»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:05 nif left 00:16 nif joined 00:24 lichtkind left 00:29 [particle] joined 00:31 wolfman2000 joined 00:36 replore_ joined, havenn left 00:44 plobsing joined 00:48 whiteknight left 00:51 xinming joined, vlixes joined 00:53 xinming_ left 00:56 lichtkind joined 01:01 lichtkind left 01:11 jaldhar joined 01:12 scott___ joined 01:15 tokuhirom left 01:16 tokuhirom joined 01:21 tokuhirom left 01:29 liuchong joined 01:39 vlixes left 01:44 wooden joined, wooden left, wooden joined 01:46 nif left 01:52 retupmoca joined 01:54 xinming left 01:56 xinming joined
moritz Juerd: nope, seems back to normal 01:57
02:18 orafu left, orafu joined 02:27 thelazydeveloper left 02:51 xinming_ joined, fgomez joined, fgomez left, fgomez joined 02:53 xinming left 03:12 sunnavy joined 03:39 replore_ left 03:41 replore_ joined 03:54 skids joined 04:44 SHODAN left, SHODAN joined 04:52 farber7 joined 04:55 sudokode left, DarthGandalf left, DarthGandalf joined
farber7 Hello I have a few questions about rooting the MotoActv, if I go for a run with it do I need to bring my phone as well to keep them linked for apps and internet. ? I know it notifies you of text messages but how close will my phone need to be. I have a iPhone 4 now but I had the droid razor and would go back to it but its way to big to take on runs with me. 04:56
04:58 lestrrat left
farber7 Also if I got the watch and made it work how I needed with heart rate monitor and all can I have help rooting it I will pay for it to be rooted if I cannot do it. I have about 10 other people on my team who will do the same thing if possibel 04:58
04:59 lestrrat joined 05:04 xinming_ left 05:06 alvis` joined 05:08 NamelessTee joined
bonsaikitten farber7: I think you're not in the right place 05:08
05:09 birdwindupbird joined 05:11 scott___ left 05:12 lestrrat left 05:13 lestrrat joined 05:14 havenn joined 05:15 sudokode joined 05:21 alvis` left 05:23 sftp left 05:24 spaceships left 05:25 alvis joined 05:26 farber7 left 05:28 replore_ left 05:31 havenn left
moritz good morning 05:36
05:56 havenn joined 05:57 havenn left 05:58 NamelessTee left 06:02 havenn joined 06:07 birdwindupbird left 06:19 jferrero left 06:27 xinming joined 06:33 jferrero joined
jnthn morninä 06:35
geekosaur בוקר טוב 06:38
(...whatever font my client substituted there is hideous)
06:44 tarch joined
cosimo morgen 06:47
shachaf geekosaur: You're allbery_b?! 06:55
moritz \o cosimo
cosimo: LWP::Simple works again on newest Rakudo :-)
06:56 thou left
jnthn wonders if cosimo will be at the Oslo hackathon 06:57
06:58 birdwindupbird joined
cosimo moritz: that's great to hear 07:00
jnthn: i'll try to show up, however my mum is flying to Oslo on saturday morning :) 07:01
geekosaur shachaf, yes 07:03
shachaf geekosaur: I thought you had disappeared forever or something.
07:09 fhelmberger joined 07:20 havenn left 07:22 fglock_ joined 07:31 havenn joined 07:32 nwc10 joined 07:41 havenn left, havenn joined
tadzik hello there 07:46
07:49 havenn left 07:50 pomysl left 07:51 havenn joined, pomysl joined, pomysl left, pomysl joined 07:52 havenn left 07:53 havenn joined 07:55 havenn left 08:13 pomysl left, pomysl joined, pomysl left, pomysl joined
masak morning, #perl6. 08:14
arnsholt o/ 08:15
moritz \o 08:18
masak .oO( I said hello, and all they did was raise arms )
:P
moritz as long as they don't fire... :-) 08:19
tadzik :P
hello hello
08:19 pomysl left
masak it's great to be back. my offline acquaintances are silly too, but not like you guys. :) 08:20
08:21 pomysl joined, pomysl left, pomysl joined
arnsholt I'm still working on my first cup of coffee, so I'm still not quite verbal yet =) 08:21
Also, too many still-s
08:25 pomysl left 08:26 pomysl joined, pomysl left, pomysl joined 08:32 pomysl_ joined 08:33 pomysl left 08:37 pomysl joined, pomysl left, pomysl joined
masak re blogs.perl.org/users/shlomi_fish/20...-been.html -- I think Shlomi was funnier back when we tried to fork Perl 6 and make his own. 08:37
s/we/he/
08:39 pomysl_ left
masak ah, it's still online: www.shlomifish.org/rindolf/rindolf-spec/ 08:42
moritz well, funniness wears off after a while 08:43
masak I think it's a somewhat interesting thought experiment -- what if the Perl 6 transition had been more on the order of Python 3k?
moritz probably like kurila 08:44
masak url?
moritz ie an interesting experiment that doesn't really take off
masak: github.com/ggoossen/kurila#readme 08:45
08:45 bbkr joined
masak I think Perl 6, much like Esperanto, survives a lot on its clear vision. 08:48
i.e. a better something.
moritz so, what's the "something"? 08:50
"everything"? :-)
masak fsvo "something", yes ;) 08:57
actually, I think we're whirlpooling on the "something".
by the end of the process we'll jump out and say "a-ha! Perl 6 is finished, and this is the kind of something it's a better one of!" :P 08:58
moritz I hope it won't come out as "better PHP" or "better awk" :-) 09:00
masak well, it'll be that too. but yeah, I'd be surprized if we landed on only that :P 09:05
09:07 havenn joined 09:09 havenn left
moritz masak: btw have you seen gist.github.com/2346494 ? 09:10
masak I hadn't. nice. 09:14
s/Gas/Vapor/ :)
moritz :-) 09:15
masak also, that's a narrower definition of "slushy" than I've been using.
masak checks his own definition in a blog post
"Slushy is good; it means that we have some backing from implementations, but the spec and the implementations don't quite agree yet." 09:16
to me "slushy" is not about stability so much as "we know that we're at least a little bit right, but we suspect things are still a bit wrong, too".
moritz that's a better phrasing, yes 09:17
masak I guess we're saying the same thing, though.
the numbering for the phases confuses the heck out of me. 09:18
I think both because it feels inverted (solid things are colder, no?) and because the numbers don't tell me anything. I think you need something spatial rather than numbers. something progress-bar-like.
moritz oh right, non-physicsts think that high temperature values are hotter 09:19
masak :P
09:20 TESTKEYz joined
masak but maybe this is "raw source" and it could be rendered as HTML. 09:20
fwiw, I think this should go up alongside the compiler comparison thingy.
moritz if you do statistical physics, you usually use 1/KT as the "temperature"
maybe think of it as Moh's scale instead of temperature 09:21
:-)
anyway, I don't have a problem with renumbering
masak heh, sludgy minerals. 09:22
moritz or maybe use letters, if we can find two terms for "solid" and "slushy" that don't start with the same letter
masak if the letters sort "right" lexicographically, it's fine.
moritz re showing it on perl6.org, I wanted to get some more feedback first 09:23
masak you're looking for that sweet spot of a model that's both simple and intuitive.
moritz because partially that rating is *very* subjective
09:24 havenn joined
masak it is? maybe it has an inherent fuzziness of one point on the scale... 09:24
09:26 havenn left
moritz some random observations so far: 09:26
S02 and S03 have a lot more subheadings than S04 09:27
masak maybe you could re-use the PSA scale from A01? "Accepted", "a few Buts", "major Caveats", "Rejected"
moritz they also have (or hopefully had) a lot more crackpot stuff than S04 (and, I suspect, S05)
masak I think diakopter++ went and added lots of subheadings to S02 at one point.
moritz right 09:28
masak S05 has its fair share of crackpot stuff ;)
moritz it has?
masak .oO( three kinds of <?after > )
moritz I'm not too deep into it
masak .oO( say hello to the ~~ operator in regexes ) 09:29
moritz ah well, some possibly unimplementable builtins are harmless compared to some of the parsing stuff I've deleted from S02 and S03 09:30
masak .oO( Conjecture: we don't know how crazy the <*abc> syntax actually is, so some bits of it might be physically impossible )
moritz masak: <*abc> is on a sound theoretical foundation
masak aye. 09:31
moritz masak: it allows only a regular language, so you can simulate it as a DFA that accepts in every possible state
masak yes, I know.
moritz (but still carries on as far as it can)
so I don't expect much in the "physically impossible" department 09:32
masak right. the conjecture in question just says "could get expensive".
moritz std: /<*a||b>/ 09:33
p6eval std 1ad3292: OUTPUT«ok 00:00 41m␤»
moritz masak: I'm more concerned about it being expensive in terms of compiler writer hours :-) 09:34
masak luckily, the feature as such is not on a critical path of any kind.
it's more of a "nice syntactic sugar" thing.
moritz more like "nice sugar" 09:35
it's not as if you could easily get the same functionality if you were willing to use ugly syntax
masak you could. S05 gives examples.
moritz well, but that requires you to spell out the regex. Which is much harder if some part of it is user input. 09:36
masak &eval
:)
r: my $rx = ''; for 'xyz'.comb.reverse { $rx = "[$_ $rx]?" }; say $rx 09:39
p6eval rakudo 3bd91f: OUTPUT«[x [y [z ]?]?]?␤»
masak r: my $rx = ''; for 'xyz'.comb.reverse { $rx = $rx ?? "[$_ $rx]?" !! "$_?" }; say $rx
p6eval rakudo 3bd91f: OUTPUT«[x [y z?]?]?␤»
moritz ok, you have won :-)
masak \o/
moritz r: my $rx = ''; for 'xyz'.comb.reverse { $rx = $rx ?? "[$_ $rx]?" !! "$_?" }; say so 'xy' ~~ /<?{ $rx }/ 09:40
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤regex assertion not terminated by angle bracket at line 1, near "/"␤»
moritz r: my $rx = ''; for 'xyz'.comb.reverse { $rx = $rx ?? "[$_ $rx]?" !! "$_?" }; say so 'xy' ~~ /<?{ $rx }>/
p6eval rakudo 3bd91f: OUTPUT«True␤»
moritz r: my $rx = ''; for 'xyz'.comb.reverse { $rx = $rx ?? "[$_ $rx]?" !! "$_?" }; say so 'x' ~~ /<?{ $rx }>/
p6eval rakudo 3bd91f: OUTPUT«True␤»
moritz r: my $rx = ''; for 'xyz'.comb.reverse { $rx = $rx ?? "[$_ $rx]?" !! "$_?" }; say so 'xyza' ~~ /<?{ $rx }>/
p6eval rakudo 3bd91f: OUTPUT«True␤»
moritz r: my $rx = ''; for 'xyz'.comb.reverse { $rx = $rx ?? "[$_ $rx]?" !! "$_?" }; say so 'xyza' ~~ /^<?{ $rx }>$/
p6eval rakudo 3bd91f: OUTPUT«False␤»
moritz r: my $rx = ''; for 'xyz'.comb.reverse { $rx = $rx ?? "[$_ $rx]?" !! "$_?" }; say so 'xyz' ~~ /^<?{ $rx }>$/
p6eval rakudo 3bd91f: OUTPUT«False␤»
masak mischivous idea: we could prototype an implementation by just throwing in this kind of code :P 09:41
moritz I just thought about that
but unfortunately it's not that easy
masak dang. :/
well, could always prototype it in a module for the time being.
moritz consider <* [xyz | a ] b >
that shouldn't match xyb 09:42
but if you just recursively apply your transformation to all literals, it would match
masak right.
OH BTW. after thinking some more about how best to create a unity matrix in S09-enabled Perl 6. 09:48
09:49 TESTKEYz left
masak I decided I was on far too many high horses at the same time, grinding far too many theoretical axes. 09:49
moritz use matrix; my @m = unity(5); # for a 5x5 unity matrix 09:50
masak how about 'sub I($n) { my Num @matrix[$n][$n]; @matrix[$_][$_] = 1 for ^$n; return @matrix }'? :)
moritz: sure, that's the API. I'm talking about the implementation.
moritz CHECK failed: assignment to @matrix[...][...] will never work. Type constraint: Num. Got: Int 09:51
masak er, s/Num/Real/, perhaps.
I meant Numeric but wrote Num.
that seems to happen to me (and others) a lot. :)
if 'my Numeric @matrix[$n][$n]' works, then I'll take that.
hm. I just realized I wrote [$n][$n] instead of [$n;$n]. 09:52
does S09 explicitly support the former form?
moritz masak: that's kinda cute. But for "real" linear algebra applications you probably want some sparsish matrix types
masak sure.
but I consider that a representation, not a part of the algorithm. 09:53
moritz masak: in fact, all serious linear algebra libs that I know have separate (upper triangle, lower triangle, diagnoal matrix, sparse matrix, dense matrix) types
masak *nod*
and that's fine.
anyway, I realized I was out in APL-land or FP land looking for nice ways to create a unity matrix. 09:54
moritz well, the two are related. If you know that the matrix always stays diagonal, it gives you an easier API
masak but it should probably just be the above.
moritz use Matrix; my @a := diagonal(1 xx $n);
masak oh, troo.
as long as it upgrades seamlessly, that's probably the cutest implementation possible.
moritz++
sub unity($n) { diagonal 1 xx $n } 09:55
moritz seamlessly, yes. Not sure if it should implicitly though
though there should be coercions available 09:56
masak if the matrices are immutable, it's not much of a problem.
then we can Liskov freely and happily.
moritz after all, my @a = 1, 2, 3; coerces the RHS to array
likewise if we allow my @a as DenseMatrix; @a = unity(5) to coerce, we should be fine 09:57
which makes me think we maybe should have a MutableMatrix vs. ImmutableMatrix distinction like Array/List and Hash/EnumMap 09:58
09:58 brrt joined
masak still thinks the Array/List distinction will forever confuse newcomers, and that the name "EnumMap" is too narrow/wide 09:59
moritz knew that masak's next line would be about the "EnumMap" name
but then you also have to complain about Pair/Enum 10:00
:-)
masak maybe I will... :)
masak eyes his collection of sledgehammers
moritz is glad he's not the only one complaining 10:01
on this occasion I'd also like to note that the multiple uses of the smartmatch operator block us from having $str ~~ m:g/.../ working
(because if m:g/../ returns a list, it desugars to do { my $_ := $str; (m:g/.../).ACCEPTS($_) } 10:03
and sine List.ACCEPTS already has a perfectly valid meaning that doesn't work in this case, we have a problem)
s/sine/since/ 10:08
10:11 GlitchMr joined 10:17 mucker joined
moritz I'd like to try a constrained writing exercise in Perl 6 one day. Something along the lines of "write a short, text based game that doesn't use the letter 'e' in its source code" or so 10:24
maybe something less drastic for a start :-)
10:24 envi_ left
jnthn "e" is great because then your code can't just use eval(...decode stuff...) too :) 10:26
10:26 daxim joined
moritz it *can* use ::('&' ~ 'd'.succ ~ 'val').($stuff) :-) 10:29
jnthn right :)_
Can't use "." would be painful :)
moritz oh yes
jnthn Can't use parens would make people learn the precedence table :P
moritz I wonder how far you can get without curlies 10:30
all your blocks written as whatever-curried expressions... :-)
jnthn ooh :D
moritz without 'e' you can't declare methods 10:31
classes, attributes and subs would be fine though
jnthn You can add code objects via the MOP...but BEGIN and CHECK are out for doing it :P 10:32
moritz but 'constant' forces compile time evaluation too
jnthn Unless capital E is OK :)
ooh, yes!
moritz you see, we already get creative just by thinking about the problem, without even doing anything in it 10:33
jnthn nom: class A { constant = A.HOW."add_m{'d'.succ
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 2␤»
jnthn gah
nom: class A { constant = A.HOW."add_m{'d'.succ}thod"(A, 'foo', sub ($slf) { say 42 }); }; A.foo 10:34
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤invoke() not implemented in class 'Undef'␤»
10:34 fsergot joined
jnthn Hm :) 10:34
We may also find bugs :P
nom: class A { constant x = A.HOW."add_m{'d'.succ}thod"(A, 'foo', sub ($slf) { say 42 }); }; A.foo
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤invoke() not implemented in class 'Undef'␤»
jnthn wtf 10:35
jnthn gotta teach stuff, will leave masak to golf and file rakudobug :P
fsergot hi #perl6 o/ 10:36
tadzik hey fsergot 10:39
10:40 NamelessTee joined 10:43 bluescreen10 left 10:50 PZt left
gfldex masak: i believe the following is asking for your attention: rosettacode.org/wiki/Assertions#Perl_6 10:51
10:52 PZt joined 10:55 tarch left
masak gfldex: nah, that's a textual macro. :) 11:01
gfldex should it be? 11:02
masak also, the one who wrote that thinkgs that "non-hygienic" means "textual".
well, textual macros fall above AST macros on the power spectrum.
which means, as usual, that you should use them sparingly, only if you can't use the things below on the scale.
here I kinda feel an AST macro would've been nicer, actually. 11:03
but then, the specifics of how to pass an expression into an AST macro are not necessarily nailed down.
