»ö« | perl6-projects.org/ | nopaste: sial.org/pbot/perl6 | evalbot: 'perl6: say 3;' | irclog: irc.pugscode.org/ | UTF-8 is your friend!
Set by Tene on 14 May 2009.
00:04 DanielC left 00:07 snarkyboojum joined 00:09 _Chillance_ joined 00:10 _Chillance_ left 00:12 kst left 00:13 kst joined 00:15 Chillance left 00:24 bacek joined 00:53 FurnaceBoy_ is now known as FurnaceBoy 01:02 snarkyboojum left 01:05 jferrero left, snarkyboojum joined
sjohnson will Perl 6 (or even perl 5), have, or does have, an easy way to read a single key in without the ENTER key, and do something? 01:06
i found Term::ReadKey as it sounds like something that would do it, but upon reading its CPAN page, it doesn't sound like it 01:07
or it's heavily obfuscated
TimToady is CTRL a key? 01:08
sjohnson not one that i really care to trap, but maybe in the future 01:09
just looking to trap keys like 1 to 0
and a few letters
i've written a flashcard program in perl for kanji review
and it would definitely speed it up if i didnt have to hit ENTER all the time 01:10
01:10 Whiteknight left
TimToady yes, well, it's a complicated problem, even on one OS, let alone cross-OS 01:10
what OS are you one?
*on?
sjohnson ubuntu linux 01:11
01:11 exodist left
TimToady I'd suggest (for p5) running stty to set your terminal to cbreak mode or raw, and then use sysread 01:11
sjohnson unfortunately it looks like what i am asking is like pulling a tooth out of a sleeping tiger's mouth without waking it up
TimToady if you use raw, make sure your exit code works first :) 01:12
sjohnson TimToady: sysread() being a function i should use in p5?
TimToady yes, bypasses STDIO, another wrinkle
sjohnson perhaps i could run an alias that does all the comamnds for me 01:13
alias flashprogram='stty cbreak && flashprogram.pl && stty normal'
or something
TimToady mebbe
pmichaud "stty sane", often. 01:14
TimToady in my program I just use system "stty raw -echo opost";
pmichaud that sounds familiar :-)
TimToady then while (sysread(STDIN, $buf, 1)) { 01:15
sjohnson Term::TermKey; from CPAN might do the trick as well
without me busting out the dusty unix manuals
though i should try the "l33t" timtoady way, as i wuold probably learn something in the process too 01:16
TimToady then when it exits, system "stty echo cooked -istrip";
sjohnson FYI: search.cpan.org/~pevans/Term-TermKe...TermKey.pm 01:17
01:19 ispy_ joined
sjohnson ahhh i knew it was too good to be true. cpan won't install it for me, some error 01:19
i will do it the Official Larry Wall way
thanks Tim's toady 01:20
oops, 01:30
i should have taken your CTRL-C advice seriously
heh
well i did, but at least i should have remembered it
thankfully i wrote a "kill" wrapper in perl too 01:31
to handle these "unforseeable" incidents
but the technology worked! 01:33
many thanks. that would have probably taken me a month to figure out on my own 01:34
pmichaud: your sane thing saved the day too 01:54
01:59 unitxt left 02:02 xinming_ joined 02:03 agentzh left 02:04 eternaleye_ is now known as eternaleye, agentzh joined 02:05 agentzh left, luqui joined 02:06 xinming left, agentzh joined
sjohnson Q: is there a universal $dummy variable in Perl 6 or 5? 02:25
maybe ()? 02:26
the Camel book said to set () = &something;
but i have found cases where that fails
also, undef = &something; fails too sometimes
(or maybe all the time)
i'd like to always be able to set $dummy stuff, without having to do "my $dummy;" all the time 02:27
literal *, I think
sjohnson am i dummy for asking this question?
literal ($foo, *, $bar) = something();
sjohnson will try it out
sjohnson hopes the universe doesn't implode
literal $ if it's in a declaration: my $foo, $, $bar = something(); 02:28
sjohnson i have found that doing undef in that case works
literal in Perl 6?
sjohnson my ($foo, undef, $goose) = getFarmAnimals();
haven't tried p6
i should tho
literal oh, I was talking about Perl 6
in Perl 5 you'd use undef
sjohnson rakudo: my ($foo, undef, $bar) = <1 2 3>; say $foo, $bar; 02:29
p6eval rakudo 10a9b2: OUTPUT«Unable to parse declarator; couldn't find final ')' at line 2, near ", undef, $"␤in Main (src/gen_setting.pm:0)␤»
sjohnson rakudo: my ($foo, undef, $bar) = <1 2 3>; say $foo.$bar;
p6eval rakudo 10a9b2: OUTPUT«Unable to parse declarator; couldn't find final ')' at line 2, near ", undef, $"␤in Main (src/gen_setting.pm:0)␤»
sjohnson hmm, now i've done it
rakudo: my $foo, undef, $bar = <1 2 3>; say $foo.$bar;
p6eval rakudo 10a9b2: OUTPUT«Symbol '$bar' not predeclared in <anonymous> (/tmp/N8Ew8cisvi:2)␤in Main (src/gen_setting.pm:3207)␤»
sjohnson throws his hands in the air
literal my $foo, $, $bar = <1 2 3>; say $foo, $bar;
rakudo: my $foo, $, $bar = <1 2 3>; say $foo, $bar; 02:30
p6eval rakudo 10a9b2: OUTPUT«Statement not terminated properly at line 2, near "$, $bar = "␤in Main (src/gen_setting.pm:0)␤»
luqui rakudo: my $foo, $, $bar = <1 2 3>; say $foo~$bar;
p6eval rakudo 10a9b2: OUTPUT«Statement not terminated properly at line 2, near "$, $bar = "␤in Main (src/gen_setting.pm:0)␤»
literal hm
sjohnson we broketed the compiler
literal rakudo: my $foo, $bar; $foo, *, $bar = <1 2 3>; say $foo, $bar;
p6eval rakudo 10a9b2: OUTPUT«Symbol '$bar' not predeclared in <anonymous> (/tmp/ETNvJ8wmyF:2)␤in Main (src/gen_setting.pm:3207)␤»
literal rakudo: my ($foo, $bar); ($foo, *, $bar) = <1 2 3>; say $foo, $bar; 02:31
p6eval rakudo 10a9b2: OUTPUT«13␤»
02:31 luqui left
literal rakudo: my ($foo, $, $bar) = <1 2 3>; say $foo, $bar; 02:31
p6eval rakudo 10a9b2: OUTPUT«13␤»
literal needs the parens, I guess
sjohnson oh right 02:32
i was doing the undef thing that wouldnt work in p6 02:33
literal: do you know how to do it in p5?
literal yeah, by using undef: my ($foo, undef, $bar) = qw<1 2 3>;
$ perl -E'my ($foo, undef, $bar) = qw<1 2 3>; say $foo, $bar' 02:34
13
sjohnson oh 02:37
what about the... other scenario
like undef = 5+5;
() = 5+5; works, but i tried it in other cases and it fails i think 02:42
02:42 alanhaggai joined 02:43 Limbic_Region left 02:44 beggars joined
s1n what's the difference between method and submethod? 02:59
pmichaud submethods aren't inherited
s1n pmichaud: thanks
03:02 amoc left 03:12 ispy_ left 03:15 aindilis joined
TimToady they're not inherited, but they must be accessible directly as $obj.ExactMatch::meth() 03:15
even if $obj isn't an exact match; the submethod is really only concerned with the infrastructure of the object relating to ExactMatch, but others can invoke that
sjohnson TimToady: do you know an answer to the $dummy variable question? 03:16
i will ask again 03:17
TimToady so, for instance, BUILDALL can invoke the BUILD submethod on the object multiple times, once for each thing that the object is
sjohnson is there a better way to write, sysread(STDIN, my $dummy, 1);
for a "press any key" action
TimToady in p5? 03:18
not really 03:19
sjohnson yeah, and p6 if you know the future
in the camel book , it says to use () for non-important returned values, but this fails from time to time
TimToady I only know the future to the extent people let me create the future
sjohnson () = 5+5; works
but not in that sysread thing
TimToady that's only for assigment, not for in/out parameters
sjohnson i thought you were the king of the future, though
TimToady I am the servant of the future 03:20
sjohnson and other people follow suit
TimToady If anyone is to be great among you, he must become least of all. --Jesus
sjohnson hence being one of Tim's toadies :) 03:21
TimToady but I can be strong on behalf of people who do not (yet) have a voice 03:22
cj is happy to note this he finally got UTF-8 working with irssi. in case anyone was still worried about that.
TimToady 楽 03:23
sjohnson music
testing my kanji knowledge?
:)
TimToady no, testing cj's utf-8
sjohnson oh
i wrote my kanji thing thanks to your tip, in TimToady(TM)-approved coding techniques 03:24
TimToady well, just because I suggested it doesn't mean I approve :)
sjohnson i suppose I should have said... inspired 03:25
TimToady inspiration is another tricky concept...
03:26 snarkyboojum left
sjohnson may i ask, what inspired you to learn some japanese? 03:27
is it because of your background in linguistics? 03:28
and/or do you think it's a cool language to tackle (as I do)
TimToady both those, plus we love anime 03:30
sjohnson me too
03:30 frew|work joined
sjohnson that is why I am learning it 03:30
TimToady also, started on my 2nd trip to Japan
sjohnson have you met Matsumoto? 03:31
TimToady a friend took me around Kyoto
and started teachin me :)
sure, we get along fine
imitation is the sincerest form of flattery, after all :)
sjohnson it is kind of neat when the "big names" of the computer world meet other big names 03:32
well, at least i think it is
and i mostly mean the heros of the computer stuff... not so much millionaire CEOs
TimToady I enjoy the Hacker's Conference because we can all get together and be "normal" with each other 03:33
sjohnson what do you guys usually talk about?
TimToady cool gadgets, works in progress, social hacks, and anything else we like, which is pretty much everything 03:34
sjohnson have you heard of the anime Rurouni Kenshin (TV, not OVA)? 03:36
TimToady sure, we've watched all of Kenshin 03:37
sjohnson wow that is suprising. it's an older one so many haven't seen it. it is my favourite 03:38
do you have any favourites?
TimToady oh, lots
Last Exile is cool
Utena 03:39
fushigi yuugi
sjohnson ya i like that Otome Ranman song they played in fushigi yuugi 03:40
TimToady his and her circumstances
sjohnson www.youtube.com/watch?v=cUpDECjbbSE
what about Hunter x Hunter
TimToady samarai champloo was pretty good, gad gaurd, escaflowne, (anything with music by Yoko Kano) 03:41
cowboy beebop, trigun, irrisponsible captain tylor 03:42
most of the old classics
03:42 FurnaceBoy left
TimToady stellvia is good 03:42
sjohnson Legend of Basara
?
that's around that time too
TimToady yes, pretty okay
sjohnson i liked that one 03:43
bought the dvds so a fansubbing group could re-release, which is ongonig right now
TimToady oh, my favorite favorite is still probably Azumanga Daioh
sjohnson im noting these, as i am in the mood for another "good show" 03:44
and its hard to find something that REALLY touches you
TimToady crest of the stars, banner of the stars was pretty good
sjohnson other than champloo, escaflowne, and fushigi yuugi, i havent heard of any of these 03:45
sjohnson is frothing at the mouth
you must have seen... Love Hina, right?
TimToady sorry, have to go 03:47
sjohnson i copide all you suggested
thanks for the tips!!!
03:54 amoc joined, Util joined
Util rakudo: my $i = 4; my @a = ($i); $i += 2; print @a[0]; 04:00
p6eval rakudo 10a9b2: OUTPUT«4»
Util rakudo: my $i = 4; my @a; push @a, $i; $i += 2; print @a[0];
p6eval rakudo 10a9b2: OUTPUT«6»
Util Why does the second version print "6" instead of "4"? Is it a known bug?
04:07 unitxt joined 04:12 davidad joined
s1n Util: looks like it's pushing the reference/pointer, src/classes/Array.pir:196 might be a good place to start 04:13
Util s1n: thanks, that looks promising. 04:17
s1n Util: np, i can't say whether it's the correct behavior, but that's probably what's happening (args might be a ref pmc, if such a thing exists) 04:18
Util I cannot believe that "6" is the correct behavior, but I see no explicit example in the S32 tests. 04:21
04:30 eternaleye left 04:48 Util left 04:55 unitxt left 04:57 unitxt joined 05:08 snarkyboojum joined 05:17 frew|work left 05:19 alanhaggai left
japhb Tene: ping 05:56
05:58 sparc joined 06:03 snarkyboojum left 06:06 eternaleye joined 06:16 viklund_ joined 06:17 viklund_ left
Tene japhb: pong 06:17
japhb: what's up?
06:41 [particle] joined 06:49 justatheory left 06:52 alanhaggai joined 06:59 [particle]1 left 07:00 beggars left 07:01 finanalyst joined 07:02 drbean_ joined 07:04 DemoFreak joined 07:07 drbean__ joined 07:11 barney joined 07:19 dakkar joined 07:20 drbean left 07:22 drbean_ left 07:25 unitxt left 07:46 masak joined
Matt-W Good morning 07:47
masak good morning! 07:48
today is Rakudo day! \o/ 07:53
and Form hackathon day! \o/
07:54 cognominal left, agentzh left 07:55 agentzh joined 07:57 lichtkind_ joined
masak lichtkind: run.pugscode.org hasn't been working in a long time. 07:57
lichtkind_: it would be really cool to re-create something like that, but for Rakudo.
Matt-W yes indeed it's form hackathon day 08:04
I fixed the failing test last night
So we're on a clean slate and ready to go
masak Matt-W++ 08:05
Matt-W So hopefully it will be an exciting day
I shall have to blog about it when it's done 08:06
And get my perl 6 category feed on planet perl six, maybe 08:07
08:08 kst left 08:09 kst joined 08:10 davidad left 08:11 lichtkind left 08:23 ssm joined, donaldh joined 08:24 ejs joined 08:28 DanielC joined 08:29 ejs left, ejs joined
DanielC rakudo: for (0..9):by(2) -> $i {say $i} 08:30
p6eval rakudo 10a9b2: OUTPUT«Statement not terminated properly at line 2, near ":by(2) -> "␤in Main (src/gen_setting.pm:0)␤»
DanielC buubot: :by
08:30 riffraff joined
buubot DanielC: hi 08:30
DanielC Damn it, my firefox just froze. 08:31
masak DanielC: :by not implemented yet in Rakudo. you want 'buubot: spack'
DanielC thanks 08:32
buubot: spack :by
buubot DanielC: S01-overview.pod:3 S02-bits.pod:119 S03-operators.pod:109 S04-control.pod:58 S05-regex.pod:95 S06-routines.pod:85 S07-iterators.pod:3 S09-data.pod:46 S10-packages.pod:5 S11-modules.pod:15 S12-objects.pod:62 S13-overloading.pod:9 S16-io.pod:17 S17-concurrency.pod:10 S19-commandline.pod:5 S22-package-format.pod:8 S26-documentation.pod:67 S29-function
DanielC buubot says that :by is in S02 but he's lying.
buubot DanielC: Stored says that :by
DanielC Either that, or Firefox text search is lying. 08:33
masak DanielC: I find one instance using ack locally.
DanielC tries searching for just "by" 08:34
masak S02:2647.
DanielC I'm looking a the web page, so I don't know where line 2647 is. 08:35
masak it's in the HTML source, if you insist on using the web page. :) 08:36
DanielC 1 .. 100 :by(3)
I assume that the spaces are optional, right? In Perl 6 spaces are almost always optional. 08:37
1..100:by(3) <-- ok
masak std: 1..100:by(3); 1 .. 100 :by(3)
p6eval std 27000: OUTPUT«ok 00:02 36m␤» 08:38
masak DanielC: STD.pm says yes.
DanielC What does "00:02 36m" mean?
masak two seconds, 36 megabytes. 08:39
DanielC STD took 36MB just to do 1..100:by(3) ?
masak DanielC: yes, well... feel free to write your own more memory-efficient Perl 6 parser... :P 08:40
masak adds STD.pm and its underlysing technology to queue of things to review, after PGE and the Perl 6 test suite 08:42
std:
std: *
p6eval std 27000: OUTPUT«ok 00:02 35m␤» 08:43
08:44 krunen joined 08:45 bacek left
Matt-W NO NO NO NO NO! 08:45
Excuse me
Colleague committing code that DOESN'T COMPILE 08:46
Matt-W fumes quietly in the corner
DanielC Matt-W: One of your co-workers?
Matt-W yes
Matt-W twitches
DanielC You'd think he'd at least make the code compile before he commits to the main branch. 08:47
Matt-W must... not... shatter... skull...
yes, you'd think
08:47 snarkyboojum joined
DanielC Matt-W: What programming language do you work in? 08:48
Matt-W C++
DanielC I thought it would be. That or C.
Matt-W occasionally I do some Java or Perl
but it's mostly C++ for me
DanielC I know C but I've never tried to learn C++. It looks complicated. 08:49
Matt-W IT is
extremely
masak C-- C-- C-- C-- C--
Matt-W but no more so than Perl 6
What's different is that the complexity hits you in different places 08:50
DanielC But I don't find Perl 5 (and so far Perl 6) complicated.
Matt-W Perl 6 is better at hiding it
and provides a lot of simple things which are easy
C++ doesn't have any of those
DanielC I tend to like simple languages. That includes C, Ruby, Perl, and others.
re Perl: You don't need to learn much to start writing useful Perl programs. Most of the language is just alternative ways of saying things. 08:51
08:51 kst left
DanielC e.g. If you only know one way to do a for loop, you'll get your work done just fine, even if there is a other way that would look nicer. 08:52
08:52 kst joined
Matt-W yeah 08:54
C++ is not too bad
until you start playing with templates
DanielC I've heard bad things about templates. Honestly, I don't know what they are.
Matt-W unfortunately, creative abuse of templates is where all the fun bits are :)
they're C++'s means of doing generic programming 08:55
classes with type parameters
etc.
DanielC Like a + function that works for both matrices, vectors and complex numbers?
08:55 riffraff left, jferrero joined
DanielC Wikipedia: "Generic programming [allows] some code to effectively circumvent the static typing" 08:58
08:59 payload left
DanielC wonders, why don't they use a dynamicly typed language if they are just going to circumvent the static types. 08:59
09:00 payload joined
DanielC Ok, I've read a bit about C++ templates and generic programming, so I have an idea of what you are talking about. 09:04
09:04 cognominal joined 09:08 xomas left, synth left 09:09 davidad joined 09:10 synth joined, alanhaggai left 09:11 xomas joined, xomas left 09:12 xomas joined 09:17 kst left 09:18 viklund_ joined, snarkyboojum left, viklund left, viklund_ is now known as viklund 09:19 viklund_ joined 09:20 kst joined 09:24 pjcj joined
Matt-W DanielC: C++ templates do not allow you to circumvent static typing in any way 09:25
they're like another preprocessor language 09:26
a templated class is regenerated for each type your code uses it with
it remains very definitely statically typed
DanielC Ok... a pre-processor...
