:/tmp/gBZkgzfxzz␤ in main program body at line 11:/tmp/gBZkgzfxzz␤»
bbkr rakudo: class X { has Int $.id; method BUILD { $.id = 666 } } X.new
p6eval rakudo 05684c: ( no output )
bbkr is confused. why adding semicolon after class definition made $.id read-only? 11:20
moritz_ bbkr: I guess without the semicolon it is mis-parsed 11:21
bbkr: and then for some strange reason X.new isn't actually run, and $.id is always read-only
bbkr rakudo: class X { has Int $.id; method BUILD { $.id = 666 } } say X.new.id # checking for misparsing 11:22
p6eval rakudo 05684c: OUTPUT«===SORRY!===␤Confused at line 11, near "class X { "␤»
moritz_ kaare_: I've created trac.parrot.org/parrot/ticket/1692
bbkr moritz_: you're right (again). is this a bug? spec doesn't force semicolons after class blocks. 11:23
moritz_ std: class X { has Int $.id; method BUILD { $.id = 666 } } X.new 11:24
p6eval std 31481: OUTPUT«===SORRY!===␤Method call found where infix expected (omit whitespace?) at /tmp/KVuVqGABwZ line 1:␤------> nt $.id; method BUILD { $.id = 666 } } X⏏.new␤ expecting infix or meta-infix␤Parse failed␤FAILED 00:01 114m␤»
moritz_ bbkr: spec says that after a } either a newline or a ; is required
bbkr: but it's a bug that rakudo doesn't complain about it 11:25
bbkr reports
moritz_ bbkr++
cognominal is there a way to print the rakudo git version and the parrot svn version from rakudo? 11:31
moritz_ cognominal: nope; there's a patch for that in RT, but it's not applied yet
parrot version is accessible though
rakudo: say %*VM<config><revision>
p6eval rakudo 05684c: OUTPUT«47723␤»
11:32 mmcleric left
cognominal I realize that my bug reports art about the "current" version which is not very helpful down the chain. 11:33
moritz_ cognominal: the rakudo git hash is usually enough, if you use the recommended parrot revision from build/PARROT_REVISION
kaare_ moritz_: thanx 11:34
moritz_ kaare_: I've also submitted a fix in parrot r47900 11:35
arnsholt .u U+FEFC 11:36
phenny arnsholt: Sorry, no results
arnsholt .u FEFC
phenny U+FEFC ARABIC LIGATURE LAM WITH ALEF FINAL FORM (ﻼ)
imarcusthis .u BEEF 11:39
phenny U+BEEF HANGUL SYLLABLE BBEGS (뻯)
imarcusthis .u CAFE
phenny U+CAFE HANGUL SYLLABLE JJWAELM (쫾)
imarcusthis boring
moritz_ .u DEAD 11:40
phenny U+DEAD (No name found)
imarcusthis .u FEED
phenny U+FEED ARABIC LETTER WAW ISOLATED FORM (ﻭ)
imarcusthis .u FADE
phenny U+FADE (No name found)
kaare_ moritz_: downloading now 11:41
moritz_ kaare_: mind you, I only tested the reduced PIR form... no idea if the Perl 6 example works now 11:42
11:42 felliott joined
kaare_ moritz_: That's why I download now :-) 11:42
moritz_ :-)
imarcusthis .u ACED 11:43
phenny U+ACED HANGUL SYLLABLE GOLT (곭)
colomon So, I just forked LastOfTheCarelessMen's Vector and ABC to try to port them to current Rakudo. github.com/colomon/Vector has my first wave of changes. ufo++ 11:45
phenny colomon: 27 Jun 06:15Z <ciphertext> tell colomon i've got series passing all (nonfudged) tests, and conforming to almost all of the spec. See nopaste.snit.ch/21602 for details. (especially note the section at the bottom).
colomon But I get this error:
error:imcc:syntax error, unexpected '\n'
in file 'EVAL_5' line 58
===SORRY!===
syntax error ... somewhere
kaare_ moritz_: I'll give you a ping when it's tested
colomon make: *** [lib/Vector.pir] Error 1
anyone have a clue what could be doing that?
11:49 envi^home joined 11:52 kfo left, xinming joined
jnthn colomon: Not without seeing the PIR that's generated 11:53
11:55 xinming_ left
takadonet morning all 11:55
11:56 ruoso joined
moritz_ kaare_: oh, it seems that the load_bytecode("postgres.pir") is interfering 11:57
kaare_: without it I have something working
kaare_ moritz_: ok, i just assumed it was necessary. 11:59
moritz_ kaare_: nopaste.snit.ch/21629
says
Pg;Conn
whih means $con actually holds a connection object
kaare_ OK, I'll play with it later. Thanks 12:01
say $dbh.WHAT; still doesn't work though 12:02
moritz_ yes
that's because it's not a Perl 6 object
but a parrot object
colomon jnthn: --target=PIR ? 12:03
jnthn --target=pir
moritz_ colomon: isn't that a .pir file alredy?
jnthn (think it only likes lowercase)
moritz_ 13:45 <@colomon> make: *** [lib/Vector.pir] Error 1 12:04
Vector.pir looks like pir :-)
colomon fair enough, I don't know what magic ufo performs. ;) 12:07
errr, no it isn't.
12:07 xinming left
colomon there doesn't seem to be any Vector.pir file. 12:08
12:08 ruoso left
colomon It gets that error generating Vector.pir 12:09
moritz_ ah
colomon env PERL6LIB='/Users/colomon/tools/Vector/lib' /Users/colomon/tools/rakudo/perl6 --target=pir --output=lib/Vector.pir lib/Vector.pm 12:10
12:10 xinming joined
colomon is the command which generates the error. 12:10
jnthn colomon: Does Vector.pm have any "use" of an already pre-compiled module? 12:11
colomon MONKEY_TYPING doesn't count, does it?
moritz_ nope 12:12
colomon then no, it's just
use v6;
use MONKEY_TYPING;
jnthn Ah :-( 12:13
My best guess was that it was loading a pre-compiled module that contained bad PIR.
colomon I'll try mucking about with the source and see if that affects anything. 12:14
in the meantime, if you'd like to play with it, the exact same source is up at github.com/colomon/Vector ... ;)
moritz_ what you can always try is to comment out large chunks of code and see when the error goes away 12:15
colomon moritz_: that's exactly what I just did to make the error go away. 12:16
I suspect it might be the use of multi sub circumfix:<⎡ ⎤>
moritz_ I think that's NYI in master
rakudo: multi sub circumfix:<[ ]>($x) { say $x }; [5]
p6eval rakudo 05684c: OUTPUT«5␤»
moritz_ huh 12:17
colomon if that's the problem, worst error message ever....
but that definitely does it. If I comment out just that, the build continues to work.
arnsholt rakudo: multi sub circumfix:<⎡ ⎤>($x) { say $x }; ⎡5⎤ 12:18
p6eval rakudo 05684c: OUTPUT«error:imcc:syntax error, unexpected '\n'␤ in file 'EVAL_1' line 24665226␤===SORRY!===␤syntax error ... somewhere␤»
colomon now to see how badly make test splodes....
arnsholt++
arnsholt Looks like it doesn't want to define new operators
colomon wow, make test really goes splode in a big way.
arnsholt rakudo: multi sub circumfix:<£ ¤>($x) { say $x }; £5¤ 12:19
p6eval rakudo 05684c: OUTPUT«error:imcc:syntax error, unexpected '\n'␤ in file 'EVAL_1' line 24665226␤===SORRY!===␤syntax error ... somewhere␤»
arnsholt Or it could be a Unicode problem (come to think of it)
moritz_ rakudo: multi sub circumfix:<foo bar>($x) { say $x }; foo 5 bar
p6eval rakudo 05684c: OUTPUT«error:imcc:syntax error, unexpected '\n'␤ in file 'EVAL_1' line 24665226␤===SORRY!===␤syntax error ... somewhere␤»
kaare_ moritz_: I did get a real connection. At least I could create a table.
moritz_ kaare_: \o/ 12:20
kaare_: do you have a github ID?
kaare_ dont know of it's usable for anything though :-)
moritz_ well, it's a start.
kaare_ moritz_: Yes, my user id is kaare. Is that enough info? 12:21
moritz_ hugme: add kaare to fakedbi 12:22
hugme hugs kaare. Welcome to fakedbi!
moritz_ kaare_: you can put a FakeDBD::pg into the fakedbi repo :-)
kaare_ hugs hugme
moritz_ github.com/mberends/fakedbi/
12:23 filius joined, ruoso joined 12:24 filius left
colomon huh. seems like user-defined operators in a module don't work at the moment? 12:25
mathw don't work inside the module, or don't export properly? 12:27
12:27 ruoso left
colomon don't export properly 12:28
mathw huh
somehow I'm not surprised
colomon don't even parse properly after export, which maybe is a clue?
moritz_ colomon: not exporting is a known to me, but dunno if there's a ticket for it 12:29
colomon: problem is that 'use' doesn't really happens at compile time right now 12:30
maybe that's not the whole problem
it creates stubs for the lexicals it exports
at that part it could modify the grammar
12:31 ruoso joined
pmichaud good morning, #perl6 12:33
takadonet pmichaud: morning 12:34
mathw lolitspmichaud! 12:35
12:35 kfo joined
jnthn morning, pmichaud 12:36
12:36 pnate joined
jnthn colomon: That's known (and why they're not marked done in ROADMAP) 12:37
12:38 masak joined
pmichaud 11:19 <bbkr> rakudo: class X { has Int $.id; method BUILD { $.id = 666 } } X.new 12:40
is being parsed as &infix:<X>(class ..., $_.new)
12:41 audreyt joined, ChanServ sets mode: +o audreyt
mathw oh my 12:41
12:41 audreyt is now known as audreyt__
pmichaud (hint: don't name your classes X, Z, Q, etc. :-) 12:41
mathw Well I wasn't planning to, but it's a bit surprising 12:42
pmichaud std: class X { has Int $.id; method BUILD { $.id = 666 } } X.new # curious
p6eval std 31481: OUTPUT«===SORRY!===␤Method call found where infix expected (omit whitespace?) at /tmp/1M7OQvOM75 line 1:␤------> nt $.id; method BUILD { $.id = 666 } } X⏏.new # curious␤ expecting infix or meta-infix␤Parse failed␤FAILED 00:01 114m␤»
pmichaud std++ again
mathw interesting 12:43
I'd think that once it saw 'class' it'd know that there's an indentifier or a block coming
colomon jnthn: any idea when we might have that? It makes Vector a complete non-starter.
jnthn colomon: I may get to work on it this week; if not, I can do it at the weekend 12:44
colomon \o/
jnthn is has a hackathon at the weekend o/
12:44 skids joined
colomon I will be off-line for a week starting sometime over the weekend, but if you get it fixed after I'm gone, maybe someone else can pick up the ball on this one. 12:46
pmichaud colomon: could we move the str2num et al methods into cheats? 12:47
for compile-time dec-number constants, I'd like to see us do the calculations at compile-time instead of runtime 12:48
12:49 moritz_ sets mode: +oo pmichaud mathw
colomon I don't think we can unless we move Rat into NQP. 12:49
pmichaud no
colomon well, I guess we could for Num constants.
pmichaud the Rat construction can continue to be runtime
but the arguments to Rat could be compile-time
colomon sure.
moritz_ as long was we can calucate numerator and denominator without using Rats
pmichaud we can
the str2num stuff just uses native parrot registers anyway 12:50
we'd save a *lot* of time on dec-number constants if we didn't have to convert them from strings (every time they're used)
colomon the PIR routines should go into NQP very easily, no?
pmichaud I don't understand the "go into NQP part" 12:51
12:51 audreyt__ left
masak I was sure we had one like rt.perl.org/rt3/Ticket/Display.html?id=76226 , but I can't find it. 12:51
colomon I mean, ....
hmmm.
pmichaud I want to move the str2num-* subs into cheats
give them ! prefixes (so they don't appear in the global namespace)
moritz_ masak: so was I
pmichaud then NQP can call them at compile time
colomon I'm trying to figure out what the non-cheating solution is.
pmichaud colomon: why? it's a cheat no matter what. 12:52
and the logic still has to be available at runtime.