11:05 brrt left
gfldex so this is for later 11:06
masak I guess Pugs has textual macros. 11:07
pugs: macro foo($x) { "$x; $x" }; foo('say "OH HAI"') 11:08
p6eval pugs: OUTPUT«OH HAI␤OH HAI␤»
11:08 benabik left
masak pugs: say "runtime"; macro foo($x) { say "compile time"; "$x; $x" }; foo('say "OH HAI"') 11:08
p6eval pugs: OUTPUT«compile time␤runtime␤OH HAI␤OH HAI␤»
masak \o/
11:09 thelazydeveloper joined
tadzik do we have textual macros too? 11:10
masak "we" as in Rakudo? no. 11:11
tadzik yes, I meant Rakudo
masak you'll have to find someone crazier than me to implement textual macros. :)
tadzik :P
masak there's a fundamental impossibility in there somewhere. strings are immutable, and yet textual macros literally change the source code being parsed. 11:12
which is problematic, because all the cursors prior to the macro application refer to the old source code, and suddenly there's a new source code.
AST macros are much easier, because ASTs are meant to be transformed in varous fun ways. that's why we have a middle end. 11:13
11:15 spider-mario joined 11:18 lestrrat left 11:19 lestrrat joined 11:20 pernatiy joined
fsergot Which file is responsible for running perl6's scripts? 11:25
masak heh, Perl 6 discussion on #perl-help :)
fsergot In nqp, right? 11:26
Oh, found. :) 11:27
moritz masak: if we want to implement the :rw regex modifier at some point, we need to reference variables and not values in cursors and match objects 11:30
11:31 root____1 joined
masak yeah. 11:31
jnthn $!orig already does hold the original thingy fwiw 11:32
root____1 hi. i'm trying to read up on perl6.. and there's one thing that i can't really figure out: does Perl 6 support DSLs, similar to Ruby or something else? 11:35
11:35 root____1 is now known as poladark
moritz well, (nearly) every language supports DSLs in the ruby sense (that is, normal code looking cute if you give the subroutines and methods cute names) 11:36
Perl 6 also allows you to extend the syntax in lexical scopes, for example by introducing new operators 11:37
masak poladark: you can pass a block to a sub, and have that sub dynamically declare some stuff, such that your DSL only works inside the block.
poladark i really loved hacking Rebol and i got kind of addicted to DSLs there
moritz there's also support for macros, but it's not quite there yet, being worked on
jnthn rakudo: sub postfix:<!>($n) { [*] 1..$n }; say 10! 11:38
p6eval rakudo 3bd91f: OUTPUT«3628800␤»
poladark i *suppose* Rebol syntax is just a giant DSL engine
masak poladark: then you can write stuff like 'recipe { ingredients onion => 1, milk => 2dl; ... }'
moritz poladark: I don't know much about Rebol, but I do know that the term "DSL" is overloaded and highly ambigous 11:39
masak I think there's a useful core to the DSL concept. you can find it around Martin Fowler and his writings.
but there's a lot of hot air around that core.
poladark my experience learning perl regex properly convinced me of how having domain-specific languages for solving particular problems well has value 11:42
moritz yes, but regexes are a very different beast of DSLs as what ruby folks mean when they say "DSL" 11:43
arnsholt And what the Lispers mean when they say is a third thing, I think 11:44
moritz ie it's really a different language within a quoting construct
11:44 tokuhirom joined 11:45 snearch joined 11:46 bluescreen10 joined
masak moritz: regexes are an "external DSL", according to the terminology. recipe/ingredients using well-named subs is an "internal DSL". 11:46
I'm not keen on the terminology, mostly because it glorifies internal DSLs a bit much. but it seems pretty established. 11:47
moritz masak: is there a term for something inbetween? like, well-named subs plus custom operators?
masak don't think so. but in Perl 6 it's all the same :) 11:48
jnthn How about we call it, "using your language's features appropriately"? :)
masak in masak.org/carl/yapc-russia-2011-ath...g/talk.pdf I prefer the distinction "growing language inwards"/"growing language outwards".
so, anything nested in something else is a language that grows inwards into some other language. 11:49
11:49 NamelessTee left
masak whereas various prepropressing or compiling is growing a language outwards. 11:49
moritz
.oO( does that mean that SQL is cancer? )
11:49 simcop2387 left 11:50 simcop2387 joined
masak when you compare Perl's regex support (syntax-wise) to Python's or PHP's, you can see that Perl groks "inner languages" at a wholly different level than Python or PHP. 11:50
11:51 fgomez left
poladark that's what initially confused me about Ruby DSLs too. i thought "wow, i can make my own database query or text processing language like SQL or Regex now!" 11:51
turns out it wasn't as impressive as that
but it works reasonably well for some things like Puppet
tadzik do you know Dancer/Sinatra? 11:52
masak DSLs do have a place. they're not just toys. but it's prudent not to let the hype and enthusiasm carry the baby out on a tidal wave of bathwater. ;) 11:53
poladark i only heard of Sinatra. never used it. 11:54
masak I've landed on a view where I consider myself the most productive and happy when I create little "internal APIs" for myself inside of an application. basically build the syntactic abstractions I need, as I need them.
it's extremely liberating to keep in mind that the programming language basically lets you do *anything* to make the solution more easy to express.
poladark over the years, i've fallen in love with the way different programming techniques are good at certain problems. i'd really love it if there was one language that was flexible enough to contain the possibility of using all of them when coding 11:55
masak welcome to #perl6 :)
tadzik :)
poladark :O 11:56
tadzik I always thought Sinatra/Dancer/Bailador is pretty much what DSL is supposed to look like
masak it's a good example, I think, of a well-balanced internal DSL.
your domain is HTTP verbs. so you get them as subroutines. 11:57
11:57 benabik joined
poladark once in a while, i'm really pissed off because i think "i could do this in a row of code with Forth!" 12:00
but i'm not sure if a ruby-like DSL would allow me to create a Forth "dialect". 12:01
[Coke] I remember back in the day when a DSL was a standalone thing. I worked with a one off language to generate medical billing app/reports aeons ago. 12:05
masak well, you'd need a dedicated parser and interpreter for something like a Forth sublanguage. 12:06
but under the final vision of Perl 6, it should definitely be possible.
poladark :)
moritz well, Perl 6 does make parsing easy
tadzik aye 12:07
gfldex i'm quite sure it will work with a version prior to Final Perl 6
masak poladark: if you want to experiment with it in limited forms now, I recommend looking into parsing with grammars.
moritz though simple DSLs are already easy to parse. Like, splitting on newlines and blanks
masak p6: say "testing a simple Forth lexer.words.perl
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤Confused␤at /tmp/cVUxgd6vFk:1␤»
..niecza v15-6-gefda208: OUTPUT«===SORRY!===␤␤Unsupported use of bare 'say'; in Perl 6 please use .say if you meant $_, or use an explicit invocant or argument at /tmp/OmaKL9jVJj line 1:␤------> say⏏ "testing a simple Forth lexer.words.per␤␤Confused at…
..pugs: OUTPUT«*** ␤ Unexpected end of input␤ expecting block, "\\", variable name or "\""␤ at /tmp/1x5N5nh6h3 line 2, column 1␤»
masak p6: say "testing a simple Forth lexer".words.perl
p6eval niecza v15-6-gefda208: OUTPUT«["testing", "a", "simple", "Forth", "lexer"].list␤»
..rakudo 3bd91f: OUTPUT«("testing", "a", "simple", "Forth", "lexer").list␤»
..pugs: OUTPUT«*** No such method in class Str: "&words"␤ at /tmp/SnmpTj4GQN line 1, column 5 - line 2, column 1␤» 12:08
12:08 brrt joined
masak adds .words to the Pugs LHF TODO 12:08
tadzik hmm, there could be a bot for this
.LHF pugs .words
or such
random idea, maybe useless :) 12:09
moritz we've already discussed integrating p6eval with ticket submission and test creation
but I guess we were reaching too high
so not much has happened 12:10
tadzik probably the cost outweights the benefits
[Coke] open pugs issue #LHF Str.words
moritz tadzik: well, it was hard to find a user interface that was both powerful and easy to use 12:11
masak [Coke]++ # github.com/perl6/Pugs.hs/commit/25...8fe9912103 12:12
what a delightfully simple patch.
moritz there's some pretty neat DSL around that patch :-) 12:13
[Coke] masak: sadly, it got me that one test file, but not much else. (I need to ratched down autounfudge's pugs a little bit more, as a lot of the stuff we've skipped is hanging when we try to unfudge it.) 12:15
masak I think the "feeling" around DSL (and the reason for much of the hype) is "whoa, I can make my code look more like the solution space than like code!"
[Coke] there's probably some manually unfudging to do, though.
moritz masak: which is also the drive behind data driven programming 12:17
masak aye. 12:18
it's also some of the drive behind employing a top-down approach to coding. 12:19
12:19 envi_ joined
masak use API -> defined API -> implement underpinnings of API. 12:19
define*
repeat until you hit the metal :) 12:20
spider-mario I’ve read that Perl 6 was expected to have faster implementations than Perl 5 thanks to optional typing and all that, 12:23
do you think we’ll get there one day?
moritz maybe 12:24
rakudo already benefits from gradual typing
but it still has to make up lots of ground
it's probably too early to tell
spider-mario probably
still, the tendency is encouraging. :) 12:25
masak I think we'll land in a place where Rakudo (say) will be slower for most things, faster for some specific things, but fast enough for it not to matter because you'd generally want the things Perl 6 has over Perl 5.
flussence the numeric stuff is getting lots of optimisation lately but I find anything using regex to still be painfully slow :/
spider-mario rakudo’s startup time already has nothing to do with that of its 2012.02 star release. 12:26
12:26 user joined
tadzik in what way? 12:26
spider-mario a good way
a very good way
12:27 user left
tadzik perl6 -e 1 0.24s user 0.14s system 81% cpu 0.459 total 12:27
not bad indeed
dalek ast: a84fe9b | coke++ | S (3 files):
pugs unfudge (so)
tadzik used to be about 0.5 seconds iirc
spider-mario (sorry, some of my sentences may seem awkward)
(I’m not always comfortable with English)
tadzik: with * 2012.02, a hello world took 3 seconds to execute on my machine
now, about 0.3s 12:28
[Coke] \o/
spider-mario (maybe 0.5 actually)
(but still!)
[Coke] I assume that's due to jnthn++'s bs branch? 12:29
flussence cold start on a netbook: 0m1.593s
moritz flussence: and warm start?
spider-mario it seems so
masak [Coke]: aye.
tadzik spider-mario: oooh
oh, Star was before bs?
flussence warm start... 1.319. I thought it'd be faster
spider-mario Star, cold: 2,95s user 0,17s system 64% cpu 4,811 total 12:30
masak [Coke]: at GPW, I was demo-ing some Rakudo stuff in the space before my talk, and was surprised, live, at the improved startup time.
spider-mario Star, warm: 2,97s user 0,17s system 98% cpu 3,189 total
git, cold: 0,47s user 0,14s system 65% cpu 0,930 total
git, warm: 0,46s user 0,15s system 97% cpu 0,620 total
masak a six-fold speedup.
that's significant.
spider-mario it is, it makes rakudo much more usable
moritz now I just wish we could have have module loading 12:32
masak it makes it usable sooner :P
moritz s:2nd/have/fast/
which probably requires working precompilation
masak well, really fast compilation wouldn't be so bad either. 12:33
people will start to feel the pain for slow compiles as application sizes scale up. 12:34
s/for/of/
moritz aye
[Coke] p6: use Test; ok(any(1|0), "how many tests?")
p6eval niecza v15-6-gefda208: OUTPUT«ok 1 - how many tests?␤»
..rakudo 3bd91f: OUTPUT«===SORRY!===␤Twigil-Variable constants not yet implemented. Sorry. ␤at lib/Test.pm6:3␤»
..pugs: OUTPUT«pugs: *** Unsafe function 'use' called under safe mode␤ at /tmp/kAdBVcKtTv line 1, column 1␤»
moritz [Coke]: 1 12:35
[Coke]: test functions shouldn't autothread
masak did we spec that?
[Coke] moritz: was trying to demonstrate that pugs authothreads.
*autothreads
locally I get:
ok 1 - how many tests?
not ok 2 - how many tests?
masak we did spec that. 12:36
[Coke] masak: there's a pugsbug ferya.
12:36 fgomez joined
moritz masak: where? 12:36
masak submits pugsbug
moritz: github.com/perl6/specs/blob/master...esting.pod
[Coke] makes it somewhat hard to fudge tests properly. I end up skipping everything with ajunction.
masak lots of Mu in the signatures.
moritz masak: ah, right
it was even me who specced it :-) 12:37
12:38 tyatpi left
[Coke] masak: pugs might need Mu before that's possible. 12:38
moritz [Coke]: you can probably write Object until then 12:39
12:39 frank59 joined
masak frank59: hi! 12:39
12:41 frank59 left
masak bye. 12:44
12:44 _ilbot left 12:45 _ilbot joined, bluescreen10 left 12:46 bluescreen10 joined, mucker left 12:53 tyatpi joined 12:54 JimmyZ joined
masak a long detailed list of the mistakes of PHP: me.veekun.com/blog/2012/04/09/php-a...ad-design/ 12:54
tadzik I wonder why would one spend so much time writing down the downsides of a language
masak perhaps interesting for language designers. also slightly depressing reading. :)
tadzik I laughed at "foo" == TRUE, "foo" == 0, 0 != TRUE :)
flussence that one about long numeric strings is new to me... 12:55
(and scary) 12:56
masak tadzik: yeah, because having more than one equality operator would just confuse people ;)
brrt i work in php and it is bad enough without snobs like you pointing it out :-p
anyway 12:57
tadzik masak: yeah, better bury the fundamental logic :)
brrt what would it take for coercive type definitions to work in rakudo
tadzik Spock will not approve
(would?)
brrt they work?
12:58 havenn joined
brrt p6: my Num(Cool) $x = "2390"; say $x.WHAT; 12:58
p6eval pugs: OUTPUT«*** ␤ Unexpected ")"␤ expecting word character, "|", ":", "*" or parameter name␤ at /tmp/Y421SmPwiD line 1, column 12␤»
..niecza v15-6-gefda208: OUTPUT«Unhandled exception: Nominal type check failed for scalar store; got Str, needed Num or subtype␤ at /tmp/6mhhKddrhv line 0 (mainline @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3842 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3843 (…
..rakudo 3bd91f: OUTPUT«===SORRY!===␤Confused␤at /tmp/MTe0Kmsg9j:1␤»
brrt as per S02: my Num(Cool) $x = prompt "Gimme a number";
masak brrt: NYI. 12:59
rakudo: sub foo($x as Num) { $x }; say foo("2390").WHAT
p6eval rakudo 3bd91f: OUTPUT«Str()␤»
masak huh.
b: sub foo($x as Num) { $x }; say foo("2390").WHAT
p6eval b 922500: OUTPUT«Num()␤»
brrt masak: i reckoned, so what would it take to implement it? 13:00
masak jnthn or pmichaud :)
brrt :-) and if I where to offer my services, where should I look
masak seems to me it would affect both Grammar.pm and Actions.pm 13:04
you might want to dig up how Rakudo Beijing did its conversion in sub signatures.