I saw the Swap() example. 09:27
template<typename T>
void Swap(T & a, T & b) { ... }
Matt-W yeah
you'd get a new definition of Swap for each type T you use it with in the code
DanielC Ok.
Matt-W templates cause a few problems with the c++ compilation model
because you have to be able to see function bodies when the templates are instantiated
so you tend to end up with all the code in the header files 09:28
DanielC ah
Matt-W not particularly pleasant
One day I'd quite like a job where I don't have to worry about that anymore
i.e. a Perl 6 job :)
DanielC A Perl 6 job would be nice. 09:29
Matt-W not very likely for a while though
DanielC I currently work with PHP, and while I don't hate PHP, it certainly doesn't excite me.
Matt-W urgh
poor you
DanielC Strange that the most popular languages seem to be the crappy ones.
Matt-W I think what most interests me about C++ is finding clever ways to circumvent its restrictions and do very cool things
and also how you can write code that's actually quite nice, but still manages to run really fast 09:30
DanielC I find nothing interesting about PHP. I think that embedding code in HTML is a great idea for making templates (Mason does that too), but that's it really.
Matt-W as a language, it's entirely unremarkable
DanielC nods
That's a good summary: unremarkable. 09:31
ugh! My boss just asked me if we could make the website give an audio warning before it logs you out automatically. 09:32
You know, because the website is not annoying enough as it is.
Matt-W you can, if the user's using firefox 3.5 :) 09:34
DanielC ??
Matt-W <audio> element :)
DanielC ah
I'd rather tell him it's impossible :)
I can just imagine a classroom fool of beeping computers. 09:35
(our software is used in classrooms)
Matt-W ouch
Just say it's impractical
And also highly unwise 09:36
Ill-considered
And it would increase the global production of Evil by nearly 0.02% year on year
DanielC He rarely listens to me when it comes to design. He figures I'm a technie so I can't know anything about design.
09:37 ejs1 joined
DanielC I told him I have studied design as long as I have studied programming (which is true), but you know that design is one of those soft topics where everyone thinks they are an expert. 09:37
jnthn ahojte...to je Rakudo den :-) 09:38
Matt-W Yes, it's a bikeshedding topic
DanielC yeah
Matt-W but I've always felt, from my limited experience, that there are certain things which are objectively true 09:39
And one of those is that making noises is a bad idea in almost every circumstance
DanielC and you would be right.
There is actually a lot of research on usability design.
Matt-W yes
I've read some of it
I thought it was very interesting
DanielC I just follow Jakob Nielsen's blog, and read a few books. 09:40
s/blog/web page/
Much of design is subjective, but much of it isn't. But my employer thinks it's all subjective. 09:41
09:42 ejs left 09:44 ilogger2 joined, hudnix joined 09:45 p6eval joined
DanielC Hmm... maybe we'll start selling auto-logout ring tones... Just to make the classroom experience /really/ annoying. 09:45
Matt-W no don't do it!
think of the children!
DanielC heh
jnthn (finishes backlogging) 09:48
DanielC jnthn: Is this your Rakudo day?
masak DanielC: yes, he just said that. :) 09:49
DanielC: "ahojte...to je Rakudo den :-)"
DanielC masak: What language is that?
masak DanielC: slovak.
DanielC Do you know Slovak too?
masak it's our second language here at #perl6.
DanielC: no, I just pretend to know it.
DanielC Ich wünsche, dass ich konnte Slowakisch sprechen. 09:50
09:50 ejs2 joined
masak well, at least you know German. :) 09:51
DanielC :-)
I'm gradually learning German.
It'll be important, since my wife-to-be is German and much of her family doesn't speak English. 09:52
jnthn Yes, it's Slovak. :-)
masak: Second language? I thought that was lolspeak? ;-)
Matt-W I always thought that a good way to improve my German would be to marry a German lady 09:53
DanielC :-)
jnthn wouldn't be entirely opposed to taking such an approach to improve his Slovak...
Matt-W But I don't meet many, and it seems a strange way to go searching for somebody
I don't really care what language she speaks, to be honest
Perl 6 would be good :) 09:54
DanielC heh
Matt-W jnthn: what's on the plan for rakudo day?
jnthn Matt-W: first, git pull && make coffee
DanielC I actually met my wife-to-be in the OpenOffice.org project. We were both volunteers there.
jnthn After that, I'm open to suggestions, but am sure I'll find plenty in RT. :-)
We currently have 344 tickets. Wow. 09:55
masak jnthn: it's probably a pmichaud thing, but I'd really like to get the Perl6::Grammar.parse thing working...
09:56 Muixirt joined
viklund is there anyway of specifying encoding for a string in rakudo (or change)? 09:57
Matt-W I'm sure there was something I really wanted to ask you to do jnthn 09:58
But now I can't remember what it was
09:58 xinming joined
Matt-W So I suppose it can't be all that important :) 09:58
jnthn masak: Do we know why it's broken?
rakudo: Perl6::Grammar.parse('42')
p6eval rakudo 10a9b2: OUTPUT«Method 'parse' not found for invocant of class ''␤» 09:59
jnthn Ah.
Matt-W rakudo: class A { has @!l; method B { @!l[*-1] = 4; } }; say A.new.B;
p6eval rakudo 10a9b2: OUTPUT«4␤»
Matt-W cool that got fixed
DanielC Question: Is there a difference between || and ?| aren't they both just boolean OR ? 10:00
jnthn rakudo: my ($meth) = Grammar.WALK(:name<parse>); $meth(Perl6::Grammar.new, '42')
p6eval rakudo 10a9b2: OUTPUT«Null PMC access in find_method()␤in method Perl6Object::WALK (src/gen_setting.pm:3207)␤called from Main (/tmp/l9smLp1BMW:2)␤»
jnthn Eww. 10:01
OK, oddness.
DanielC: || short-circuits.
DanielC do this || die ...
Muixirt the exit status of rakudo doesn't reflect if sth. went wrong while compiling, it's always? zero, will that change? 10:02
DanielC jnthn: thanks
Matt-W jnthn: sub/method return types
masak submits rakudobug 10:03
Matt-W aah the first masakbug of the day
jnthn Muixirt: That should be fixed...
Matt-W: Yes?
10:05 tulcod joined
jnthn Matt-W: We have sub/method return types afaik. I seem to remember there being a ticket of some kind about 'em though... 10:05
masak s1n: maybe it might be a better idea to do the Dallas.p6m announcements on perl6-users? 10:06
Matt-W jnthn: I'm sure there was a problem using them in Form.pm, you get type mismatch errors all over the place
jnthn Matt-W: I'm happy to help work out what's going on. 10:08
Let me see if I can spot any RTs...
I'm sure there was one on return value type checking.
Matt-W I'm looking...
masak s1n: p6l is more for language discussion, p6u more for community discussion... 10:09
Matt-W rt.perl.org/rt3/Public/Bug/Display.html?id=58022 is just an 'implement' task
jnthn ah, that is out of date since we already have "of"
"as" we don't have, mostly due to lack of co-ercion at the moment.
10:10 eiro joined, eiro left
Matt-W rt.perl.org/rt3/Public/Bug/Display.html?id=65128 might be related to what I remember from Form 10:11
jnthn std: my Num List sub f () { return () }; 10:12
p6eval std 27000: OUTPUT«##### PARSE FAILED #####␤Multiple prefix constraints not yet supported at /tmp/ZpyIfLZt4o line 1:␤------> my Num List sub f () { return () };␤ expecting typename␤FAILED 00:02 35m␤»
Matt-W I can't find it exactly 10:13
Sorry :)
jnthn OK. Well, that above bug needs a fix anyways...
10:13 eiro joined
DanielC while $fh.get -> $line { say $line } only returns one line. ??? 10:13
jnthn rakudo: my Num List sub f () { return () }
p6eval rakudo 10a9b2: ( no output )
DanielC Strange, because the while loop works fine with $*IN. 10:14
Matt-W DanielC: you want for $fh.lines, I think 10:16
although maybe while $fh.get should work... 10:17
and it should if it works with $*IN.get...
Matt-W smells herring
DanielC Yes, it should work.
masak Matt-W: maybe take a shower?
:)
DanielC I know that $fh.lines works, but if $*IN.get works then so too should $fh.get 10:18
masak jnthn: what about rt.perl.org/rt3/Ticket/Display.html?id=65960 ?
jnthn: #65390 is kinda strange too... 10:20
jnthn std: Role
p6eval std 27000: OUTPUT«ok 00:02 35m␤»
jnthn rakudo: role A {}; say A ~~ Role; say A.WHAT
p6eval rakudo 10a9b2: OUTPUT«1␤A()␤»
jnthn masak: Seems like it's fixed now. :-) 10:21
masak jnthn: indeed. closing. :)
jnthn masak: wait wait wait...need a regression test!
masak jnthn: #64948 would be nice.
jnthn: ok, I'll add that first.
Matt-W tests++ 10:22
masak phenny: tell s1n maybe it might be a better idea to do the Dallas.p6m announcements on perl6-users? p6l is more for language discussion, p6u more for community discussion...
phenny masak: I'll pass that on when s1n is around.
jnthn masak: OK, or I can add it? But go ahead. :-)
masak I'll add it. :) 10:23
I'm already in ur test suite, adding a test.
10:25 rjh left, rjh joined
DanielC rakudo: run "ls" 10:25
p6eval rakudo 10a9b2: OUTPUT«operation not permitted in safe mode␤in Main (lib/Safe.pm:25)␤»
pugs_svn r27001 | masak++ | [t/spec/S12-role] added role-does-role.t, with one test 10:26
jnthn masak: S12-role? 10:27
Roles are in S14...
masak oh.
jnthn I thought the tests had all been renamed.
masak moves the test
jnthn Also, maybe you could put it in with an existing file? 10:28
10:28 snarkyboojum joined
masak jnthn: so that's why the S12-role dir was empty! :) 10:28
jnthn S14-roles/basic.t is probably good.
masak: ah, lol :-)
DanielC if $fh.get { say "yes" } else { say "no" } --> prints "no" 10:29
that can't be right.
(ie. the file does have lines)
jnthn DanielC: If the line is "0" or "" that's false...
Otherwise, maybe bug. 10:30
DanielC wait... now it's saying "yes", something is fishy here.
The line is not zero. But this is really weird. Some times it says "yes" and some times "no". I'm trying to narrow it down... one minute. 10:31
jnthn: Actually... it behaves the *opposite* than what you'd expect. 10:32
pugs_svn r27002 | masak++ | [t/spec] moved test from S12-role/ to S14-roles/
DanielC my $fh = open('foo.txt', :r); if $fh.get { say "yes" } else { say "no" }
says "yes" 10:33
my $fh = open('foo.txt', :r); say $fh.get; if $fh.get { say "yes" } else { say "no" }
says "no"
masak lunch &
jnthn Is $fh.get slurping up all of the file, by any chance?
DanielC No, just one line.
Or at least... say $fh.get only prints one line. 10:34
Wait... I need to test a bit more... 10:36
#!/usr/bin/perl6 10:37
It looks as iff I remove that line from the file it behaves correctly. But that can't be right...
DanielC is bewildered
Matt-W weird 10:38
DanielC Ok, I can make a report. 10:39
My open() line looks like this: my $fh = open('test.pl', :r); 10:40
Where 'test.pl' is the file of the program. Yes, the file is reading itself (I was lazy, didn't want to make another file).
Matt-W out of interest, does it behave if you read a different file?
DanielC When test.pl starts with #!/usr/bin/perl6 the program behaves weird.
Matt-W: It behaves correctly when you read a different file (!!!). 10:41
When test.pl does not start with #!/usr/bin/perl6 the program behaves right.
But when I read a different fie, the program always behaves right.
jnthn DanielC: Erm. It isn't by any chance being stupid enough to run the program when it has the #! line at the top, is it? 10:42
(If it is, then epic ouch...)
DanielC Can you think of a way to test that?
I'll put a sample program in pastebin...
rjh strace it? 10:43
you could also add a sleep in the code, then run pstree
pasteling "DanielC" at 92.227.64.74 pasted "Weird behaviour with open() and $fh.get" (15 lines, 232B) at sial.org/pbot/37044 10:44
10:44 pmurias joined
jnthn fixes one half of RT#65128 10:44
rakudo: sub Int f() { return "a" }; f() 10:45
p6eval rakudo 10a9b2: OUTPUT«Malformed routine definition at line 2, near "Int f() { "␤in Main (src/gen_setting.pm:0)␤»
jnthn rakudo: sub Int f() { "a" }; f()
p6eval rakudo 10a9b2: OUTPUT«Malformed routine definition at line 2, near "Int f() { "␤in Main (src/gen_setting.pm:0)␤»
jnthn oh gah
rakudo: our Int sub f() { "a" }; say f()
p6eval rakudo 10a9b2: OUTPUT«a␤»
DanielC There's the sample code. If you either change the file that is being opened, or remove the #! line, the program works correctly.
jnthn Gah, sucky.
DanielC: Looking 10:46
DanielC: Reproduced. 10:47
DanielC: It doesn't by any chance come back if you have a blank line at the start of the file?
DanielC Good to know I'm not crazy :)
jnthn DanielC: Erm, two blank lines in a row... 10:48
DanielC you are right...
two blan lines => outputs "NO"
jnthn DanielC: Right. Because $fh.get for a blank line returns ""
Which is false.
DanielC s/blan/blank/
rjh DanielC: am I an idiot or does your first 'say $fh.get' not snaffle the #! line
leaving a blank line
jnthn rjh: That's what I'm thinking too. 10:49
rakudo: my $x = ""; if $x { say "yes" } else { say "no" }
p6eval rakudo 10a9b2: OUTPUT«no␤»
jnthn It's exactly like that.
Remember that Perl 6 auto-chomps
So the \n is gone and you just have an empty string
DanielC rjh: You are correct, it seems.
jnthn You probably want more like
rakudo: my $x = ""; if defined($x) { say "yes" } else { say "no" }
p6eval rakudo 10a9b2: OUTPUT«yes␤»
rjh or maybe 10:50
DanielC So maybe this is the problem with the while loop that the guy reported on the list.
Maybe his input file has a blank line.
rjh my $line = $fh.get; say $line; if $line { ... }
so you don't read twice
DanielC So the while() gets "false"
jnthn DanielC: I very much suspsect so...
q1faq!
That's probably going to catch a few people out.
But so far as I can see, Rakudo is behaving as spec'd in your program. 10:51
DanielC So it seems now.
jnthn Which list by the way? p6u?
jnthn should maybe subscribe to that
DanielC yeah, users 10:52
jnthn yeah, not subscribed there
But anyway, seems you have an answer.
DanielC yeah
jnthn OK, let me try and fix up type checking of implicit return values...
DanielC Oh, and I just noticed that he used for loop instead of while. Is that wrong too?
for $fh.get -> $line 10:53
10:53 xinming left
jnthn DanielC: Yes 10:53
DanielC thanks
jnthn for $fh.lines -> $line { } # will work
DanielC ok, thanks for the clarification
jnthn And will not be tripped up on the blank ines issue.
DanielC I suspect that ' while $*IN.get -> $line ' will be a very common construct. 10:54
jnthn I hope not. 10:55
DanielC Is there a way to make it behave as expected?
jnthn No
DanielC :-(
Matt-W for $*IN.lines -> $line is equivalent
to what people want to do with while
jnthn People should learn to use for $*IN.lines -> $line { } instead
DanielC Matt-W: I know...
jnthn while for IO is a p5ism.
Matt-W and really
it makes more sense
sounds like a good article for someone's Perl 6-related blogging 10:56
DanielC But the $*IN.get is something I learned right here in this channel.
jnthn DanielC: Yes - for reading one line. 10:57
That's fine.
DanielC No, in a while loop I mean.
jnthn Ah.
DanielC So people in this channel are making the same error.
Matt-W well it works fine as long as none of the lines you're reading are false
so maybe the person who told you never had that
jnthn :-)
Matt-W and hadn't realised
jnthn Most likely.
DanielC: At some level, everyone is still learning Perl 6. 10:58
Matt-W now I suppose it's possible that .get could return $line but True, but... well
messy
DanielC yeah
Matt-W and .lines is great when we have lazy evaluation
jnthn Matt-W: I pondered that for a moment but soon stopped pondering and realized it would probably cause more harm than good.
Matt-W when are we getting lazy?
jnthn Matt-W: We're putting it off until we have to do it.
Matt-W jnthn: I think it would ripple unexpectedly outwards, people wouldn't realise it was stuck on True
jnthn ;-) 10:59
DanielC Matt-W: Is $fh.lines lazy?
Matt-W DanielC: it will be
DanielC good!
Matt-W nothing's lazy in rakudo at the moment
but there will be lazy one day
DanielC Matt-W: I was just thinking that the main problem with $fh.lines is that it might suck up the entire memory if you are going through a large file.
jnthn Matt-W: Seriously though - it depends on a list refactor that I think pmichaud++ already has underway, or at least knows how to do.
DanielC That's my concern with $fh.lines ... memory. 11:00
Matt-W DanielC: yeah, which is why the idiom in so many languages is 'while I can read one more line'
DanielC yeah
jnthn Matt-W: After that, the major roadblock is probably cleared.
Matt-W but with lazy evaluation, that goes away
jnthn: I'm glad to hear it. I had assumed it was a distant thing
woohoo! 11:01
jnthn Matt-W: At NPW there was a discussion about laziness, and a lot of things became clearer on that area to pm and myself.
Matt-W lunch &
jnthn: hurrah for NPW :)
I'm aware that laziness is not particularly easy to implement 11:02
jnthn Matt-W: So there's at least a plan for moving forward now
Matt-W So I'm trying not to be impatient for it
jnthn It's not, but the fact that much of it hangs off gather/take helps.
Matt-W yeah
once that works, a lot of stuff will be lazy automatically
jnthn Right.
Matt-W lunch really &
jnthn That's why I'm really happy many things in the setting are using it.