(for conversion of Str to numbers)
masak moritz_: it's a bit hard to search for.
colomon pmichaud: I guess I don't quite understand the ramifications of moving the routines around. but I certainly have no objections to do it, as long as numeric constants continue to actually work. 12:55
12:56 steffan joined 12:57 jaldhar left 12:58 perlygatekeeper joined
pmichaud I suppose I could also put some better str-to-num conversion code into nqp-rx itself, then re-use it from rakudo. 12:59
13:00 jaldhar joined
moritz_ when I do a 'my Int $x' declaration, where is the type constraint stored at compile time? 13:00
pmichaud on the container for $x
oh, wait
at compile time
moritz_ I kinda expected it in the Past::Var
pmichaud I'm not sure it's stored at the moment. It will likely be in the symbol table that holds the reference to $x
Parrot doesn't know about type constraints. 13:01
13:01 masonkramer joined
pmichaud also, there's a PAST::Var for each usage of $x in a program, not a global one 13:01
it really belongs in the block's symbol table.
(the symbol table could hold a link to the PAST::Var defining $x, but that's entirely up to the HLL)
moritz_ so I need to find the block which encloses the PAST I'm looking at
pmichaud in rakudo we have a subroutine for that 13:02
...well, we used to. 13:03
NQP has an example of a way to do it. essentially, you walk through @BLOCK looking for the first one that has an entry in @BLOCK.symbol 13:04
13:04 Zapelius joined
moritz_ is @BLOCK accessible after the action methods ran? 13:04
pmichaud oh, you mean from the already-created PAST tree 13:05
moritz_ ie in a compilation stage after PAST
yes
pmichaud nope
moritz_ :(
pmichaud PAST::Compiler handles it by keeping a stack of nested Blocks as it traverses the tree.
it also keeps a stash of all of the in-scope symbols 13:06
moritz_ so, the current state for compile-time optimizations in rakudo is "don't do", it seems
or analysis, for that matter
what I wanted to do is catching things like my Int $x = 'foo'
and the code for finding &infix:<=>($var, $typed_thing) works 13:07
but since there's no good way to get the type constraints of $var, a static check seems not a LHF at all
pmichaud adding a type constraint to the symbol isn't too difficult 13:08
I'm more curious about how one would detect/handle: my XYZ $x = 'foo'
I guess we'd only check the known builtin types 13:09
moritz_ look up XYZ, if possible
pmichaud (for a first pass)
moritz_ and only if the lookup works, do the check
pmichaud where "do the check" is actually calling '&infix:<=>' ? 13:10
moritz_ no, I thought about .ACCEPTS
we should really have a long talk at YAPC::EU 13:11
:-)
pmichaud definitely. :)
anyway, off the top of my head, here's one thing I'd do 13:12
one choice would be to add symbol table entries for each PAST::Block's outer block 13:13
that would make walking up-the-tree simpler.
this could be done as part of an optimization pass as well
i.e., while walking down the past tree, establish outer links as appropriate. 13:14
13:14 kaare_ left
pmichaud another possibility would be to build a stash in the block's symtable that keeps a list of all of the symbols in scope and references to their PAST::Var decl (if available) 13:15
there's nothing that says that the HLL compiler has to (or even should) build these structures... in some senses it's more generic for the optimizing passes to do it. 13:16
moritz_ depends on how much analysis it wants to do
it just bothers me that we have to rebuild data structures that we already have in the action methods 13:17
13:17 masonkramer left
moritz_ feels like there could be a better long-term solution 13:17
pmichaud I'm not entirely sure about that (more)
the action methods already have enough to keep track of, without also trying to manage a lot of cross-referencing structures 13:18
13:18 mmcleric joined
pmichaud one advantage of doing cross-referencing after building the entire PAST structure is that it's "static" at that point 13:18
moritz_ well, if the PAST tree held more information, the action methods could use that, and keep less state on their own 13:19
pmichaud especially for Rakudo, while in the midst of a copmile the PAST structure is incredibly dynamic as the action methods are taking place. Keeping up with the cross-referencing structures (and patching them whenever a section needs to be moved/refactored) would be very difficult to keep straight.
the PAST tree can already hold more information, yes.
nothing blocks that.
moritz_ I guess I need to read more about PAST make useful changes (or suggestions) 13:20
pmichaud keep in mind that PAST allows a HLL to put its own annotations in place
PAST only puts minimal expectations on what goes in a block's symtable (i.e., it expects the symtable to hold "user-defined" information as well as the basic symbol list)
it also allows each note to have its own set of annotations -- i.e., each PAST::Node is really a hash. 13:21
s/note/node/
for example, it's perfectly okay for a PAST::Block's symtable to maintain a list of all symbols in scope, not just those defined within the block itself. Rakudo doesn't do this yet, but PAST doesn't prohibit it. 13:23
moritz_ nods
pmichaud but I figured that requiring every HLL compiler to supply this information in order to make use of an optimization might place some extra burden on the hll compiler author. In some sense I think the optimization passes ought to be smart enough to deduce some of these things on their own. 13:24
or, more precisely, iwbni the optimization passes could deduce these things on their own whenever the hll compiler author doesn't provide them as hints. (If the hll compiler author provides appropriate hints, then the optimization passes can work better.) 13:27
13:27 slavik joined, plobsing joined
jnthn rakudo: {;}.WHAT.say 13:28
p6eval rakudo 05684c: OUTPUT«Block()␤»
13:30 gbacon left
moritz_ pmichaud: I'll talk to tcurtis about that (the gsoc student working on the parrot optimization framework) 13:31
13:34 redicaps joined, steffan left 13:35 redicaps left 13:36 JimmyZ joined
pmichaud rakudo: " { a => 4 } ".say 13:37
p6eval rakudo 05684c: OUTPUT« a 4 ␤»
bbkr rakudo: (...).WHAT.say # WHAT returning empty string? seems weird, because spec says everything is derived from Mu. bug?
pmichaud rakudo: " { } ".say
p6eval rakudo 05684c: ( no output )
rakudo 05684c: OUTPUT«Null PMC access in get_string()␤ in main program body at line 11:/tmp/usmP5SQR0Q␤»
pmichaud interesting.
rakudo: say (...) ~~ Failure
p6eval rakudo 05684c: ( no output )
pmichaud bbkr: ... is a failure 13:38
bbkr: invoking it causes the program to stop at that point
(well, it causes the routine to return a failure)
so the .WHAT is never being executed
moritz_ rakudo: say sub {...}.() ~~ Failure
p6eval rakudo 05684c: OUTPUT«===SORRY!===␤Confused at line 11, near "say sub {."␤» 13:39
13:39 mberends left
moritz_ rakudo: say (sub {...}).() ~~ Failure 13:39
p6eval rakudo 05684c: OUTPUT«1␤»
pmichaud rakudo: say (sub { ... }).().WHAT
p6eval rakudo 05684c: OUTPUT«Failure()␤»
bbkr pmichaud: not a bug then, thanks.
hmm, but Failure returned from main block should IMO set $? bash variable to 1, am I right? otherwise there is no way of checking if code failed in main block. 13:42
pmichaud it should at least not fail silently 13:44
bbkr pmichaud: right. should i put it in RT queue for discussion? 13:45
pmichaud bbkr++
(+1)
13:49 RAT joined
masak rakudo: my $a = [1, 2, 3]; foo($a); bar($a); sub foo(@a) { say @a.perl }; sub bar(@a is copy) { say @a.perl } 13:50
p6eval rakudo 05684c: OUTPUT«[1, 2, 3]␤[[1, 2, 3]]␤»
masak that one's new, isn't it?
masak submits rakudobug
RAT Did my suggestion of the converse of slurp=burp get into the standards yet? ;-) 13:51
jnthn I thought splat was the latest suggestion. ;-) 13:52
pmichaud masak: that one is interesting. I'm not sure it's a bug.
masak RAT: what should such a function do if it resks overwriting an existing file?
mathw pmichaud: it is quite surprising...
masak pmichaud: oh, come on! :/
mathw although I suspect flattening would hide it in many cases 13:53
masak pmichaud: going against my expectations is *always* a bug! :P
moritz_ rakudo: sub bar(@a is copy) { say @a.perl }; say bar [1, 2, 3] 13:54
masak pmichaud: er. put differently, where does the spec say "putting 'is copy' on an array gives you an extra layer of array"?
pmichaud my $a = [1,2,3]; my @b = $a; say @b.perl;
p6eval rakudo 05684c: OUTPUT«[[1, 2, 3]]␤1␤»
moritz_ pmichaud: that's assignment
pmichaud it's the same logic by which
moritz_ pmichaud: signature binding is *binding*
pmichaud moritz_: except for is_copy
it's the same logic by which
my $b = [1,2,3]; sub xyz($x is copy) { $x[1] = 'b'; }; xyz($b); say $b.perl; 13:55
rakudo: my $b = [1,2,3]; sub xyz($x is copy) { $x[1] = 'b'; }; xyz($b); say $b.perl;
p6eval rakudo 05684c: OUTPUT«[1, "b", 3]␤»
masak rakudo: my $a = 5; sub bar(@a is copy) { say @a.perl }; bar($a)
p6eval rakudo 05684c: OUTPUT«Nominal type check failed for parameter '@a'; expected Positional but got Int instead␤ in 'bar' at line 11:/tmp/zpKCBPqCXv␤ in main program body at line 11:/tmp/zpKCBPqCXv␤»
pmichaud anyway, I can fix it.
I agree that the other interpretation likely makes more sense.
masak pmichaud: if the above doesn't bind, why does my first line give double arrays?
seems very arbitrary. 13:56
pmichaud I just said I wasn't certain of it. :-)
masak just trying to make you even less certain :P
pmichaud the "is copy" logic is still not completely nailed down, based on discussion with TimToady++ at yapc::na
masak I do see the parallel with attr arrays and :attr[]. by the way. 13:57
colomon was wondering about laziness and "is copy".... 13:58
pmichaud anyway, the one thing I *am* certain of is that "is copy" is not "binding". Indeed, the whole point of "is copy" is to avoid the bind. :-) 13:59
mathw yes that's what I thought :)
moritz_ I thought it copied, and then binds 14:00
pmichaud okay, one can think of it that way. :-)
but it's not binding to the argument
in the same way, my @b = $a; is a bind only in the sense that we bind the symbol '@b' to an array before doing the assign 14:01
anyway, I know the fix. Should be one-line :) 14:02
did the ticket get filed yet?
14:02 alester joined
masak almost. :) 14:03
stop talking about it, and I'll file it. :P
pmichaud I'm compiling a fix now. :) 14:04
1-line fix. Actually, 0-line fix, just need to move a statement to a new location. :)
14:04 skids left
mathw I think git might count that as two lines 14:05
one delete, one insert :)
masak :)
mathw obviously this is enormously important for pmichaud to know, lest he misunderstand how big his patch is
masak pmichaud: should become #76242.
14:05 mmcleric_ joined 14:06 mmcleric_ left
masak yup :) 14:06
14:06 PacoLinux joined, mathw sets mode: +o masak, mmcleric_ joined, mmcleric left, mmcleric_ is now known as mmcleric 14:09 skids joined, cygx left, Guest23195 joined
pmichaud > sub xyz(@a is copy) { say @a.perl; }; xyz([1,2,3]); 14:10
[1, 2, 3]
mathw \o/ 14:11
pmichaud > my $b = [1,2,3]; sub xyz(@a is copy) { @a[1] = 'x'; }; xyz($b); say $b.perl;
[1, 2, 3]
spectesting now. 14:12
mathw pmichaud++
masak pmichaud++ 14:13
pmichaud is Bool still an enum? (I'm looking at RT #76238) 14:16
we keep trying to make it into one, and failing.
masak by spec, it is.
I've always thought that it's hard because Bool is so central, and keeps getting in the way of the circularity saw.
jnthn There's that.
masak enums are quite high-level in the implementaion. 14:17
jnthn There's also been in the past other weird requirements on Bool that make it not just an enum iirc
masak Bools are quite low-level.
jnthn: like what?
jnthn Like .succ and .pred on it
Which I don't think would be on enum elements.
masak why not?
that's be kinda nice.
colomon they need to be there, actually. 14:18
for range
and probably series.
jnthn Further violates the "enum values behave like their underlying value" thingy.