...but I fear this might be a non-trivial patch, especially considering that the question "how first-class are these things?" never really got a satisfactory answer, if I recall correctly. :/ 13:05
i.e. can you declare an attribute to be 'Num(Cool)'? (probably yes). can you instantiate a parametric role with they type 'Num(Cool)'? (no frickin' idea) 13:06
the type*
13:08 tarch joined 13:10 tokuhirom left 13:11 tokuhirom joined
brrt hmm... okay 13:11
masak I want the 'Num(Cool)' syntax too. but it'll be non-trivial to get it in place, I think. 13:13
PerlJam that rant on PHP oddly leads me to somewhat of an opposite conclusion than the author. 13:14
(not that PHP isn't braindamaged, but that this is partly what makes it successful :)
masak: the *syntax* is easy, it's the semantics that are non-trivial :-) 13:15
13:15 tokuhirom left
jnthn Coercion types are non-trivial because they need to be first class 13:16
Which likely means CoercionHOW
And diddling how type checks are handled and stuff.
It goes deep...
masak yeah. we haven't explored how deep yet. 13:17
13:17 thelazydeveloper left
jnthn It mostly needs me to get some design tuits to explore it, but mostly those are tied up in other things. 13:17
masak maybe we should start by implementing it on a small scale -- for example, forbidding Role[Num(Cool)]
jnthn Like, QST
masak QAST!
\o/
jnthn *QAST
masak \o/ 13:18
jnthn masak: Why would be forbid that? :)
s/be/we
masak jnthn: to make it easier to implement a first cut of it.
jnthn masak: Your "make it easier" sounds like "do it innevitably wrong so we have to rip out a bad implementation later".
masak but maybe it'll all be so unified that it doesn't much matter...
jnthn And given the places that it'll touch, I really, really don't fancy that approach. :) 13:19
[Coke] jnthn: ... I'm confused. doesn't that describe the last 11 years pretty well?
[Coke] ducks.
jnthn [Coke]: :P
masak jnthn: no, I was just thinking there might be unintended consequences lurking, and we could shield ourselves from some of the corner cases by putting up limits. but feel free to ignore it if it doesn't sound like a good idea. :)
13:19 thelazydeveloper joined
[Coke] hugme: hug jnthn 13:20
hugme hugs jnthn
[Coke] hugme: kick jnthn
13:20 bluescreen10 left
PerlJam masak: seems like you'd want to find the unintended consequences right away so that you can fix the design 13:20
(which might mean "forbid this construct" :-)
[Coke] hugme: hug hugs 13:21
hugme hugs hugs
tadzik hugme: hug pugs
hugme hugs pugs; tadzik++
tadzik tee-hee
jnthn would be a little bothered if we had to forbid that case.
PerlJam in any case, finding them sooner means that we don't put in something that sort of works and then people rely on it before we find out it's really broken
jnthn Mostly because Str(Mu) is the default kash hey type 13:22
...hash key!
.oO( wtf... )
tadzik kash, hey
13:22 fsergot left
masak oh? not Str(Any)? 13:24
so junctions won't autothread in a hash lookup?
jnthn Oh, good question... 13:25
I think postcircumfix is where the autothreading happens, though.
masak p6: my %h = foo => 0, bar => 0, baz => 1; say "OH HAI" if %h{"foo" | "bar" | "baz"}; say "!"
p6eval pugs: OUTPUT«!␤»
..rakudo 3bd91f, niecza v15-6-gefda208: OUTPUT«OH HAI␤!␤»
masak oh, ok. 13:26
spider-mario p6: (1 .. 5)».say
p6eval pugs: OUTPUT«2␤3␤4␤5␤1␤»
..niecza v15-6-gefda208: OUTPUT«1␤2␤3␤4␤5␤»
..rakudo 3bd91f: OUTPUT«1␤4␤5␤3␤2␤»
brrt seriously, how does pugs come up with that? 13:27
masak it rotates it one step. 13:28
all of those are "correct", though. you don't get promised execution ordering with »
spider-mario p: (1 .. 10)».say 13:29
p6eval pugs: OUTPUT«1␤2␤3␤4␤5␤6␤8␤9␤10␤7␤»
spider-mario 7
:D
r: (1 .. 10)».say
p6eval rakudo 3bd91f: OUTPUT«2␤8␤6␤10␤5␤9␤3␤4␤1␤7␤»
arnsholt brrt: IIRC, pugs intentionally does it out of order to keep the end-user "honest"
That is, not rely on ordered execution =)
jnthn Rakudo the same, fwiw
PerlJam pugs++ rakudo++ 13:30
spider-mario rakudo really shuffles them
brrt i see
spider-mario (or does it actually do multithreading?)
[Coke] p: (1..1)».say
p6eval pugs: OUTPUT«1␤»
[Coke] #HA.
jnthn Rakudo doesn't multithread yet, no
arnsholt [Coke]: Nice one! ^_^ 13:31
brrt p6: (1..1)>>.say
p6eval pugs, rakudo 3bd91f, niecza v15-6-gefda208: OUTPUT«1␤»
13:32 birdwind1pbird joined 13:35 bluescreen10 joined 13:36 fsergot joined
colomon n: (1 .. 10)».say 13:38
p6eval niecza v15-6-gefda208: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤»
13:39 JimmyZ left
colomon :( 13:40
[Coke] colomon: pugs is starting to catch up with niecza. 13:46
colomon: up to 40%!
colomon pugs++
felher Hey folks: I have a segfault on my machine when doing 'echo -e "~Date.new\n" | nom' with latest nom. Is that known? (the same happens if i type it directly into the repl, fwiw) 13:47
[Coke] hey, I'm trying to troll you over here.
nom: ~Date.new
p6eval rakudo 3bd91f: OUTPUT«Cannot unbox a type object as a native int␤ in any <anon> at src/gen/BOOTSTRAP.pm:97␤ in sub sprintf at src/gen/CORE.setting:1949␤ in method Str at src/gen/CORE.setting:9453␤ in method Stringy at src/gen/CORE.setting:674␤ in sub prefix:<~> at src/gen/CORE.sett…
moritz doesn't get a segfault either 13:48
felher gets a segfault in the nom repl if he types "~Date.newn^D"
strange :) 13:49
gfldex no segfault here on cygwin
13:50 fgomez left
jnthn Can one of you that segfautls possibly gdb/bt it? 13:51
moritz oh right, I get one at REPL exit too
jnthn REPL...exit?!
moritz aye
in nqp_op_info_table 13:52
jnthn: gist.github.com/2351524
[Coke] some non sixian return value?
moritz I wonder if Date.new() should the same as Date.today() 13:53
or if it should die with "Which date do you want?" 13:54
jnthn That segfault is quite odd - it's not even dying in an op itself...
nom: Date.new('blind')
p6eval rakudo 3bd91f: OUTPUT«Invalid Date string; please use the format "yyyy-mm-dd"␤ in method new at src/gen/CORE.setting:9397␤ in block <anon> at /tmp/Spzu5eSx2P:1␤␤»
moritz (or just return some random date, just to mess with people a little)
jnthn aww
masak moritz: epoch date? :) 13:55
it's the zero of dates.
[Coke] TimToady's birthday.
masak :)
felher +1 for the random date solution :D
masak Mayan end-of-the-world date.
felher Oh, thats a good one too :)
moritz or maybe the birth year of Perl 6 13:56
masak founding of Republic of Austria.
moritz always the next Christmas :-)
brrt +1 for next christmas
flussence next christmas, unless the VM's version is >= 6.0, in which case return last christmas :)
masak "Last Christmas, I gave you Perl 6..." 13:57
jnthn "...and the very next day, you segfaulted it..."
[Coke] mmhehehe. 13:58
fsergot :D
[Coke] nice little Wham! reference.
13:59 liuchong left, PacoAir joined
brrt the day the bikeshed was built 14:00
gfldex n: my @a = 1..5; my @b = 'a'..'e'; my %h = @a Z=> @b;
p6eval niecza v15-6-gefda208: OUTPUT«Potential difficulties:␤ %h is declared but not used at /tmp/u3Ww_ItoJH line 1:␤------> my @a = 1..5; my @b = 'a'..'e'; my ⏏%h = @a Z=> @b;␤␤» 14:01
gfldex n: my @a = 1..5; my @b = 'a'..'e'; my %h = @a Z=> @b; say %h.perl;
p6eval niecza v15-6-gefda208: OUTPUT«{"1" => "a", "2" => "b", "3" => "c", "4" => "d", "5" => "e"}.hash␤»
masak "This year, to save me from tears, I'll just pass a continuation."
gfldex should that really be "1" ?
moritz brrt: did you mean: the day the /green/ bikeshed was built? :-)
masak I heard it was blue.
14:02 NamelessTee joined
masak gfldex: hashes are Str-keyed by default. 14:02
brrt basically, it should be green
i'm synesthic, e is a green letter
perl is a green word 14:03
its bikeshed is green
(which is really stupid, really, now i think of it)
moritz brrt: how does perl6.org look and sound for a synesthic?
brrt not green enough :-p 14:04
masak like a symphony of pastel vibrations.
brrt the pastel makes it doable, really 14:06
dalek kudo/nom: b8fdbea | moritz++ | src/core/Temporal.pm:
Date.new should not die with LTA error
14:07 havenn left
spider-mario “The functional parts of PHP are designed like C” 14:07
procedural*?
masak I thought PHP was all procedural.
maybe it's "functional" as in "it works"? 14:08
spider-mario the rest of the sentence is  “but the objectional (ho ho) parts are designed like Java”
14:08 mucker joined
brrt maybe the writer isn't natively english 14:08
moritz there's a part of the world that says "functional" when they mean they write and call functions.
masak yeah, some people like to distinguish procedural and OO, and call the former "functional". mostly non-FP people :) 14:10
gfldex PHP got cloasures that don't relly work. That might be the functional part.
masak to me, distinguishing procedural and OO in the first place is a bit of a stretch... :)
moritz masak: then you haven't written nearly enough QuickBasic 14:11
14:11 havenn joined
moritz or any other language without objects and nothing-like-objects 14:11
spider-mario gfldex : C doesn’t have it.
brrt masak: try squeak
masak I have written quite enough of QuickBasic. I know what you mean, but I still think OO is a subset of procedural, not distinct from it.
gfldex spider-mario: that might be the reason why then dont work :) 14:12
spider-mario :D
masak brrt: yeah, squeak looks attractive.
moritz uhm, to me OO is an addition to procedural
masak confirm segfault on Rakudo HEAD with '~Date.new' on REPL followed by Ctrl+D.
masak submits rakudobug
brrt the thing about OO... in theory, not in practice
spider-mario to me, it’s a parallel concept and can be applied to functional programming as well
cf. OCaml and Scala 14:13
brrt is that you have this 'network' of objects
moritz masak: HEAD~1
brrt and pushing one object causes some kind of 'wave' of messages
and that does work
and that works very well, but it requires a lot of world to set up
moritz masak: b8fdbeaa avoids the segfault, but it's no real fix
brrt because many people don't do that, their networks are simple
spider-mario oh, it reminds me of that saying that Erlang has great support for OOP. :d
it’s kind of true. 14:14
brrt and then, it really is procedural
masak moritz: ok, that goes in the ticket as well.
moritz: do you know a way to cause the segfault after that commit went in?
spider-mario erlang’s por
oops
erlang’s processes can be seen as objects
moritz # 14:15
masak brrt: I think I have never worked in an OO project that does that idealized network thing.
brrt neither
spider-mario erlang projects do
masak brrt: it tends to be just a little bit of runtype polymorphism here and there.
spider-mario :D
moritz masak: no :/
except by reverting it
masak moritz: any guess at the underlying cause?
moritz "programming error"
brrt masak: i know :-) 14:16
14:16 bluescreen10 left
brrt especially in scripting / web development, where big-world-setup is inpractical 14:16
masak the most interesting thing I heard about last year was CQRS. it wraps some ordinary things inside-out in an interesting way. 14:17
tadzik moritz: ooc, why 24-12 for Date.new()? 14:18
masak tadzik: that's Christmas in Germany. 14:19
tadzik: also, I've never seen that date format.
dd-MM
is it a Polish thing?
tadzik I don't think it's a format of any sort, but it's precisly the information I wanted to pass :) 14:20
moritz tadzik: feel free to change to the 25th if that is more christmassy to you
tadzik :) 24 is fine
masak tadzik: I found ISO 8601, and I never looked back. 2012-12-24.
to me, 25th is Christmas in the US and the UK, and 24th is Christmas in Sweden. 14:21
spider-mario I don’t understand the need for the special case 14:22
brrt 25th (and 26th) in the netherlands, although the 'church event' - the 'kerst mis' is on the 24th
spider-mario if I call it on 2012/12/26, I get a Date that represents 2013/12/24? 14:23
why?
oh, ok
sorry.
masak spider-mario: "next Christmas"
spider-mario I just got it. :D
brrt also, cqrs is interesting 14:24
spider-mario I’m wondering whether I should rebuild rakudo just to get that constructor.
my current rakudo is HEAD^ 14:25
(on nom)
14:30 bluescreen10 joined
masak I'd be thankful if someone could extract Date from src/core of HEAD~1, reproduce the error, and golf it down to something minimal. 14:30
my bet is it involves native attributes somehow.
fsergot r: my Str $s; $s.chars;
p6eval rakudo 3bd91f: OUTPUT«use of uninitialized value of type Str in string context␤»
fsergot It should print line number, shouldn't it?
masak this is your chance to catch a segfault!
14:31 tyatpi_ joined
fsergot ;:) 14:31
14:31 tyatpi left, Trashlord joined
spider-mario masak : paste.pocoo.org/show/zOfIYY9ZjS5rN1ITYTVm/ 14:31
14:32 sw joined
tadzik fsergot: I wish it would, yeah 14:32
14:33 sw left
masak spider-mario: yes, and that's now rt.perl.org/rt3/Ticket/Display.html?id=112376 14:33
fsergot perl6 test.p6
use of uninitialized value of type Str in string context
0
tadzik: it doesn't. :)
masak spider-mario: but... what is it in the Date class that causes the segfault? 14:34
14:34 REPLeffect left
tadzik yeah, I know. I have few of those warnings in Panda and I wasn't able to track them down 14:34
spider-mario oh
fsergot tadzik: is there any possibility to change this? :)
tadzik of course there is. Turing completeness, parallel universes and all this :P 14:35
frankly, I don't know
it seems to come from Mu.pm, which uses warn() 14:36
I don't see why warn() would not print a line number
r: warn "foobar"
p6eval rakudo 3bd91f: OUTPUT«foobar␤»
tadzik gah
masak arguably it should. 14:37
if you just want to print to $*STRERR, use 'note'.
tadzik sure
but I think precisely about the "use of uninitialized value" warning 14:38
14:38 birdwind1pbird left
fsergot Can I find somewhere the list of all built-in variables? :) 14:39
jnthn Note it's $*ERR 14:41
fsergot: iirc S28
masak oh right.
fsergot jnthn++ thanks 14:42
Maybe in this error message variable name should be printed? It'll let us track line down easier? :) 14:49
Or this is a bad idea? :)
is this*
tadzik I think it does that for some variables 14:50
fsergot It does it only if we use uninitialized variable, not value.