:-)
DanielC jnthn: How about a "warnings" mode for Rakudo that detects probable bugs like 'while $*INC.get'? 11:04
It could say "Excuse me, but this line might not do what you think it does"
rjh I don't know, I'm going to feel weird about for $*IN.lines -> $line 11:05
jnthn DanielC: Maybe more something for a kind of "lint" program.
rjh you can stream into STDIN
'while' seems more sensible to me
jnthn DanielC: But thanks to having the Perl 6 grammar available that shouldn't be so hard.
DanielC I'd bet my house that other people will make exactly the same mistake. Using 'for $fh.lines' actually looks wrong (it looks like it reads the entire file into memory). 11:06
11:06 snarkyboojum left
DanielC jnthn: What rjh said :) 11:06
11:06 nsh_ joined
DanielC jnthn: Perl Lint sounds like a great idea. 11:06
jnthn DanielC: Heh, don't tell me, I just implement the spec. :-P
DanielC :-) 11:07
jnthn oh, hmm...checking implicit return types...less easy than first expected.
DanielC Should someone mention this to TimToady?
rjh perl 5 people are going to have to get used to lazy evaluation :)
jnthn DanielC: He backlogs, I believe. 11:08
DanielC rjh: Problem is that you don't notice lazy evaluation most of the time... until it bites you. I'ts like Haskell where you *have* to learn about laziness before you can get anything done. 11:09
rjh perl 6 is still an eager language by default, though?
DanielC jnthn: Maybe he'll notice I mentioned his nick.
rjh: Rather, some things are lazy but most is eager. 11:10
rjh: Only lists are lazy.
jnthn rjh: Different operations have different levels of laziness/eagerness.
For example, array assignment is mostly eager.
DanielC I thought it was binary: lazy or eager. Lists are lazy, everything else is eager.
jnthn No. 11:11
It's more broken down into 4 groups
rjh :/
this is going to _more_ confusing than Haskell :) 11:12
jnthn Array assignment is mostly eager in so far as my @a = 1,2,3,4; will be eager but my @a = 1..Inf; will not be.
rjh jnthn: is @a = 1..Inf lazy list assignment, or is it a Range object?
jnthn rjh: As far as I remember the discussion though, strict eagerness and strict laziness are things that are generally explicitly asked for. 11:13
1..Inf is a range object.
@a = 1..Inf puts the 1..Inf into the "unevaluated" section of @a. 11:14
lambdabot Maybe you meant: activity activity-full admin all-dicts arr ask . ? @ v
rjh @shutup
lambdabot Unknown command, try @list
jnthn I think we mostly at the moment are tending to see arrays as having an evaluated and unevaluated part.
Either of which may not exist.
At least, that's the thinking at an implementation level. 11:15
rjh So you have iterators masquerading as arrays too?
jnthn It's more that the array contains the iterator. 11:16
And just knows that if you ask for something beyond the "already evaluated" portion of the array, it should go and ask the iterator to give it some values.
rjh ...so if you do an operation (map? grep?) on an array, you might get an iterator back? 11:18
11:18 hanekomu joined
rjh iterator-inside-an-array I mean 11:18
jnthn Right. 11:19
But if you then assign it to another array, it's going to be mostly eager and evaluate what it can (e.g. non-infinite bits) at that point. 11:20
Non-infinite is really the wrong term though.
More like "not readily available" or something.
11:21 donaldh joined
rjh I don't see much coverage in the spec 11:22
jnthn S07 is about the best there is at the moment.
rjh There's going to be real fun if you take an array/iterator and start pushing values into it 11:23
jnthn But I agree it needs some filling out.
rjh I guess you need to immediately coerce to a eager array at that point
jnthn That or you push onto the ungenerated part if it contains infinite bits. 11:24
erm
lazy bits
I'm not sure what push should do.
my @a = 1..Inf; push @a, 42; # might be a little too surprising.
(if it works)
(in finite time ;-)) 11:25
rjh for me it appears to go into an infinite loop
as expected
Might be worth adding protection against this in the obvious cases
jnthn Aye. 11:26
rjh rakudo: my @a = 1..Inf; say @a[0];
p6eval rakudo 10a9b2: ( no output )
jnthn Note Rakudo doesn't handle laziness at all yet.
rjh ah 11:27
11:29 pmurias left, eiro left 11:30 pmurias joined, eiro joined, cls_bsd joined, frew joined, pnu joined, pugs_svn joined, antiphase joined, patmat joined, Caelum joined, jiing joined, zev joined, pmichaud left, presh joined 11:31 pmichaud joined
masak young wippersnappers... first time I did 'while =$*IN' (as it was called at the time), I got a segmentation fault from Parrot. that's how I learned not to use that idiom! :) 11:31
jnthn lol 11:32
masak I remember it clearly (thanks to 'git log'). it was on the 23rd of July last year.
we've sure come a long way since then. 11:33
11:33 nsh- joined
jnthn Less segfaults generally, methinks. 11:33
masak oh yes.
especially of the 'your program is very large and complex so I think I'll just segfault here' kind.
very happy about that.
11:34 pmurias left
jnthn Yes, Parrot has certainly improved over the course of Rakudo's development. 11:34
masak aye.
jnthn factors out some duplicated code 11:35
DanielC I wish I could contribute so Rakudo. It would be so cool to work on a Perl 6 compiler.
jnthn DanielC: You've already made a start. :-) 11:36
DanielC :)
jnthn (Bug reports are a useful contribution. :-))
DanielC I've always thought of compilers as the sort of thing that only über-hackers do. Kind of like kernel development.
jnthn Well, at the end of the day, compilers just take one thing and translate it into another. 11:37
The complexity is that the thing we're translating from is Perl 6. ;-) 11:38
11:38 kst joined
jnthn Really though Rakudo mostly boils down to, a parser, a bunch of tree transformations, some built-ins and a few bits of C that do occasionally mind-bending things. 11:39
(Thus why they're written in C. ;-))
DanielC I'll make a point to finish S05. That should help me understand Rakudo. 11:40
jnthn If you want to learn more and get a higher level overview, Pm and I both have slides on Rakudo too.
DanielC I didn't know that Rakudo had anything in C. I thought it was all Parrot+Perl.
jnthn That give a kind of overview.
Rakudo has some little bits in C.
DanielC I would love to see the slides. 11:41
jnthn Mine are on www.jnthn.net/articles.shtml
DanielC *click*
jnthn Probably easiest is "Rakudo: Implementing Perl 6, in Perl 6, on Parrot"
A little out of date in some spots no doubt, but probably not so far.
DanielC found it... 11:42
jnthn Also "All Your Dynamic Language Are Belong To Us" is an overview - not specific to Perl 6 - of the Parrot Compiler Toolkit, and goes through buliding a toy language.
Rakudo is based on that same toolkit, and looking at a smaller example is probably a less overwhelming introduction to it.
11:43 xinming joined
DanielC Thanks. I've downloaded them. 11:43
11:43 nsh_ left
DanielC jnthn: Where are you located? Sweden? 11:43
Nordic Perl Workshop, Stockholm University, etc 11:44
masak :)
11:45 zev left, antiphase left, pugs_svn left, frew left, patmat left, jiing left, Caelum left, pnu left, cls_bsd left, eiro left, presh is now known as bigpresh, bigpresh left, eiro joined, cls_bsd joined, frew joined, pnu joined, pugs_svn joined, antiphase joined, patmat joined, Caelum joined, jiing joined, zev joined
DanielC Actually, the lectures seem very spread out. 11:45
11:45 presh joined
jnthn DanielC: I'm located in Slovakia at the moment. 11:46
But yes, I've talked in a lot of places.
11:46 presh is now known as Guest92010
DanielC k 11:46
jnthn I travel quite a bit. :-)
DanielC :-)
11:50 snarkyboojum joined 11:53 zev left, antiphase left, pugs_svn left, frew left, patmat left, jiing left, Caelum left, pnu left, cls_bsd left, eiro left 11:55 eiro joined, cls_bsd joined, frew joined, pnu joined, pugs_svn joined, antiphase joined, patmat joined, Caelum joined, jiing joined, zev joined, Guest92010 is now known as presjh, presjh is now known as presh
pmichaud www.pmichaud.com/2009/pres 12:02
(for my slides)
DanielC *click*
thanks
Matt-W will have to read those at some point
jnthn pmichaud: Ah, that's the URL. :-)
pmichaud npw-pct and npw-rakudo are the latest
Matt-W but in the mean time, I venture that for $fh.lines reads very well to me
jnthn pmichaud: I didn't find it linked anywhere on pmichaud.com
morning, btw.
pmichaud yeah, pmichaud.com is _way_ out of date :-(
Matt-W I don't think it's a serious problem that people might have to learn about laziness. Oh, the horror!
pmichaud good morning! 12:03
Matt-W hi pmichaud
jnthn > my Int sub foo() { "x" }; foo() 12:04
Type check failed on return value
lambdabot <no location info>: parse error on input `"'
jnthn woo
masak makes a mental note to sprinkle pointers to $fh.lines in appropriate places in u4x...
jnthn Apart from it only works for that case...
12:04 pmurias joined
Matt-W jnthn: does it work if you give it an int? 12:04
jnthn Matt-W: Yes. :-)
masak pmichaud: you mentioned you could probably get #66270 to work... 12:05
pugs_svn r27003 | pmurias++ | [re-smop] punned roles call BUILDALL during new if it's present
r27004 | pmurias++ | removed debugging say
Matt-W jnthn: excellent!
jnthn++
pugs_svn r27005 | pmurias++ | [re-smop] default values from params
jnthn (by that case I meant that sub foo($x --> Int) { } does work like that yet...)
pmichaud jnthn: obtw... it's doubtful that Parrot calling conventions refactor will land before 1.4 release 12:06
jnthn That one should be a sub call addition away though.
pmichaud: omgz I'm like totally shocked... :-/
pmichaud (based on latest perl 6 design meeting)
jnthn (not)
1.4 is like, the release this month though? 12:07
pmichaud no, July
June release is 1.3
jnthn Oh. That's more ouch.
12:07 xomas joined
jnthn Did you get much indication of the time scale we're looking at? 12:07
DanielC jnthn: I'm looking at "Implementing Perl 6 in Perl 6". Why do you need the outer $( ) in the statement $then := $( $<block> ). 12:08
pmichaud Allison said 1.5 (Aug) but I very much doubt that either unless someone else takes it up
jnthn DanielC: That bit of the slides is kinda out of date, we write $<block>.ast now 12:09
DanielC ok
jnthn Which is a bit more self-explanator.
DanielC yes, it is
jnthn But menas "give me the PAST tree for this parse tree node"
jnthn much prefers the new syntax 12:10
DanielC I forget what AST means, but I figure it is "A... Syntax Tree"
Abstract
pmichaud yes, I prefer the new syntax also. 12:11
DanielC What is the difference between a PAST and an AST? 12:12
P = Parrot?
jnthn ye
s
oh noes, here's where we find out that the setting has type constraint issues.
our Bool multi method true() {
self!to_test($.from_exclusive ?? ++($.from.clone) !! $.from)
}
I'm guessing to_test doesn't return a bool... 12:13
Matt-W clearly it should
beat it up?
jnthn Yeah, I'm curious how many things I'm going to have to beat up now... ;-) 12:14
Example
our Str multi method Str() {
$.list
}
masak :)
jnthn .list is *not* going to return a Str!
masak jnthn: it's all for the greater good... 12:15
pmichaud Yeah, needs to be ~$.list
jnthn Right.
pmichaud can I see the diff?
jnthn sure 12:16
pmichaud I'm curious how the exception-return versus non-exception is being handled (more)
jnthn You might be able tot hink of something better...
pmichaud I've been thinking of refactoring PCT a bit there
I probably don't do anything immediately; just curious.
s/don't/won't
jnthn gist.github.com/124262 12:17
12:17 nsh- left
jnthn pmichaud: Basically I force a return *if* we have a return type constraint (and only if we have one). 12:17
pmichaud: I'm sure there's a more efficient way, but couldn't think of one without dipping into :inline's with a goto or PCT changes, which I'd rather you did.
pmichaud agreed. This looks good, actually. 12:18
obtw....
jnthn ya?
pmichaud (checking something)
there's no more trait_auxiliary or trait_verb :-)
DanielC proto infix:<+> is precedence('t=') <--- What does "t=" mean?
pmichaud DanielC: each predefined operator has a precedence level given by a string 12:19
jnthn pmichaud: Oh, Larry is has a decision?
pmichaud string comparisons are used to decide if one operator is tighter than another
so, for example, an operator with precedence 'v=' would be tighter than infix:<+> 12:20
DanielC pmichad: So it's like alphabetical order? a= is a higher precedence than b= ?
pmichaud a= is lower than b= (less tight)
DanielC ok, got it backwards
pmichaud the '='s are there to make it easier for us to dynamically create new precedence levels based on existing ones
for example, if someone does
DanielC Why do you bother with the = sign? Why not just call it "a" or "u" ?
pmichaud proto infix:<blah> is tighter(&infix:<+>) { ... } 12:21
then infix:<blah> ends up with a precedence level of t>=
if someone then does
proto infix:<bar> is looser(&infix:<blah>) that one ends up with a precedence level of t><=
and if we sort/compare those, it turns out that we get the right ordering :-) 12:22
DanielC Interesting.
Thanks.
pmichaud so, we insert a '<' to mean looser and a '>' to mean tighter
jnthn oh wow, I didn't know it did that. That's cool! :-) 12:23
pmichaud (and it works because '<' lt '=' and '>' gt '='
DanielC This is indeed cool. Almost magical that it actually works.
I's an interesting coincidence that in ASCII < comes before = which comes before > 12:24
pmichaud well, that's why I chose those chars :-)
if it hadn't worked out that way, I would've picked some different sequence
afk, kids to school
DanielC for looser(&infix:<blah>) I don't think you need t><= I think t<= should be enough. 12:25
in fact... that's probably what pmichaud meant to type because ><= does the wrong thing :) 12:26
jnthn our Range multi method reverse() { 12:28
@.list.reverse;
}
Is that really right?
masak no. 12:29
jnthn I mean, should a range being reverse return a range with the endpoints flipped?
Or a reversed list of the range's elements?
masak oh wait.
Matt-W is glancing through pmichaud's PCT slides with his jaw on the floor 12:30
masak rakudo: say (1..4).reverse.perl
jnthn I'm not sure if the return type is wrong or the body is wrong. :-)
p6eval rakudo 10a9b2: OUTPUT«[4, 3, 2, 1]␤»
Matt-W pmichaud: I didn't realise PCT did quite that much of the work
jnthn Matt-W: PCT is The Awesome. :-)
masak jnthn: this is what I'd expect, at least.
jnthn masak: OK, in that case the return type wants to change.
Matt-W jnthn: it is really really awesome
masak jnthn: maybe some Range wizardry with :by(-1) could be used to acheive the same effect.
Matt-W now I want to write a compiler 12:31
jnthn Matt-W: Health warning: highly addictive
;-)
Matt-W: Not sure. 12:32
erm, masak
:-/
masak: Changing return type for now.
masak aye.
I can't think of a case where it matters.
it's not like people would be trying to reverse 0..Inf anyway.
Matt-W jnthn: if I got addicted to working on rakudo, I don't think that would be a problem 12:33
jnthn rakudo: say (?1) ~~ Bool
masak Matt-W: famous last words.
p6eval rakudo 10a9b2: OUTPUT«1␤»
jnthn rakudo: say (1..7).true.WHAT
p6eval rakudo 10a9b2: OUTPUT«Int()␤»
jnthn oh.
rakudo: say (?((1..7).true)).WHAT 12:34
p6eval rakudo 10a9b2: OUTPUT«Bool()␤»
jnthn rakudo: say (?((1..7).true)) ~~ Bool
p6eval rakudo 10a9b2: OUTPUT«1␤»
Matt-W masak: indeed. Let's get our Form hackathon done first shall we? 12:35
masak Matt-W: deal.
jnthn Hmm. Well, checking implicitly returned values is going to make the setting a little more honest, at least...
12:36 clintongormley joined
Matt-W jnthn: no bad thing 12:36
masak it's a kind of in-code tests, actually. 12:37
pmichaud no, t><= is correct. 12:40
t<= would make the other operator looser than infix:<+>, which it shouldn't be. 12:41
Range.reverse returns a Range. However, we can't really do that until we implement :by 12:42
jnthn sticks some bacon in the frying pan
pmichaud so we've been returning a list of the reversed elements
(so that it at least works somewhat)
for now, eliminate the return type constraint
or leave it as an embedded comment 12:43
masak (...and maybe add a comment...)
pmichaud our #(Range) multi method reverse() { ... }
masak ooh, nice!
jnthn pmichaud: Done that, and added a comment to the body too.
pmichaud (feel free to pick a different bracketing char :-)
jnthn pmichaud: Saying what it should do eventually.
pmichaud jnthn: excellent.
jnthn goes looking for the weirdest unicode bracketing char he can find
;-)
12:44 meppl joined
Matt-W bacon++ 12:44
jnthn eww we fails a bunch of test scripts from type checks being enforced 12:46
jnthn fixes up
pmichaud Matt-W: (from backscroll) yeah, I really like PCT. Much of it was inspired by audreyt's work on Pugs and PIL.
Matt-W pmichaud: good to know there are influences from that marvellous project 12:47
I remember when I heard about Pugs, I was overjoyed
To think that such an interesting language as Perl 6 was going to be implemented in a language I like as much as Haskell
pmichaud (more backscroll) difference between ?| and || is (1) short-circuit and (2) ?| always returns a boolean 12:50
actually, ?| can short circuit if it wants 12:51
oh, perhaps it can't. hmm.
anyway ?| returns a boolean. || doesn't.
(er, || doesn't have to)
12:52 skids_ joined
jnthn Bacon. Inside fried bread. 12:53
...musteatandforgetimmediatelyhowgooditis... 12:54
Matt-W || will if the side that comes out is a Bool :)
jnthn: mmmmm!
Matt-W goes to jnthn's for dinner
12:57 agentzh joined, FurnaceBoy joined 13:14 hv2 joined 13:17 snarkyboojum left
DanielC Shoot. I burnt my pizza. 13:17
I got distracted with Rakudo stuff.
jnthn :-( 13:18
DanielC puts another pizza in the oven
pmichaud It's important to keep priorities straight. Pizza > Rakudo
DanielC :)
I find Rakudo interesting, so I get distracted.
pmichaud Beer > Rakudo 13:19
DanielC I get distracted eas... oh look at the bird!
pmichaud I did get distracted by Rakudo while trying to cook some rice, however, so Rakudo > Rice
jnthn Beer > * 13:20
PerlJam Dr Pepper > *
:-)
(hello btw)
DanielC o/
13:22 beggars joined, rjh_ joined, rjh left
DanielC Regex: Can someone explain the difference between a rule and a token? When do you use each? 13:25
jnthn pmichaud: www.pmichaud.com/perl6/rakudo-tests...-06-02.png is has a shrunk.