Unless we decide we want that in this case.
colomon Range is full of examples like Wed..Fri
jnthn Oh
Then we probably do want to define them somehow 14:19
But ugh
How the heck to implement that nicely...
colomon Are you worried about the middle cases, or just the ends? :)
jnthn We'd make the enum values by taking the value and mixing some role into them, I guess 14:20
How does it then find its next value?
Maybe go and look through the .mappings...
colomon I was under the vague impression they mapped to Ints, but I admit I haven't paid full attention to the enum spec updates. 14:21
pmichaud can we go enum => Int => .succ => enum ?
i.e., convert an enum to its int represetation, take the .succ of that, then go back to the enum? 14:22
mathw eeew
jnthn pmichaud: no
pmichaud why not?
jnthn consider the pair case.
pmichaud Day(3) doesn't work?
jnthn enum foo ( :a(1), :b(42) ); 14:23
1.succ isn't 42 :-)
pmichaud in that case, I'm not sure that a has a .succ
mathw that would lead to two kinds of enum
and you'd never know what you had
jnthn If the succ was always "the next thing in the enum" then it can work
mathw yes 14:24
jnthn But that means you can't go round-tripping to the base-type.
[particle] does p6 have sparse arrays natively?
jnthn S09
14:24 IllvilJa left
moritz_ decommutes for a few hours 14:24
mathw Personally I'm all for anything that divorces enums from the base type as much as possible, provided the conversion's still possible if you really need it. I've seen far too much code which abused enums as ints to be comfortable with it.
[particle] if so, there must be some way to get to the next non-empty slot in the sparse array
mathw must there? 14:25
masak senses another slight tweat to the enum spec coming up...
bbkr rakudo: my @t=1; sub foo(@u is copy) { }; foo(@t[0..*]); # hmm
[particle] in order to efficiently traverse them, yes
p6eval rakudo 05684c: ( no output )
jnthn masak: Heh. You start to understand why I avoid working on enums like the plague yet? :-) 14:26
pmichaud anyway, back to my original question, is Bool ever going to make it to be an enum? ;-)
[particle] if my array has items at indexs 1, 1_000_000 and 1_000_000_000_000, i sure don't want to iterate naively
mathw I think a clearer idea of what enums are is required first...
Bool might have to become a... something else
masak jnthn: nobody said you have to work on it like the plague... :P
bbkr rakudo: my @t=1; sub foo(@u is copy) { }; foo(@t[0..*]); say "works";
14:27 Zapelius left
p6eval rakudo 05684c: ( no output ) 14:27
mathw jnthn: please don't, I like having enums and wouldn't want them all to die of a horrible fever
jnthn pmichaud: I've only been asking that for a couple of years. ;-)
pmichaud jnthn: me also :)
jnthn pmichaud: And enums are a moving target.
mathw that's the problem
if enums were pinned down, then somebody could say 'yes, Bool is an enum'
or 'no, Bool is not an enum'
jnthn The enum spec is like the IO spec imnsho.
mathw nah, the IO spec's worse :) 14:28
jnthn Full of speculative "ooh wouldn't this be nice"
mathw and much, much more minefieldy IMO
pmichaud well, afaict, the reason that enums are a moving target is because we keep asking "so, how is it that Bool is an enum again?" 1/2 :-)
jnthn Rather than something straightforward and implementable.
And then we can add things as people actually need them.
pmichaud i.e., every enum spec change has been in response to trying to figure out how Bool fits with it :)
mathw That might be a hint
Maybe it means Bool isn't an enum
masak I mostly like the current enum spec very much. it's an improvement on the previous one.
pmichaud I like it as well. 14:29
So the iterative refinement is very good.
But we're still left wondering about Bool :-)
masak neither of pmichaud and jnthn seem to think that Bool is an enum.
maybe that's a hint too.
jnthn +1 to "it' snot an enum"
pmichaud I can think of Bool as an enum, I just keep getting nicked by the circularity saw
jnthn pmichaud: Exactly.
er, it's not 14:30
EPLAGUETHEMEWON'TGOAWAY
pmichaud actually, perhaps the answer is that while Bool is an enum, "bool" is not.
jnthn std: my bool $does-it-exist; 14:31
pmichaud and we should be returning low-level bools
p6eval std 31481: OUTPUT«ok 00:01 111m␤»
jnthn pmichaud: That's also possible.
pmichaud: In that case, how would bool and bit differ?
std: my bit $does-it-exist;
p6eval std 31481: OUTPUT«ok 00:01 108m␤»
mathw jnthn: the difference is... er... symbolic constants?
'bool' makes people think 'true or false'
pmichaud well, 'true' isn't a constant :) 14:32
mathw the difference is mind control
pmichaud (in P6)
oh, at least it wasn't.
mathw pmichaud: pfft you could make it one if you wanted to
pmichaud I think we s/true/so/
it used to be a low-level prefix
mathw oh yes, I remember that, dimly
umm, I've lost the current enum spec 14:33
masak the main thing I like about the current enum spec is that you can get at the underlying EnumMap.
I find it slightly... odd... that EnumMap ends up being the parent class of Hash in Rakudo. 14:34
but oh well.
14:35 plobsing left
mathw ah, found it 14:35
masak maybe it's just named unfortunately.
*ill-named 14:36
frettled mathw: we had a horrible punning session, involving also «make it so», of course.
jnthn masak: Well, Mapping was the parent class of Hash in master
masak jnthn: that's OK, methinks.
jnthn: but a Hash has nothing to do with enums.
jnthn masak: But Mapping is gone, long live EnumMap
masak: Tell that to anonymous enums.
:P
frettled EA = Enumns Anonymous 14:37
masak I can see a Hash generalizing to non-Str keys. can't see the same with EnumMap.
mathw well that's why Hash derives from, it adds stuff
like mutability
although it is kind of weird, I just accept it 14:38
frettled Nah, it's normal OO, ain't it?
mathw yeah, that's why I just accept it :)
frettled heh
arnsholt masak: Hash should definitely support non-Str keys. But that'd entail adding some methods to Any or some such I think
mathw I'm convinced that there's something wrong with OO pretty much everywhere, I just haven't figured it out yet 14:39
It might be because virtually everything isn't Smalltalk
masak indeed.
bbkr rakudo: perl6 -e 'class Maybe is Bool::True is Bool::False {}; say Bool::True; say Maybe'; # stupid example :P but Maybe calles in ~ context should return the same output as Bool::True 14:43
p6eval rakudo 05684c: OUTPUT«===SORRY!===␤Confused at line 11, near "perl6 -e '"␤»
bbkr rakudo: class Maybe is Bool::True is Bool::False {}; say Bool::True; say Maybe; # stupid example :P but Maybe calles in ~ context should return the same output as Bool::True
p6eval rakudo 05684c: OUTPUT«1␤Maybe()␤»
frettled mathw: Smalltalk is just so unamerican, Bigtalk is more like it! ;)
pmichaud decides he needs to re-read the updated enum spec 14:44
masak maybe the "why can't Bool be an enum is really two questions": (1) "what prevents us, with the current Rakudo, from creating an enum type MyBool which does all we want Bool to do?", and (2) "once we have such a type, what are the obstacles to putting it where Bool is now?" 14:45
all the circularity issues end up in the second part. 14:46
pmichaud masak: I was just trying that (1)
masak oh, nice.
broquaint docs/vendadocs/VendaProduction/Wire...ic_Content 14:47
mathw I'm wondering where the circularity issues come from
does enum itself require Bool in order to work?
broquaint very ww :S
pmichaud mathw: no, but lots of builtins and internals want True and False to work. 14:48
mathw and enum requires some of those
hmm
True and False are pretty fundamental really
14:49 skids left
mathw which leads back to 'bool' 14:49
pmichaud rakudo: enum MyBool { MyFalse MyTrue }; say MyTrue.succ;
p6eval rakudo 05684c: OUTPUT«Could not find sub &MyBool␤ in main program body at line 11:/tmp/k0WSSSWH7a␤»
pmichaud rakudo: enum MyBool < MyFalse MyTrue >; say MyTrue.succ; 14:50
p6eval rakudo 05684c: OUTPUT«2␤»
pmichaud rakudo: enum MyBool < MyFalse MyTrue >; say MyTrue.name;
p6eval rakudo 05684c: OUTPUT«Method 'name' not found for invocant of class 'Integer'␤ in main program body at line 11:/tmp/ebKr3V6byi␤»
pmichaud rakudo: enum MyBool < MyFalse MyTrue >; say MyBool; 14:51
p6eval rakudo 05684c: OUTPUT«MyFalse 0MyTrue 1␤»
pmichaud rakudo: enum MyBool < MyFalse MyTrue >; say MyTrue ~~ MyBool;
p6eval rakudo 05684c: OUTPUT«0␤»
14:51 skids joined
mathw this looks unencouraging 14:51
dalek kudo: c18d372 | pmichaud++ | src/binder/bind.c:
Fix @a is copy (and %h is copy) when passed a scalar array or hash. Fixes RT
mathw & 14:52
jnthn pmichaud: Rakudo's current idea of enums is very much like what S12 has at the start of the enums seciton
e.g. just sugar for a bunch of constants
pmichaud jnthn: that's fine with me, I'm exploring the spec. :-)
14:52 timbunce joined
pmichaud rakudo: enum MyBool < MyFalse MyTrue >; say 0 ~~ MyBool; 14:53
p6eval rakudo 05684c: OUTPUT«0␤»
pmichaud hmmm.
masak I suspect that one is filed.
masak checks
pmichaud rakudo: enum MyBool < MyFalse MyTrue >; say MyBool.WHAT;
p6eval rakudo 05684c: OUTPUT«EnumMap()␤» 14:54
masak pmichaud: guess it's a manifestation of rt.perl.org/rt3/Ticket/Display.html?id=75296
it's not even a bug really, more like a TODO.
jnthn++ for making it work at all in the meantime.
JimmyZ hello, what's wrong with my bug report? 14:55
phenny JimmyZ: 22 Jun 11:23Z <masak> tell JimmyZ if moritz_ is right, and I think he is, a BOM must've snuck into your src/core/metaops.pm -- maybe you've edited that file?
masak JimmyZ: hi. what do you mean "what's wrong..."? 14:56
JimmyZ masak: I did a clean clone, and it builds now
... I don't know...