Or i'm wrong? :) 14:51
tadzik well, that's expected :)
see github.com/rakudo/rakudo/blob/nom/...Mu.pm#L137
it depends on nqp::iscont()
fsergot I see. :)
tadzik ...whatever it is 14:52
jnthn nom: my $a; say $a
p6eval rakudo 3bd91f: OUTPUT«Any()␤»
jnthn nom: my $a; say +$a
p6eval rakudo 3bd91f: OUTPUT«use of uninitialized variable $a of type Any in numeric context␤0␤»
masak niecza: warn "foo"
p6eval niecza v15-6-gefda208: OUTPUT«foo␤ at /home/p6eval/niecza/lib/CORE.setting line 1226 (warn @ 3) ␤ at /tmp/pzJMHodAjo line 1 (mainline @ 2) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3842 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3843 (module-CORE @ 65) ␤ at /hom…
fsergot I thought We can add printing variable name in value context too. 14:53
tadzik why doesn't warn print a line number anyway>
14:54 brrt left 14:55 sisar joined 14:57 brrt joined 14:59 thou joined
moritz because nobody has implemented it 15:01
15:01 poladark left
masak because the source code doesn't say to print a line number. 15:01
tadzik :) 15:02
I thought there's maybe a specced reason 15:03
masak nope. S32/Basics talks about warn, but not how its output is supposed to look. 15:06
tadzik I see 15:07
15:08 REPLeffect joined
tadzik r: gist.github.com/2352043 15:10
p6eval rakudo 3bd91f: OUTPUT«foobar at line 5␤»
moritz yes, but that's the wrong approach 15:11
it's rather that the code that catches the exception should introspect the backtrace
tadzik probably, yes 15:12
what catches warns? 15:13
moritz good question 15:16
r: return 15:17
p6eval rakudo 3bd91f: OUTPUT«Attempt to return outside of any Routine␤ in block <anon> at src/gen/CORE.setting:340␤ in block <anon> at /tmp/FBoK_sKATj:1␤␤»
jnthn It's configured in Exceptions.pm iirc
15:17 snearch left
moritz probably sub print_control 15:17
tadzik yeah 15:18
Exception.pm#102
tadzik is testing a patch 15:19
15:20 pomysl left
moritz fwiw the "proper" approach is to reuse method nice() from class Backtrace 15:20
needs a wee bit of refactoring, but I can do that
and then simply take the first line for warnings, and all lines for errors 15:21
15:21 pomysl joined, pomysl left, pomysl joined
tadzik want to do it? 15:22
dalek kudo/nom: 32c674e | moritz++ | src/core/Backtrace.pm:
Backtrace.nice can now produce single lines
 15:24
moritz tadzik: you can now use .nice(:oneline) to get the line to append to the warning 15:25
tadzik I shall, thanks moritz++
moritz r: sub f() { }; say &f.get_namespace.get_name 15:27
p6eval rakudo 3bd91f: OUTPUT«Method 'get_namespace' not found for invocant of class 'Sub'␤ in block <anon> at /tmp/J7A_vQcVxZ:1␤␤»
15:32 wolfman2000 left
moritz jnthn: the .get_namespace thing on the parrot subs that Perl 6 Code objects wrap just gives 'perl6' 15:33
jnthn sure
moritz jnthn: is there any way to get the Perl 6 level package name (for backtraces)?
jnthn we dont' use Parrot nanespaces
moritz: Not without storing it in the code objects. 15:34
Which is doable.
And easy. Edit BOOTSTRAP.pm and add an extra field at the end of Routine 15:35
Then in create_code_object if it's a Routine (already a conditonal for that in there somewhere) then nqp::bindattr that field to $*PACKAGE
Then a method in Routine to introspect it
moritz sounds really doable :-)
jnthn It should be easy. :) 15:36
15:36 havenn left
jnthn s/field/attribute/ # in .Net terminology mode today due to teaching :) 15:36
moritz it's OK, I understood you :-)
15:36 havenn joined 15:37 tyatpi_ left
moritz isn't quite monoglot 15:37
jnthn Well, the mistake is worse in the opposite direction 15:38
moritz what's an attribute in .Net?
jnthn (The term "attribute" means something very different in .Net, and I've confused people terribly.)
It's compile-time meta-data added to a declaration of soem kind.
Think traits, but less powerful :) 15:39
moritz :-)
masak those are called "annotations" in most languages, I think.
moritz r: { ... } 15:41
p6eval rakudo 3bd91f: ( no output )
jnthn masak: In Java and...which other ones? :) 15:42
moritz jnthn: I don't see a fleg that tells us if something is a routine 15:43
just $is_dispatcher
jnthn moritz: no no, it's not a flag, ti's the type
moritz: I think an if block in create_code_object looks for it. 15:44
moritz jnthn: how do I type-check this early in the bootstrap? .HOW.isa ?
jnthn if nqp::istype($code, $routine_type) { 15:45
lien 718
moritz ah, ok
thanks
jnthn sorry, I forgot create_code_object was broken out into parts now
moritz not a problem 15:46
I figured it must have been the bigger one :-)
jnthn Well, maybe a problem with my memory given I only did the refactor within the last month or so :) 15:47
moritz Argentinia might have flushed a few buffers :-9
15:47 Psyche^ joined 15:50 tyatpi_ joined 15:51 Patterner left, Psyche^ is now known as Patterner 15:52 vlixes joined
timotimo can i refer to a classes "things" with the :: syntax somehow? 15:54
masak you mean "Foo" and "Bar" in Foo::Bar? 15:55
TimToady only if you need indirect lookup of the name, then you can use ::(), but otherwise stuff is in the metaobject 15:56
unless explicitly declared 'our'
masak ah, I misunderstood the question. TimToady++
TimToady so I agree with masak++, what are you really asking?
moritz $ ./perl6 -e 'class A { sub f { }; say &f.package} '
A()
timotimo OK. i'm reading an article about php weirdnesses and one thing he mentions is that he knows of no language that has ClassName::something for static access
moritz \o/
PerlJam thought he was asking about attributes
timotimo i thought maybe perl6 has something vaguely similar
masak yes, it turns out A::foo is only visible if the sub foo in A was given 'our' scoping. 15:57
moritz timotimo: well, both Perl 5 and Perl 6 have $ClassName::OurVariable
which is somewhat similar, I assume
masak r: class A { our &x }; A::x = 42; say A::x
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤Preceding context expects a term, but found infix = instead at line 1, near " 42; say A"␤»
masak r: class A { our $x }; A::x = 42; say A::x
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤Preceding context expects a term, but found infix = instead at line 1, near " 42; say A"␤»
TimToady but then, technically, we wouldn't say the sub is in the class, but in the classs's package
moritz masak: A::x is a listop 15:58
masak oh, of course.
15:58 PacoAir left
masak r: class A { our $x }; $A::x = 42; say $A::x 15:58
p6eval rakudo 3bd91f: OUTPUT«42␤»
masak \o/
TimToady classes can have subs, but then they're called methods
masak what? no.
timotimo "Object attributes are $obj->foo, but class attributes are $obj::foo. I’m not aware of another language that does this or how it’s useful." - so it's like "our" variables in perl5 and 6? 15:59
masak classes can have subs, but then the classes are acting more like packages than like classes :)
jnthn The class doesn't really "have it" either, if you mean the meta-object :)
TimToady you are talking about the syntax, I'm talking about the semantics
masak r: class A { our sub foo { say "OH HAI" }; method foo { say "OH HAI again!" } }; A::foo; A.new.foo
p6eval rakudo 3bd91f: OUTPUT«OH HAI␤OH HAI again!␤»
masak TimToady: no, I'm talking about what hook the sub hangs itself on.
the A::foo thing is because classes are packages, and packages share their 'our'-y things. 16:00
TimToady well, that's what I said, it's in the class's package, not the class
the class's package isn't part of the class, from an OO point of view 16:01
masak ok. it sounded like you were saying that whenever a class has a sub, that's a method, which is wildly wrong.
TimToady I'm using the metaobject as the boundary of the actual class here
and when you poke a sub into it, it's a method
masak but yeah, the class's "class" aspect can't have subs in that sense, I guess. only the class's "package" aspect can.
and if you don't 'our' the sub, it's the class's "block" aspect that gets the sub :) 16:02
TimToady methods are just subs that are "mislocated" from some other scope into the metaobject
16:02 birdwindupbird left
jnthn -> hotel, bbl 16:02
tadzik hm, Backtrace.new($ex) segfaults Rakudo for some reason
PerlJam masak: is that aspect oriented programming? ;)
16:03 brrt left
masak PerlJam: more like role deconstruction of OO in Perl 6 ;) 16:03
16:03 Ingis joined
masak => grocery_store => home 16:05
16:06 IngisKahn left 16:07 Ting_ joined
Ting_ ls 16:07
sjohnson ls -l
Ting_ ls -lh
16:07 havenn left
PerlJam total 0 16:07
tadzik :P 16:08
16:09 Ting_ left
tadzik clearly he didn't find what he was looking for 16:10
dalek kudo/nom: edfd0f2 | moritz++ | src/Perl6/World.pm:
[World] avoid a lookup for which we already have a variable
 16:12
kudo/nom: 8144a04 | moritz++ | src/ (3 files):
implement Routine.package
16:12 plobsing left
TimToady moritz: under the OKness proposal, irclog.perlgeek.de/perl6/2012-04-10#i_5421023 would return something like OK[list], which would not rerun the .ACCEPTS since the OK bit is already considered success; ironing out smartmatching is the main point of that proposal, but people keep getting distracted by the less important bits of the proposal 16:13
16:14 plobsing joined
tadzik gist.github.com/2352529 makes rakudo segfault, if anyone wants to take a look 16:15
16:15 envi_ left
moritz tadzik: probably related to $ex being a parrot exception, not a Perl 6 type 16:16
tadzik oh, must be, yes
moritz tadzik: try EXCEPTION($ex) instead
tadzik yeah, everything uses nqp::atkey
tadzik tries 16:17
moritz or even more efficient
spider-mario moritz: since $rtype is just an alias for $routine_type and is only used in the line below its definition, we might as well get rid of it, might we not?
(or am I missing something?)
moritz Bactrace.new($ex.backtrace, 2) 16:18
spider-mario (I’m talking about github.com/rakudo/rakudo/commit/edfd0f216f )
16:18 envi_ joined
tadzik well, this way we miss the whole point of reusing nice(), no? 16:18
moritz or maybe Bactrace.new($ex.backtrace, 0).nice 16:19
or maybe Bactrace.new($ex.backtrace, 0).nice(:oneline)
thing is, if you wrap the parrot exception in an Exception, then the first thing that Backtrace.new does is unwrap it to get at its .backtrace
tadzik right 16:20
moritz spider-mario: you're right
16:20 nwc10 left
moritz hey, people actually read my patches. 16:20
that's both awesome and scary :-) 16:21
spider-mario :D
dalek kudo/nom: 72e8fe8 | moritz++ | src/Perl6/World.pm:
[World] further cleanup by spider-mario++
16:23 jaffa4 joined 16:26 envi_ left 16:27 envi_ joined 16:34 envi_ left 16:38 havenn joined 16:43 envi_ joined 16:44 s1n left 16:54 xinming left 16:55 xinming joined 16:56 apejens left 16:57 fsergot left
moritz $ ./perl6 -e 'sub f() { warn "foo" }; f' 16:57
foo in sub f at -e:1
16:58 s1n joined
masak home 16:59
17:00 pernatiy left
moritz $ ./perl6 -e 'sub f() { +Any }; f' 17:00
use of uninitialized value of type Any in numeric context in method Numeric at src/gen/CORE.setting:650
17:00 sftp joined
moritz these are not the line numbers you are looking for :/ 17:00
jaffa4 what is perl6? 17:03
17:03 s1n left
TimToady perl6 is what. 17:03
jaffa4 WHat is executed under the hood of perl6? 17:04
moritz rakudo
jaffa4 perl6:say <1 2 3>
PerlJam niecza
moritz pugs
jaffa4 perl6: say <1 2 3>
p6eval pugs: OUTPUT«123␤»
..rakudo 3bd91f, niecza v15-6-gefda208: OUTPUT«1 2 3␤»
jaffa4 all of them 17:05
TimToady but not when someone sez: $ ./perl6 -e ...
which is why moritz++ said rakudo
17:05 PacoAir joined
benabik ponders installing a perl6 script that calls all three. 17:06
TimToady of course, ./perl6 depends on which directory you're in
masak jaffa4: if you look far enough down, it's all just electrons in circuits executing. but that can't be what you're asking. what is your question -- I mean, what kind of answer do you expect to get?
jaffa4 that answer you happened to give, taht kind of answer.
benabik Rakudo is the only project that I know being arrogant enough of building an executable simply named perl6. ;-) 17:07
17:07 fsergot joined
PerlJam benabik: hubris is one of the 3 virtues 17:07
TimToady well, Perl builds one called "perl" :)
masak jaffa4: much of Rakudo and Niecza runs Perl 6 under the hood.
TimToady but I guess it's not arrogant enough to build one called "perl6" either... 17:08
moritz benabik: not arrogance; hysterical raisins
TimToady Rakudo and Niecza runs much of Perl 6 under the hood.
benabik moritz: Stupid raisins. Have to pick them out of food all the time, now they're in my compilers? ;-)
PerlJam
.oO( Where is this "hood" anyway? )
masak I don't get the hangup people have with the name of the 'perl6' executable. there are several implementations of Ruby. at least one has an executable called 'ruby'. people expect this. 17:09
benabik: well, they *are* hysterical, so no wonder they get in everywhere.
TimToady I think when there are multiple of a particular program in your path, it should just pick one randomly and run it
masak PerlJam: the "hood" is the part of the executable that keeps memory from leaking out into the rest of the system. 17:10
it's a technical term.
TimToady I thought that was the bonnet...
masak no, the bonnet keeps away the bees.
moritz which botnet?
TimToady otnet, oviously 17:11
moritz 17:12
mm
17:15 fsergot left
masak twitter.com/sixthformpoet/status/18...0331586562 -- autopun, or just regular pun? or both? 17:17
17:17 s1n joined
PerlJam heh 17:17
jnthn v hotele :)
pmichaud commit b8fdbea (Date.new) is cute :) 17:18
jnthn pmichaud! \o/
masak this one definitely is an autopun, though: twitter.com/sixthformpoet/status/18...8033349634
pmichaud good afternoon, #perl6
17:19 sisar left 17:24 s1n left
pmichaud reads backscroll 17:25
(php language rant): I read all of that and the author doesn't really have a conclusion? Sheesh. 17:26
17:27 Chillance joined 17:28 fsergot joined
jnthn
.oO( was pmichaud's comment a meta-rant? )
 17:34
pmichaud Yes. 17:35
I wonder if Backtrace.nice(:$oneline) should instead have an argument indicating the number of lines wanted 17:38
instead of choosing between "1" and "all" 17:39
17:41 s1n joined
jnthn I can imagine cases where something could only want the first n lines. 17:42
PerlJam making a parameter for the number of lines is less special-casey and results in a net increase in functionality 17:45
17:45 s1n left
mucker TimToady: Curious doubt. Allowing ' as a legal character in a variable name, was that a linguistic decission ? -- $foo's_bar indicating possesion 17:46
17:46 sisar joined
pmichaud mucker: also things like $L'Hopital 17:47
mucker :)
pmichaud or even $L'Hôpital
daxim well, isn::t that cute… 17:48
PerlJam mucker: the answer to your question is definitely "yes" :) 17:49
17:49 fhelmberger left
spider-mario btw (regarding the “perl6” executable name), 17:49
S01 says that Perl 6 code should include a v6 header,
what is the preferred one?
I use #!/usr/bin/env perl6
currently.
pmichaud "use v6;" is the canonical one, I think. 17:50
spider-mario ok, thanks.