DanielC I'm looking at the RPN calculator from jnthn's talk.
PerlJam DanielC: rules have significant whitespace, tokens do not.
DanielC thanks
pmichaud jnthn: yes, I know. That's what GD::Graph does, apparently.
DanielC In the RPM calculator operators are tokens but values are rules. 13:26
pmichaud I'm not sure what to do about it.
13:26 beggars left
DanielC rule Value { \d+[\.\d+]? }; 13:26
pmichaud DanielC: it has to do with where we allow whitespace
Oh, I don't know why that would be a rule. Should be a token.
DanielC Ok, so I'm not crazy :-)
pmichaud I don't have enough evidence to judge that. 13:27
:-P
PerlJam rakudo: say "hello".trans( l => 'L' ); say "hello".trans( 'l' => 'L' );
DanielC :-)
p6eval rakudo 10a9b2: OUTPUT«hello␤heLLo␤»
13:27 tulcod left, mizioumt joined
PerlJam What's happening on that first one? Is rakudo interpreting a lowercase ell as something different? 13:27
pmichaud named parameter vs Pair
jnthn PerlJam: that looks...odd
oh!
pmichaud the first is a named param. The second is a Pair.
PerlJam ah. 13:28
jnthn And methods ignore named params.
oh huh?
PerlJam (It's still surprising)
jnthn # Looks like you planned 52 tests, but ran 52
...gee!
dalek kudo: 056847f | pmichaud++ | src/classes/Grammar.pir:
Add .parse to base Perl6::Grammar and other non-Rakudo grammars. Fixes #66270.

we can inherit that one and omit this one.		 13:39
jnthn Yay :-) 13:41
jnthn spectests, hoping he's cleared up the setting's return type constraint issues now. 13:43
DanielC jnthn: I'm having trouble getting your RPN calculator (from your slides) to work.
pasteling "DanielC" at 92.227.64.74 pasted "RPN Calculator" (36 lines, 870B) at sial.org/pbot/37046
DanielC Error: Statement not terminated properly at line 34, near "else {\n\tsa" 13:44
Line 34 is the "else" right at the end of the file.
hv2 Hi guys, just trying to get up to speed on Rakudo, got 3 failures in the spectests. The first is one I think is known about: a coredump after the end of t/spec/S09-subscript_slice/slice; the second I don't see mentioned, a (different) coredump after the end of t/spec/S10-packages/basic; something similar to the third was reported and fixed before, in t/spec/S02-lexical-conventions/unicode: 13:47
error:imcc:syntax error, unexpected USTRINGC ('unicode:"\x{e4}\x{f6}\x{fc}"')
in file 'EVAL_27' line 2553
Null PMC access in type()
in Main (t/spec/S02-lexical-conventions/unicode.rakudo:100)
FAILED tests 30-31
Failed 2/31 tests, 93.55% okay (less 4 skipped tests: 25 okay, 80.65%)
Which ones should I open tickets for? 13:48
jnthn hv2: I see the t/spec/S10-packages/basic one too 13:50
hv2: What platform are you on?
hv2 linux, FC6
jnthn Ah, different form me then...so it ain't platform specific.
The unicode one - that's known too, actually a Parrot issue. We filed a Parrot ticket on it.
t/spec/S10-packages/basic I'm not sure if fails for everyone? 13:51
PerlJam DanielC: rather than $expr ~~ RPNExpression, try RPNExpression.parse($expr) or $expr ~~ /<RPNExpression>/;
jnthn Feel free to file a ticket on that one.
DanielC: What PerlJam said...
pugs_svn r27006 | pmurias++ | [re-smop] fix bug in ReadonlyParam
hv2 Oh ok, most recent ref I could find for the unicode one was a bug closed by chromatic last month, for an earlier parrot revision than I fetched.
jnthn DanielC: That older syntax is outdated.
hv2 ack
DanielC jnthn/PerlJam: thanks
jnthn hv2: I *think* pmichaud filed a new one recently for the unicode method names bug. 13:52
hv2 ok, Google might not have indexed it yet, I didn't search directly on the database
jnthn Ah, OK. 13:53
It was recent.
trac.parrot.org/parrot/ticket/730 13:54
DanielC Method 'postcircumfix:{ }' not found for invocant of class 'Failure'
:-(
hv2 Is there a standard set of information to include, maybe even a rakudobug script? rt.perl.org just says "send a detailed email". 13:55
jnthn Include your platform, the name of the spectest that fails and if you know how to (since it's segfaulting) then a gdb backtrace can be useful. 13:56
hv2 ok
Oh, how do I get the rakudo version that I git-cloned to? 13:57
PerlJam git rev-parse HEAD 13:58
hv2 ty
13:58 nsh_ joined, ruoso joined
ruoso HellO! 14:00
PerlJam greetings ruoso
14:04 Chillance joined
DanielC Having trouble with the RPN calculator. 14:06
sial.org/pbot/37047 14:07
14:07 Scorp1us joined
DanielC for $!parsetree<Expr> -> $cur { say $cur.perl } <-- why is this wrong? 14:07
I'm just trying to figure out how the jnthn's RPN calculator is supposed to work. 14:08
PerlJam What makes you think there's somethign wrong with that line?
DanielC The output is: undef and the -1
But the expression is "2 3 +"
s/and the/and then/ 14:09
14:09 snarkyboojum joined
PerlJam DanielC: maybe put a $!parsetree.perl.say in there some where to get a look at the whole parse tree ? 14:10
DanielC I did that, and the whole tree looks fine (keeping in mind that I don't know exactly what it should look like).
But to my untrained eyes it looked perfect. 14:11
pasteling "DanielC" at 92.227.64.74 pasted "Parse tree" (52 lines, 1K) at sial.org/pbot/37048 14:13
DanielC There's the output of $!parsetree.perl.say
14:13 nsh_ left
DanielC The undef and -1 at the end are just from the 'say $cur.perl' 14:13
14:15 alester joined
jnthn DanielC: moment, will look 14:15
jnthn is momentarily distracted
pmichaud might need for $!parsetree<Expr>.list -> ... 14:17
it would help to see the full code
there are also some known bugs for .perl on Match objects 14:18
DanielC pmichaud: Right now the full code for the RPN calculator is at sial.org/pbot/37047
I'll try .list
no change 14:19
Ultimately I'm just copy-pasting from Jonathan's talk.
pmichaud note that the tree is actually in $/<RPNExpression>, not just $/
so there's no $/<Expr> -- it would actually be $/<RPNExpression><Expr> 14:20
DanielC Replacing $/ by $/<RPNExpression> made a big difference.
pasteling "DanielC" at 92.227.64.74 pasted "Replacing $/ by $/<RPNExpression>" (33 lines, 522B) at sial.org/pbot/37049 14:21
DanielC The 'undef' is now a parse tree.
PerlJam DanielC: what is it you expect to be output btw?
DanielC PerlJam: Honestly, I don't know. I got this code from jnthn's slides and I'm trying to figure out how it works. 14:22
PerlJam: The first link I posted is a copy of Jonathon's code.
Except I changed =$*IN to $*IN.get 14:23
jnthn is undistracted
DanielC It seems to come from an older version of Perl 6.
jnthn And spectests pass \o/
DanielC jnthn++
pugs_svn r27007 | pmichaud++ | Add a test for STD.parse() from RT #66270. 14:24
Matt-W DanielC: the loss of prefix:<=> was quite recent, so there'll be a lot of example code floating around which will require that change 14:26
pugs_svn r27008 | pmurias++ | [re-smop] :($foo) creates a new Signature
[particle]- anyone offhand know the sive of an SV in p5?
*size
pmurias ruoso: hi 14:27
ruoso hi pmurias 14:28
PerlJam [particle]-: didn't it change from 5.8 to 5.10?
[particle]- ooh, yeah. i need 5.8. 14:29
14:29 decasm joined
pmurias ruoso: re your lazyness post on p6l sometimes you can easily calculate a large chunk of a lazy list e.g. 1..100000 but you don't want to waste time & memory on unused values 14:29
14:30 hv2 left, nihiliad joined
ruoso pmurias, that's why lazy lists should probably have buffer sizes 14:30
so you only calculate some of it
like Unix Pipes
PerlJam ruoso: so ... laziness is a whole other form of IO ? :)
dalek kudo: 636a2b6 | jnthn++ | src/parser/actions.pm:
Fix one half of what is reported in RT#65128 - our T1 T2 sub f() { ... } now gives a compile time error like STD.
 14:31
kudo: 7a40779 | jnthn++ | src/setting/ (2 files):
Fix up various return type constraints in teh built-ins.
ruoso yeah... the way I see it is really in the same logic as unix pipes
[particle]- pj: what header is SV defined in? i'll try to look it up on github
dalek kudo: 8c09b53 | jnthn++ | src/parser/actions.pm:
Refactor setting of return types to eliminate duplicate code, and implement checking of implicitly returned values for the my T sub foo() { ... } and sub foo(--> T) { ... } cases.
DanielC Ok, if I use $/<RPNExpression> but remove the .list the code seems to behave reasonably.
kudo: a89161f | jnthn++ | :
Merge branch 'master' of [email@hidden.address]
ruoso except you exchange objects instead of strings
PerlJam [particle]-: I haven't really looked at p5 in a while, but I would guess sv.h :) 14:32
[particle]- pj: yes, that was a silly question....
14:32 snarkyboojum left
Matt-W Is there any semantic difference between saying my Int sub foo() { ... } and my sub foo (--> Int) { ... } 14:33
PerlJam Matt-W: I hope not.
jnthn Matt-W: No.
Matt-W jnthn: great
DanielC What does the ~ do in ~$var ?
Matt-W what about foo() returns Int
PerlJam DanielC: stringifies
jnthn Matt-W: In fact I just refactored Rakudo to call exactly the same some path for the two...
returns Int - doesn't enforce on implicits yet. 14:34
That's what I have left to fix now.
DanielC PerlJam: thanks
jnthn But it should be the same
Also .of
erm,
sub foo() of Int I mean
Matt-W hmm
so why do we have four ways to say the same thing
jnthn It's like...there's only 4 ways to do it. :-)
Heh
Matt-W I know TMTOWTDI but this seems a bit silly
jnthn Yes, I haven't found it too amusing trying to keep all four ways in sync. 14:35
Matt-W 'as' is a different semantic, isn't it?
jnthn as is coercion rather than check, yes 14:36
Matt-W that's okay then
it's not five :)
PerlJam Matt-W: perl always optimizes for expressivity.
Matt-W so if you say sub foo() as Str, you basically get prefix:<~> called on anything you return?
jnthn No 14:37
IIRC, you get .Str called on everything you return
Matt-W jnthn: that would make more sense
it's just that prefix:<~> usually ends up calling .Str :)
but I suppose you could change that
jnthn It's either that or a multi-dispatch sub call Str($ret_val) 14:38
14:38 frew left
jnthn Oh, or is it not quite like that... 14:38
Hmm
Whatever co-ercion syntax would be anyway. I should check up on that.
14:39 frew joined
pugs_svn r27009 | jnthn++ | [t/spec] Tests for one issue reported in 65128. 14:40
DanielC The RPN calculator works!!! sial.org/pbot/37050
pmichaud multi-dispatch sub call, I think. 14:41
(rakudo doesn't do it this way yet; it's one of those areas that I sense some spec volatility)
DanielC jnthn: Do you mind if I post your RPN calculator in my blog?
jnthn DanielC: Sure, fixed version I hope ;-) 14:42
pmichaud: Yeah.
DanielC jnthn: Thanks.
jnthn pmichaud: Rakudo doesn't do coercion at all yet though AFAICT. 14:43
pmichaud not through 'as', no.
but obviously prefix:<~> coerces to a Str :-)
PerlJam Does augment work in rakudo yet? 14:45
jnthn pmichaud: Yes, I meant the general case.
pmichaud no, it's still 'isa also'
*'is also'
PerlJam thanks
jnthn pmichaud: We can't in Rakudo currently write Str($x) afaik.
pmichaud jnthn: correct. I'm not sure what to do with that one.
I'm _very_ tempted to put some out-of-band handling for 'VTABLE_invoke' to get the original object. 14:46
jnthn pmichaud: Well, if it just be multi sub I'd guess that it's only the result of 'is export' on multi-methods, no? 14:47
class MyType { multi method Str() is export { return "badger" } }
pmichaud no, I suspect people can write them as plain multisubs also
14:47 kst left
jnthn Oh, for sure. 14:47
pmichaud but that's not the difficulty
jnthn But I don't see how it's any more difficult than a plain 'ole multi.
pmichaud it's not 14:48
the difficulty is vtable_invoke on protoobjects
er, type objects
jnthn Oh!
14:48 kst joined
jnthn Ah, and this means we need to start storing subs as &foo before we can fix this too... 14:48
pmichaud not necessarily 14:49
although that might be where we end up. That's a conversion I'm not particularly looking forward to.
jnthn Otherwise how will the proto Str and the multi Str($foo) co-exist in the namespace?
pmichaud the proto can forward.
jnthn Huh? 14:50
pmichaud (yes, that's not clean or nice)
jnthn For sure it can when the sub is &Str
But I can only at the moment see the imported multi sub conflicting with the Str proto.
e.g. trying to both go into the namespace slot Str.
pmichaud I'm sure there are ways around it. That said, I'm not sure I want to do any of those ways. 14:51
jnthn rakudo: multi Str($x) { say "badger" }; Str(42)
p6eval rakudo a89161: ( no output )
pmichaud std: multi Str ($x) { say 'badger' };
p6eval std 27009: OUTPUT«ok 00:02 37m␤»
pmichaud hmmm.
std: multi Str foo($x) { say 'badger' };
p6eval std 27009: OUTPUT«##### PARSE FAILED #####␤Malformed block at /tmp/kFD2rGlZWR line 1:␤------> multi Str foo($x) { say 'badger' };␤ expecting any of:␤ blockoid␤ routine_def␤ terminator␤ trait␤FAILED 00:02 35m␤» 14:52
jnthn (btw, I just did that in my local Rakudo. It segv'd.)
Yes, that's a syntax error
pmichaud std: multi sub Str ($x) { say 'badger' };
p6eval std 27009: OUTPUT«ok 00:02 37m␤»
pmichaud std: multi sub Str foo($x) { say 'badger' };
p6eval std 27009: OUTPUT«##### PARSE FAILED #####␤Malformed block at /tmp/QWHcfuOrFq line 1:␤------> multi sub Str foo($x) { say 'badger' };␤ expecting any of:␤ blockoid␤ routine_def␤ terminator␤ trait␤FAILED 00:02 35m␤»
pmichaud okay.
jnthn std: our Str multi sub foo($x) { say 'badger mushroom' }
p6eval std 27009: OUTPUT«ok 00:02 37m␤»
jnthn Needs mushroom
pmichaud std: our Str foo($x) { ... } 14:53
jnthn erm, I mean, needs a scope decl.
p6eval std 27009: OUTPUT«##### PARSE FAILED #####␤Whitespace is required between alphanumeric tokens at /tmp/t5ynAqTx3j line 1:␤------> our Str foo($x) { ... }␤ expecting any of:␤ declarator␤ multi_declarator␤ typename␤FAILED 00:02 35m␤»
jnthn And a routine decl. :-)
pmichaud indeed.
jnthn Which is afaik what Rakudo needs too.
pmichaud anyway, converting to use '&' on subnames is going to pose some serious library conflicts.
(with PGE and PCT) 14:54
jnthn rakudo: multi Str($x) { say "badger" }; say "alive"
p6eval rakudo a89161: ( no output )
jnthn That segv's too
Ah? Can PGE and PCT switch too?
pmichaud likely no.
jnthn Youch.
pmichaud I'm also not sure if method names should get &'s as well 14:55
i.e., does it become $P0.'&foo'(...) ?
I suspect not.
jnthn I suspect not too.
pmichaud (an exported method should definitely have the & on its symbol table entry, but I'm not sure about the methodn name itself)
jnthn But OTOH they should get entries in the lexpad or namespace with an &.
OK, looks like we're agreeing. :-) 14:56
pmichaud switching to & also means we have to be a lot more rigorous about when we stick things directly into a namespace (via namespace_keyed) and when we use 'add_sub' 14:57
er, set_pmc_keyed
jnthn Ah yes, the typed interface fun. 14:58
pmichaud and set_*_global
jnthn Yay, of and returns now check implicit return values too.
jnthn does a spectest run in case there are any waiting surprises 15:00
15:02 Jedai left
pmichaud jnthn: likely to need my suggestions/ideas for anything else this morning? I'm going to be having to do some errands shortly 15:07
(which will require ~3 hrs total)
15:07 iblechbot joined
jnthn pmichaud: I've not got much planned other than looking through tickets today. 15:07
pmichaud: Unless there's anything you'd like me to work on. 15:08
pmichaud nothing specific
jnthn Oh, I might hack on the Perl6MultiSub hll_map branch too...
pmichaud looking through tickets sounds really good -- I did some of that earlier in the week.
I found many that could be closed, or easily closed.
jnthn Aye, we've plenty of 'em.
pmichaud okay, I'm afk for a while 15:14
bbl
pugs_svn r27010 | jnthn++ | [t/spec] Correct a type constraint. 15:20
15:20 donaldh left
pugs_svn r27011 | jnthn++ | [t/spec] Tests for type checking of implicitly returned values. 15:20
15:21 donaldh joined 15:22 Jedai joined 15:23 unitxt joined
dalek kudo: fb2fd43 | jnthn++ | src/parser/actions.pm:
Make sure we type check implicit return values for of and returns also. Completely resolves RT#65128.
 15:24
DanielC What does "$sel:" mean in somemethod($self:) mean?
What does "$sel:" mean in somemethod($self:)?
s/sel:/self:/
DanielC can't type today
15:25 ejs1 joined
[particle]- it's the invocant 15:27
DanielC If I don't plan to use it, do I still need it?
[particle]- $self.somemethod($a) == somemethod($self: $a)
DanielC is having some trouble here 15:28
[particle]- it's not much of a method if you don't need the invocant
15:28 ejs1 left
jnthn DanielC: You can access it via self normally 15:28
DanielC: It's useful if you want to take the invocant into a variable of a different name, or apply extra constraints to it. 15:29
[particle]- oh, right, you can name it by whatever ...
jnthn++
DanielC What do you mean "access it via self"?
jnthn The self keyword
DanielC ah
jnthn self.foo within a method calls foo on self
As in, the current instance 15:30
DanielC So I can write self.somemethod() inside a class.
jnthn Right.