呵呵 14:57
masak heh. someone on Twitter thinks Rakudo Star released already: twitter.com/exussum/status/17256454939
14:57 steffan joined
jnthn omg tweets from the future! :-) 14:58
masak I told him this: twitter.com/carlmasak/status/17256876238
14:58 patspam joined
pmichaud it looks to me as though enum Day <Mon Tue Wed>; should create Day as a subclass of Int 14:59
or subtype
(could be a role)
ash_ a more styled web shell for try.rakudo.org, with an index of the tutorial: greaterthaninfinity.com/frontend/ma...shell.html any comments?
jnthn pmichaud: I was expecting mixing in a role
pmichaud jnthn: okay, so it subclasses Int and mixes in a role? 15:00
jnthn pmichaud: mixing in to an int automatically derives a subclass
*Int
So no need to do that first
colomon ash_: I like it, and my 22 month old does too (he's pointing in admiration at Camelia)
masak colomon: lol! 15:01
ash_ if you type "chapter index" plus enter it shows the chapters now
jnthn That is, 42 does role { method enum { return $enummap } } or some such creates a subclass that composes the role, re-blesses the 42 object into it
pmichaud: See operators.pm, infix:<does>
masak ash_: I would strongly advise against using hovering to show/hide things. that's been tried far too many times, and has led to madness in all documented cases. 15:02
jnthn pmichaud: You may find anonymous roles are handy here, anyway
pmichaud jnthn: that feels odd. I have to think on it a bit.
jnthn pmichaud: Which bit? :-)
ash_ masak: alright, i'll try to find a better way of doing that
masak ash_: to be precise, showing/hiding things which *displace* other text is Wrong.
ash_ ya, i know what you mean
it makes it feel jump
jumpy*
masak also, it can lead to infinite update loops. 15:03
pmichaud anyway, I'm going to leave Bool alone for now; will write a response to the recent ticket.
masak sort of like the CSS version of time paradoxes.
pmichaud++
ash_: apart from that, I'm positively surprised (this is the first time I look) at how nice the design is, and how well it fits with the rest of perl6.org's "look". 15:04
ash_++
ash_ thats where i stole the styles for part of it :P (hope no one minds)
pmichaud on a related question.... what do we do with %hash.pick ?
masak ash_: au contraire, I suspect.
pmichaud: pick a pair? 15:05
jnthn ash_: It looks nice
ash_++
ash_ i modified the colors some, but yeah, its the same basic stuff
masak I guess I *could* argue for "pick a key" as well, but pair feels righter.
pmichaud so, Bool.pick returns a bool, while Bool.enums.pick would return a Pair.
masak pmichaud: that makes sense. 15:06
pmichaud (assuming Bool is an enum)
masak :)
pmichaud that's a pretty convincing case that an enum type is not an EnumMap :)
(it may contain an EnumMap, but it isn't one)
jnthn pmichaud: Agree. 15:07
pmichaud rakudo: enum Day <Mon Tue Wed>; say Day.pick;
p6eval rakudo 05684c: OUTPUT«Wed 2␤»
pmichaud rakudo: enum Day <Mon Tue Wed>; say Day.pick;
p6eval rakudo 05684c: OUTPUT«Tue 1␤»
JimmyZ rakudo: enum Day <Mon Tue Wed>; say Day.pick.WHAT; 15:08
p6eval rakudo 05684c: OUTPUT«Pair()␤»
JimmyZ looks like array of pair?
masak bug? 15:09
oh wait.
it's the same one again :P
15:12 songmaster joined 15:13 tadzik joined
pmichaud ticket #76238 updated 15:13
masak \o/ 15:14
I'm still in the camp favoring Bool as an enum type.
pmichaud I'm idly wondering if int ~~ Int 15:15
(and, by analogy, if bool ~~ Bool)
masak pmichaud: based on what sorear said TimToady said the other day, yes.
jnthn Given comments by TimToady recently, I suspect so.
pmichaud I'm pretty sure it would have to be that way.
but I guess that also implies that int ~~ Cool and int ~~ Any 15:16
masak sure.
jnthn that's fine
They just have to be boxed before you can call any methods on them.
So by the time they're seen as an invocant in a Cool or Any method, they're an Int.
pmichaud so I'm wondering if we can make 'bool' work.
or even 'int', for that matter.
then our builtins could start returning bool 15:17
masak how does one create a bool?
pmichaud iiuc, it's just a bit
masak `my bool $b = 1`?
pmichaud internally, we'd probably use an int in parrot
i'm more curious about what happens with &prefix:<?> then 15:18
can it return a bool?
ash_ if anyone else wants a copy of the code i am using to do basic highlighting in the web shell, let me know, i can make it a separate entity
pmichaud (same for the other boolean-returning methods) 15:19
in fact, I wonder what would happen if we simply converted our existing "Bool" type to instead be "bool", and then (for now) derived Bool from it. 15:20
15:20 jaldhar left
pmichaud do typed enums work in Rakudo yet? i.e., my Str enum .... ? 15:20
actually, deriving Bool from bool can't work, because bool ~~ Bool 15:21
(at least, not in the inheritance notion of "derive") 15:22
jnthn pmichaud: The trouble is, I think we're not all the way there with these definitions yet. 15:24
pmichaud jnthn: right, I'm trying to explore it a bit.
jnthn Take Num, for example
(Because it's easiest one to discuss in isolation.)
pmichaud in particular, I think I can find the contradiction that results from trying to make a Bool an enum.)
or at least find *a* contradiction that has to be resolved. 15:25
jnthn We'd say num ~~ Num
masak ash_: I typed 'my @a = <a b c>', and got 'my @a = ' in the history. perhaps needs an escaping step?
ash_ yup, haven't hit that issue yet, thanks
jnthn In that num is a constrained version of Num (constrained not by range of representable values in this case, but by representation)
BUT
I would also expected to have
class Num { has num $.num; ... } 15:26
That is, it has a low-level num inside it.
pmichaud oh, I don't quite expect that second part.
I think that builtin types are allowed to cheat quite liberally on internal representation.
I think they have to.
15:26 redicaps joined
jnthn *sigh* 15:26
No, that's just nasty.
masak ash_: "123.456" is color-coded, but not "123e42"
jnthn OK, on what num means under the hood, yes, some cheating is fine. 15:27
masak ash_: the history doesn't scroll to the bottom when there's more than a screenful.
ash_ hmm, i don't do the e syntax yet, i'll add that too
masak ash_: STD.pm6 is a good hint :)
ash_ which browser?
jnthn But we should know how, in Perl 6, to write class Num { ... }
And define that it has some underlying storage
masak ash_: Google Chrome on a Mac.
jnthn Because, after all, we need to be able to define compact structs made up of natively typed attributes too.
ash_ its using javascript to scroll to the bottom... so that might be tricky
hmm, i am using chrome on a mac too, i'll look into it 15:28
pmichaud jnthn: sure, but compact structs aren't invoking the circularity saw
masak ash_: it seems to scroll a part of the way down.
pmichaud defining Num in terms of num is.
jnthn pmichaud: Only because we define num in terms of Num
pmichaud: That's the bit that makes me uncomfortable.
pmichaud it's a similar problem with int and Int 15:29
masak ash_: I'm on the fence whether errors should have the little '→' prefix just like the results do.
jnthn It feels like the primitive types should be, well, the primitives.
pmichaud well, there's certainly little to prevent Num.ACCEPTS from special-casing num
anyway, that wasn't where I was headed. 15:30
for any my X enum Y ....
jnthn Yes, I wanted to avoid int vs Int at this point because Int has the added complexity of needing to handle big-ints too, iirc.
pmichaud we have that elements of Y ~~ X
for example: my Int enum Day <Mon Tue>; Mon ~~ Int # true
TiMBuS kraih So #rakudo doesn't even support sockets yet and they are already porting cgi frameworks? I smell a marketing desaster. 15:31
jnthn wtf
TiMBuS what is up with this FUDdery
tadzik no sockets?
jnthn Rakudo supports sockets.
tadzik trololololo…
jnthn Nice spelling desaster too ;-)
TiMBuS trolling on twitter? why i never
pmichaud I think the original author might be confusing "sockets" with "non-blocking sockets" 15:32
jnthn pmichaud: I had once envisioned it along the lines of...
TiMBuS thats what im thinking
15:32 pmurias left
pmichaud so, if we do my bool enum Bool <False True> 15:32
jnthn Oh, hm 15:33
pmichaud ah, it works differently for native types
okay, so that's plausible
I wish I knew what to call the "false" and "true" values of "bool"
(i.e., as constants) 15:34
are they just 0 and 1? since native values don't know their own type?
jnthn For non-native ones, I expected the implementation may be that the enum type name itself is a role that we generate. We then mix it in to each of the values, but with a different "current value" set
pmichaud: Yes, I believe so
pmichaud: I think the enum spec has some language on that 15:35
pmichaud jnthn: that's what I'm reading now :)
masak pmichaud: that's what I meant earlier with "how do you create a bool"?
pmichaud masak: yes
ingy I find it odd that this fails: nopaste.snit.ch/21637
jnthn oh eww
3 ~~ Day # True, using Day as a subset of Int
masak ingy: you must declare before you use. 15:36
ingy: it's the one-parse thing, you know.
ingy It masak that's what I find odd
pmichaud also, can't have two semicolon declarations in a single file.
masak ingy: Perl 6 is strictly one-pass.
ingy: I wrote a long mail on p6l about it some months ago, and TimToady wrote a long email back :)
pmichaud i.e., it would need to be class Document::Section { ... } and not class Document::Section;
jnthn ingy: If you want to go "out of order" in terms of definitions, it's fine, but you need to declare a stub earlier. 15:37
15:37 macdaddy joined
masak ingy: here: www.nntp.perl.org/group/perl.perl6....33141.html 15:38
15:38 macdaddy is now known as Guest3330
ingy so when does the 'has' assignment actually happen? 15:38
pmichaud okay, well, thinking about enums is very interesting, but it's not on the critical path for R* so I think I'll suspend it for now and work on something else :)
ingy: when an instance is created, iirc.
ingy that's good
at least what I wanted
jnthn ingy: What pm said - the thingy on the RHS is actually "thunked"
Gets promoted to an anonymous method. 15:39
So you can do like has $.a; has $.b; has $.c = $.a + $.b;
15:39 sftp left 15:40 steffan left
ingy thanks :) 15:40
masak rakudo: class A { ... }; class B { has A $.a }; class A { has B $.b }; say "alive"
p6eval rakudo 05684c: OUTPUT«alive␤»
jnthn Nice try. :P 15:41
masak wasn't actually trying to break things :P
ash_ masak: i can change the error prefix to be something more noticeable so you can tell the difference easier, maybe a ! or ‽
masak just showing how to make two classes use each other.
ash_: is there a skull in Unicode? :)
.u skull 15:42
phenny U+2620 SKULL AND CROSSBONES (☠)
masak ash_: use that :)
ash_ ☠
lol
k
masak ash++
ash_ html entity &#9760;
masak er, ash_++
rakudo: say :16<2620>
p6eval rakudo 05684c: OUTPUT«9760␤»
masak yup.
_sri TiMBuS: kraih is me, and i meant to write non blocking sockets but ran out of chars :) 15:43
btw. the offer still stands, my firstborn for non blocking sockets!
pmichaud 15:41 <masak> wasn't actually trying to break things :P 15:44
....somehow I don't quite believe that. :-)
I think masak is always on the lookout for possible breakages :)
masak pmichaud: I swear, most of the time when I do break stuff, it's not deliberate! :P
pmichaud masak: I *do* believe that. :)
ash_ when is the 12e123 valid? (i guess i could find it in the spec... S02?)
masak ash_: anywhere a term is valid, I guess. 15:45
ash_ but can you do 0b01210e123
is more my point
masak ash_: you do realize that you're about to write a Perl 6 parser here?
ash_ shhh only select bits
masak ash_: how will you decide which bits to fail on?
ash_ std: 0b01210e123 15:46
p6eval std 31481: OUTPUT«===SORRY!===␤Whitespace is required between alphanumeric tokens at /tmp/tdYazMUpXi line 1:␤------> 0b01⏏210e123␤Two terms in a row at /tmp/tdYazMUpXi line 1:␤------> 0b01⏏210e123␤ expecting any of:␤ POST␤ bracketed
..infix␤ i…
ash_ dumbly
masak ash_: I admire your optimism.
ash_ 123e123
std: 123e123
p6eval std 31481: OUTPUT«ok 00:01 106m␤»
masak ash_: here's hoping that what you end up with won't be too messy.
ash_ github.com/moritz/try.rakudo.org/bl...l.html#L14 lines 14~35 are all of my highlighting to date 15:47
its pretty dumb so far, but i am not trying to hard to make it perfect, just good enough that the web shell looks alright
pmichaud while I definitely do not want to stifle ambition here, I do hope that the effort to syntax highlight doesn't distract too much from the overall goal of having an interactive web-based perl 6 shell. 15:48
ash_ cygx is doing the backend of it, so i am waiting on him to evaluate the content, i am just working on the front end since its something i am capable of 15:49
15:49 ruoso left, rv2733 joined, Su-Shee left
pmichaud ash_++ sounds like you have it well-handled then :) 15:49
ash_ right now, aside from the highlighting corrections masak++ found, i am adding a tutotrial thing, mostly I am planning on expecting a certain result, so if the example says "enter 1 + 2", i'll look for 3 in the stdout 15:50
15:51 Ross joined
ash_ i am trying to come up with a good way of designing those steps, plus i have the index of the chapters now so I can figure out what goes where when I need to build examples, if you type "chapter index" in the terminal it will show you the chapters 15:51
i'll probably also start with a brief tutorial on how to use the terminal before starting perl6, (a chapter 0 really) 15:52
pmichaud there should always be a chapter 0. :)
bbkr rakudo: say "foo" ~~ /<[d..b]>? foo/ 15:53
p6eval rakudo 05684c: OUTPUT«foo␤»
ash_ should i say "perldoc" instead of "help" for looking up definitions? (since that might be more like what you would do in the terminal)
TiMBuS _sri, i asked a few places (perl 5 hackers) what they thought about the perl 6 logo and um. 15:54
they were very honest.
they prefer yours to say the least
15:55 JimmyZ left
_sri TiMBuS: i'll release it in a different form than originally planned, but under cc-share-alike to allow remixing and stuff 15:56
15:56 IllvilJa joined
TiMBuS good idea 15:56
15:57 sftp joined
masak wow, subs defaulting to 'my' feels more and more right every day! 15:57
15:57 plainhao joined
mathw masak: yes it is nice isn't it 15:57
masak mathw: it's like having Scheme, but with a sane syntax. 15:58
ash_ _sri, TiMBuS: its not done yet by any means, but have you guys seen: greaterthaninfinity.com/frontend/ma...shell.html ? its the current design I am working with for an interactive web terminal for rakudo star
tadzik subs defaulting to 'my'?