17:50 brrt joined
spider-mario it does have the advantage of producing a useful error message in perl 5. 17:51
TimToady mucker: yes, it was primarily a decision from traditional word-based orthography 17:53
PerlJam at least until someone breaks it with a v6 module on CPAN
mucker TimToady: love that feature. I find my self using lots of variables names implying possession of some sort. 17:54
TimToady unfortunately it won't allaw things like 'twas
*allow
17:54 lichtkind joined
TimToady or Charles' 17:55
but then, I always write Charles's anyway
TimToady wonders if there are any 0-width letters in Unicode... 17:56
spider-mario ·
oops
 
ah, letter?
17:57 envi_ left
PerlJam my $_'twas-the-night-before-Christmas; _ is almost invisible sometimes 17:57
spider-mario (I’m not sure “ ” is actually supposed to have a null width but it does on my machine)
jnthn .u  
phenny U+202F NARROW NO-BREAK SPACE ( )
spider-mario we french-speaking people call it « espace insécable fine » 17:58
17:58 envi_ joined
TimToady perl6: my $ʹtwas = 'the night before Christmas'; say $ʹtwas 17:59
p6eval pugs, rakudo 3bd91f, niecza v15-6-gefda208: OUTPUT«the night before Christmas␤»
PerlJam well then
TimToady perl6: constant ʹtwas = 42; say ʹtwas
p6eval pugs: OUTPUT«*** ␤ Unexpected "="␤ expecting ":" or "("␤ at /tmp/zQ9wrugMqh line 1, column 16␤»
..rakudo 3bd91f, niecza v15-6-gefda208: OUTPUT«42␤»
18:01 s1n joined
mucker The spec says atleast one alphabet must be preceding ' :) perlcabal.org/syn/S02.html#Apostrophe_separator 18:01
TimToady yes, but I wasn't using ', but ʹ 18:02
.u ʹ
phenny U+02B9 MODIFIER LETTER PRIME (ʹ)
spider-mario .u ’
phenny U+2019 RIGHT SINGLE QUOTATION MARK (’)
TimToady which is officially a letter
spider-mario r: constant ’ = 42; say ’
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤Missing initializer on constant declaration␤at /tmp/C3EJ1bsyeF:1␤»
spider-mario r: constant ’x = 42; say ’x
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤Missing initializer on constant declaration␤at /tmp/LxlR5r1QiZ:1␤»
mucker oo .. okay
TimToady r: constant ʹ = 42; say ʹ 18:03
p6eval rakudo 3bd91f: OUTPUT«42␤»
TimToady n: my \ʹ = 42; say ʹ
p6eval niecza v15-6-gefda208: OUTPUT«42␤»
TimToady r: my \ʹ = 42; say ʹ
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤Malformed my␤at /tmp/uJSI_Smad9:1␤»
TimToady r: my \a = 42; say a 18:04
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤Malformed my␤at /tmp/rC2cylvC8G:1␤»
jnthn Ah, i still need to add that syntax.
TimToady r: my (\a) = 42; say a
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤Malformed parameter␤at /tmp/vka_kHZy2x:1␤»
jnthn ETOOMUCHTODO
pmichaud TimToady++ seems to use it often :)
TimToady I want a story for people who have an irrational dislike for sigils 18:05
18:05 vlixes left
pmichaud TimToady: you mean a story other than "Your dislike of sigils is irrational?" ;-) 18:05
TimToady a TMTOWTDIish story
PerlJam the story seems to be "your dislike of sigils is well-founded" so far ;) 18:06
pmichaud r: say <$ @ % &> ~~ Rational
p6eval rakudo 3bd91f: OUTPUT«False␤»
spider-mario :D
pmichaud r: say <$ @ % &>.any ~~ Rational
p6eval rakudo 3bd91f: OUTPUT«False␤»
TimToady though the function in rosettacode.org/wiki/Runge-Kutta_method#Perl_6 reads rather nicely with the sigilless forms 18:07
mikemol discovers he doesn't have whatever font is necessary for 𝛿 18:08
TimToady .u 1D6FF 18:10
phenny TimToady: Sorry, no results for '1D6FF'.
TimToady .u U+1D6FF 18:11
phenny TimToady: Sorry, no results
TimToady hmm
daxim www.fileformat.info/info/unicode/ch...upport.htm
out of that list, I recommend the Code200x tribe of fonts. 18:12
TimToady well, firefox seems to get it fine on my machine
18:12 birdwindupbird joined
TimToady part of the reason that function works well though is that the italic delta functions visually as a mock sigil 18:13
which is kind of how mathematicians use them
18:13 plobsing left
TimToady n: my \dy₃ = 42; 18:17
p6eval niecza v15-6-gefda208: OUTPUT«===SORRY!===␤␤Term definition requires an initializer at /tmp/lERpq5rHLH line 1:␤------> my \dy⏏₃ = 42;␤␤Confused at /tmp/lERpq5rHLH line 1:␤------> my \dy⏏₃ = 42;␤␤Parse failed␤␤»…
TimToady but I couldn't get it to work with subscripts, for some reason that is probably my own fault...
mikemol FWIW, my setup atm is Chrome on Win7. 18:18
18:19 envi_ left
TimToady n: say so '₃' ~~ /\w/ 18:19
p6eval niecza v15-6-gefda208: OUTPUT«False␤»
TimToady that's the problem
eval: say 'troo' if '₃' =~ /\w/ 18:20
buubot_backup TimToady: ERROR: syntax error at (eval 20) line 1, near "say 'troo'"
TimToady eval: print 'troo' if '₃' =~ /\w/
buubot_backup TimToady: No output.
18:21 envi_ joined
TimToady hmm 18:21
eval: use utf8; print 'troo' if '₃' =~ /\w/
buubot_backup TimToady: No output.
TimToady eval: use utf8; print 'troo' if '₃' =~ /\w/u
buubot_backup TimToady: No output.
lichtkind TimToady: i just dont know how to describe $*COMPILING and $*DEBUGGING
PerlJam would have thought subscripts would be used with some combining sequence to "attach" them to letters and thus they'd match w 18:22
TimToady those are placeholders for Perl 5 varialbes, and we probably have no idea what they actually do in Perl 6
funny, subscript three matches \w in my local p5
eval: use utf8; print ord '₃' 18:23
buubot_backup TimToady: 83231
TimToady well, if they've gone and redefined superscripts and subscripts, maybe we should consider whether they should be allowed specially at the ends of identifiers 18:28
lichtkind allright so i dont add them in the index 18:29
18:29 lichtkind left
TimToady (perhaps along with the real prime character, not the fake modifier letter) 18:29
.u ′″‴⁗ 18:30
phenny TimToady: U+2032 U+2033 U+2034 U+2057
sorear good * #perl6
TimToady howdy doody
TimToady still thinks it's a bit odd that super/subscript chars can't match w though 18:32
18:33 LylePerl left
TimToady now someone else should point out that y₁ should mean y[0], and someone else should point out that y₀ should mean that... 18:34
18:36 pernatiy joined, sisar left
au neither ⑦ nor ⒎ nor ㍟ match \w, either... seems a character is only considered a digit when it's denoting a number. 18:38
spider-mario r: '⑦' ~~ /\d/ 18:39
p6eval rakudo 3bd91f: ( no output )
spider-mario r: say '⑦' ~~ /\d/
p6eval rakudo 3bd91f: OUTPUT«#<failed match>␤»
18:40 lichtkind joined
spider-mario r: '⑦'.perl.say 18:40
p6eval rakudo 3bd91f: OUTPUT«"⑦"␤»
sorear .u ⑦
phenny U+2466 CIRCLED DIGIT SEVEN (⑦)
sorear .u ₁₀
phenny U+2081 SUBSCRIPT ONE (₁)
U+2080 SUBSCRIPT ZERO (₀)
spider-mario r: '⑦'.comb.perl.say
p6eval rakudo 3bd91f: OUTPUT«("⑦",).list␤»
spider-mario good. :)
r: '⑦'.chars.say 18:41
p6eval rakudo 3bd91f: OUTPUT«1␤»
sorear au!! long time no see
au indeed! how's things?
18:41 envi_ left 18:43 daxim left 18:44 GlitchMr left
masak au! \o/ 18:45
18:45 fglock left
au /o 18:45
sorear au: ok
au: lack of tuits, though :| 18:46
18:46 envi_ joined
au *nod* 18:46
18:47 s1n left
au . o O ( The good thing about reinventing tuits is that you can get a round one ) 18:48
18:49 estrabd left 19:02 s1n joined
prammer anyone interested in presenting at strange loop? thestrangeloop.com/sessions-page/c...sentations 19:02
pmichaud I'm always interested in presenting. :) 19:06
prammer I'd love to see more perl (5 or 6) at strange loop. I'm biased though since St. Louis is local for me. 19:09
19:09 mucker left
pmichaud I think I'll submit a talk. 19:09
Not sure what I'll talk about yet -- but I'll look at last year's program and figure out something useful
Perhaps a talk about Perl 6 examples from rosettacode.org 19:10
prammer Anything perl 6 related I suspect would be a good fit.
pmichaud looks like the conference is code-heavy
mikemol I'll put together an updated local mirror copy for whoever needs it. 19:11
mikemol is getting used to that.
pmichaud mikemol++
prammer Yes. A lot of JVM languages, typically.
Matt Follett gave a Perl 6 talk, in 2010 iirc
brrt wonders whether he is the only person who doesn't like jquery 19:12
mikemol brrt: No. But my reasons aren't syntactic or semantic.
brrt interesting conference though 19:13
masak brrt: well, the thing about jQuery is that it shouldn't have been necessary. JavaScript's role has changed a lot over the years.
19:14 plobsing joined
brrt .. i guess. i just see such messes made with jquery 19:14
19:17 fglock_ left
mikemol I don't care for it, because I once frequented a Wordpress-based blog that required over forty page requests to load, and at least six of them were different versions of jQuery... 19:19
masak r: macro redirect($code) { quasi { sub say(*@args) { say "Something else" }; {{{$code}}}(); } }; redirect { say "Hello world"; } # no, {{{}}} is not implemented yet. just gonna try and golf this NPMCA...
p6eval rakudo 3bd91f: OUTPUT«Null PMC access in get_pmc_keyed_str()␤ in block <anon> at /tmp/Mce8a2QGgP:1␤␤»
mikemol It's like a static-linking problem with symbol collisions.
masak r: macro r($c) { quasi { sub say(*@) {}; {{{$c}}}(); } }; r { say "!" } 19:20
p6eval rakudo 3bd91f: OUTPUT«Null PMC access in get_pmc_keyed_str()␤ in block <anon> at /tmp/fWt3K5gDaR:1␤␤»
masak r: macro r($c) { quasi { sub say(*@) {}; {{{$c}}}(); } }; r 42
p6eval rakudo 3bd91f: OUTPUT«Null PMC access in get_pmc_keyed_str()␤ in block <anon> at /tmp/uaXbPAiZLP:1␤␤»
masak r: macro r($c) { quasi { sub say(*@) {}; {$c}(); } }; r 42
p6eval rakudo 3bd91f: OUTPUT«Null PMC access in get_pmc_keyed_str()␤ in block <anon> at /tmp/5p668VUDIs:1␤␤»
masak r: macro r($c) { quasi { sub say(*@) {}; {}(); } }; r 42 19:21
p6eval rakudo 3bd91f: OUTPUT«Method 'postcircumfix:<( )>' not found for invocant of class 'Hash'␤ in <anon> at src/gen/BOOTSTRAP.pm:812␤ in any <anon> at src/gen/BOOTSTRAP.pm:808␤ in <anon> at /tmp/AGNxfdDvIJ:1␤ in block <anon> at /tmp/AGNxfdDvIJ:1␤␤»
masak r: macro r($c) { quasi { sub say(*@) {}; {;}(); } }; r 42
p6eval rakudo 3bd91f: OUTPUT«Null PMC access in get_pmc_keyed_str()␤ in block <anon> at /tmp/QoErNGV9SU:1␤␤»
masak r: macro r { quasi { {;}() } }; r
p6eval rakudo 3bd91f: OUTPUT«Null PMC access in get_pmc_keyed_str()␤ in block <anon> at /tmp/SC_GjpL3kR:1␤␤»
masak r: macro r { quasi { Nil } }; r
p6eval rakudo 3bd91f: ( no output )
lichtkind TimToady: was it right to delet $*ARCH and should i add $*XKERNEL?
masak inneresting.
r: say {;}()
p6eval rakudo 3bd91f: OUTPUT«Nil␤»
masak r: say Nil.WHICH; say Nil.WHICH
p6eval rakudo 3bd91f: OUTPUT«Nil␤Nil␤»
masak r: say {;}().WHICH; say Nil.WHICH 19:22
p6eval rakudo 3bd91f: OUTPUT«Nil␤Nil␤»
lichtkind jnthn: i have real problems 19:31
even if adding simple things i got runtime hackles
masak lichtkind: what did you write, what result did you get, and what were your expectations? 19:35
be precise.
lichtkind i just added my $KERNEL = 4; and i get runtime error evenif my program is ";"
masak added... to what? 19:36
anyone else seeing the 12th test (which is TODO'd) in t/spec/S32-list/uniq.rakudo pass? 19:37
dalek ast: 110d111 | masak++ | S06-macros/macros-d1.t:
[S06-macros/macros-d1.t] added Null PMC access test

discovered by samlh++		 19:44
19:45 brrt left
[Coke] brrt: I love jquery. but I'm comparing my love to raw javascript. 19:49
(messes with jquery) ... that sound slike the common argument against p5. ;)
r: my $KERNEL = 4; 19:50
p6eval rakudo 3bd91f: ( no output )
[Coke] nope, WMF.
er, WFM.
tadzik masak: yes 19:52
masak so, it turns out the condition in github.com/rakudo/rakudo/blob/nom/...s.pm#L3080 is never true. 19:53
it looks reasonable. any idea why it doesn't work as advertized? 19:54
[Coke] that is a very long file
19:54 colomon left
[Coke] are you sure that $nil_class is actually finding that symbol? 19:55
masak no.
not sure that's the proper check either. 19:56
pmichaud does find_symbol find any currently in-scope symbol, or only those that are in the bootstrap? 19:58
I'm guessing it finds any currently in-scope symbol 20:00
masak hm, sometimes it's :scope('lexical') and sometimes :scope('lexical_6model')
PerlJam What's $*W ? 20:01
masak the world.
pmichaud I'm guessing it's "world" or somesuch
PerlJam ah
[Coke] .u 20DD
phenny U+20DD COMBINING ENCLOSING CIRCLE (⃝)
flussence
.oO( doesn't look all that enclosing to me... )
pmichaud looks like the circles that I find in the windmills of my.... encoding
[Coke] enofont?
pmichaud I see the circle okay in my font, but it's offset to the left of the parens 20:02
[Coke] was trying to figure out how to get a unicode checkmark in a combining circle.
I see, basically: []()
flussence .u 2713 20dd
phenny flussence: Sorry, no results for '2713 20dd'.
flussence bah
r: say "\x[2713]\x[20dd]"
p6eval rakudo 3bd91f: OUTPUT«✓⃝␤»
flussence :/
it's enclosing the wrong thing...
flussence decides urxvt just sucks at unicode 20:03
tadzik wfm
felher Hey folks. Suppose i have a class A::B::MyClass and a class A::B::SecondClass. Is it possible to create a module A::B::ALL that exports those classes? So that i could write something like: "use A::B::ALL; SecondClass.new()".
PerlJam masak: I'd think they should all be lexical_6model (unless there are really 2 Nils, but that makes my brain hurt)
flussence it displays around the « on my screen, maybe there's a font mismatch there... oh well, not that important 20:04
au .u 24E5
phenny U+24E5 CIRCLED LATIN SMALL LETTER V (ⓥ)
flussence felher: I had that exact problem a while back. It's specced (use :EXPORT) but it doesn't work in anything
masak all of mine are lexical_6model, but that code is never reached.
pmichaud afaik, there's only one Nil 20:05
src/core/Nil.pm
tadzik moritz: (warnings), yeah, same results here; seems that it's all trickier than it seems to be. Or maybe just 'is hidden-from-backtrace' for all those subs will do?