[particle]- method foo { say self.bar() }
jnthn But $.somemethod() is shorter and usually just fine. :-)
[particle]- method foo { say $.bar() }
DanielC Ok, that looks better than the exclamation mark way: $self!bar()
[particle]- method foo($this) { say $this.bar() }
er
jnthn That's for calling your privates.
DanielC self.bar() <-- better
[particle]- method foo($this:) { say $this.bar() }
DanielC jnthn: I'm trying to see if I can simplify your RPN calculator a bit so it is easier to explain how it works. 15:31
[particle]- . is used to call public methods
! is used to call private methods
DanielC ok
and "my method" makes it private. 15:32
jnthn DanielC: I think by now that's outdated.
DanielC Ok.
jnthn I think you write method !secret_thingy() { }
DanielC It seems to work though.
jnthn Aye, I think Rakudo trails the spec there. 15:33
DanielC ok
!secret_method() is more consistent with the syntax for public and private variables.
jnthn *nod*
DanielC Do you write 'method !secret()' or '!method secret()' ? I assume the former. 15:34
15:34 ejs2 left
jnthn rakudo: class Foo { method !affair() { say "shhh" }; method what_you_doing() { self!affair(); return "washing up" } }; say Foo.new.what_you_doing 15:35
p6eval rakudo a89161: OUTPUT«shhh␤washing up␤»
jnthn Like that.
DanielC thanks
pugs_svn r27012 | pmurias++ | [re-smop] :($foo is ref) 15:42
pmurias ruoso: is copy should work by calling .clone? 15:45
15:47 hudnix left
pmurias ruoso: it seems it must deal with readonly container specially so it won't be enough 15:47
15:51 nihiliad left, sri_kraih joined
TimToady I don't think Str should be a multi 15:55
Str produces a type object, which responds to .()
that's all
well, that's the top view, I guess you still need a way to specify different coercions by type 15:56
jnthn TimToady: OK fine but then what does it do in response the .()? 15:57
Call $the-thingy-passed.MyTypeName?
15:57 justatheory joined
TimToady not sure, lemme drink some coffee before spouting off further 15:57
jnthn :-) 15:58
DanielC jnthn: Is your RPN calculator supposed to be able to handle a more complex expression like "3 4 - 5 +" ? 15:59
TimToady I think Str's .($x) calls $x.Str as a normal method
jnthn TimToady: OK.
TimToady: How does this play out with multi-joined names? 16:00
TimToady because Str is a known type, it ignores any following () syntactically
jnthn DanielC: I can't remember now...
DanielC ok
TimToady multi-joined names?
jnthn class Foo::Bar { }
Foo::Bar($x) # I want to coerce $x to a Foo::Bar 16:01
$x."Foo::Bar" # but how to define that method?
TimToady good question, more coffee
jnthn :-)
CDD # Caffeine Driven Development 16:04
16:04 hercynium joined 16:05 tulcod joined
pmichaud yes, I remember TimToady saying once before that Str responded to .(), which is why I was thinking that & wasn't an issue for us here. 16:06
(back from errands part 1)
however, that does mean that Str() acts somewhat differently from the now-normal notions of if(), loop(), e(), rand(), etc. 16:08
(i.e., that immediate parens always indicate a sub call)
jnthn pmichaud: In src/parser/expression.pir we have a comment: 16:12
TimToady if we think of everything in terms of an abstract multi coerce(From,To), then .() is coerce(Any,Str)
jnthn (FIXME Parrot bug RT#53296 prevents us from using :optional
on the C<tighter> argument along with :slurpy :named parameters,
so we use :multi as a temporary workaround.)
pmichaud I think that bug has since been fixed. Is it still using :multi?
jnthn pmichaud: Do you know if that workaround can go away?
pmichaud: It is still using multi, and it just exploded onto the scene in my hll_map Perl6Multi branch. :-) 16:13
pmichaud if you can get it to work with .param pmc tighter :optional, then yes, it can go away.
jnthn pmichaud: OK, I'll try that.
DanielC jnthn: Is this valid? rule Expr { <Value> | <Expr> <Expr> <Op> };
pmichaud DanielC: it'll infinitely recurse 16:14
jnthn pmichaud: Not if it parses a <Value> first, no?
DanielC pmichaud: Until it hits values.
TimToady | is parallel
DanielC :-(
jnthn Or do we need || for that?
pmichaud If it doesn't find a <Value>, it'll infinitely recurse (in PGE)
jnthn Right, I suspect that's what it relies on.
DanielC yes 16:15
jnthn If that was taken from the RPN example.
DanielC Yes. It's a variation on jnthn's RPN example.
You'd need to write a different class to manage the parse tree, but I figure that the above (recursive) regex is the "correct" parse tree.
jnthn pmichaud: Does this mean we're saving ourselves a Parrot multi-dispatch per time we enter the expression parser? :-)
DanielC rule Expr { <Value> || <Expr> <Expr> <Op> }; 16:16
pmichaud jnthn: yes, but I don't think that's significant from a time perspective. (I'd be happy to be wrong about that, though.)
jnthn pmichaud: I doubt it'll be a big difference. 16:17
pmichaud jnthn: same here
jnthn pmichaud: Just a little nice to have.
pmichaud I'm looking for the original rt ticket that it was supposedly referring to
I guess I typoed it
jnthn pmichaud: What might be a nicer win is that currently every infix:= to a scalar has to check it isn't really a Perl6Array or a Perl6Hash and then re-dispatch. 16:18
Because Parrot's MMD didn't always get us the correct answer.
pmichaud yes, that will indeed be nicer
the original bug was 53926 16:19
(transposed 2 and 9)
jnthn We lose two isa-s and gain multi-dispatch caching.
Ah, OK.
pmichaud did you see that I split deref_objectref into two different opcodes? 16:20
jnthn Yes
Makes sense.
Guess they are two different operations. 16:21
pmichaud well, sometimes I need to completely dereference something (e.g., for dispatch)
other times we need to dereference only ObjectRefs
jnthn *nod*
pmichaud (so that we stop at a scalar)
ruoso pmurias, re: is copy.... I think that's a question for TimToady... I'm not sure the spec is very clear on that... 16:25
16:28 pmurias left 16:29 Util joined
jnthn suspects a lot of the things in the PIR built-ins that are currently marked as multis may not be in the spec... 16:31
16:31 hanekomu_ joined
jnthn oh no, a lot of them are... 16:32
TimToady I think maybe any surface coercion syntax ends up calling COERCE($x,Str), where class Str redirects &.() into an autogenerated COERCE(Any,Str), but class Rat redirects method Str into an autogenerated COERCE(Rat,Str) 16:33
and the optimizer is always free to call COERCE directly if it can deduce that
16:33 Psyche^ joined
pmichaud notes his "spec volatility" comment was spot-on again :-) 16:34
TimToady but my $s = Str; $s.($x) still works, as does $x.$s 16:35
or maybe COERCE is spelled infix:<as>
[particle]- hrmm... we should have a spec volatility metric generated regularly 16:36
pmichaud it still needs to be a trait or something, yes?
my sub foo($x as Str) { ... }
16:36 rocket_guatemala joined
TimToady that just turns any return $x into return $x as Str 16:36
16:36 kst left
pmichaud I mean from a parsing perspective. 16:36
TimToady yes, the trait is just factoring out the operator
16:36 rocket_guatemala left
pmichaud okay. 16:36
TimToady as a convenience 16:37
pmichaud and actually, it's modifying $x, not the return value, right?
16:37 kst joined
pmichaud sends TimToady some more caffeine. 16:37
TimToady why would $x need to be modified?
pmichaud isn't my sub foo($x as Str) different from my sub foo($x --> Str ) { ... } 16:38
or am I way off again?
I thought that $x as Str coerced $x into a Str
TimToady oh, on the input side
16:38 rocket_guatemala joined
TimToady sorry, didn't have a good night 16:38
I was thinking sub foo($x) as Str
pmichaud yes, in that case it parses "traitish" as well. 16:39
I'm fine with infix:<as> -- just wanted to make sure the trait existed still.
TimToady I view the parameter trait as binding to an anonymous temporary, then $x := $temp as Str 16:40
pmichaud agreed
that's the way I've tended to think of it.
TimToady but $x never can be rw
well, can't refer to the arg 16:41
as rw
pmichaud agreed again.
TimToady as implies copy
pmichaud 16:35 <TimToady> but my $s = Str; $s.($x) still works, as does $x.$s
$x.$s confuses me.
16:41 kst left
rocket_guatemala hi, I am doing some work on the tests and i have a question: should the t/run be moved to the t/spec/S19 folder? 16:41
pmichaud rocket_guatemala: if you're reasonably sure it's up-to-date with respect to the specs, then yes, please do so. 16:42
16:42 donaldh left, kst joined
rocket_guatemala great, ill check each one before moving them 16:42
TimToady pmichaud: it confuses me too
pmichaud rocket_guatemala: that's the correct process, many thanks!
rocket_guatemala: if you end up with questions, [particle] is the current lead on that synopsis :-) 16:43
TimToady just trying to indicate that $x."Str" would still work even if the optimizer doesn't turn it into infix:<as>
pmichaud (but you can ask any of us and we'll venture an opinion)
I'm fine with $x."Str", so we're good there.
rocket_guatemala thanks, ill chime in if i have any other question 16:44
TimToady my basic point is that defining either .() or .Str still allows that syntax, but also guarantees the corresponding deep infix:<as> exists with appropriately typed args 16:45
pmichaud it seems to me that there's also a difference between .() on a type object and .() on an instance 16:46
TimToady sure, it's the abstraction thing again
pmichaud i.e., if I define method postcircumfix:<( )> { ... } then I want to distinguish whether my invocant is the type object or an instance.... right
16:47 hanekomu left 16:49 Psyche^ is now known as Patterner
pmichaud would something like $x as 3 result in coercing $x to Int (the same type as 3) or a "no matching sub"? 16:49
i.e., is infix:<as> typically infix:<as>($x, Abstraction $y) ?
TimToady no matching, I'd think 16:50
pmichaud actually, I have a more general question
sometimes it seems as though methods delegate to operators 16:51
other times it seems as though operators delegate to methods
16:51 ashizawa joined
pmichaud in this particular instance, if I have 16:51
class A { method Str() { ... } }
then how does my $a = A.new; $a as Str
play out?
TimToady if you have a more general question then I'll have a less specific answer :) 16:52
thinking...
pmichaud my more general question would be "is there a guide for methods calling operators or vice versa"?
This also relates to the relationship of prefix:<?>, .bool, .Bool, .true, etc (more) 16:53
TimToady yes, that's a question that's been niggling my subconscious for some time too
pmichaud but we also have prefix:<~>, .Str, and now infix:<as>
and similar issues with prefix:<+>, .Num, .Int, ....
16:53 mizioumt1 joined
pmichaud fwiw, the model with .ACCEPTS and how it relates to type matching and infix:<~~> seems to work very well in practice. 16:54
I don't know if there are parallels to be drawn here.
TimToady overall my impression is that single dispatch tends to be more primitive than multi, but either there exceptions, or we're missing a level somewhere 16:55
*there are
pmichaud I'm fine with single dispatch being more primitive than multi -- I'd just be curious which are the single dispatchy things
the operators? the methods?
TimToady all surface operators are conceptually multi dispatch, even if unary 16:57
pmichaud then if I'm understanding correctly, that would seem to argue in favor of operators-defined-in-terms-of-methods 16:58
TimToady multi vs single is where we distinguish language-forcing from OO-forcing semantics
in general
and as does look a lot like the reverse dipatch of ACCEPTS
pmichaud (side note: we might want to think about how someone defining .ACCEPTS for a class keeps from overriding the typecheck form of .ACCEPTS for the type object. Does the class definer have to remember to constrain the .ACCEPTS invocant to avoid grabbing the type object?) 16:59
TimToady with secondary dispatch on the first arg
which is why I wanted some easier syntax than Abstraction 17:00
that can be negated
pmichaud makes sense. Although we could have Concrete that is just the opposite of Abstraction :-)
TimToady where { $_ !~~ Abstraction } is much too heavy
pmichaud method ACCEPTS(Concrete $x: ...) { ... } 17:01
TimToady well, something also to be said for that being the default
jnthn multi method...
TimToady and something else that means Abstraction|Concrete
jnthn Object :-)
pmichaud Maybe Abstraction isn't a subclass of Any
or Any is somehow constrained to !Abstract 17:02
jnthn I think the thing to also keep in mind is that this is kinda "special" in that it almost breaks the rules of "no multiple prefix constraints"
TimToady starts picking up the pieces of his brain
jnthn And also it's something that we want to be able to recognize and optimize (heavily) in the dispatcher.
TimToady I was thinking of syntax more on the line of ! and ? 17:03
pmichaud I thought about that also. 17:04
Seems like there might be cases where we'd want to say "anything _except_ this type :-)"
TimToady wheren't Abstraction
[particle]- prefix:<*>
pmichaud exclude 17:05
sub foo($x excluding Abstraction ...)
TimToady these are all pretty toxic
pmichaud still probably too long :-)
well, errands part 2 is upon me, so I'll come back and read the solution in backscroll (hope hope)
17:06 tulcod left
[particle]- toxic, for sure. 17:07
TimToady $x sans Abstraction is betterer
[particle]- oh, great. french. 17:08
TimToady $x sans { $_ % 2 }
[particle]- what's the texas form of that?
TimToady $x without_none_o'_that_thar Abstraction 17:09
though that's more hillbilly, I guess
shower & 17:11
17:13 rocket_guatemala left 17:15 M_o_C joined, mizioumt left, nihiliad joined 17:23 broquaint left
pmichaud 16:35 <TimToady> but my $s = Str; $s.($x) still works, as does $x.$s 17:27
oops
ignore that past-o
$x whole-nuther-thing Abstraction # probably 17:30
or maybe 17:31
oh, never mind. :-)
rjh_ $x not { $_ % 2 }, $x not Abstraction :)
pmichaud $x butnot Abstraction 17:32
rjh_ yeah, the negation should be longer than 'where'
larry loves his huffman encoding
pmichaud oh? Then why is "no" shorter than "use"? ;-) 17:33
it's 33% shorter, in fact.
and "not" is shorter than "true" (25%)
rjh_ that's a point
can we finally have don't { } blocks
pmichaud and ! has a lot fewer 1-bits than ?
rjh_ you need to be less optimistic and start using "no" and "not" more often :) 17:34
pmichaud even the frowny :-( comes before the smiley :-)
rjh_ std: my $don't = 0; 17:35
pmichaud rakudo: say ":-(" lt ":-)"
p6eval std 27012: OUTPUT«ok 00:02 36m␤»
rakudo fb2fd4: OUTPUT«1␤»
jnthn gaaahhh 17:36
rjh_ rakudo: sub don't (*@) { }; don't { say "Hello!" }
p6eval rakudo fb2fd4: ( no output ) 17:37
jnthn changes Signature!add_param to not require a junction to be passed if if there's just one type...
pmichaud: I implemented the "auto-generated a Signature by arity" thing, btw.
pmichaud: Got that bit working now. :-)
pmichaud oh, nice! 17:38
jnthn Not caching 'em for sharing purposes yet - will do that later once I get things working. 17:39
(Working in a branch)
Ouch. 17:45
(sitting in an infinite loop at startup)--
Tene rakudo: sub circumfix:<:-( :-)> { say "☹ $^msg ☺" }; :-( "hi guys" :-) 17:47
p6eval rakudo fb2fd4: OUTPUT«☹ hi guys ☺␤»
jnthn omg
Tene I read "the frowny :-( comes before the smiley :-)" at the top of my buffer before scrolling up to read context, and thought of that. 17:48
jnthn It's sick and wonderful that Rakudo can do that. :-) 17:49
Tene TimToady: I see you using "use Foo:from<ruby>;" or some such in scrollback a while ago. Is that actually specced? Rakudo currently uses :lang<...>; 17:50
17:54 nsh- joined 17:55 PacoLinux joined 18:01 PacoLinux left
DanielC rule Expr { <Value> || <Expr> <Expr> <Op> }; <--- doesn't do what I want because Rakudo seems to match the smallest possible token. 18:01
So if your string is "2 2 +" Rakudo only matches the first "2" instead of the whole string. 18:02
Tene DanielC: put it inside of a rule that's like:
rule Foo { ^^ <Expr> $$ }
DanielC ok
Well, that did something... Now it doesn't recognize "2 3 +" as an expression... 18:03
if $expr ~~ / ^^ <Expr> $$ /
jnthn DanielC: I think I originally had it as rule TOP { ^ <Expr>+ $ } 18:04
TimToady Tene: see S11:421
Tene Ah. 18:05
Thanks.
DanielC jnthn: I'd have to remove the + because with the new Expr rule I wrote, the whole thing is an expression. 18:06
TimToady DanielC: you really mustn't do it with left recursion; that doesn't work in recursive descent
DanielC ok
TimToady then you've written an ambiguous grammar, probably
DanielC So I'm barking up the wrong tree here. What is the correct way to do it?
You'd think a grammar for a RPN calculator would be simple... 18:07
TimToady to stick with recursive descent, you need to split the Expr concept into can-be-first, can't be first
18:07 PacoLinux joined
TimToady or you use a bottom up parser as a rule like STD does 18:07
"as simple as possible, but no simpler"
DanielC :-) 18:08
I think I have a lot to learn about grammars.
This is my first attempt to write a grammar for anything.
TimToady figure out what distinguishes the first Expr from the second Expr, then you can do recursive descent 18:09
DanielC ok
DanielC starts thinking
TimToady but as it is, it's like writing a factorial function without an end condition
18:09 nsh- left
DanielC was hoping that <Value> would be the end condition... 18:09
TimToady what's the end condition between the two Exprs though, is the point 18:10
but more generally, it's kinda strange to use a top-down parser to parse something that is designed to be processed bottom-up 18:11
(reverse polish)
DanielC I don't even know the difference between a top-down parser and a bottom-up parser. 18:12
18:12 nsh- joined
PerlJam DanielC++ trying to do something beyond your understanding. 18:12
:-)
DanielC :)
TimToady: The objective of this exercise is for me to learn something about grammars, so it's not surprising that I'm making elementary mistakes. 18:13
Can anyone give me a link that explains the difference between a top-down and a bottom-up parser? 18:14
PerlJam google can help with that.
or wikipedia
DanielC I find that wikipedia often gives very broad explanations for a concept, so I may have trouble relating it to Perl. 18:15
But I'll try wikipedia.