TiMBuS heey thats cool 15:59
_sri TiMBuS: btw. i'm honestly worried about all the cgi stuff going on, it could totally backfire marketing wise
pmichaud _sri: what would you suggest instead?
mathw masak: an altogether completely more awesome syntax
TiMBuS he wants nonblocking sockets. id settle for threads :3 16:00
ingy is there a way to get .perl to dump objects deeply?
mathw if you have threads, you can write a non-blocking I/O layer
pmichaud (I mean to avoid the "marketing disaster") 16:01
_sri pmichaud: non blocking sockets and a sweet http 1.1 api (which i'll be able to provide)
16:02 ashleydev joined
pmichaud _sri: and you're saying that without that, we'll have a disaster? 16:02
ingy when I say Document.new.perl I get >>Document.new()<<
_sri pmichaud: thats not what i said
pmichaud _sri: okay, I don't mean to misquote or misinterpret. 16:03
ingy even if the object has been populated
_sri pmichaud: i said by nurturing a culture of cgi hacking you are likely to have a disaster
arnsholt ingy: Not at the moment, AFAIK 16:04
pmichaud ingy: we don't have a generic .perl for objects yet, no.
ingy ok
pmichaud I suppose we could see about writing one that introspects the attributes and dumps them. 16:05
ingy :)
one more newbie q
how do I get a unique id on an object
is there a .id?
pmichaud _sri: I don't see the disaster, but accept that there could be one. 16:06
ingy: perhaps .WHICH ?
rakudo: my $a = Int.new; my $b = Int.new; say $a.WHICH; say $b.WHICH;
jnthn _sri: I'm not sure I follow. Yes, there are better ways to do things than CGI, but it's hardly like nobody ever writes/deploys CGIs any more.
p6eval rakudo 05684c: OUTPUT«0␤0␤» 16:07
16:07 songmaster left
pmichaud oh, bad example. 16:07
[particle] ha, really bad example.
pmichaud rakudo: my $a = Array.new; my $b = Array.new; say $a.WHICH; say $b.WHICH;
p6eval rakudo 05684c: OUTPUT«80887408␤80885392␤»
jnthn _sri: Plus there *is* a basically working HTTP server.
I think mberends++ is hacking on stuff in that area.
So CGI is not the only option Rakudo offers either.
_sri: But granted I may have missed your point entirely. 16:08
pmichaud besides, the way to get non-blocking sockets implemented will be for us to have enough people interested in Rakudo to hit upon someone that decides they want it badly enough to investigate what's needed to implement them. :) 16:09
ingy pmichaud: nice. it's interesting that WHICH just gives a stringification for strings...
is that on purpose?
pmichaud .WHICH on value types basically returns the values
see S12? S13? for a more thorough description of what .WHICH actually does.
ingy fair enough 16:10
thanks!
_sri /me gives up 16:11
16:12 justatheory joined, ash_ left
_sri i accept that marketing has no place in this channel 16:13
masak it might.
pmichaud _sri: I'm trying to understand, really. 16:14
16:15 ash_ joined
masak _sri: my reaction to that tweet was "huh, someone who thinks we're more organized and less volunteer-based than we actually are" 16:16
_sri i could argue all day about perl5 still being haunted by the 90s... but i think we would keep spinning round and round...i'm tired :(
ash_ non-blocking IO, do you mean like using select? or fork?
masak _sri: seems no-one really is in disagreement with you.
ash_ can current rakudo fork?
pmichaud ash_: I'm guessing it's along the lines of select. 16:17
_sri non blocking sockets + poll/epoll/kqueue
pmichaud I don't really know what Parrot has in mind for that. 16:18
16:20 Mowah left, skids left
_sri without you can pretty much forget about websockets 16:21
pmichaud _sri: I'm speculating that somewhere you've read or seen something that implies that we're trying to market Rakudo to cgi hackers. 16:22
perhaps others are doing that, but I don't think I've done anything to market specifically to cgi.
ash_ the only reason i have tried using rakudo with cgi is cgi is dead simple
_sri pmichaud: my fear was triggered by moritz_ mumbling about porting cgi frameworks
pmichaud ...which iirc was itself in response to someone saying we need a Perl 6 forum written in perl 6. 16:23
which, btw, I totally disagree with.
jnthn Doesn't the world have enough forum software? :-)
masak if I use the fat comma for formatting purposes, am I totally evil? :>
pmichaud I'm all in favor of new forum software, written in Perl 6.
But if we're looking at that as a short-term project, I think it a bit unrealistic. 16:24
so I think we need to set expectations. If we want a forum -soon- for Perl 6 specific discussions, we shouldn't require it to be written in Perl 6.
masak I just realized that instead of `say something(), "\t", something-else();`, I could write `say something() => something-else();` with the same effect.
jnthn masak: heh
pmichaud If we want to have some forum software written in Perl 6, that's great too, but we shouldn't expect to have it soon.
masak mwhaha.
jnthn masak: Relying on how pairs stringify. 16:25
masak aye.
jnthn masak: I guess that's spec'd, tested and kosher.
:-)
masak jnthn: probably not very maintainable, though.
jnthn masak: It will, however, probably confuse some newbie who reads your code. :-)
masak right.
pmichaud decides to comment to the perl6 mongers post 16:26
er, perlmonks
16:26 skids joined, Su-Shee joined 16:31 redicaps left
ash_ masak: sub amp { }; say &amp; # that works in my console now, that one was trixy 16:33
masak ash_: as in 'trixy to color-code'?
ash_ yeah, since &amp; is html for & 16:34
masak ah, right. 16:35
yes, that one is often tricky to get right, when implementing an escaper. :)
it does help that Str.trans does all the replacements in one go...
...but I guess this is js, not p6. :)
pugssvn r31482 | bbkr++ | t[t/spec/S06-signature/slurpy-params.t] tests for RT #72600 if statements clobber @_ in Rakudo 16:38
16:39 thebird left, envi^home left
ash_ str_obj.replace(/reg/g, "Replacement"); works in javascript, but since i have multiple passes to do various things like highlight $var's it was seeing &amp; as not a variable (since &amp; gets translated to &amp;amp; it was trying to highlight the wrong part of it) 16:39
masak swimming &
16:39 masak left 16:40 hercynium joined 16:41 Trashlord joined 16:43 kloeri joined 16:48 hercynium left 16:56 cdarroch joined, cdarroch left, cdarroch joined 16:58 dakkar left
pugssvn r31483 | bbkr++ | [t/spec/S06-multi/type-based.t] tests for RT #72750 Make Inf and NaN acceptable constants parameters in Rakudo 17:04
bbkr tickets closing day :) 17:05
colomon bbkr++ 17:06
pmichaud yes, I tried to do things so that we'd have some tickets to close :)
bbkr pmichaud++ 17:07
17:08 meppl joined 17:15 timbunce left
bbkr rakudo: say (4...^5).perl # is this correct output? 17:17
p6eval rakudo c18d37: OUTPUT«((4, 3, 2, 1, 0), [1, 2, 3, 4])␤»
pmichaud bbkr: I claim the current definition of .perl in the spec is a bit ambiguous.
cxreg so what's the deal with STM? looks like it was removed from parrot last year? is it still part of the design? 17:18
bbkr rakudo: say (4 ... 0 .. 5)
p6eval rakudo c18d37: OUTPUT«4321012345␤»
pmichaud anyway, testing a result based on .perl is generally not the best way to test
rakudo: say (4...^5).elems # just curious 17:19
p6eval rakudo c18d37: OUTPUT«9␤»
pmichaud good.
bbkr pmichaud: thanks, I was checking if rt.perl.org/rt3/Ticket/Display.html?id=72912 can be closed
pmichaud since .perl now works on &infix:<...>, I think so. 17:20
17:21 mberends joined 17:25 cygx joined 17:30 TiMBuS left
[Coke] cxreg: parrot questions are generally better asked in #parrot on irc.parrot.org - but yes, I don't think STM is currently expected to make a comeback. I'm not sure that the version we had was ever in trunk or finished. 17:31
17:39 Mowah joined 17:40 cygx left 17:41 [Coke] left 17:42 [Coke] joined
ash_ ping cygx ? 17:47
doh i think he left already
17:47 proller left 17:48 proller joined 18:06 hejki_ joined, spinclad_ joined, hejki_ left, Ross left, patspam left, diakopter left, takadonet left, breinbaas left, elmex left, saaki left, silug left, szabgab left, kolibrie left, _sri left, ashleydev left, Patterner left, sorear left, rhr left, jjore_ left, Khisanth left, astinus left, renormalist left, jql left, spinclad_ left, Trashlord left, dimid left, revdiablo left, cdarroch left, tadzik left, mmcleric left, clintongormley left, Helios left, christine left, lue left, frettled left, Tene left, Su-Shee left, pragma_ left, azert0x left, dju left, nimiezko left, sukria left, c1sung left, zostay left, szbalint left, slavik left, jedai left, mikehh left, gfldex left, allbery_b left
ash_ phenny: tell cygx how is the backend going? Is there anyway I could start testing any of it with the backend? I can run it on my server, also i don't mind if some of the responses are faked for now 18:06
phenny ash_: I'll pass that on when cygx is around.