.Numeric etc
20:05 s1n left
felher flussence++: thanks for the answer :) 20:05
masak let's see. Nil isn't mentioned in BOOTSTRAP.pm. 20:07
pmichaud right, it's just a normal type object 20:08
masak Actions.pm only mentions it as PAST nodes ordinarily. I might be the only one in there trying to fetch it from a lexical environment.
pmichaud it is a lexical class
my class Nil is Iterator {
masak right, but isn't src/core "too late" for Actions.pm? 20:09
I mean, Actions.pm is nqp...
and src/core is Perl 6...
masak tries to make the levels fit together in his head
src/core isn't even compiled at the time Actions.pm is being compiled! 20:10
pmichaud no, but it is at the time when Actions.pm is being run
(except when compiling src/core itself)
20:11 havenn left
masak so Actions.pm can pull type names from src/core? just checking. 20:11
r: say Nil ~~ Nil 20:12
p6eval rakudo 3bd91f: OUTPUT«True␤»
masak r: say {;}() ~~ Nil
p6eval rakudo 3bd91f: OUTPUT«True␤»
PerlJam asserts that both PAST::Var.new(:name('Nil'), :scope('lexical_6model')) and PAST::Var.new( :name('Nil'), :scope('lexical') ) can not both be simultaneously correct
masak r: say nqp::istype(Nil, Nil)
p6eval rakudo 3bd91f: OUTPUT«Method 'gist' not found for invocant of class 'Integer'␤ in sub say at src/gen/CORE.setting:6258␤ in block <anon> at /tmp/vZLARDUYVe:1␤␤»
pmichaud r: say nqp::p6box_i(nqp::istype(Nil, Nil))
p6eval rakudo 3bd91f: OUTPUT«1␤»
masak ok, so outside of Actions.pm it works... 20:13
masak drops it for now 20:14
pmichaud PerlJam: I don't know what's up with the :scope('lexical') in Actions.pm 20:15
I wonder if it's fossil
PerlJam I'd wager so
TimToady
.oO(there is no bootstrapping problem in computer science that cannot be solved by adding another level of bootstrap...)
 20:16
pmichaud I find 13 instances of :scope('lexical') or :scope<lexical> in src/Perl6 20:17
PerlJam
.oO(... or by being crushed under the heels of the boot that was strapped)
pmichaud it's entirely possible that :scope('lexical') works for lexical lookups in common cases 20:18
although it might not be as efficient as 'lexical_6model' or might fail in certain instances
pmichaud tries s/lexical/lexical_6model/ to see what happens 20:19
tadzik loliblogged! ttjjss.wordpress.com/2012/04/10/pla...n-in-oslo/ 20:20
20:21 s1n joined
masak \o/ tadzik++ 20:22
20:22 havenn joined
masak tadzik: if you put energy at bringing Hitomi back to relevance, I'm willing to chip in with additional tuits. 20:22
PerlJam tadzik++
pmichaud lol! "Whatever but Cool" 20:23
masak I have a vague feeling my main focus at the Oslo hackathon will be QAST/macros, though.
tadzik masak: that sounds like a plan
yet I really want to refactor Bailador, before or during, not after
(for the reasons I stated)
masak pmichaud: yeah, tadzik got the best blog name. someone inadvertently suggested it to him :P
tadzik shush!
PerlJam tadzik: is bailador a dancer port or really just a dancer-a-like for perl 6 heavily inspired by dancer?
masak tadzik: also, remember our highly secret but very exciting project. I've already done a few tentative commits on it. 20:24
tadzik PerlJam: it's really a Dancer lookalike, but I've started digging into Dancer2 source code recently and been trying to make it something closer to a real Dancer port
arnsholt tadzik: Sounds cool. I've been pondering various things along similar lines as well, so starting to really look forwards to the Hackathon now!
tadzik masak: sure thing
pmichaud I haven't decided what I'll focus on in Oslo. I was thinking of working a bit on the Perl 6 module universe
jaffa4 What is Dancer 2?
tadzik Dancer 2 is a rewrite of Dancer :) 20:25
jaffa4 and Dancer?
pmichaud I was also thinking of looking at zavolaj-like things a bit
tadzik Dancer is perldancer.org/ a nice web framework for Perl
PerlJam jaffa4: Dancer is a web framework inspired by Sinatra
arnsholt (My recent interest in Zavolaj being prerequisitely related to my web musings)
spider-mario I like Mojolicious 20:26
pmichaud although maybe I'll just focus on closing RT tickets and providing suggestions/guidance/distractions to others :)
PerlJam Dancer 2 has the advantage that mst is helping with the rewrite
masak pmichaud: yay \o/
20:26 uvtc joined
pmichaud I'll definitely help out with QAST if folks are working on that. 20:27
PerlJam pmichaud: Rakudo profiling tools would be nice too :)
tadzik I have an affection for Dancer. It showed me that web development doesn't have to be a horrible torture :)
pmichaud oh, I can do those. I had a start on some profiling stuff in nqp that was never finished.
uvtc Hi #perl6. Was just going to install Rakudo * on a newly-reinstalled OS (Lubuntu). Looking at the R* readme, it says that I need to have subversion installed ... I think that's there for Parrot ... but isn't parrot on github now? 20:28
tadzik or nqp profiling. I would love to see some compilation speed improvements
uvtc: what version of R* is taht?
PerlJam tadzik: me too
pmichaud uvtc: that's.... surprising
uvtc 2012.02
tadzik oh gosh, one so needs to fix that README :) 20:29
tadzik looks
uvtc: certainly you don't need svn for anyting these days
pmichaud pmichaud@kiwi:~/p6/star/rakudo-2012.01$ ack -i -a subversion
docs/announce/2009-04:58: spectests, instead of obtaining checkout copies via Subversion.
CREDITS:9: description (D), subversion username (U) and snail-mail 20:30
pmichaud@kiwi:~/p6/star/rakudo-2012.01$
ohwait
spider-mario mojolicio.us/perldoc/Mojolicious/Gu...Guidelines
“Code should be written with a Perl6 port in mind.”
\o/
pmichaud wrong repo
tadzik yesh
masak spider-mario: whatever the heck that means :P
uvtc The readme also says that I should have libreadline5-dev, but libreadline-dev pulls in ... lessee ... libreadline6-dev. If that's ok, perhaps the readme could be amended there as well.
spider-mario sure :D
tadzik __sri++ has stated an interest in a Perl 6 port for some time now. Last time I checked he was missing docs
PerlJam maybe someone should put that "rule" to the test and try to port mojo to Perl 6 20:32
:-)
masak we should write docs with __sri++ in mind :)
PerlJam though, I think sri is the right person for that task
pmichaud uvtc: I'm guessing we should remove the 'subversion' requirement
I'm not sure what libreadline library should be requested now 20:33
dalek ar: 789872f | tadzik++ | skel/README:
Remove unnecessary dependencies from README, uvtc++
uvtc I'll try just libreadline-dev, get libreadline6-dev, and see how that works.
Oooh, thanks, dalek.
tadzik I'm having libreadline6-dev installed and it works fine 20:34
(debian)
uvtc tadzik, great.
Argh, gotta go. Will check back later.
masak o/
tadzik thanks for noticing!
uvtc :)
20:34 uvtc left
felher Is there a nice way to write 'my Str $string = "long string that doesn't get into one line"'? Maybe some kind of quoting form that melts a newline and following whitespaces into one space, so that one could nicely align those strings? I now about heredocs, but i think they require an extra line, which is no problem for large strings, but not so nice for strings that fit into two lines. :) 20:38
masak felher: <<'EOT'.subst("\n", " ", :g); 20:39
er. q:to/EOT/ :)
PerlJam felher: There's also ye old standby: my Str $string = "line one " ~ "line two " ~ "line three";
masak felher: more generally, there are many times when heredocs "almost" do what I want, and I just make a final transformation to make them fit nicely. 20:40
flussence std: @(q:to/EOT/).join # would be nice if this works :) ␤EOT
p6eval std 1ad3292: OUTPUT«ok 00:00 41m␤»
20:40 birdwindupbird left
masak felher: like in Perl 5 where indentation is always significant. I just remove it afterwards. sometimes with a custom sub. 20:40
[Coke] S02 has "the .perl method"/"the .pretty method" ... should it not also have "the .gist method" ? 20:41
masak flussence: you can do .lines>>.join
[Coke] (and do we really need all threde?
masak er, .lines.join
[Coke] *three
[Coke] hates twiddling strings at runtime! inefficient!
masak pfah. premature optimization. 20:42
[Coke] hey, is perljam's example optimized at compile time?
masak [Coke]: I can see "do we really need all N" about a lot of things in the spec these days :/
felher PerlJam: this is what i currently have. The problem is that if i add a word i can't let my editor re-indent it, because it then moves the '"' on the next line :)
masak [Coke]: ideally, yes.
[Coke] masak: there is such a thing as pessimistic optimization. :P
er, what is the opposite of premature... hurm. 20:43
PerlJam cotto: "belated"
er, [Coke]
[Coke] overdue?
PerlJam "over the hill" 20:44
masak tardy.
[Coke] so you are.
PerlJam [Coke]: there are no hills here :)
[Coke] <van halen>I don't FEEL tardy</van halen>
fsergot Good night #perl6 o/
felher masak: two things: first: i hoped there is already something for this, as i find me having that problem quite often. Second: Doesn't EOT have to be on a extra line? :)
masak 'branocm
'branoc, fsergocie 20:45
pmichaud I can vouch for PerlJam's statement -- definitely no hills in CRP
flussence ... S02:3739 :)
[Coke] PerlJam: I remember it well, you're right. highway onramps are the biggest hills.
pmichaud HOU is hilly compared to CRP, fwiw :)
20:45 tarch left
masak felher: just because you find yourself having a problem often doesn't mean it deserves a feature that's almost but not quite like heredocs :) I don't understand the second question. 20:45
[Coke] I disbelieve you, but only because houston is so flat to begin with! 20:46
the end of the heredoc is a line by itself, yes.
masak right. is that a problem?
[Coke] it is verbose. I'd go with perljam's explicit ~ for now and eventually the compiler will optimize it away. 20:47
pmichaud there's always q(...)
masak consider [~]
20:48 fsergot left
masak r: say [~] "This is a ", 'string in several ', q[parts.] 20:48
p6eval rakudo 3bd91f: OUTPUT«This is a string in several parts.␤»
pmichaud rakudo: my $str = q(( This is a string that includes a ) closing paren )); say $str
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 2␤»
pmichaud oops
moritz q() is a function call
masak q() is a sub call.
pmichaud rakudo: my $str = q{{ This is a string that includes a } closing brace }}; say $str
p6eval rakudo 3bd91f: OUTPUT« This is a string that includes a } closing brace ␤»
pmichaud yes, I know -- been doing too much p5 lately
masak :) 20:49
[Coke] lichtkind: I wonder if you'd be better off moving the tablets into a github repo.
pmichaud and with q{} you can pick your closing delimiter also
masak +1
PerlJam Still ... a quoting construct that would turn interstitial whitespace into a single space char seems like it would be useful.
[Coke] e.g.: www.perlfoundation.org/perl6/index....ics_tablet has a bunch of small typos and fixes that I'd be happy to give you a pull request for.
pmichaud my $str = q{{ This is a string }}.cleanup
masak PerlJam: .subst(/\s+/, ' ', :g)
PerlJam masak: yeah, but implicit :) 20:50
masak I'm all for utility methods and stuff, but really.
come on. define it in your code and move on in life ;)
pmichaud well, I'm not sure of the NYI status of .indent
PerlJam masak: are you saying that would be *too magical* for Perl 6? ;)
pmichaud otherwise I'd recommend it :)
[Coke] (e.g. "embedd", "I many lines" "it's own")
masak PerlJam: I'm saying I think the use case is very weak.
pmichaud anyway, I always find q{{{ ... }}} to be a reasonable long quoting mechanism 20:51
masak PerlJam: I'd include primes and fib sequences ten times before .cleanup
[Coke] indent is implemented in rakudo & niecza, no?
I swear I stole it from one for 'tother.
masak not in nom.
moritz not in nom
PerlJam [Coke]: was.
masak moritz: stop doing that! :P
[Coke] ah, so it's only in niecza now! MUAHAHAHA.
moritz masak: you're creepy :-)
masak it's that obvious? :/ 20:52
pmichaud I wasn't seriously suggesting ".cleanup" as a builtin
I was simply suggesting it as a place holder for "whatever sort of cleaning up you wish to do"
PerlJam masak: primes and fibonacci works for me too ;)
pmichaud .indent isn't in nom because .... ? 20:53
masak PerlJam: just saying I can see the use case for them. not for "oh hey sometimes I want to collapse spaces, why isn't that in core?"
moritz blug: perlgeek.de/blog-en/perl-6/2012-hac...tions.html
lichtkind [Coke]: why?
felher masak: right. Thats no reason. But if i have a problem quite often the possibility that there is already a solution for it is higher than for problems i encounter just once. So i thought it might be a good idea to ask. Sure there is no problem with the methods available. But i find all of them kinda ugly. Its just about aesthetics :) 20:54
Defining a own quoting form would be a nice solution to me :)
masak felher: my_fixup(q:to/EOT); # there. pretty. 20:55
er, but syntactically correct.
PerlJam I wonder how many of these "little things" are missing from Perl 6? Seems like their absence contributes to the barrier to entry. 20:56
felher yeah, there is nothing wrong with that solution. I just don't like a third line after a two-line string :) 20:57
masak I think lack of docs contributes to the barrier to entry. lack of noise outside the echo chamber contributes to the barrier to entry. lagging superstitions about the status of Perl 6 contribute to the barrier of entry.
I don't think the lack of a .cleanup method contributes to the barrier to entry. 20:58
pmichaud 20:52 <pmichaud> I wasn't seriously suggesting ".cleanup" as a builtin
masak felher: make the stopper a dot. you'll hardly notice it :)
pmichaud (oops, sorry for double-paste)
masak pmichaud: I'm intentionally using your silly name because I think the idea is silly.
PerlJam masak: no, not that ... I was thinking about q:to/EOT/
moritz pmichaud: sheesh, you think people will lsiten to you? :-)
pmichaud moritz: not anymore. :-) 20:59
PerlJam masak: and .indent
masak: and all of the other "little things"
pmichaud For a two-line string constant, I'd just use ~ 21:00
my $str = "first line'
~ "second line";
PerlJam sure lack of docs is a problem, but if Perl follows the least surprise principle, then they can discover many things on their own (if they are implementeD)
[Coke] lichtkind: as I said, for ease of maintenance/patches. in any case, there are the three issues I found in a quick glance. 21:01
felher masak: i think it will be possible one day to define own quoting-operators (maybe through macros? :) ) that handles that stuff? Until then i may very well use your solution. Thanks :)
masak PerlJam: if you mean that it's surprising/disappointing to discover that things are NYI, then yes, I think you have a point.
21:01 djanatyn left
masak we should implement stuff as fast as we can to avoid disappointment. 21:01
PerlJam masak: no just any things; simple, common, every-day things.
s/no/not/
benabik Does $foo.&bar.baz mean bar($foo).baz ? 21:02
masak felher: probably quotes will be extensible in various ways, yes. it's just a slang, after all.
lichtkind [Coke]: what hinders you to edit?
masak benabik: yes.
PerlJam It wouldn't surprise or disappoint most people if junctions didn't really exist because, for most programmers, the whole idea is outside the realm of their experience anyway. But the lack of heredocs freaks them out.
[Coke] lichtkind: personal dislike of that wiki software.
PerlJam (for instnace) 21:03
pmichaud is it SocialText? If so, same here.