TimToady www.google.com/search?q=top-down+bo...-up+parser
DanielC starts reading
TimToady don't get distracted :) 18:16
18:16 nsh- left 18:26 DemoFreak joined 18:29 bkeeler joined 18:32 justatheory left
bkeeler rakudo: grammar Foo { rule TOP { <ww> } }; Foo.parse("foo") 18:32
p6eval rakudo fb2fd4: OUTPUT«Unable to find regex 'ww'␤Null PMC access in invoke()␤in regex Foo::TOP (/tmp/Iw6E4bKMzm:2)␤called from Main (/tmp/Iw6E4bKMzm:2)␤»
bkeeler rakudo: grammar Foo is PCT::Grammar { rule TOP { <ww> } }; Foo.parse("foo")
p6eval rakudo fb2fd4: ( no output ) 18:33
18:33 gegloe joined
bkeeler I shouldn't have to sub-grammar PCT::Grammar to get <ww> should I? 18:33
PerlJam is <ww> spec? 18:34
pmichaud you might -- I don't think <ww> has been added to PGE yet.
Yes, <ww> is now spec.
at least, it was spec last time I checked.
bkeeler It's in S05 for sure
pmichaud I can add <ww> to PGE pretty simply. 18:35
and then it'll show up by default.
bkeeler That would be nice
18:35 synth joined 18:37 justatheory joined
PerlJam Is there a way to see what rules a grammar has? 18:37
pmichaud .methods?
rakudo: Grammar.methods>>.say 18:38
p6eval rakudo fb2fd4: OUTPUT«Method 'methods' not found for invocant of class ''␤»
jnthn meta
pmichaud hmmm.
jnthn .^
pmichaud rakudo: Grammar.^methods>>.say
rjh_ rakudo: Grammar.^methods>>.say
p6eval rakudo fb2fd4: OUTPUT«parse␤parsefile␤»
18:38 ispy_ joined
rjh_ snap! 18:38
pmichaud there are likely more than this -- right now I'm not sure how Rakudo's Grammar type is interacting with PGE::Grammar
I'm a little surprised it only came back with parse and parsefile, though. 18:39
TimToady thinking more about abstraction and failure, I'm wondering if "while foo() -> $var {...}" should be defined as running until the binding fails, where presumably an undefined return from foo() fails to bind 18:40
pmichaud that somewhat goes against the notion of if $x { } else -> $y { } those
*though
not a big loss, but different. 18:41
TimToady else -> $y where Abstraction
or whatever sugar it turns out to be
pmichaud it does seem like a good idea to not propagate failures except where explicitly allowed 18:42
TimToady it's almost a variant of else -> $y?
pmichaud i.e., passing a Failure to a function that isn't prepared to handle it could be a form of "using the failure should throw the exception"
TimToady could we overload ? to mean undef allowed? hmm
pmichaud that seems odd.
TimToady or -> $y// somehow 18:43
pmichaud I did (75% jokingly) suggest prefix:<//> in the meeting the other day. :-)
TimToady I'm thinking more suffix, since stands in for a where
pmichaud sure.
18:43 ispy_ left
pmichaud that could work, but it doesn't have that "ring of truth" about it to me yet. 18:44
DanielC PerlJam: Well... I think I have a bottom-up implementation. It works, but it's not very elegant. sial.org/pbot/37055
[particle]- $m//
DanielC PerlJam: Oops, that's missing the do_op() function, but you can guess what it does. 18:45
[particle]- $y?
that doesn't really work in sigs 18:46
pmichaud I wonder if/how this relates to the fact that we allow any scalar to hold undefs regardless of type constraint 18:47
the fact that it wouldn't bind by default but would assign by default seems... weird.
[particle]- wait, what?
my int $x = 3; $x = undef; # no earth-shattering kaboom?
pmichaud rakudo: my Int $x = 3; $x = undef; 18:48
p6eval rakudo fb2fd4: ( no output )
pmichaud rakudo: my Int $x = 3; $x = undef; say "no kaboom"
p6eval rakudo fb2fd4: OUTPUT«no kaboom␤»
[particle]- that's Int, and that i get.
i'm talking int.
TimToady that's defined to kaboom anyway
pmichaud sure, that kabooms
[particle]- ok
that matches my expectations. 18:49
is 'int' not 'Scalar'?
TimToady it's essentially the Just vs Maybe distinction of Haskell, we need a notation on the type itself as to whether it should be considered a subset that disallows undef 18:50
and a decision on whether that should be default
StephenPollei I thought section 02 said int couldn't hold a undef, I don't recall for certain 18:51
TimToady because native types are naturally subset types
and int in particular has no representation for undf 18:52
*undef
whereas num could translate at least some undefs to NaN
StephenPollei thats what I recall S02 saying
pmichaud S02 says that non-native typed variables can hold undef 18:53
native typed variables cannot (with the num exception that TimToady++ just cited)
StephenPollei my int $y = undef; # dies
TimToady the can, but that's independent of whether the type check acts as a subset constraint or not
StephenPollei my Int $x = undef; # works
pmichaud My original statement should've probably said "any non-native type constrained scalar"
18:53 ruoso_ joined
pmichaud In general I'm not referring to native types in any way, shape, or form unles I explicitly say so :-) 18:54
*unless
my original point was that we allow
my Int $x = undef;
but would seemingly disallow
[particle]- (Just vs Maybe)++
pmichaud sub foo(Int $x) { ... }; foo(undef) 18:55
[particle]- so, in general, core non-native types follow 'maybe' semantics 18:56
native types follow 'just' semantics
but there are exclusions, in that some non-native types want 'just' semantics
TimToady so even if we leave that default, we could tell people to write something like: while foo() -> just $x 18:57
[particle]- and there needs to be a way to distinguish the semantics
18:57 masak joined
masak oh hai 18:57
TimToady but really just Str $x
if it modifies the type constraint
well, I suppose just $x means just Any $x
pmichaud I'm afraid I don't completely understand the meaning behind 'just' and 'maybe'
StephenPollei multi sub foo(Int $x) { ... }; multi sub foo(Str $x) { ... }; foo(undef) 18:58
TimToady as opposed to "not just any $x" :)
[particle]- Just = only this type and not undefined things
TimToady haskell doesn't have the anti-just sense, afaik
pmichaud that doesn't match my sense of "just"'s english meaning, though.
TimToady the the don't care sense of maybe
pmichaud To me, "just Int" would mean "Int and only Int,"
TimToady s/the/just
not suggesting that syntax 18:59
just trying to nail down the fact that it's a modifier on the type, not the $x
19:00 ruoso left
pmichaud agreed. 19:01
TimToady it would be possible to interpret Int differently by default for assignment than for binding
pmichaud yes, I thought of that -- that seems strange to me also.
TimToady if we had a way to mark it either way
jnthn -> dinner, bbiab 19:02
masak eagerly waits for the Matt-W/masak hackathon to commence
Matt-W is just putting on some hacking music 19:04
masak is putting on his robe and wizard hat
TimToady just don't put on some putting off hacking music
Matt-W Right 19:07
There we go
masak Matt-W: I'm looking at search.cpan.org/dist/Perl6-Form/For...cal_fields right now. you got an attack plan, or do we just wing it?
Matt-W I think we throw together a plan, and then we grab hold of it and leap off into the unknown, attempting to use it as a parachute 19:08
There are clearly two areas required here
masak sounds reasonable to me. and perfectly safe.
Matt-W One is the extension of the grammar and actions and data structures to parse these fields and create the objects to describe them
The other end is the routines which actually do the work - a new Form::NumberFormatting module, I suspect 19:09
masak aye.
Matt-W And then we just stick them together with a little glue
masak runs './proto update form'
Matt-W So do you feel more like grammar hacking, or writing formatting routines? 19:10
masak I feel like writing tests! 19:11
Matt-W You don't get out of writing tests either way :)
masak oh. :)
can I at least write the tests first? :)
Matt-W Tests go hand in hand!
If you like
masak visualises the tests going hand in hang
er, s/hang/hand/
Matt-W yes, tests are very friendly and slightly archaic 19:12
19:12 bkeeler left
Matt-W culturally, that is, not at all archaic in software usefulness 19:12
Matt-W makes sure form works with latest rakudo
19:13 bkeeler joined
Matt-W masak: you should be able to push to mattw/form btw 19:13
all 100 form tests passing
better fix that
masak :)
19:13 payload joined
Matt-W I'll do Form::NumberFormatting 19:14
masak ooh, pmichaud++ # Perl6::Grammar.parse
19:14 justatheory left
masak I'm not sure I'll get any sleep tonight. :) 19:14
eiro hello guys
pmichaud note that it will change to STD.parse at some point.
masak eiro: o/ 19:15
pmichaud: yes, of course.
I've lived through change before, I can do it again.
pmichaud I suppose that as a temporary fix I can do: grammar STD is Perl6::Grammar { }; 19:16
:-P
Matt-W :)
masak oh, you don't have to. I can do it in my script instead.
sjohnson hey ma 19:19
Matt-W:
i mean
Matt-W considers how to do number formatting 19:20
sjohnson: hello
masak Matt-W: number format parsing, does it go in its own .t file, or in an existing one, such as 01-fieldparsing.t? 19:21
japhb Tene: ping again # sorry I missed you before
Tene japhb: hi
japhb Yay! You're here.
.oO( Nice, reception, huh? )
 19:22
Matt-W masak: I'd put it in with the parsing tests for text
Tene AFK, internet sabbatical for the next week.
japhb OUCH
Tene lies.
Anyway, what's up?
japhb I guessed. :-)
I'm trying to get OpenGL working with Rakudo again. Last time I had it working was before the HLL namespace change and your use Foo:from<bar> work. 19:23
So now I need to figure out how to get a Parrot module to load into current Rakudo.
Do you know offhand the right incantation, and if not, where in the code do I start looking for answers?
Tene japhb: I know exactly what you need to do! 19:24
japhb Yay again!
Tene Check out... uh... 'sec
pmichaud is very interested to learn this also. :-)
Tene pastes 19:25
Matt-W do we have an integer division operator?
Tene gist.github.com/124453
masak hm, my Rakudo doesn't work. :/
Tene japhb: like that. Make sure that the namespace matches the file path, in the usual way, including case 19:26
TimToady we have a type-specific div that could do that
Matt-W masak: noooooooooo
TimToady but it's not div in the Pascal sense
masak pmichaud: has anyone reported the error message C<"load_bytecode" couldn't find file 'perl6.pbc'> lately?
Matt-W TimToady: okay
pmichaud masak: no.
masak pmichaud: I had it earlier today on feather too.
19:26 [particle]1 joined
masak pmichaud: I'm thinking it might be because I insist on running bleeding-Parrot/bleeding-Rakudo. 19:26
pmichaud masak: if running bleeding parrot, that could be an issue, yes. 19:27
I'll try bleeding parrot.
TimToady div could integer division, but it wouldn't coerce to Int unless you supplied the extra multis
masak rebuilds
TimToady *could do
japhb Tene: When you say the namespace matches the file path ... where exactly does it need to be now? For OpenGL, is the current location in parrot/runtime/parrot/library/OpenGL.pbc correct?
Tene masak: I always run latest rakudo and parrot. Where are you seeing that problem?
TimToady div is really generic division, not integer 19:28
masak Tene: when running the fakecutable.
Tene japhb: if you have that path, you need to use the namespace ['OpenGL']
masak Tene: nice to know I'm not alone in this habit. :)
pmichaud the fakecutable of perl6, or some other fakecutable?
masak pmichaud: the perl6 one.
pmichaud okay.
TimToady Matt-W: but otherwise you currently want Int($x/$y) 19:29
or floor maybe
Matt-W TimToady: thanks
I thought about it and I think floor is more appropriate
japhb Tene: Line 81 of OpenGL.pir says .namespace ['OpenGL'] , so I'm good there.
pmichaud (in rakudo, Int($x/$y) wont work. use floor, or .int)
TimToady .int should coerce to the native type, of course... 19:30
pmichaud in our case, it probably does exactly that :-)
we can likely add a .Int, though. 19:31
japhb Tene: does c.'export'('subname') have a lot of set up overhead? In other words, do you expect a big difference between doing c.'export'('subname') several thousand times, versus doing names = join(list-of-thousands-of-names); c.'export'(names) ? 19:34
19:34 EvanCarroll joined
EvanCarroll I just had a brilliant idea. 19:34
Rather than unshift/shift, we should have pull/shove.
;))
pmichaud masak: the perl6 fakecutable appears to be working fine for me, bleeding rakudo/parrot
rjh_ EvanCarroll: beautiful 19:35
EvanCarroll because it takes more force to shove something under an array, then it takes to push it on the top of an array...
Tene japhb: I'm not sure whether the former will work at all
rjh_ can we use those for gather/take semantics as well?
Tene lemme check
japhb take($this_job) && .shove($_)
masak pmichaud: I get the C<"load_bytecode" couldn't find file 'perl6.pbc'> error when running outside of the parrot/languages/rakudo dir. when I run from within it, it works for -e scripts.
19:35 hercynium left
pmichaud sounds to me like parrot may have changed its load_bytecode semantics. 19:35
masak pmichaud: for Matt-W's first test file, I get a really weird error: C<A method named 'clone' already exists in class 'Perl6Object'. It may have been supplied by a role.>
EvanCarroll rjh_: I'm being serious here... there is an easy mental connection between force and the begining of an array.
Tene japhb: looks like the former would work (thousands of calls), but I wouldn't be too surprised if it was kinda slow. 19:36
japhb Tene: OK, thanks
EvanCarroll I'm just saying it would certainly be better than a confusing inconstant un- syntax.
rjh_ EvanCarroll: I actually think there's a clear mental distinction between push/pop and shift/unshift
japhb Syntax tends to be very inconstant.
pmichaud Tene: doesn't 'export' take an array also?
Tene pmichaud: it should. It currently doesn't. 19:37
rjh_ whereas push/shove are strongly associated words
EvanCarroll but a shove is always a stronger form of a push.
rjh_ i don't really see the relation
end-users don't need to be concerned that adding to the other end of a stack is more 'difficult'
pmichaud masak: how long since a realclean? 19:38
EvanCarroll no they don't. but humans naturally think about this...
masak pmichaud: a couple of minutes ago. I just rebuild everything.
pmichaud hmmm.
rjh_ Perl6 arrays are presumably designed to work as queues or stacks
EvanCarroll 100% of people view arrays as laying horizontally or vertically, no one thinks of them us internally ordered, non-sorted abstract lists.
masak s/d/t/
pmichaud I wonder how I fall outside of the 100%, then :-) 19:39
rjh_ imho, push/pull/shove are too similar to each other
Matt-W pmichaud: it worked for me when I updated about 20 minutes ago
EvanCarroll I'm just saying the connection is definitly there, I remember being confused about ``shifting onto an array'' or ``shifting off of an array'' niether of those made any sense to me when i start, and they still don't know
now*
pmichaud masak/Matt-W: I'm a bit stumped then. It all appears to be working fine for me. 19:40
rjh_ I'll admit I only know what 'shift' does because of Perl 5 :)
pmichaud fwiw, I knew of 'shift' from shell programming.
EvanCarroll see.
masak I admit it seems to be something on my system that fails.
pmichaud "shift" is the way to process command line arguments in shell scripts.
masak I'll try a clean download and see if that helps.
pmichaud masak: that would be good.
rjh_ seems python doesn't have shift/unshift. it uses pop(0) to pop from the left side 19:41
perhaps they agree with you
pmichaud "shift" is also what we do with bit shifting , although it's normally called left-shift and right-shift.
japhb Speaking of shell scripts, I've clearly spent too much time around you guys, because yesterday I yak shaved for a couple hours figuring out how to do functional programming in bash. Quoting was an issue.
19:41 kate21de joined
jnthn back 19:42
pmichaud japhb: hey, don't blame me! Most of my code is highly dis-functional. :-)
jnthn I eated a cheezburger!
Greatest achievement all day...
japhb mmm, did that last night
TimToady Int:D is constrainted to defined, Int:U is constrained to undefined, and Int:P is constrained to stick out its tongue
pmichaud I had a cheezburger for lunch
sjohnson heheh
"eated"
jnthn TimToady: Oh, pretty.
masak I baked some delicious bread tonight.
pmichaud Once again, TimToady gets the colon. 19:43
TimToady I especially like :D
masak and then I eated it.
TimToady for a value that makes you happy
rjh_ :U looks aghast
japhb Once again, my IM client is going to mangle my code ...
TimToady it does at that
well alloweing either is probably :B or some such 19:44
EvanCarroll TimToady: shift/unshift, vs pull/shove ?
pmichaud :U looks like it's not quite complete. Like it's somehow poorly defined... :-)
rjh_ "An undefined parameter? In MY subroutine?"
pmichaud "Excuse me, you seem to be missing part of your mouth."
japhb "Oh, Programmer, how *could* you?!?"
TimToady then we just make -> $x default to Any:D and while -> work on binding success rather than boolean 19:45
japhb "Any:D" ... isn't that a famous rapper?
pmichaud ....binding success?
oh, just for while.
hmmm.
TimToady and if
pmichaud is that a special syntactic case, then? 19:46
TimToady but it's specially generally, not just special specially :D
pmichaud okay, I'm confused.
rjh_ DWIM++
TimToady while -> is already a special syntactic form
pmichaud while $a != 'hello' { ... } # is that binding failure?
TimToady or at least if -> is 19:47
masak how so?
TimToady since the if changes the default arity of the block
masak ah.
TimToady so does while
EvanCarroll How would i change perl6 to use pull/shove rather than shift/unshift? This is very important to me. 19:48
masak nice to know there are still things Perl 6 sweeps under the carpet for me! :)
pmichaud currently rakudo does if/while without changing the arity of the block.
and this sounds to me more like the arity of the block changing the meaning of the if/while
(unless it's truly syntactically special)
skids_ EvanCarroll: with macros, once they work (that is, not currently)
StephenPollei EvanCarroll, probably just make alias ::= or something 19:49
sjohnson what does the p6 community think of a $DUMMY compiler variable
TimToady I have no idea what the p6 community thinks of anything :)
most of them don't know they'll be a part of it someday
sjohnson much like the other $VARIABLES_TO_USE in Perl
pmichaud EvanCarroll: class Array is also { method pull(*@_) is export { self.unshift(@_); } ... }
EvanCarroll: (only works in Rakudo at the moment, will eventually change to 'augment') 19:50
sorry, s/unshift/shift/
rjh_ lol his point has been proven 19:51
pmichaud not really.