ash_ phenny: cygx also greaterthaninfinity.com/frontend/ma...shell.html has a more up-to-date shell example
18:06 mdxi_ joined, pjcj_ joined, hatsefla1s joined, Mowah left, ash_ left, felipe left, colomon left, eiro left, thepler left, tomaw left, Gothmog_ left, opx^away left, pjcj_ left, kloeri left, alester left, pmichaud left, PZt left, buubot left, IllvilJa left, rv2733 left, PacoLinux left, RAT left, synth left, meppl left, rcsheets left, [Coke] left, perlygatekeeper left, frew left, chitragupt left, mdxi_ left, sftp left, foodoo left, pnu left, literal left, fda314925 left, TimToady left, justatheory left, Guest23195 left, bbkr left, baest left, PerlJam left, Maddingue left, f00li5h left, Guest3330 left, shade_ left, dual left, cono left, zamolxes left, pugssvn left, nothingmuch left, awwaiid left, bkeeler left, moritz_ left, mberends left, pnate left, xinming left, felliott left, jnthn left, buu left, sbp left, yves__ left, hatsefla1s left, proller left, plainhao left, ciphertext left, Juerd left, frodwith left, cxreg left, m6locks left, phenny left, mdxi left, hatseflats left, ascent_ left 18:07 Ambiguity joined, dimid joined, cuppe_ joined, spinclad joined, imarcust1is joined, patch_ joined, ascent_ joined, m6locks joined, phenny joined, yves__ joined, sbp joined, cxreg joined, buu joined, moritz_ joined, jnthn joined, TimToady joined, card.freenode.net sets mode: +vooo phenny moritz_ jnthn TimToady, f00li5h joined, Maddingue joined, frodwith joined, bkeeler joined, Juerd joined, fda314925 joined, PerlJam joined, awwaiid joined, baest joined, nothingmuch joined, pugssvn joined, zamolxes joined, literal joined, pnu joined, chitragupt joined, frew joined, gabiruh joined, wolverian joined, lucs joined, BinGOs joined, dukeleto joined, card.freenode.net sets mode: +oovo Juerd PerlJam pugssvn dukeleto, sunnavy joined, athomason joined, Grrrr joined, nsh_ joined, cognominal joined, Woody2143 joined, charsbar joined, kst joined, dalek joined, card.freenode.net sets mode: +v dalek, zakame joined, lestrrat joined, Solarion joined, cosimo_ joined, yahooooo joined, knewt2 joined, mantovani joined, LionMadeOfLions joined, cls_bsd joined, sjohnson joined, sykes_ joined, constant joined, Lorn joined, japhb joined, stepnem joined, hugme joined, ingy joined, sawyer_ joined, exodist joined, skangas joined, stef_ joined, ciphertext joined, cono joined, buubot joined, dual joined, bbkr joined, PZt joined, rcsheets joined, foodoo joined, pmichaud joined, colomon joined, eiro joined, thepler joined, tomaw joined, Gothmog_ joined, opx^away joined, synth joined, felipe joined, shade_ joined, felliott joined, xinming joined, pnate joined, perlygatekeeper joined, RAT joined, alester joined, PacoLinux joined, card.freenode.net sets mode: +oo pmichaud colomon, Guest23195 joined, Guest3330 joined, rv2733 joined, IllvilJa joined, sftp joined, plainhao joined, justatheory joined, ash_ joined, kloeri joined, meppl joined, mberends joined, Mowah joined, [Coke] joined, proller joined, mdxi_ joined, pjcj_ joined, hatsefla1s joined, hejki_ joined, cdarroch joined, Su-Shee joined, ashleydev joined, Ross joined, tadzik joined, patspam joined, mmcleric joined, slavik joined, pragma_ joined, azert0x joined, clintongormley joined, Helios joined, Patterner joined, jedai joined, christine joined, dju joined, mikehh joined, lue joined, nimiezko joined, saaki joined, cotto joined, araujo joined, simcop2387 joined, dmpk2k_ joined, nadim joined, Sanitoeter joined, arnsholt joined, _ilbot2 joined, card.freenode.net sets mode: +v _ilbot2, p6eval joined, cj joined, tylerni7 joined, [particle] joined, s1n joined, Sarten-X joined, estrabd joined, drbean joined, broquaint joined, aesop joined, Util joined, mtve joined, rgrau joined, IRSeekBot joined, mathw joined, card.freenode.net sets mode: +vovo p6eval [particle] IRSeekBot mathw, betterworld joined, huf joined, meteorjay joined, metaperl joined, gfldex joined, sorear joined, silug joined, rhr joined, jjore_ joined, diakopter joined, sukria joined, c1sung joined, takadonet joined, szabgab joined, Khisanth joined, frettled joined, kolibrie joined, breinbaas joined, astinus joined, _sri joined, zostay joined, elmex joined, renormalist joined, jql joined, Tene joined, allbery_b joined, szbalint joined, card.freenode.net sets mode: +ooo sorear diakopter Tene, kfo left, Ambiguity left, kfo joined 18:08 Trashlord joined 18:10 Ambiguity joined 18:23 stef_ left 18:29 stef_ joined, timbunce joined 18:35 cuppe joined
cuppe rakudo: my @squares = gather for 0 .. Inf { take $_*$_ }; say 1; 18:36
p6eval rakudo c18d37: ( no output )
cuppe hm 18:37
18:37 ash_ left
cuppe shouldn't that work and give 1 18:37
mberends some of it is not lazy yet
for, probably 18:38
18:38 ash_ joined
cuppe hm 18:38
18:38 dakkar joined
[particle] i don't like the lack of spaces there 18:39
rakudo: my @squares = gather for 0 .. Inf { take $_ * $_ }; say 1;
p6eval rakudo c18d37: ( no output ) 18:40
cuppe that's how I would type it in code
colomon it should not work.
or at least, it shouldn't work yet.
cuppe should it work...eventually?
colomon I don't know.
pmichaud?
cuppe it's a wikipedia example
colomon For perl 6? 18:41
cuppe en.wikipedia.org/wiki/Perl_6#Gather
colomon rakudo: say $squares = gather for 0 .. Inf { take $_ * $_ }; say 1; # should work 18:42
p6eval rakudo c18d37: OUTPUT«===SORRY!===␤Symbol '$squares' not predeclared in <anonymous> (/tmp/q49fB8Dkng:11)␤»
[particle] rakudo: say 0 * 0
p6eval rakudo c18d37: OUTPUT«0␤»
colomon rakudo: my $squares = gather for 0 .. Inf { take $_ * $_ }; say 1; # should work
[particle] that's not a square.
p6eval rakudo c18d37: OUTPUT«1␤»
cuppe rakudo: my $squares = gather for 0 .. Inf { take $_ * $_ }; say $squares.perl; 18:43
p6eval rakudo c18d37: ( no output )
ash_ rakudo: my $squares = gather for 0 .. Inf { take $_ * $_ }; say $squares.WHAT;
[particle] rakudo: my @squares = gather for ^Inf { take $_ * $_ }; say $squares[3];
p6eval rakudo c18d37: OUTPUT«List()␤»
rakudo c18d37: OUTPUT«===SORRY!===␤Symbol '$squares' not predeclared in <anonymous> (/tmp/EuTmu3056z:11)␤»
colomon cuppe: you have to watch for trying to evaluate infinite lists.
[particle] rakudo: my $squares = gather for ^Inf { take $_ * $_ }; say $squares[3];
p6eval rakudo c18d37: OUTPUT«9␤»
cuppe colomon: of course
pmichaud right now, rakudo stops eagerly evaluating only on those things that it can provably determine are infinite 18:44
colomon $squares.perl will do that.
[particle] rakudo: my $squares = gather for 1 .. Inf { take $_ * $_ }; say $squares[3];
p6eval rakudo c18d37: OUTPUT«16␤»
[particle] yep, works
pmichaud a gather/take loop isn't provably infinite, for the most part :-)
cuppe goes to read up on sigils again
colomon cuppe: what pmichaud is saying is that @a = infinite list will only work in certain limited cases. 18:45
very limited at the moment, as far as I know.
pmichaud yes
we might be able to get .map and .grep to transitively determine infinite-ness
cuppe why is this working with $squares
pmichaud because assignment to a scalar isn't eager 18:46
assignment to an array is eager
18:46 armagad joined
colomon when you assign to @squares, it tries to find the first value, then second, etc. 18:46
ash_ can you do something like: (gather for 1..Inf { take $_ * $_ }).lazy; ?
or lazy gather ...
colomon when you assign to $squares, it just puts the List object in $squares.
ash_ (with list assignment) 18:47
pmichaud ash_: I suppose that .lazy could somehow be a list that always reports true for infiniteness
but that gets tricky with things like .elems then
18:47 armagad left
colomon seems like that is confusing laziness with infinite-ness. 18:47
pmichaud it's really a question of "mostly eagerness"
and when something that is "mostly eager" (such as array assignment) should stop evaluating. 18:48
ash_ i expect the lazyness of gather/take to catch some people by surprise, especially with 1..* ranges
pmichaud I think there are surprises no matter what we do. 18:49
So the question is where to put the surprises, or how to minimize them.
ash_ rakudo: my @a = 1..*; for @a { last }; # what if someone tries that? 18:50
pmichaud that works out fine.
p6eval rakudo c18d37: ( no output )
pmichaud rakudo: my @a = 1..*; for @a { say $_; last; }
colomon rakudo: my @a = 1..*; for @a { last }; say "got here"
the suspense is killing me.
pmichaud I'm wondering why there's suspense. :) 18:51
ash_ i think its timing out with p6eval (it might work locally, haven't checked)
p6eval rakudo c18d37: ( no output )
colomon ash_: it should be very fast
works locally
pmichaud > my @a = 1..*; for @a { say $_; last; }
1
works here.
ash_ ah, i didn't try it locally, i kinda assumed it was going into an infinite loop
pmichaud rakudo: my @a = 1..*; 18:52
something isn't right there.
p6eval rakudo c18d37: ( no output )
pmichaud rakudo: my @a = 1..*; say 'alive';
ash_ maybe p6eval is out of date?
colomon c18d37 is latest
p6eval rakudo c18d37: ( no output )
pmichaud yeah, I'm thinking p6eval is out of date, even though it appears to be reporting the correct git revision
ash_ rakudo: say 'testing'
p6eval rakudo c18d37: OUTPUT«testing␤»
colomon rakudo: say 4 %% 3 18:53
p6eval rakudo c18d37: OUTPUT«===SORRY!===␤Confused at line 11, near "say 4 %% 3"␤»
colomon oooo, definitely out of date.
pmichaud it's definitely out of date.
colomon that's an ugly p6eval bug.
ash_ its lying about its version, eh?
colomon yes
pmichaud where's the RT for p6eval?!? I want to channel masak++!
ash_ maybe someone could add the git id to $?PERL.version 18:54
?
or is that not worth the effort
pmichaud ash_: where would we get it from?
ash_ git? (if its installed? at compile time?)
if not thats fine 18:55
pmichaud and why would we believe it any more than whatever p6eval is reporting now? presumably p6eval gets it from the same place :-)
ash_ just thinking
well, if its added to the settings or something at compile time... it shouldn't change...?
pmichaud maybe we can do something with git describe 18:56
ah, we can put git-describe into a generated file 18:57
and use it
ash_ rakudo: say $?PERL.version
p6eval rakudo c18d37: OUTPUT«===SORRY!===␤Symbol '$?PERL' not predeclared in <anonymous> (/tmp/kyzYCGPscl:11)␤»
cuppe so I have another thing that's confusing me. I noticed that BUILD kills any initialized values you gave variables. so I can create them in the BUILD, except they don't seem to keep values
rakudo: class Data { submethod BUILD { has $.user = 13;}}; say Data.new.user; 18:58
p6eval rakudo c18d37: OUTPUT«Any()␤»
colomon rakudo: class Data { has $.user; submethod BUILD { $.user = 13;}}; say Data.new.user;
p6eval rakudo c18d37: OUTPUT«Cannot assign to readonly value␤ in '&infix:<=>' at line 1␤ in 'Data::BUILD' at line 11:/tmp/SJlRxmNmmd␤ in main program body at line 11:/tmp/SJlRxmNmmd␤»
ash_ rakudo: class Data { has $.user is rw; submethod BUILD { $.user = 13;}}; say Data.new.user; 18:59
pmichaud inside of build, use $!user
p6eval rakudo c18d37: OUTPUT«13␤»
colomon rakudo: class Data { has $.user; submethod BUILD { $!user = 13;}}; say Data.new.user;
p6eval rakudo c18d37: OUTPUT«13␤»
18:59 shade_ left
cuppe oh okay. my problem that I was leading at was creating readonly variables with initial values 19:00
and that'll solve that
19:00 masak joined
pmichaud afk 19:00
bbl
19:01 hejki_ left, hejki joined, emadum joined 19:02 pyrimidine joined 19:03 shade_ joined, emadum is now known as __eric__ 19:04 rv2733 left 19:11 rindolf joined
cuppe if a scalar can hold a List type...what's special about arrays 19:12
masak cuppe: they convey list context. 19:14
cuppe: they also narrow things down to something Positional, often an Array.
think of it as a small, cute type system :)
hm, I seem to have interpreted your question as "what's so special about the @ sigil?" -- hope that's what you meant. :) 19:15
cuppe but surely I can force list context in some other way, like () in p5
masak sure.
it's just for convenience.
but the convenience is quite significant, I would say.
cuppe yeah
mberends \o masak 19:16
cuppe so I can create scalars of hashes too then. or any type
masak cuppe: yes, scalars are the generic type.
mberends: o/
mberends masak: after porting half of HTTP::Server::Simple::PSGI, I discovered that I'd misunderstood the intend of HTTP::Server::Simple :/ 19:17
masak mberends: :/ 19:18
mberends: in what way?
dakkar grammar Foo { rule TOP { <[a..z]> } };my $match=Foo.parse('a');.perl.say for @($match) 19:19
rakudo: grammar Foo { rule TOP { <[a..z]> } };my $match=Foo.parse('a');.perl.say for @($match)
p6eval rakudo c18d37: OUTPUT«Match.new(␤ # WARNING: this is not working perl code␤ # and for debugging purposes only␤ from => 0,␤ orig => "a",␤ to => 1,␤ positional => [␤,␤ ],␤)␤»
dakkar rakudo: my $match=Perl6::Grammar.parse('$a+$b');.perl.say for @($match)
p6eval rakudo c18d37: OUTPUT«Method 'perl' not found for invocant of class 'Regex;Match'␤ in main program body at line 11:/tmp/c6s7sZA5NA␤»
mberends masak: I thought the latter was a working, self contained class. But it's only abstract :(
masak oh.
dakkar ehrm… what am I doing wrong? 19:20
masak dakkar: that shouldn't happen, I think.
dakkar oh :)
rindolf Is rakudo-2010.04-1mdv2010.1 good enough or should I use the git version? 19:21
masak at least if moritz_++'s efforts to make Match a pure-Perl 6 class were ever finished, and I think they were.
rindolf: please be more specific.
rindolf: you do know that we add and change things on a daily basis? :P
rindolf masak: ah.
masak rindolf: sometimes I take for granted that people know that. Rakudo has a fairly high development speed. 19:22
rindolf: the last release added lazy lists.
dakkar masak: well, Match seems to work. Perl6::Grammar is not returning Match objects, though
masak dakkar: I wish I knew more. moritz_ will, though.
dakkar I'll stay around 19:23
sorear hello #perl6 19:25
sorear hugs _sri 19:26
Negativity is not tolerated here.
masak backlogs
_sri hugs sorear back
[Coke] everybody's always hating on negativity. maaaan. 19:29
rindolf Hi sorear
github.com[0: 207.97.227.239]: errno=Connection timed out - sucks.