[Coke] pmichaud: aye.
benabik Thanks, moritz++
21:03 havenn left
moritz huh? what have I done? 21:04
[Coke] lichtkind: so, certainly don't feel like I'm applying pressure. I just don't find it easy to help you the way things are now. (but I have many other things to devote my time to, so that's ok.)
pmichaud SocialText (or at least that installation of it) has always been too slow for me, among other annoyances.
lichtkind [Coke]: yes socialtext sucks but i rather use better wiki that github
masak moritz: stealing my karma! >:( :P
benabik moritz: 1) been generally useful, 2) have a name close enough to masak++ that I don't pay attention when I autocomplete
lichtkind [Coke]: so jsut tellme the errors i can change it
[Coke] lichtkind: I just did. 21:05
pmichaud lichtkind: it might be better off not as a wiki, but as a set of files that can be cloned, edited, and changes merged
lichtkind [Coke]: i requested several times tha something better should be installed
moritz well, sometimes you have take action yourself
pmichaud it might naturally fit in the perl6 repo, for example.
lichtkind pmichaud: maybe but where put it nonline? 21:06
pmichaud s/repo/account
moritz lichtkind: tables.perl6.org?
masak I think more people here would pay attention to changes/commits if it were on github.
pmichaud lichtkind: perhaps something like we do for the features page
or perl6.org
both of which are maintained on github
PerlJam masak: I know I would
[Coke] lichtkind: irclog.perlgeek.de/perl6/2012-04-10#i_5424740 - there are the errors.
masak PerlJam: me too.
moritz lichtkind: if there is a simple-ish script that generates HTML, we can throw it into a directory or subdomain of perl6.org
pmichaud github also supports some sort of markdown rendering of files 21:07
(probably not sophisticated enough for this... but maybe)
lichtkind moritz: even handwriting html does less pain than now because then i would have clear rules how to handle special character
moritz lichtkind: well, then it would be even easier to stick them into perl6.org 21:08
21:08 PacoAir left
lichtkind moritz can you init tablets.perl6.org? 21:09
jaffa4 Is there an advent calendar each year?
lichtkind yes
PerlJam jaffa4: so far
dalek kudo/nom: 4bacf1a | moritz++ | src/core/Exception.pm:
add line numbers for warnings

does not yet do the right things for warnings thrown from inside the setting, but warn() calls should be fine
moritz lichtkind: yes, I can
pmichaud lichtkind: take a look at perl6.org/about/ to get some ideas about how it's done for the perl6.org site
(the box that says "Contributing to perl6.org")
moritz lichtkind: give me a repo with .html files and I'll set it up, along with a cron job that updates it once per hour or so
pmichaud I'm sure that once sometihng is started in github, you'll get lots of people submitting patches 21:10
tadzik moritz++
pmichaud (that's what happened with the features page, for example)
lichtkind mostly i prefer bitbuckte, has it to be github?
moritz doesn't have to be, but all our stuff is there already 21:11
so if you want people to contribute easily, github is the way to go
[Coke] I think it if it's git, that's fine, though - nice thing about not under the perl6 umbrella is that he can maintain editorial control, if that's important.
PerlJam lichtkind: git bitbucket or hg?
what Coke said 21:12
[Coke] if it's git bitbucket, we could probably make the github one a mirror, no?
moritz aye
[Coke] and then I can fork from there, and /i/ win, anyway. ;) 21:13
lichtkind non github is ok
i hust try to start a new project under perl 6
moritz only admins can do that 21:14
wait a sec
lichtkind but i always thought its easier for other to partisipate with a wiki
PerlJam lichtkind: not necessarily.
lichtkind "tablets" not tables
21:14 plobsing left
PerlJam lichtkind: we're programmers; we like to hack code :) 21:14
lichtkind it was for normal people
[Coke] exactly! none of us is that.
PerlJam heh
moritz lichtkind: github.com/perl6/tablets
pmichaud since in the Perl 6 world most things are in some sort of git repository (compiler source, document specifications, book text, examples, Perl 6 modules, etc.) -- it seems natural to expect that most people involved in Perl 6 programming in any form will end up with some familiarity with git 21:16
and by Perl 6 programming I also include "writing Perl 6 programs"
21:16 Araq joined
pmichaud not just the language implementors/designers 21:16
moritz well, p5 has a very similar trend 21:17
pmichaud yes, that too.
PerlJam lichtkind: were you hoping for a shared knowledge base built from people's experiences trying to use Perl 6 ?
[Coke] realizes he's got some spare time coming up and wonders what he can do on p6.
moritz all the "big" modules (DBIx::Class, Moose, etc) are in git
benabik github also makes it to edit these things "in place". They added simple editing capabilities to the site.
[Coke] mmm. if you squint, github IS a wiki.
pmichaud we certainly use it like one for specs / websites / book 21:18
moritz lichtkind: I accidentally added you to the 'nqp' team on github. Don't worry about it, no harm done :-)
lichtkind moritz: you will regret that :) 21:19
pmichaud moritz: did you see my comment above about :oneline versus something more generalized? or did you reject it already?
moritz lichtkind: je ne regrette rien :-)
benabik (cd nqp; git push origin :master)
moritz pmichaud: yes, I#ve seen it
pmichaud: thing is, I want it to be a bit smarter when :oneline is passed 21:20
lichtkind haha
moritz pmichaud: well, not smarter, more aggressive in search for user code
benabik That will, of course, fail because it rejects non-fast-forward pushes. :-)
pmichaud okay. could key that off of $lines == 1, but I don't have a strong opinion
lichtkind moritz: i have my own sorta tutorial which waist for some updates and translation should it be also in a repo?
but its other wise finished 21:21
PerlJam moritz: for smartness, let the user pass a custom formatter (with a reasonable default one that you provide too)
moritz pmichaud: in a multi line backtrace, it often makes sense to include setting functions. Not if we only ever print a single line
PerlJam: well, that's always possible, this is just a default formatter
21:22 lichtkind_ joined
pmichaud as I said, I don't have a strong opinion; :oneline just looks a bit smelly to me 21:22
moritz pmichaud: we might make the two things (no of lines and setting fucntions) orthogonal at some point. Or not, we'll see
jaffa4 Are there records in Perl6 now?
moritz pmichaud: but a better name wouldn't be amiss
jaffa4 Are there records/structs in Perl6 now?
moritz jaffa4: we certainly set records in terms of development time :-)
[Coke] jaffa4: there are classes and hashes. 21:23
PerlJam jaffa4: do you mean "packed data structures"?
jaffa4 on change in this respect
benabik wonders how hard it would be to do something like ruby's OpenStruct
jaffa4 yes, something like that
lichtkind_ moritz: allright we can worrylater about the tutorial 21:24
moritz lichtkind_: in general, putting stuff into repos is good :-)
pmichaud jaffa4: perlcabal.org/syn/S09.html#Compact_structs
NYI, but part of the spec
moritz lichtkind_: you can also start repos as your own user, and we can then fork them into the perl6 organization
21:25 lichtkind left, lichtkind_ is now known as lichtkind, lichtkind left
dalek kudo/nom: fa2d53c | moritz++ | / (2 files):
first shot at bringing Str.indent back
 21:25
21:26 lichtkind joined
pmichaud moritz++ moritz++ 21:26
dalek ast: 7051d3a | moritz++ | S32-str/indent.t:
fudge and correct indent.t
21:26 bluescreen10 left
moritz ok, there's one new regression in there, which I don't have the energy to track down now 21:26
OTOH we can now properly test for warnings
with CONTROL { default { pass('we got a warning') } }
lichtkind since i use github ff crashes all the time
Araq interesting so apparently perl6 has everything that C has too 21:27
moritz wow. Complain to the Mozilla folks.
Araq: no pointer arithmetic
Araq: no inline assembler
pmichaud There will be modules.
:-)
Araq what? no inline assembler? that's a strange omission 21:28
PerlJam Araq: yet.
pmichaud We do have Q:PIR { ... } :-P 21:29
Araq somehow against the pilosophy of perl6, isn't it?
moritz well, rakudo has Q:PIR { }
[Coke] can you write C# in niecza?
or haskell in pugs?
21:30 lichtkind_ joined
moritz I think you can write C# in niecza (with that CgOp thing) 21:31
not sure if that works outside the setting though
21:32 lichtkind left 21:34 jaffa4 left 21:35 lichtkind_ left 21:36 tokuhirom joined 21:40 lichtkind joined
lichtkind shit crashed fifth time and tkom made 24h reset 21:40
i could shit bullets
[Coke]: sorry i could not find there any problems under this link
[Coke] "embedd" is not a word. 21:42
"I many lines" is not meaningful. maybe you meant "If" instead of "I" 21:43
"it's own" is grammatically incorrect, it should be "its own" (it's == a contraction of it is; its == belonging to it) 21:44
those are the 3 issues found in a quick glance.
(doesn't address any issues of style, or coverage of the source material, or whether it's useful from a teaching perspective, etc.) 21:45
lichtkind ah all right 21:46
just a min
21:48 lichtkind left 21:49 lichtkind joined
lichtkind that was srash number 7 i could nuke github out of orbit 21:49
[Coke] lichtkind: what version of FF are you running?
(on what platform, and how much memory?)
lichtkind 11 and newest ubuntu 21:50
benabik has never found Firefox to be the most stable of browsers, although he hears it's gotten better.
21:51 lichtkind left
jnthn back from beer :) 21:52
21:52 colomon joined
pmichaud newest ubuntu == 11.10 or 12.04 ? 21:52
I'm running 11.10, my firefox is firefox 10
jnthn My word, I go for 3 pints and you guys generate HOW much backlog? :) 21:53
pmichaud jnthn: We could stop.
jnthn No, it's fine :P
I may just leave it until tomorrow while @student are doing exercises to catch up :) 21:54
[Coke] jnthn: drink more beer, an all'll be well.
21:54 lichtkind joined
lichtkind allright 3 crashes later i have another problem 21:54
i did the first steps like proposed on github 21:55
21:55 Araq left
lichtkind but now he wants some [email@hidden.address] pwd 21:55
wtf
jnthn [Coke]: I'd love to have, but I am teaching tomorrow, so should be vaguely responsible. :) 21:56
pmichaud: You're curious to look at Zavolaj? Nice :)
pmichaud: It's not the epic hack the version for ng was :)
benabik lichtkind: I would guess that that's the identification for your SSH key.
lichtkind benabik: where i can change that? 21:57
jnthn pmichaud: In Oslo I'd like to go through some of the list code with you.
pmichaud: I feel I'm missing some things in there.
benabik lichtkind: Change the key associated with github, or change the ident of the key?
lichtkind: The latter is `ssh-keygen -c -C "new ident" -f keyfile 21:58
lichtkind: The former is github.com/settings/ssh
jnthn re perlcabal.org/syn/S09.html#Compact_structs I'd say that the CStruct REPR partially implements this stuff :) 22:00
lichtkind thank you very much
benabik: WHERE FIND THE KEYFILE? 22:01
sorry
benabik lichtkind: It is usually in ~/.ssh and is one of identity, id_dsa, or id_rsa 22:02
lichtkind i look
[Coke]: all 3 things fixed 22:08
pmichaud jnthn: (review list code) aye, no problem
I may try to get the .munch fix in place first -- that may resolve quite a few issues 22:09
tadzik good night #perl6
lichtkind good night tadzik
how can i so git push -u origin master and explicitly tell the pwd for that command 22:11
22:11 vlixes joined 22:13 plobsing joined
jnthn pmichaud: Wouldn't surprise me if we can knock qbootstrap's remaning blockers out between us with a bit of time on it too 22:17
pmichaud jnthn: I suspect that depends on qregex improvements; I'm not sure I'll have a lot of tuits for it before the hackathon, and it's hairy enough that doing much *at* the hackathon is not a good use of my time 22:18
i.e., if I'm focused on qregex, I'm not really available for much else
jnthn pmichaud: mark_commit
pmichaud: and protoregexes incorporating other protoregexes into their NFA
pmichaud yes, that's still in my head. don't we also need the ltm stuff too?
jnthn pmichaud: Those are the two big ones I know of. 22:19
pmichaud right. those are a little hairy.
22:19 alester joined, vlkv joined
jnthn Yeah, because the naive solution explodes as soon as you hit recursion. 22:19
pmichaud mark_commit is less so, but getting the nfa stuff completely integrated in takes a bit of work
22:19 vlkv left
jnthn pmichaud: What do you mean by "completely integrated in"? 22:19
pmichaud afaik, qregex doesn't have any ltm semantics at the moment 22:20
jnthn pmichaud: It does.
nom: grammar Foo { proto TOP { * }; rule TOP { \w+ }; rule TOP { ab } }; say Foo.parse('abc') 22:21
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤Method 'match' not found for invocant of class 'String'␤»
22:21 alester left, alester joined
jnthn nom: grammar Foo { proto token TOP { * }; rule TOP { \w+ }; rule TOP { ab } }; say Foo.parse('abc') 22:21
p6eval rakudo 3bd91f: OUTPUT«===SORRY!===␤Method 'match' not found for invocant of class 'String'␤»
jnthn huh...
oh, duh :) 22:23
pmichaud wouldn't it need to be something like proto token abc { * }; rule abc:<w> { \w+ }; rule abc:<ab> { abc };
?
jnthn nom: grammar Foo { proto TOP { <...> }; token TOP { \w+ }; token TOP:sym<w> { \w+ }; token TOP:sym<ab> { ab } }; Foo.parse('abc') 22:24
p6eval rakudo 3bd91f: ( no output )
jnthn nom: grammar Foo { proto TOP { <...> }; token TOP { \w+ }; token TOP:sym<w> { \w+ }; token TOP:sym<ab> { ab } }; say Foo.parse('abc')
p6eval rakudo 3bd91f: OUTPUT«=> <abc>␤␤»
jnthn Right.
That is using the NFA.
The qregex branch is also using the NFA engine for such things.
pmichaud okay, so you've extended it beyond where I left off then
jnthn gha
the *qbootstrap* branch
Yes, I've done quite a bit of stuff.
pmichaud okay
checking: 22:25
jnthn Also the qbootstrap branch contains a dumb port of the NFA runner to C that runs 15 times faster than the PIR version.
er
s/PIR/NQP/ 22:26
pmichaud nom: grammar Foo { token TOP { <abc> }; proto abc { <...> }; token abc:<www> { \w+ }; token abc:<ab> { ab }; }; say Foo.parse('abc');
p6eval rakudo 3bd91f: OUTPUT«Method 'abc' not found for invocant of class 'Foo'␤ in regex TOP at /tmp/jsVMkgq9c1:1␤ in method parse at src/gen/CORE.setting:8415␤ in block <anon> at /tmp/jsVMkgq9c1:1␤␤»
jnthn Needs to be proto token 22:27
or proto regex
pmichaud nom: grammar Foo { token TOP { <abc> }; proto regex abc { <...> }; token abc:<www> { \w+ }; token abc:<ab> { ab }; }; say Foo.parse('abc');
p6eval rakudo 3bd91f: OUTPUT«#<failed match>␤»
jnthn proto abc { <...> } is a subroutine.
pmichaud the version you had seemed to have an extra TOP rule
proto TOP { <...> }; token TOP { \w+ }; # error 22:28
jnthn oh
nom: grammar Foo { proto token TOP { <...> }; token TOP:sym<w> { \w+ }; token TOP:sym<ab> { ab } }; Foo.parse('abc')
p6eval rakudo 3bd91f: ( no output ) 22:29
jnthn nom: grammar Foo { proto token TOP { <...> }; token TOP:sym<w> { \w+ }; token TOP:sym<ab> { ab } }; say Foo.parse('abc')
p6eval rakudo 3bd91f: OUTPUT«=> <abc>␤␤»
jnthn That's what I meant.
pmichaud so why does mine faile?
*fail?
oh, no sym
nom: grammar Foo { token TOP { <abc> }; proto regex abc { <...> }; token abc:sym<www> { \w+ }; token abc:sym<ab> { ab }; }; say Foo.parse('abc');
p6eval rakudo 3bd91f: OUTPUT«=> <abc>␤ abc => <abc>␤␤»
pmichaud okay
jnthn Anyway, it's using the NFA engine.