I understand shift/unshift just fine, I got pull backwards.
rjh_ no, i was just goading you
sorry :/
TimToady he was shifting your leg
rjh_ heh :)
Matt-W I found a bug! 19:52
TimToady is it a butterfly?
Matt-W rakudo: my $a = 0.14; $a *= 100; my $b = $a.floor; say $a == $b 19:53
p6eval rakudo fb2fd4: OUTPUT«0␤»
TimToady welcome to the world of floating point
pmichaud anyway, back to if/while
I have trouble seeing if foo() -> $y { ... } as meaning "does foo() bind to $y" as opposed to doing a boolean check 19:54
Matt-W rakudo: my $a = 0.14; $a *= 100; my $b = $a.floor; say $a == $b; say $a; say $b
p6eval rakudo fb2fd4: OUTPUT«0␤14␤14␤»
Matt-W I find it very hard to blame floating point arithmetic when they stringify to the same string
jnthn pmichaud: fwiw, the same...
TimToady Matt-W: 0.14 is not exactly represented in floating point
Matt-W okay
so can I have a sane number system?
TimToady rakudo: say 0.14.fmt("%16f") 19:55
StephenPollei Matt-W, it happens that sometimes the stringify rounds some
p6eval rakudo fb2fd4: OUTPUT« 0.140000␤»
pmichaud rakudo: my $a = 0.14; $a *= 100; say $a.fmt('%.15f');
p6eval rakudo fb2fd4: OUTPUT«14.000000000000002␤»
masak pmichaud: both errors still persist after clean install.
TimToady rakudo: say 0.14.fmt("%.16f")
p6eval rakudo fb2fd4: OUTPUT«0.1400000000000000␤»
TimToady rakudo: say 0.14.fmt("%.20f")
p6eval rakudo fb2fd4: OUTPUT«0.14000000000000001332␤»
Matt-W I think I'll just rip all my hair out now
pmichaud Matt-W: but if stringification is your criteria, then 19:56
StephenPollei p6 has Rat if you need it
pmichaud rakudo: my $a = 0.14; $a *= 100; my $b = $a.floor; say $a eq $b; say $a; say $b
p6eval rakudo fb2fd4: OUTPUT«1␤14␤14␤»
masak StephenPollei: p6 yes, r no...
TimToady pugs: my Rat $a = 0.14; say $a.fmt("%.20f");
p6eval pugs: ( no output )
StephenPollei otherwise with float point it's good idea never to test equality but if it's within epsilon
TimToady pugs: my $a = 14/100; say $a.fmt("%.20f");
pmichaud ...is 0.14 a Rat?
p6eval pugs: ( no output )
Tene japhb: how's it going? 19:57
Matt-W okay
so
using string operations to manipulate numbers
japhb Tene: delayed by sudden need for sandwich.
pmichaud I sometimes find myself wanting an "approx equal" operator. I've been using ~==~ for that lately.
Matt-W I've never been bitten by floating point like this before
pmichaud Unfortunately it looks too much like a longhorn cattle.
StephenPollei by p6 I mieant perl6 and I thought it was in S02 and others, unless I'm wrong
TimToady what's the non-Texas form of that?
Tene ≈ 19:58
pmichaud There is no non-Texas form. :-)
Tene ?
TimToady nothing else approximates Texas...
pmichaud longhorn cattle are unique to Texas :-P
japhb Tene: but I did think of a possible issue. How would I import the OpenGL functions in *pir*? (For example, if I wanted to write OpenGL::Simple, say, and I decided to write it in PIR, how would I do the import after loading OpenGL.pbc? Right now, there's a hand-coded import function in the OpenGL that you can call, but I'm hoping your work makes that redundant.) 19:59
StephenPollei ≈ and ~==~ might be useful but often you have to specify the epsilon
pmichaud StephenPollei: adverb
$a ~==~ $b :epsilon(0.1)
StephenPollei yes I keep forgeting , I need to get used to perl6ish stuff 20:00
Tene japhb: I can make it redundant shortly... speaking of which, do you expect to ever have any interest in importing a library written in a HLL from PIR?
Matt-W pmichaud: would that get given to the infix op as a named parameter?
japhb Tene: yes
Tene japhb: Okay.
pmichaud Matt-W: yes.
Matt-W pmichaud: cool. does that work in rakudo? 20:01
pmichaud Matt-W: no.
Tene japhb: I'm thinking an 'import' method on the 'parrot' compiler... what sort of API do you expect?
Matt-W aww
pmichaud Tene: my expectation has been 'import' method.
Tene: however, I need to review TimToady's latest changes to S11
StephenPollei rakudo: my Rat $a = 14 div 100; say $a
p6eval rakudo fb2fd4: OUTPUT«Malformed declaration at line 2, near "Rat $a = 1"␤in Main (src/gen_setting.pm:0)␤»
pmichaud rakudo doesn't have Rats yet. 20:02
Matt-W puts Rat on his christmas list
pmichaud You want a Rat for christmas?
Tene import(lang,list)? import(lang,delimited-string)?
pmichaud hmmmm....
Matt-W I want a non-floating-point non-integer numeric type :)
pmichaud Tene: why 'lang'?
Tene pmichaud: japhb just said the he expects to want to import foreign libraries from PIR. 20:03
masak pmichaud: I'm one step closer to encircling my problem. it doesn't trigger for small things like -e 'say "OH HAI"', but it triggers for a precompiled lib/Form/Grammar.pir in Matt-W's Form.
pmichaud: when I remove the .pir file, everything works.
Tene pmichaud: I could have a multi, though
Matt-W masak: not a very productive hackathon really, is it
StephenPollei also is S02 serious that there is going to be a rat and a Rat .. I don't know of any handware where rat is native, or am I overreading into what native means
masak Matt-W: you don't get this error?
Matt-W masak: works fine here 20:04
masak Matt-W: how bleeding is your Rakudo?
Tene masak: is the directory that contains perl6.pbc in your parrot search path?
masak ditto Parrot.
Tene masak: if not, how do you expect 'load_bytecode' to find it?
pmichaud Tene: I think I need to review S11 a bit more first before committing to a design. But I'll do that tonight.
masak Tene: what's a Parrot search path?
pmichaud Tene: the 'load_bytecode' is working for me no matter what directory I'm in.
Tene masak: the places that parrot looks for stuff when you say 'load_bytecode'
japhb Tene: even if I personally didn't think I would use it, I would still want to see it supported. It should be possible to do from PIR anything an HLL do. IOW, it should be a first-class language, just special to Parrot because it's "core". 20:05
Tene pmichaud: you must have a different setup from masak
skids_ Speaking of Rat, do we have a "require BigNum" or something specced? Or is it default, in which case is there a "no Bignum"?
pmichaud Tene: sure, we're wondering what that difference might be
because I don't have anything special in my setup afaik
Tene masak: run:
strace -estat -o perl6.log perl6 whatever.pl
then nopaste the perl6.log
pmichaud my guess is that masak might be getting a different perl6 binary 20:06
i.e., one that is looking for perl6.pbc in the wrong place.
masak I need to install strace, it seems.
pmichaud masak: how are you invoking perl6?
with a full path, or just as 'perl6'?
Tene pmichaud: I'll add a method to languages/parrot/parrot.pir, and if you don't like it, you can change it later, yes?
pmichaud Tene: sure, as long as it doesn't become spec.
masak pmichaud: I symlink from /usr/local/bin/perl6 into the fakecutable in the Rakudo directory. 20:07
Tene pmichaud: the 'parrot' compiler isn't specced at all anywhere yet.
pmichaud: this is just a utility method for when writing PIR.
pmichaud I know, but we still have to be careful about what people come to reply upon.
StephenPollei skids, I think Int in Bignum-ish , not sure about Num
Tene Okay.
masak pmichaud: that's how I usually do. but here I call the perl6 fakecutable directly, just to be safe.
pmichaud Technically the PGE stuff isn't specced anywhere either (outside of S05), but I still have to fundamentally deal with deprecation issues.
masak: and there's not some other libparrot floating around on the system that might be getting in the way? 20:08
masak pmichaud: how do I test that?
I really don't think so.
pmichaud I don't know. I really don't think so either, but I'm at a loss as to why you might be running into these issues. 20:09
I need a break -- bbiab 20:10
[particle]1 back from power-outage-induced lunch 20:11
masak I have four Rakudo/Parrot setups (locally) built on my computer right now: the one I usually run, one installed by proto, another installed by another copy of proto, and the one I just built.
pmichaud maybe try from a fresh shell? Perhaps it's hashed the wrong copy.
masak neither should interfere with any other.
masak tries from a fresh shell
pmichaud I'm grasping at straws, yes. 20:12
masak it works now. :) 20:13
pmichaud++
Matt-W: now, where were we? :) 20:14
TimToady std: $a == $b :ϵ(1e-10)
p6eval std 27012: OUTPUT«Potential difficulties:␤ Variable $a is not predeclared at /tmp/B1GzYvDNha line 1:␤------> $a == $b :ϵ(1e-10)␤ Variable $b is not predeclared at /tmp/B1GzYvDNha line 1:␤------> $a == $b :ϵ(1e-10)␤ok 00:02 36m␤»
masak oh right, I was to add a test.
20:15 pmurias joined
pmichaud having: $a == $b :ϵ might be nice :-) 20:15
masak oh wait. :/ I tested the wrong thing. it still fails when I precompile...
I guess I'll simply have to run un-precompiled tonight.
pmichaud masak: what are you precompiling?
skids_ wonders whether TimToady forgets the "my" like everyone else, or just likes to show off STD's error handling.
masak pmichaud: Matt-W's Form. 20:16
pmichaud: you can easily compile it by doing './proto install form'
pmichaud masak: and then you're running it directly with Parrot?
masak pmichaud: then I'm running 'perl6 -e "use Form::Grammar"'
pmichaud Hmm.
StephenPollei $a == $b :ϵ(1e-10) looks nice TimToady++
Tene masak: strace log?
masak so, .pm file works, .pir file doesn't.
pmichaud I didn't try that on my system. 20:17
masak Tene: I'll try to find strace for Mac OS X.
Tene masak: oh, OS X. No idea.
TimToady skids_: never attribute to malice that which can be easily explained by stupidity :)
Tene I think they have something else there...
bkeeler dtrace
japhb Tene: Does c.'export'() just add subnames to the DEFAULT tag for now? If so, what will be the syntax to add something to a different tag, or to multiple tags?
masak ah, dtruss.
skids_ Does :ϵ have a default. I mean other than 50 points for the medium speed space invader? 20:18
pmichaud japhb: they'll be additional parameters to 'export'
I don't know yet if they'll be named or positional
japhb pmichaud: OK, thanks. 20:19
Tene japhb: it just adds to the 'DEFAULT' and 'ALL' tags. The syntax for anything else is whatever you want it to be. :)
skids_ Oh I guess :ϵ would be ϵ => 1. Alas.
japhb heh
20:20 mizioumt1 left
pmichaud actually, :ϵ is ϵ => True 20:20
(which numifies to 1, but "true" could be taken to mean some other constant in this case)
s/constant/value/
TimToady so $a == $b :ϵ is integer comparison :) 20:21
pmichaud oh?
TimToady I'm just shifting your leg
pmichaud 2.3 == 1.4 :ϵ # not integer comparison
2.3 == 1.4 :ϵ(1) # not integer comparison 20:22
TimToady I agree it could sigwise match Bool and do something else
StephenPollei almost but not quite integer comparisom Int(4.5) == Int(5.2) and 4.5 == 5.2 :ϵ -- differ
masak Tene: I don't really expect this to help/enlighten: gist.github.com/124489
skids_ Could be taken as "reasonable defloatification" for (Int, Num)
pmichaud I like the sigwise match bool
Tene No, not helpful. :)
pmichaud and it could use $*EPSILON or something as the default :-) 20:23
TimToady well, some small percentage of the large magnitude value
japhb TimToady: How would you express matching to a particular number of significant bits, so that the looseness of the equality match is automatically proportional to the magnitude of the operands?
TimToady *larger
StephenPollei well I see $.EPSILON being part of class more helpful
TimToady yes, though for floaters you want sigbits as japhb suggests 20:24
StephenPollei japhb, I think yet another adverb might be needed
TimToady just different for single vs double precision, for instance
japhb Right.
skids_ There is a letter for relative error I think...
StephenPollei num is always double iirc, and Num I don't know 20:25
japhb skids_: that would be perfect
TimToady yes, but num32 is not always double :)
japhb StephenPollei: smaller nums are spec'ed.
Down to num16, at my suggestion. :-)
StephenPollei yes I show the native by size ones
I also wondered what that mean for cpus without floating point, I guess the compiler decides 20:26
masak Matt-W: just made my first commit. :)
TimToady is looking forward to ubiquitous num128s so time doesn't have to return a Rat
skids_ en.wikipedia.org/wiki/Relative_error
TimToady Num is defined to allow Rat internals lazily converted to Num, though it's still more than a bit handwavey 20:27
StephenPollei nad it isn't speced how much is matissa and how much is exponent .. hopefully terms right
TimToady anyway :ϵ is close enough for now; I'll leave the rest to smart people :) 20:28
20:28 gegloe left
pmichaud Isn't that the definition of :ϵ ? ;-) 20:29
TimToady 1..10 :Δ(3)
skids_ Rat still needs some sort of .num, .den specced IIRC. Also, have the advatages/disadvantages of Rat does Pair been weighed?
pmichaud note that ".num" is taken.
japhb skids_: so 1.1 == 1.0 :η(...) then, but what should the ... be?
pmichaud TimToady: in S11, how much do you think is likely to change...?
(w.r.t need/defines/etc)
TimToady defines is still a bit wobbly, syntaxwise 20:30
pmichaud is it "generally correct" now, or do you think there's a good chance it will major shift?
TimToady thinking about "defines *"
or maybe making that the default
it's all just syntax now
pmichaud also, are tags simply identifiers that begin with a colon?
skids_ japhb: I meant there is a greek letter used for that, not that it's specced -- but that would be a number < 1 usually. 20:31
Tene japhb: do you expect that foreign namespaces will be imported too?
TimToady they're just adverbials, syntactically
japhb Tene: urm. Explain what you mean by that, please.
pmichaud the spec doesn't show them parsing as adverbial, though.
Tene japhb: if I do:
TimToady are you referring to :MY<:ALL>? 20:32
japhb skids_: I am just wondering if it's more useful to have a decimal number in the eta adverb, or some number of sigbits.
pmichaud yes
Tene compiler.import('perl6','Foo::Bar') would you expect that ['perl6';'Foo';'Bar'] is also available as ['parrot';'Foo';'Bar'] ?
pmichaud I was also referring to is export(...), but I see there they could be adverbial
Tene I guess that's not as much of an issue for PIR, but other languages often expect that.
TimToady it's handling the string ':ALL' specially
pmurias ruoso_: ping
DanielC Yay \o/ I now have full RPN calculator with bottom-up grammar. It's actually quite simple, but I had to learn the basics of grammars first.
TimToady :MY<$x $y>
ruoso_ pmurias, pong
skids_ japhb: potentially putting 1/η in there would be prettier. 20:33
TimToady '$x' and '$y' are just strings here
pmichaud right.
20:33 rocket_guatemala joined
pmichaud I was looking for ":ALL is a string there". 20:33
(which I got)
TimToady unless we go with something supra-declarational like defines (my $x, my $y)
pmichaud in the case of
sub bar is export(:DEFAULT :others) {...} # :DEFAULT, :ALL, :others
those are named parameters, syntactically?
TimToady sec... 20:34
pmichaud sorry, named arguments
pmurias ruoso_: if new signatures will be used for subs and methods how will we bootstrap them?
japhb Tene: that makes sense. Might want to be able to turn that off, but as a default it makes sense to me.
pugs_svn r27013 | rocket++ | Created the t/spec/S19-command-line-options folder and started moving the t/run tests into it.
r27013 | rocket++ | Moved 01-dash-uppercase-i.t to t/spec/S19-command-line-options and added smartlinks to it.
Tene japhb: I'll leave it commented out for now, and when someone wants it, they can work out the semantics.
japhb skids_: true. Now figuring out a good syntax for that without confusing mathemeticians ...
Tene: OK
japhb can barely keep up with IRC today and still continue to actually code ... 20:35
ruoso_ pmurias, I think it might be a mildew option like "--core-stage x" where that modifies the emitted code...
TimToady is export is sugar for :export, so the question is whether :export(:DEFAULT :others) is treated as arglist or inside ()
ruoso_ probably '--target-stage x'
masak Matt-W: oh! I see now why you dreaded the numeric parsing. :D 20:36
pmurias ruoso_: the other option would be to use the default signature when defining Signature
ruoso_ what do you mean?
pmurias using the implicit :(*@_,*%_) signature
but that seems inelegant
TimToady if :() is just (), then they're not named params, just as foo((:a :b)) aren't 20:37
pmichaud right
TimToady arguable the same for :MY()
pmichaud also, S11 doesn't seem to give an example of importing from a tagset (via use)
TimToady though :MY($x, $y) is problematic
japhb Tene: oooh, just ran across a problem. What about renames? My custom exporter currently does this because it needs to rename a couple NCI functions before exporting:
gl_namespace.'export_to'(to_namespace, export_list)
gl_namespace.'export_to'(to_namespace, export_renames)
Tene japhb: will this work for you: nopaste.snit.ch/16793 ?
pmurias ruoso_: wouldn't it be better to use a pragma instead of a compiler flag?
pmichaud unless it's 20:38
pmurias ruoso_: like use adhoc-signatures; ?
TimToady but we could go for defines my(:TAG, $x, $y) possibly
Tene japhb: That's currently not handled at all.
ruoso_ pmurias, the Perl 6 code will be the same
pmurias, it's not the code that changes, but the compiler
pmichaud use Sense <common @horse>;
TimToady and then it looks more like a real my declaration
pmichaud "common" is referring to the sub?
or is it a tag? 20:39
TimToady not a tag
was thinking sub
but that's also a problem
pmichaud yes
Tene japhb: if you work out an API for it with pmichaud, I'll implement it.