Oh wait. Firewall.
No, doesn't work. 19:31
19:31 simcop2387 left
masak _sri: I'd love to have unblocking sockets too. not sure Parrot allows this at thigs point, even theoretically. Rakudo (and Rakudo Star) is all about releasing something which is Usable Now rather than Perfect and Complete. 19:32
_sri masak: i don't think parrot supports them yet
sorear anyways, jnthn and I need your firstborn at this point 19:33
use POSIX:from<perl5>; my $s = socket(...);
masak _sri: given this, are you suggesting unblocking sockets are important enough to slip the release date for Rakudo Star?
_sri masak: thats not my decision to make 19:34
masak right. neither is it mine.
just asking about what you consider to be the priorities here.
you seem to think they are very important. 19:35
_sri i'm just lobbying from the sideline
ash_ parrot doesn't do async io yet, so you can't do async sockets either
_sri i know i want a websocket server on perl6 asap
Guest23195 hm, @b is flattened into @a if I push @a, @b. How do I make a multidimensional array? 19:36
masak _sri: I think it's very good that we have someone lobbying for that, lest we forget it among all the other stuff that's happening.
_sri short term the next big fight in the web development work will be all about non blocking io, everybody will have to compete with node.js
masak Guest23195: push @a, [@b]
_sri so thats what i'm interested in
Guest23195 masak thx 19:37
masak Guest23195: that copies. there's one variant which doesn't, too.
Guest23195: push @a, \@b
Guest23195: the copying is shallow.
_sri *world
Guest23195 ok. all perl5 is not lost :-) 19:38
masak Guest23195: we did retain some fragments :P
_sri masak: right now it looks like ruby will be in big trouble soonish, because it can't keep up with the async requirements for the next generation of web frameworks 19:39
masak: so i want to verify the same thing doesn't happen in perl6
ash_ non-blocking io and threading are big issues, especially if you don't have a GIL or a way to work around a GIL
masak ash_: what's a GIL? 19:40
sorear _sri: seriously, it disheartens me to see you passing over blizkost like this
ash_ global interpreter lock
python and ruby have them, they can use system threads, but they can only have 1 scheduled at a time
19:41 plainhao left
_sri sorear: :( 19:41
ash_ so, while they are doing multi-processing, they aren't taking advantage of multiple cores
in python though you can use the module multiprocessing, which does shared memory and forks, so you can take advantage of multiple cores, although you can't do somethings like share a file handler 19:42
_sri ruby fibers are really poor
ash_ macruby got rid of the GIL in OS X, and added libdispatch too 19:43
_sri it's a big opportunity for perl6 actually
ash_ so it can do some cool stuff with that 19:44
19:44 simcop2387 joined
_sri oh, thats interesting 19:44
ash_ but, thats mac only, and requires the llvm, its also not 100% compatible with 1.9, its a fork of matz's ruby 1.9, but it forked a while back, they have gotten almost all of the compatibility back though, it can run rack, and most gems are fine, its mostly edge cases that differ 19:45
its actually kinda cool, you can even compile your ruby code into a native executable or a .dylib, and it can natively call anything in the obj-c runtime so you can do cocoa programming in it without a bridge 19:47
thats what i want to do for parrot with part of my GSoC 19:48
add an llvm based runcore, so it can compile parrot bytecode into native code, and i am almost done with my libffi changes so you can have a lot more flexibility with NCI calls
_sri damn, if it wasn't mac specific it could totally take over the world 19:50
19:50 rindolf left
ash_ yup, they replaced the GC system in macruby, so it uses the obj-c GC system, that is the biggest reason they still have the GIL in ruby 19:51
19:51 rindolf joined
ash_ and all ruby objects are obj-c types, ([].class #=> "NSMutableArray" in macruby, Array is an alias to that) so all obj-c calls just use native obj-c objects, its pretty crazy, but part of the reason they can get away with that is ruby's object system is almost identical to obj-c, and they both allow runtime construction of objects 19:52
_sri i was already excited when i was reading up on rubinius, but this is wow
ash_ i bet it will be included by default in 10.7 since there is code for 10.7 in their repository, which is kinda funny IMO 19:54
19:54 mmcleric left
_sri one more reason to be afraid of cocotron www.cocotron.org/ 19:54
alester stackoverflow.com/questions/3135673...ons-mature FYI 19:55
masak thanks! 19:58
ash_ it would be cool if more platforms did Obj-C (only OS X 10.6 does Obj-C 2.1), the gnu obj-c runtime doesn't even do 2.0 yet 19:59
10.5 was obj-c 2.0
anyway, this is a perl channel :P 20:00
_sri hmm, apple is heavily sponsoring macruby 20:01
masak I think pmichaud would be able to write a good reply to that stackoverflow question.
Su-Shee ash_: which means we should steal the best of other ideas.
_sri i suspect it might become a first class language in the apple eco system
ash_ they have an svn host and a git host and will accept patches, but like 99% of the macruby work is done by apple employees (or they all have apple emails) 20:03
sorear sheesh 20:04
ash_ you could port macruby to other OS's, its been done on linux before, but you need to compile libautozeon (the GC system macruby and objective-c use on OSX), which is also open source, but apple specific, its drops down to assembly in places, i hear it compiles on x86, x86-64 linux, and might work on ppc, and ppc-64, but its not plain C, (its C++, C and assembly) 20:08
libautozone*
masak "Some of the smartest error messages that STD emits are actually the two-terms message transmogrified into something more specific by context." -- I like this quote. I hope I remember to integrate it into my YAPC::EU talk. it's by TimToady in www.nntp.perl.org/group/perl.perl6....33148.html 20:09
std: constant Int pi = 3; pi 'foo'; 20:10
p6eval std 31483: OUTPUT«===SORRY!===␤Malformed constant at /tmp/svPM4BwE7Q line 1:␤------> constant Int ⏏pi = 3; pi 'foo';␤ expecting trait␤Parse failed␤FAILED 00:01 109m␤»
masak TimToady, sorear: uhm... ^
sorear std: pi; 20:11
p6eval std 31483: OUTPUT«ok 00:01 106m␤»
masak oh.
sorear pi is already defined
masak std: constant Int psi = 3; psi 'foo';
p6eval std 31483: OUTPUT«===SORRY!===␤Malformed constant at /tmp/NbS2wffr4e line 1:␤------> constant Int ⏏psi = 3; psi 'foo';␤ expecting trait␤Parse failed␤FAILED 00:01 106m␤»
masak but that's not the problem.
unless psi is also already defined :P
std: psi
p6eval std 31483: OUTPUT«===SORRY!===␤Undeclared routine:␤ 'psi' used at line 1␤Check failed␤FAILED 00:01 107m␤»
ash_ should that work?
what does that mean? 20:12
3 'foo'?
20:12 ShaneC joined
masak ash_: see www.nntp.perl.org/group/perl.perl6....33148.html 20:13
ash_: TimToady theorizes that it should error out at parse time.
ash_ isn't that a parse time error? "Parse failed␤FAILED"
masak ash_: yes, but it's occurring earlier than I would expect. 20:14
ash_: take a look at the bail-out marker.
rindolf t/spec/S05-mass/properties-derived.rakudo ...................... Failed 9/256 subtests # Mandriva 2010.1 on an x86-64 machine - rakudo master.
ash_ masak: is it the type? 20:15
Int?
rakudo: constant psi = 3; psi 'foo';
p6eval rakudo c18d37: OUTPUT«Could not find sub &psi␤ in main program body at line 11:/tmp/WlvSppdD8i␤»
masak ash_: yes, it's the type. but that should parse.
ash_ std: member:constant psi = 3; psi 'foo';
p6eval std 31483: OUTPUT«===SORRY!===␤Preceding context expects a term, but found infix = instead at /tmp/uYXNEX6EcF line 1:␤------> member:constant psi ⏏= 3; psi 'foo';␤Parse failed␤FAILED 00:01 109m␤»
ash_ std: constant psi = 3; psi 'foo' 20:16
p6eval std 31483: OUTPUT«===SORRY!===␤Two terms in a row at /tmp/GuLrMgAsH9 line 1:␤------> constant psi = 3; psi ⏏'foo'␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤ statement modifier loop␤Parse failed␤FAILED 00:01 109m␤»
masak that's more what I wanted.
ash_ std: constant Int a = 123;
p6eval std 31483: OUTPUT«===SORRY!===␤Malformed constant at /tmp/1IGa5FgQ1Q line 1:␤------> constant Int ⏏a = 123;␤ expecting trait␤Parse failed␤FAILED 00:01 106m␤»
ash_ i don't think std's constant likes types
masak rindolf: some of the test files in t/spec/S05-mass/ are known to misbehave irregularly. 20:17
ash_: that's the problem, yes. correct.
rindolf masak: do you know how to solve it?
masak rindolf: no. 20:18
rindolf masak: I see.
masak ash_: ah, maybe the syntax isn't that. there's not an example in the spec that puts the type in that place.
std: my Int constant psi = 3
20:18 rindolf left
p6eval std 31483: OUTPUT«ok 00:01 109m␤» 20:18
masak std: constant psi of Int = 3
p6eval std 31483: OUTPUT«ok 00:01 109m␤»
[particle] std: constant a = Int(3) 20:19
p6eval std 31483: OUTPUT«ok 00:01 109m␤»
20:19 rindolf joined
masak either of those is fine. 20:19
the form in the email probably isn't.
ash_ got ya
masak we're lucky that TimToady wrote a program to tell us when he writes improper Perl 6 :P
ash_ so my and constant aren't interchangeable, it looks like he just replaced my with constant 20:20
std: my Int pi = 3;
p6eval std 31483: OUTPUT«===SORRY!===␤Multiple prefix constraints not yet supported at /tmp/jhP3b2jXAw line 1:␤------> my Int pi ⏏= 3;␤Malformed my at /tmp/jhP3b2jXAw line 1:␤------> my Int pi ⏏= 3;␤ expecting any of:␤ multi_declarator␤
..typename␤Pa…
ash_ ah wait, no sigil
std: my Int $pi = 3;
p6eval std 31483: OUTPUT«ok 00:01 110m␤»
masak ash_: correct. with 'constant', you get away with not having the sigil. 20:21
ash_ std: constant Int pi = 3;
p6eval std 31483: OUTPUT«===SORRY!===␤Malformed constant at /tmp/HG3ME7a3Ai line 1:␤------> constant Int ⏏pi = 3;␤ expecting trait␤Parse failed␤FAILED 00:01 106m␤»
20:27 Trashlord left 20:35 rindolf left, tadzik left, mberends left 20:37 ShaneC1 joined 20:38 ShaneC left 20:39 ShaneC joined 20:41 RAT is now known as robert3t, ShaneC1 left 20:43 ShaneC left 20:45 timbunce left 20:46 Intensity joined 20:48 Su-Shee left
masak "omg people are helpful there!" -- :) -- blogs.perl.org/users/kaare/2010/06/...-code.html 20:50
oh, and moritz_++
20:50 Gruber joined 20:53 Grrrr left
masak with #parrot's help, I just figured out a way to read a non-utf8 file with Rakudo. \o/ 20:53
[particle] does $ntuples-1 actually parse an infix:<-> there?
masak rakudo: my $ntuples = 8; say $ntuples-1
p6eval rakudo c18d37: OUTPUT«7␤»
masak [particle]: yes. 20:54
std: my $ntuples = 8; say $ntuples-1
p6eval std 31483: OUTPUT«ok 00:01 113m␤»
[particle] that doesn't mean it's correct, and i don't know how to check std... wait, i do
masak [particle]: yes, that means it's correct. 20:55
[particle] std: my $ntuples-1 = 8;
p6eval std 31483: OUTPUT«ok 00:01 110m␤»
[particle] orly?
jnthn rakudo: my $ntuples-1 = 8;
p6eval rakudo c18d37: OUTPUT«Use of uninitialized value in numeric context␤Cannot assign to readonly value␤ in '&infix:<=>' at line 1␤ in main program body at line 11:/tmp/ufkEM0BA3_␤»
jnthn Right. :-)
It's doing
(my $ntyples - 1) = 8
[particle] that still doesn't tell me whether rakudo or std is wrong
jnthn Both are right I think.