I got that in place months ago. 22:30
pmichaud okay
jnthn Alternations don't use it yet
That is, standalone ones
diakopter++ patched the NFA builder to support quantifiers.
pmichaud so, does it use the nfa engine for subrules?
jnthn Yeah
You implemented merge_subrule
It does as much as that. 22:31
But merge_subrule doesn't consider subrules that are in turn protoregexes.
Well, put better
protoregexes don't have an NFA.
pmichaud ah
protoregexes get an NFA that are the alternation of all of the candidates
jnthn Makes sense. Though NYI. :)
Also, I already implemented caching of the computed NFA. 22:32
pmichaud all you have to do is look up the NFAs for each candidate and add them as an alternation
lichtkind benabik: i got Comments are only supported for RSA1 keys.
pmichaud the caching is per-grammar, yes?
jnthn Yes.
It's done in the meta-object
lichtkind but there are no comments
jnthn So if you were to do a mixin (like we should for derived grammars) then it'd recompute things on demand.
pmichaud: fwiw, you may want to build qbootstrap and just look through the test failures. 22:33
pmichaud ummm, that part doesn't make sense to me
22:34 lichtkind_ joined
jnthn pmichaud: stage1 builds a stage2 using qregex, so when you "make test" you're testing an NQP that uses qregex for parsing. 22:34
pmichaud: Maybe I explained badly then.
22:34 gv joined
pmichaud jnthn: this is in qbootstrap branch, yes? 22:34
benabik lichtkind: Then perhaps I was wrong?
jnthn All I mean is, if you have an instance of a grammar and you mix in to it, then that's a type change, and it will not share the cache with the original class. Which means the new tokens will be considered.
pmichaud: Yes, what I just said about the stage twiddles is only true of qbootstrap branch 22:35
pmichaud okay
jnthn pmichaud: Also I created a temporary QHLL
This contains things like EXPR updated for QRegex.
pmichaud how divergent is qbootstrap from master ?
jnthn Not crazily.
I updated it recently 22:36
It incorporates the last major set of NQP changes (the bs stuff)
So it's basically some qregex fixes + QHLL + Makefile changes for the most part.
And the C port of the NFA runner, because performance is hosed otherwise. :)
lichtkind_ has here anyone a clue why git brings up wxperl.info which is in no config 22:37
pmichaud I'm a little surprised by that -- the tests I did with the NFA runner earlier weren't too bad.
but I'm fine with a C port.
jnthn pmichaud: It was nomming 50% of runtime when I used the profiler on it.
pmichaud: that is, when parsing NQP code.
22:37 lichtkind left
pmichaud okay 22:37
22:37 lichtkind_ is now known as lichtkind
jnthn pmichaud: When I ported it to C in a naive "just get it done" way, that dropped to 3% 22:38
pmichaud well, I was mainly comparing to the nqp-rx form of protoregex
jnthn That is to say, we can do better.
pmichaud not to what the fastest form would be
jnthn But it was a big enough improvement that I left it there. :)
22:38 miso2217 joined
pmichaud well, I'm fine with a C port :) 22:38
jnthn pmichaud: qbootstrap still runs a little slower than master.
pmichaud: I'm not entirely sure why yet.
22:38 spider-mario left
pmichaud lack of transitive nfas, probably 22:38
jnthn That could well be it. 22:39
pmichaud master (nqp-rx) has transitive protoregexes
22:39 spider-mario joined
jnthn aha 22:39
Another part is that the builtins are written in NQP
22:39 miso2217 left
jnthn Not PIR 22:39
benabik lichtkind: What command are you running when it asks for it?
jnthn *but* that'll get better post-QAST.
pmichaud okay 22:40
(more coming)
the approach I had been taking was not to create a separate branch, but rather create a new NQP compiler (NQPQ) that would use qregex for its regexes, and get it to bootstrap
i.e., do it within a branch instead of separate branches
I'm fine with throwing NQPQ away - it wasn't heavily patched over NQP 22:41
jnthn pmichaud: Oh, NQPQ still exists
pmichaud: But just doing it in stage 2 is not enough.
pmichaud I suspect NQPQ is way out of date though
jnthn In qbootstrap it is up to date (more) 22:42
pmichaud and I'm not planning to just do it in stage2 -- the intent was to use NQPQ to close the loop
jnthn It's just that qbootstrap builds now in stage1 *and* stage2.
And stage2 adds in QHLL
So if you "make bootstrap" the final output of this lot, you should close the loop that way.
I guess we have the same goal but reached different conclusions on how to do it... :)
pmichaud yes, and I'm way confused about the status of any of the pieces. 22:43
I'm not really able to keep all of the branches/threads straight in my head
jnthn pmichaud: I suggest build qbootstrap, make test, glance over the Makefile. :)
pmichaud does a 'git diff qbootstrap master' 22:44
jnthn pmichaud: I have the disadvantage of not having been the designer of qregex and the bootstrap system, so largely I've just tried to make the best sense of it I could and do what felt sane. I'm not going to be in the slightest surprised if I've done it differently to how you'd initially envisioned. OTOH, I *think* the path I've led it down so far can work out. 22:45
pmichaud I'm sure it can work out. I don't know how quickly *I* can make it work out.
jnthn pmichaud: That's fine. We have both of us. :) 22:46
pmichaud okay, well let me try to make sense of this -- a few q's
does NQPQ have any role in the qbootstrap building process?
jnthn Yes 22:47
Well, if you mena what I think you do. :)
pmichaud I mean: is it compiled at any point in the build chain
jnthn stage0 = the PIR files. No qregex, no NQPQ, no QHLL. Nothing new.
stage 1 = old HLL, QAST, QRegex, NQPQ
stage 2 = QHLL, QAST, QRegex, NQPQ
(because you can't meaningfully try a updated HLL until you have your grammars being compiled with QRegex) 22:48
so NQPQ is in stage 1 and stage 2
pmichaud okay, so the qbootstrap Makefile basically takes what was the NQPQ target in master and makes it the primary target
jnthn Correct
In fact, the NQP directory itself is perhaps redundant and unused in qbootstrap now 22:49
pmichaud right, that's what I was trying to articulate next. Good.
You're comfortable that NQPQ is not too far diverged from NQP, then?
jnthn I spent two hours at the weekend undiverging them. I'm very comfortable. :) 22:50
pmichaud and you undiverged them in the qbootstrap branch
?
jnthn Correct.
pmichaud checking.
22:51 NamelessTee left
pmichaud okay, I just did a recursive diff -- I'm comfortable with that also. 22:51
jnthn \o/
pmichaud there are some differences but none extensive (unrelated to QRegex stuff) 22:52
so, in qbootstrap, stage0 builds NQP from the .pir files 22:53
stage1 then uses that NQP to build QRegex, QAST, HLL, and NQPQ
jnthn right
pmichaud stage2 then builds QHLL, QAST, QRegex, and NQPQ using ... ?
NQPQ?
jnthn Note nqpq actually ends up in nqp.exe
stage2 builds everything using stage1 22:54
pmichaud okay
jnthn er, the clarify
we build an nqp.pbc at each stage
pmichaud right
jnthn in stage1 and stage2 it's just that NQP::* come from src/nqpq
22:54 spider-mario left
pmichaud okay, I got it. 22:54
jnthn And stage2 adds QHLL
I did my best shot at porting EXPR and friends. 22:55
pmichaud QHLL is an adaptation of HLL?
jnthn Yeah
In fact
22:55 whiteknight joined
jnthn I *think* only Grammar has changes 22:55
pmichaud so, in the qbootstrap branch, src/HLL/ and src/NQP/ are basically unused
jnthn Certainly only Grammar has substantive changes.
wrong (more)
src/HLL/ is still used for *stage1* 22:56
pmichaud oh, src/HLL.... right
22:56 havenn joined
jnthn Yeah, it has to come in one stage later. 22:56
pmichaud got it. makes sense.
jnthn That is a big part of why qbootstrap looks the way it does in terms of how I have got stage1 and stage2 in the Makefile.
pmichaud okay
given that you have protoregexes working in QRegex, at least non-transitively, the rest ought to be fairly straightforward 22:57
jnthn At first I thought I could get away with just doing stuff in stage2. Alas, no :)
pmichaud: Oh, I think with transitive NFA + mark commit fixes, we'll be able to count the number of failing files in t/nqp on the fingers of one hand. :)
pmichaud I think I may be able to close the loop on qbootstrap prior to oslo
then the question becomes, how much of qregex is missing to be able to compile src/Perl6/Grammar.pm 22:58
jnthn Yeah, that I didn't yet look into.
pmichaud that's the other piece that scares me a bit 22:59
let me check my calendar a sec
jnthn I'm not scared about making it compile.
I'm much more scared the impact that having real LTM will have on it. :)
pmichaud oh, that part doesn't bother me too much. :) 23:00
so, since we're each scared about different things, perhaps none of them are truly scary :)
(or both of them are)
have to check calendar with $wife 23:01
jnthn ok 23:02
pmichaud ....who went to the store :-/
jnthn also has to sleep soon
Well
Attempt to
pmichaud here's my tentative plan
jnthn listens 23:03
pmichaud tomorrow I'm basically without tuits -- we'll be at chemo all day tomorrow
thursday I have a bit of work for $otherjob, plus household stuff
so, I may dive into things a bit on Thu or Fri
iiuc, my Sat is completely clear. So assuming energy/tuit supply is still good, I'll plan to spend Sat closing the qbootstrap loop and getting protoregexes to do transitive ltm 23:04
jnthn I suspect the tran ltm is a prereq for closing the loop :) 23:05
Anyway, +1, and here's hoping :)
pmichaud I suspect not, if nqpq is already able to build itself
jnthn I'll be around on Saturday and contactable somewhat on Thu/Fri.
pmichaud in that sense, the loop is already closed, I think.
jnthn My tuit supply this week is unusually low...just result of having been away from $dayjob on vacation and having some backlog there. :) 23:06
After this week things are normal again.
pmichaud the problem isn't closing the loop, it's more that the result doesn't have all of the features needed to build rakudo
(or run all of the things that rakudo expects)
jnthn That could well be an issue, yeah.
pmichaud if nqpq builds itself, it's "bootstrapped"
jnthn Well, remember Rakudo already uses QRegex for userland regexes. 23:07
So it's only really src/Perl6/Grammar.pm that we have to worry much over.
pmichaud right, I'm mainly concerned with Perl6/Grammar and Perl6/Actions stuff
jnthn I don't see Actions being a big deal.
Grammar a bit more so.
pmichaud I don't either, unless it's depending on the old Match objects somehow.
[Coke] is excited for the perl6 weekend! 23:08
jnthn OK, my class tomorrow will be thankful if I sleep now, so I probably should. :)
pmichaud because when we switch to a qregex-based parser, the parse nodes are different from what nqp-rx was producing
(but obviously not too different, if nqpq compiles itself) 23:09
jnthn Yes, that is true.
pmichaud anyway, I'll prioritize some tuits towards qbootstrap over the next few days, if I have any
jnthn OK, nice :)
jnthn would be really happy if we get qbootstrap landed in the next couple of weeks. 23:10
pmichaud I don't know if we'll want to switch for the 2012.04 release
jnthn Me either.
pmichaud I guess we see how this weekend goes.
jnthn But if we've a branch ready to merge right after it...that's fine too :)
pmichaud yeah, that works also.
okay, sleep well. nice work on qbootstrap -- I'm not as scared as I was
jnthn qbootstrap gets rid of 1 out of our 2 major PIR dependencies. :)
pmichaud the other being PAST/QAST? 23:11
jnthn Correct.
:)
pmichaud that one ought to be a lot less difficult to fix
jnthn Yeah. It's do PAST but with better native type handling and some other twiddles. :)
And in NQP, not PIR. :)
pmichaud right
and fix some of the other buglets
jnthn yeah
pmichaud then the hard part becomes converting Rakudo to not use PAST :-) 23:12
jnthn I think Oslo will be a good chance for us to sync our ideas. :)
We can potentially do that picemeal (more)
Since you made QAST/PAST nestable in each other. :)
pmichaud that's a bit ugly (my fault!) but yes, we can go that route :)
23:13 benabik left
jnthn masak++ is also very keen for QAST because then QAST nodes are 6model objects which means they will serialize which means quasi interpolation can work with precompilation :) 23:13
Well, and AST macros generally :) 23:14
pmichaud yes, I hadn't realized that particular part
definitely another motivation for the switch
jnthn :)
Yeah, lots to be excited about. :)
pmichaud to some extent I'd also like to make sure we think of qast nodes in terms of eventual nqp:: nodes
e.g., nqp::list, nqp::hash, nqp::if, and the like 23:15
jnthn ah
pmichaud but that's likely another stage down the road
jnthn I'm a bit bothered about excessive conflation there
pmichaud if you just think of nqp::* translating directly to ast nodes, it makes sense
jnthn e.g. I'd rather we have a prefix like qast:: for QAST nodes and another one (nqp:: or vm::) for VM op abstraction 23:16
pmichaud instead of translating to opcodes
jnthn The current nqp:: opcode handling is problematic because it happesn too soon
The optimizer, for example, should see the abstract, untranslated op.
pmichaud I'd be fine with vm:: for vm opcodes
jnthn And only at the PAST -> backend stage would end trnaslation to the PIR-land or CLR-land or whatever take place. 23:17
OK, I'm not likely to argue over naming. I'm mostly bothered about being able to write a portable optimizer. :)
And ease of writing other backends. :)
pmichaud anyway, I'm fine with that -- I just like having the nqp::op(...) functional style of getting to the ast
jnthn I may arge that wants to be qast::if(...) to be really clear 23:18
but nqp:: can work too
It's just if it's AST nodes, qast:: implies that better.
pmichaud I'm fine if it's ast:: or similar
jnthn Yes, or ast:: :)
pmichaud we can figure out the needed distinction as we get to it.
nqp:: can be the conflated one where we move things around until we know what we ultimately want :)
jnthn true :) 23:19
pmichaud (at which point nqp:: probably goes away)
okay, I have to prepare for evening's activities. See you later
jnthn yeah, sleep... :) 23:20
o/
masak good night jnthn, #perl6 23:22
lichtkind o/
whats a good ssh server? 23:24
Tene lichtkind: openssh 23:27
lichtkind i mean server url 23:29
excuse me for not being clear
Tene I'm very curious about what you're looking for, now. You're looking for an open, public access shell host? 23:30
23:30 Chillance left
lichtkind Tene: i just want to set my config properly 23:31
im totally lost now
[Coke] do you mean "how do I setup ssh locally for use with github"? 23:33
r: gist 3
p6eval rakudo 3bd91f: ( no output )
23:34 nif joined
[Coke] r say gist 3 23:34
r: say gist 3 #bother
p6eval rakudo 3bd91f: OUTPUT«3␤»
[Coke] n:say gist 3
n: say gist 3
p6eval niecza v15-6-gefda208: OUTPUT«3␤»
lichtkind [Coke]: yes
[Coke]: funnily it forked for commits to spec 23:35
but now im master
and things turned ugly
[Coke] lichtkind: help.github.com/linux-set-up-git/ has initial setup instructions. if you need something other than that, asking specific questions helps.
like "I run this command and get <foo> but I expect <bar>". 23:36
p: say ~3
p6eval pugs: OUTPUT«3␤»
23:36 betterworld left
[Coke] moritz: if I rebuild pugs now, will it work? 23:39
(not ready to do it yet)
p: say gist 3 23:40
p6eval pugs: OUTPUT«*** No such subroutine: "&gist"␤ at /tmp/QluWJTy33n line 1, column 5 - line 2, column 1␤»
[Coke] question: for spec tests that compare a .WHAT.gist to a literal string (e.g. "Hash") that are not explicitly testing .gist, can those be changed to compare to Hash.gist ? 23:46
because the literal is Hash(), and pugs is giving "Hash" 23:47
... eh. nevermind.
23:49 Liutox008 joined
[Coke] eh. I WILL change it, but mainly because S02-types/hash.t is inconsistent about it. 23:49
23:49 frankaa112 joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!