[particle] yes, and you can assign to that?
jnthn STD doesn't do semantics like that. 20:56
[particle] hrmm
jnthn So it'd not know.
(Rakudo whines at you at runtime there.)
sorear std: die
[particle] ok, gotta re-read the semantics for the apostrophe
p6eval std 31483: OUTPUT«ok 00:01 107m␤»
sorear std: 2 + 2 = 5;
p6eval std 31483: OUTPUT«ok 00:01 108m␤»
[particle] s/semantics/syntax rules/ 20:57
masak [particle]: anywhere between two \w chars is fine.
std: my $_-_-_'_-_'_
p6eval std 31483: OUTPUT«ok 00:01 108m␤»
[particle] is 1 not \w?
masak [particle]: er. yes :/
[particle] see! 20:58
dalek kudo: 2acfad3 | moritz++ | src/Perl6/Grammar.pm:
complain about infix:<!%>. Closes RT #76170
kudo: cccc121 | moritz++ | (2 files):
re-enable Cool.rindex

parrot String. Also enable rindex.t
kudo: 1734e43 | moritz++ | src/core/Signature.pm:
properly reflect capture/parcel binding in signature introspection
[particle] LTM suggests $ntuples-1 is one identifier
masak then it must be <+[\w]-[\d]>
that is, alphabetics and underscore.
pugssvn r31484 | moritz++ | [t/spec] tests for RT #75956, @*INC should be writable
r31485 | moritz++ | [t/spec] test for RT #76174, "if" does not break lexical aliasing of $_ in a for-loop
[particle] token identifier { <.ident> [ <.apostrophe> <.ident> ]* } 20:59
ash_ won't it only say $ntuples-1 is an identifier if there is an identifier $ntuples-1 ?
[particle] token ident { <.alpha> \w* }
jnthn There's always reading the grammar ;-)
<.ident> [ <.apostrophe> <.ident> ]*
[particle] urk
pugssvn r31486 | moritz++ | [t/spec] test for RT #76226, all-whitespace range
r31487 | moritz++ | [t/spec] test for RT #76198, {;} should be a Block
ash_ rakudo: my $a = 1; say $a-1;
p6eval rakudo c18d37: OUTPUT«0␤»
jnthn token ident { <.alpha> \w*
}
[particle] yep
pugssvn r31488 | moritz++ | [t/spec] unfudging for rakudo
r31489 | moritz++ | [t/spec] regex unfudging for rakudo
r31490 | moritz++ | [S06] \| and | prefixes of parameters should be separately introspectable
r31491 | moritz++ | [t/spec] tests for RT #76116, parcel and capture binding in signature introspection
jnthn So it has to be alphanumeric after the - or the '
It's not about what's before
gah
alpha
[particle] yep
jnthn not alphanum... 21:00
colomon moritz_ goes crazy!
moritz_ just had a long, boring offline train ride 21:01
masak moritz_: dakkar has a Regex::Match-related question in the backlog that I suspect you might be able to answer. 21:02
or is that Regex;Match? :)
dakkar precisely the problem :)
rakudo: my $match=Perl6::Grammar.parse('$a+$b');.perl.say for @($match)
p6eval rakudo c18d37: OUTPUT«Method 'perl' not found for invocant of class 'Regex;Match'␤ in main program body at line 11:/tmp/dzxhNxBx5Z␤»
dakkar that is: why is Perl6::Grammar returning objects of that weird class? 21:03
21:03 __eric__ left
moritz_ because it derives from Regex::Cursor, not Grammar 21:03
[particle] so... i can't seem to enter unicode 2170 here 21:04
but that could be confusing after an apostrophe
masak [particle]: ⅰ 21:05
[particle] rakudo: my $x = 5; say $x-ⅰ
p6eval rakudo c18d37: OUTPUT«===SORRY!===␤Confused at line 11, near "say $x-\u2170"␤»
[particle] rakudo: my $x = 5; say $x - ⅰ
p6eval rakudo c18d37: OUTPUT«===SORRY!===␤Confused at line 11, near "say $x - \u2170"␤»
[particle] ok, still confused 21:06
.u 2170
phenny U+2170 SMALL ROMAN NUMERAL ONE (ⅰ)
masak [particle]: I don't think ⅰ is defined in Rakudo (or Perl 6)
[particle] icu should see it as numeric, no?
masak [particle]: yeah, but...
...Rakudo still doesn't know what to *do* with it.
it doesn't parse as anything. 21:07
espectially not a number literal.
[particle] hrmm
dalek kudo: d18b5e7 | masak++ | src/core/Buf.pm:
[Buf] now does Positional and .[]
 21:09
masak time to blog a little. 21:10
colomon rakudo: say 5 - 1i
p6eval rakudo c18d37: OUTPUT«5 + -1i␤»
colomon rakudo: say 5 - i
p6eval rakudo c18d37: OUTPUT«Could not find sub &i␤ in main program body at line 11:/tmp/epnqR0vEKi␤»
lue ohai
I looked once, and couldn't find it; where's the i for imaginary/complex numbers defined? 21:11
21:11 ciphertext left
colomon These days it's in Numeric, I think. 21:12
multi sub postfix:<i>(Numeric $z) {
$z * 1i;
}
lue
.oO( hooray for recursion! \o/ )
 21:13
21:14 songmaster joined
lue I look at that and am amazed it doesn't break. 21:16
moritz_ I guess there's another grammar rule for <number>i 21:17
so that the postfix:<i> only handles cases like $x\i 21:18
21:18 patch_ is now known as patch
moritz_ stackoverflow.com/questions/3135673...92#3136192 21:21
lue Because I have the beginnings of a Quaternion type, and I was wondering how to get stuff like 4+3i+2k+1j working.
21:24 Mowah left 21:26 foodoo left
sorear somebody needs to point out that [+] is implemented by calling a high level function which uses a loop 21:27
masak loliblogged! use.perl.org/~masak/journal/40422
moritz_ sorear: just updated my post
masak I really like the look-and-feel of StackOverflow. 21:29
moritz_ lolialzoblogged: perlgeek.de/blog-en/perl-6/contribu...ttery.html
ash_ masak: anymore suggestions for the try.rakudo.org shell? 21:31
masak ash_: looking.
moritz_ have prepared a post for tomorrow: next challenge is to implement $*ARGFILES, as suggested by pmichaud++
github.com/moritz/perlgeek.de/blob/...gfiles.txt # preview 21:32
21:33 whiteknight joined
masak ash_: well, unfortunately after my comment about hovering for showing things, the cammands list sticks out from under the main blue thing in a not altogether aesthetic way. 21:33
ash_ yeah, i know
lue
.oO(I thought they were clockroaches)
ash_ i am trying to figure out how i want to deal with that
masak ash_: not a very critical thing, just letting you know. 21:34
ash_ i might resize camelia and move it to fit above the terminal and move the commands list up the side some
masak ash_: I really like the skull-and-crossbones ^^
ash_ lol
masak ash_++
ash_: I bet I could make up all manner of nightmarish things for you to fix with string interpolation. are you sure you want me to tell you? 21:35
for example: "foo $bar baz" should highlight $bar, but 'foo $bar baz' shouldn't. right now, it does. 21:36
21:36 Guest23195 left
masak ditto q[foo $bar baz] 21:36
ash_ i don't do fancy quotes yet
only ' and "
moritz_ but q:s[foo $bar baz] should highlight $bar :-)
masak well, '$bar' shouldn't hightlight $bar. 21:37
21:37 robert3t left
masak ash_: it should only highlight if it actually interpolates. 21:37
ash_ i think i broke string constants...
one sec
masak in that vein, "foo @bar[] baz" should highlight @bar[], but "foo @bar baz" shouldn't highlight @bar. 21:38
ash_ hmm, string constants seem broken completely 21:39
wonder when i did that, probably when i add <a b c>
masak you had a separate color for string literals? 21:40
ash_ ya
masak ash_: make sure you have "$foo."$bar"" as a test case, then.
um.
ash_ um...
masak "$foo."$bar"()"
that's it.
ash_ k
i can live with that
(i think)
21:42 japhb left
ash_ ' and " work again 21:43
masak '$baz' still highlights $baz here. 21:44
ash_ is that wrong?
masak yes. 21:45
ash_ ya, thats wrong isn't it, ' doen't interpolate
masak single quotes don't interpolate.
"$foo."$bar"()" doesn't interpolate right either. it gets the variables right, but should probably interpolate the dot and the parens, too. 21:46
also, there's not very much difference between the non-interpolated green and the interpolated green :)
masak stops complaining and goes to bed 21:47
'night, #perl6
21:47 masak left 21:51 dakkar left 22:04 Ross left 22:06 pyrimidine left
sorear jnthn: invocant types in Perl 6 are never statically known without whole-program analysis 22:08
sorear is, as it happens, working on a whole-program optimizing compiler for Perl 6 22:09
22:11 christine left 22:13 christine joined
jnthn sorear: While that may be true for full-blown Perl 6, it doesn't mean you can't have some additional pragmas in place for your internals/setting which tells the compiler that it can asusme more when it compiles those. 22:13
Like, that Foo.new actually returns something that ~~ Foo 22:14
22:21 gfldex left, gfldex joined 22:31 tylercurtis joined 22:32 felliott left
pugssvn r31492 | bbkr++ | [t/spec] tests for RT #72856 Null PMC access when two methods from different roles collide in a class composition in Rakudo 22:33
22:42 japhb joined
pugssvn r31493 | bbkr++ | [t/spec] tests for RT #72870 Cant take .WHAT of sufficiently decimal-endowed Rat in Rakudo 22:53
22:56 rgrau_ joined 22:57 masonkramer joined 23:05 skids left 23:06 tedv joined 23:08 lestrrat is now known as lest_away 23:14 rgrau_ left
dalek meta: r352 | stefa...@cox.net++ | trunk/vicil/Kernel.cs:
[vicil] Sub gets to be a native class too for now
 23:15
23:22 bluescreen joined 23:23 bluescreen is now known as Guest99621 23:29 cdarroch left 23:31 stepnem left 23:32 azert0x left
ingy rebuilds an EC2 image and rebuilds rakudo... 23:38
I need to learn how to save this EC2 image before turning it off
23:41 stepnem joined
pugssvn r31494 | bbkr++ | [t/spec] tests for RT #72848 "0 but True" causes a no-applicable-candidates error in Rakudo 23:42
23:53 ruoso joined, Psyche^ joined 23:57 hercynium joined 23:58 Patterner left, Psyche^ is now known as Patterner

Powered by Cro and the Raku® Programming Language.

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

Thank you!