Might take me a bit.
pmurias ruoso_: could you explain? 20:40
TimToady use Sense my(&common, @horse) would also need to have the my redefined like defines would do explicitly
pmichaud Tene/japhb: I'll have an API when I get S11 worked out
jnthn rakudo: multi foo(Int, Any) { }; multi foo(Any, Int) { }; foo(1,1)
p6eval rakudo fb2fd4: OUTPUT«Ambiguous dispatch to multi 'foo'. Ambiguous candidates had signatures:␤:(Int ::TYPE_CAPTURE17, Any ::TYPE_CAPTURE18)␤:(Any ::TYPE_CAPTURE19, Int ::TYPE_CAPTURE20)␤in Main (/tmp/Ohi1kel73X:2)␤»
TimToady unless we do something like use Sense defines my(&common, @horse)
masak Matt-W: I'm wondering if not some, most or all of the 'regex' keywords in Form::Grammar::Format couldn't be 'token' instead...
ruoso_ pmurias, eventually we might recompile the CORE code using the next --target-stage and get a more high-level code redefining the types...
pmurias, so the CORE is plain Perl 6 20:41
but the compiler evolves from a stage 0 compiler
that only uses the SMOP c implemented types
to later using the types defined in the CORE to build the CORE
and that's bootstrap
so the code in the CORE is always the same
Tene pmichaud: Okay.
pmurias shower& 20:42
Tene japhb: do you need renaming to get OpenGL working at all? Or will that just be nice eventually?
TimToady Module my ($x, $y) our ($z) # infix declarators in general?
20:44 [particle]1 left
japhb Tene: (re: your PIR module loading paste): So the parent code loads the 'parrot' language, and asks its compiler object to import module 'Foo'. The parrot compiler object then knows to do a load_bytecode for the Foo.pbc. At :load time, Foo.pbc loads the 'parrot' compiler object again, registers its exports, and returns to the parrot compiler, which then does an 'import' operation, making use of the newly created export tags. Then finally 20:45
the original compiler.import('Foo') returns, and you're ready to go. Do I have that all right?
pmichaud seems like the infix declarators would want to follow 'defines'
TimToady Module defines (my ($x, $y); our $z) is more likely to extned well
japhb Tene: Actually required. Can't get the callbacks working without the rename.
TimToady nod
masak Matt-W: ping
pmichaud and tagsets? 20:46
20:46 REPLeffect joined
Tene japhb: that's right 20:46
TimToady my (:foo, $x, $y) maybe
japhb Excellent. Let's hope the compiler object doesn't leak when reentered. ;-)
20:46 ST47 joined
pmichaud is there a my/our default? 20:46
20:47 simcop2387 joined
pmichaud Module defines (:foo, $x, $y) 20:47
or is the my/our required?
TimToady defines :(:foo)
pmichaud what about caller and context?
TimToady yeah
it just doesn't quite mesh with normal binding either
pmichaud or maybe just context 20:48
Tene japhb: If you paste some code showing what you'd like to do, I'll implement it, as long as you're comfortable potentially changing it in a few days after pmichaud potentially decides on API changes here.
TimToady maybe context is a declarator then :)
pmichaud there is that possibilty.
maybe every trait is really a declarator. :-|
japhb I'm fine with potential energy.
pmichaud since "is also" became "augment"
and "is context" might become a declarator....
TimToady lets make all types into declarators, and use * to dereference things 20:49
ruoso_ enjoys the idea of using simply "my $*a" instead of "my $a is context"
pmichaud we still want the sense of context<rw> I suspect 20:50
ruoso_ "my $*a is rw"?
20:50 [particle]1 joined
pmichaud that makes it sound like the variable itself is rw 20:50
japhb Tene: It would be nice if c.import() took all of: a space separated stringified list (as it does now), a real array, or a rename hash. The latter two make it match the design of namespace.'export_to'(), which I use now.
pmichaud we run into trouble with
sub foo($*a is rw) { ... }
which is different from sub foo($a is rw is context) { ... } 20:51
Tene japhb: it's import that you need renames for?
japhb japhb: Tene: mind you, I'm not sure how you express the rename operation when you're just adding to tagsets.
20:51 rocket_guatemala left
ruoso_ hmmm... 20:51
20:51 jferrero joined
ruoso_ too tired to think 20:52
TimToady I need to go to $job
japhb Tene: It doesn't matter to me if 'export' or 'import' actually does the rename. Just that a function called ['parrot'; 'OpenGL'; 'glutcbFoo'] needs to be named ['perl6'; 'OpenGL'; 'glutFoo'] somehow.
TimToady anyway, apart from all that, it's just syntax :)(
pmichaud TimToady: ultimately, you see tags more like named arguments than strings, yes? 20:53
Tene japhb: I'd very much prefer that it happens in export, so that any other language that loads it will also see the renamed versions.
TimToady yes, though processed more like slurpy args than bound like named
pmichaud so, more like pairs than strings
japhb I guess you need to mark the rename in 'export', then 'import' knows what to do with the mark.
Tene Right. 20:54
pmichaud Tene/japhb: A rename simply means that the exported symbol has a different name in the EXPORT::* namespace
Tene Exactly
japhb pmichaud: speclink? 20:55
Tene japhb: so, maybe passing a hash to export() ?
pmichaud hash... probably not
japhb Tene: yes, that's what I meant five minutes ago, but I typo'ed confusingly.
Sigh. This is the problem with being too tired to keep up.
Should have been: "Tene: It would be nice if c.export() took all of ..." 20:56
Tene ah
:)
pmichaud I think I have enough info from TimToady now to indicate how I'd like import/export to work
I'll think it through tonight and either write up a spec or implementation tomorrow.
japhb I guess then I will put this on the back burner until then. 20:58
pmichaud it shouldn't be too difficult for us to implement
20:58 bkeeler left
pmichaud other than the syntax (in Perl 6) :-P 20:58
jnthn pmichaud: I've got some progress on the "Perl6MultiSub everywhere" branch.
pmichaud jnthn: nice
faster? slower? 20:59
jnthn pmichaud: Like, we can actually build and startup now and pass some of the sanity tests.
A tiny bit but not much slower, but I didn't do any signature caching yet.
pmichaud is there a strong reason to switch over if it's not significantly faster?
jnthn Several. 1) Being able to actually overload operators and various other builtins that we can't now. 2) Be able to toss the gen_junction special case code since it should all be handled by the auto-threading in Perl6MultiSub. 3) Correct dispatch semantics in line with the Perl 6 algorithm. 21:01
pasteling "japhb" at 76.191.190.8 pasted "Tene, pmichaud: Rough draft of OpenGL.pir symbol export (to be called by :load routine)" (49 lines, 1.3K) at sial.org/pbot/37059
pmichaud I thought we could solve (1) and (2) by writing the operators in setting
jnthn 4) Won't need the hack in infix:= any more.
pmichaud and (3) also
for (4), we could manually switch it over to a Perl6MultiSub instead of having it happen automatically, yes? 21:02
21:02 decasm left
jnthn We could do all of those things yes. 21:02
Though we would pay the price of !SIGNATURE_BIND
And also, we'll run into various other issues that I'm running into now. 21:03
Example:
pmichaud we could have a pragma/trait that removes the call to SIGNATURE_BIND
jnthn .sub 'infix:+' :multi('Complex', _)
pmichaud but in some sense we need/want SIGNATURE_BIND even for the operators
jnthn .sub 'infix:+' :multi(_, 'Complex')
We can't just translate this into Perl 6.
pmichaud why is that an issue?
jnthn Because it's an ambiguous dispatch.
If you dispatch with two complex numbers, under Perl 6 MMD that's an error. 21:04
Because the candidates are tied.
pmichaud Perl 6 already supports that, though.
jnthn ?
pmichaud (looking up reference
ruoso_ decommute & 21:05
21:05 ruoso_ left
jnthn We get away with it now in Rakudo because we're using Parrot's MMD algorithm. 21:05
pmichaud yes, I know. I'm saying that commutativity of types is already in the spec. 21:06
pmurias sleep&
21:06 pmurias left
jnthn Explain? 21:07
pmichaud there's a way to specify multiple signatures on a sub (or, at least, there was)
jnthn Ah, that.
Yes, we don't support that in Rakudo at the moment... 21:08
pmichaud all that the infix:+(Complex) case is doing is handling the case where either argument is a Complex.
jnthn Right.
pmichaud support that, and then the multi issue you just gave goes away
jnthn Sure, we'd have to write the 3 signatures, but they could share a body.
pmichaud exactly. 21:09
jnthn Hmm. I hadn't planned to do that, but maybe it's better sooner than later.
pmichaud that's all that the current :multi(...) is doing in Rakudo/Parrot
jnthn Sure, I know that.
pmichaud that might be a bigger win than trying to convert things to Perl6MultiSub, if we're not seeing a significant speed improvement
jnthn I'm just pointing out that there's some things like this that crop up when looking at making the switch to the Perl 6 multi-dispatch algorithm.
pmichaud right.
jnthn We're just at a point now where we don't need to cheat on these things any more, and I'd rather we stopped doing so and got the semantics correct. 21:10
pmichaud and I still like the idea of converting to Perl6MultiSub lazily (if that's at all possible)
I think the better non-cheat is to move them into settings
rather than try to keep them working in PIR 21:11
jnthn Well the whole point of the branch I'm working on is that we don't have to convert at all.
We just create the correct, mapped PMC in the first place.
pmichaud fair enough.
If it doesn't give us a speed hit, I'm okay with doing it.
jnthn (Since in theory, not having to convert will be less work at startup.) 21:12
pmichaud (big speed hit, small speed hit okay)
I just don't know that it's worth troubleshooting many cases that really ought to move to setting anyway
as you said, let's fix it so that we don't need to cheat at all :-)
jnthn Speed hit is like a few percent.
Or less.
pmichaud that's no problem. 21:13
I'd only be worried about 10%-20% or more.
and I'm more concerned about runtime speed than startup.
jnthn And I kinda wonder if we're losing rather than winning because hll_map can be slow (as you noted for Int etc)
pmichaud well, hll_map on subs shouldn't be an issue.
and hll_map to a PMC type shouldn't be an issue. 21:14
I think it's primarily when we hll_map to a ['parrot';'Object']
jnthn Yes
That's going to be slower. :-(
Not sure how much we can improve on that. 21:15
pmichaud I'm not sure what that's going to mean for things overall, unfortunately.
It might mean we have to come up with our own Perl6Object PMC type
masak rakudo: Perl6::Grammar.parse("say <OH HAI>", :action(Perl6::Grammar::Actions.new)); say ?$/
pmichaud which would mean ditching P6object altogether, or turning it into a PMC type
p6eval rakudo fb2fd4: OUTPUT«1␤»
masak \o/
pmichaud masak: for non-declarational things, I think :action(...) might not be needed 21:16
jnthn Well, to do a bunch of S09 things I'd figured we'd might need to do somehting like that.
masak pmichaud: now, is there any way I can look into $/ ?
pmichaud masak: sure, it's just a match object.
masak pmichaud: please show me.
pmichaud rakudo: Perl6::Grammar.parse("say <OH HAI>"); say $/<TOP>; 21:17
jnthn
.oO( .perl )
p6eval rakudo fb2fd4: OUTPUT«Use of uninitialized value␤␤»
masak jnthn: I tried. it doesn't have one.
pmichaud rakudo: my $match = Perl6::Grammar.parse("say <OH HAI>"); say $match<TOP>;
p6eval rakudo fb2fd4: OUTPUT«Use of uninitialized value␤␤»
jnthn aww.
masak neither can it be turned into a Hash.
pmichaud rakudo: my $match = Perl6::Grammar.parse("say <OH HAI>"); say $match.hash 21:18
p6eval rakudo fb2fd4: OUTPUT«No result object␤in Main (/tmp/fVyY03YoaO:2)␤»
pmichaud rakudo: my $match = Perl6::Grammar.parse("say <OH HAI>"); say $match.PARROT
p6eval rakudo fb2fd4: OUTPUT«Method 'PARROT' not found for invocant of class 'Perl6;Grammar'␤»
pmichaud okay, I have to rework the inheritance a bit more.
masak just wanted to alert you to the remaining issues. :)
jnthn rakudo: my $match = Perl6::Grammar.parse("say <OH HAI>"); say $match<statement_block>
p6eval rakudo fb2fd4: OUTPUT«say <OH HAI>␤»
jnthn pmichaud: There you go. :-) 21:19
pmichaud rakudo: my $match = Perl6::Grammar.parse("say <OH HAI>"); _dumper($match);
masak jnthn: hm, introspection by guessing... :P
p6eval rakudo fb2fd4: OUTPUT«Could not find non-existent sub _dumper␤»
jnthn masak: No, by looking at grammar.pg :-P
"Educated guessing" ;-0
masak jnthn: not sure that scales well...
pmichaud rakudo: class Perl6::Grammar is also { method perl() { Grammar::perl(self); } }; my $match = Perl6::Grammar.parse("say <OH HAI>"); say $match.perl; 21:20
p6eval rakudo fb2fd4: OUTPUT«Null PMC access in invoke()␤in method Perl6::Grammar::perl (/tmp/HDQYLrGhRA:2)␤called from Main (/tmp/HDQYLrGhRA:2)␤»
pmichaud rakudo: class Perl6::Grammar is also { method perl() { Match::perl(self); } }; my $match = Perl6::Grammar.parse("say <OH HAI>"); say $match.perl;
p6eval rakudo fb2fd4: OUTPUT«Method '!_perl' not found for invocant of class 'Perl6;Grammar'␤»
pmichaud bah. 21:21
I'll have to rework the inheritance a bit.
jnthn: I'll need to run for a bit (more)
jnthn pmichaud: OK. I'm getting tired and making dumb mistakes anyway, so think I'm going to stop coding here anyway.
pmichaud on Perl6MultiSub, to the extent that it makes existing (source) code simpler, I'm very happy with it. If it makes the source code more complicated, I'd prefer to stick with our existing approach (and work on moving things to settings) unless we get a speed improvement. 21:22
masak pmichaud: I think that last one worked, but struck an error inside the Match somewhere, like the RT bug I submitted the other day.
pmichaud it worked, but PCT::Grammar ends up not having a '!_perl' method.
because it's not really inheriting from Match
which if it did, would solve the .perl problem 21:23
thus
jnthn pmichaud: I think from what I've seen today moving things to setting is going to be a better way to go.
pmichaud I'll have to rework the inheritance a bit.
jnthn: I'm in agreement with that.
jnthn pmichaud: But I think I want to get the alternative signatures syntax in place too.
pmichaud yes, that's also very important to this.
jnthn Sure. I hadn't realized that until now.
pmichaud I'll see if I can get the "is equiv()" stuff in place.
with "is equiv()", we can move more operators to setting 21:24
jnthn OK, great.
Combining the load init blocks together would also be a win.
pmichaud I'll look at that also.
jnthn I'm also thinking we need to try and speed up signature creation some.
pmichaud that would be a big win.
jnthn Right, I'm just not sure how best to do it yet. 21:25
pmichaud we may get a bit of a win if/when I can switch ro<->rw
21:25 hanekomu_ is now known as hanekomu
pmichaud the existing sig creation stuff was more "straightforward, works" than "fast" 21:25
jnthn Aye.
pmichaud as I said, I'm much more concerned about runtime performance than startup cost.
(startup cost is easier to explain away to people who say "performance stinks" :-) 21:26
jnthn Sure, but we can't ignore startup cost completely either.
pmichaud no, but we can point out ways that it's likely to be improved
that's harder to do with runtime performance right now. 21:27
jnthn True.
We'll know better with profiling tools.
Plus when the everlasting calling conventions refactor lands.
pmichaud oh, if you have any comments to add to the emails I sent to rdice earlier, they might be helpful. :-)
21:27 gegloe joined
pmichaud chromatic expressed an interest in working on that. 21:28
jnthn I intend to, will today or probably tomorrow since I will be more coherent then.
pmichaud as in "soon"
jnthn The profiler or the claling conventions?
The profiler I know about.
pmichaud calling conventions
jnthn Oh, wow.
That would be good.
We might actually get them fast. :-)
Or fastish.
pmichaud as in "make it work immediately after 1.4 is released" (since it's unlikely to be ready before 1.3 releases, and allison doesn't want significant changes between 1.3 and 1.4)
21:29 skids_ left
jnthn Ideally, I'd like to work in August on trying to get our signature stuff straightened out a bit. 21:29
So that's just a bit after 1.4... 21:30
pmichaud sounds good, I'm not sure how calling convention will play out.
jnthn Sure, me either...gets frustrating waiting and wodnering.
pmichaud I'm more than a little disappointed that it won't be here in 1.4, especially since I flagged it as critically important at PDS
(I flagged it as critically important for 1.0, even, but it got dropped from the list somehow)
jnthn Right, it's slipped many months now. 21:31
21:32 eternaleye joined
pmichaud afk for a while 21:33
jnthn k
21:33 Whiteknight joined
masak rakudo: module A { role B {}; class C does B {} } 21:39
p6eval rakudo fb2fd4: OUTPUT«Null PMC access in find_method()␤in Main (/tmp/w0Ci3yOnXG:2)␤»
21:46 [particle]1 left
masak wow! I was naughty and wrote the implementation before the tests... so I git-stashed the implementation, wrote a few tests, made sure they failed, unstashed, and watched them pass! git++ 21:51
21:52 iblechbot left
viklund_ tdd ftw!!! 21:52
21:53 eternaleye left
masak tla ftw too :) 21:53
viklund_ what's too short for? 21:54
masak "I Think You Know". 21:55
Matt-W: I notice you're not testing for interesting failure modes in t/06-form.t. 22:09
Matt-W: I'm momentarily stuck with the numeric field stuff, so I might add a few tests for that. 22:10
22:15 nihiliad left 22:16 snarkyboojum joined
masak rakudo: say 'abcdefghi'.comb(/<[aei]>/).perl 22:18
p6eval rakudo fb2fd4: OUTPUT«["a", "e", "i"]␤»
masak rakudo: my $a="bacadae"; for $a.comb("a") -> $c { say $c }
p6eval rakudo fb2fd4: OUTPUT«No applicable candidates found to dispatch to for 'comb'␤in Main (/tmp/NDXRKfHIWJ:2)␤» 22:19
masak rakudo: my $a="bacadae"; for $a.comb(/a/) -> $c { say $c }
p6eval rakudo fb2fd4: OUTPUT«a␤a␤a␤»
22:20 M_o_C left
masak rakudo: say /foo/ ~~ Regex 22:23
p6eval rakudo fb2fd4: OUTPUT«1␤»
22:23 clintongormley left
masak the signature at src/setting/Any-str.pm:134 still says Code. should it be Regex instead, like in S32? 22:24
rakudo: say 'abcdef'.comb({1}).perl
p6eval rakudo fb2fd4: OUTPUT«Method 'to' not found for invocant of class 'Int'␤»
masak submits rakudobug 22:25
(I can change the source too, if someone just nods.)
22:37 frew|work joined, snarkyboojum left 22:44 snarkyboojum joined
jnthn bratislava.pm.org/down/doc/Perl6-Sp...r27013.pdf # jozef++ from my PM group does pod2docbook stuff and managed to PDF-ize the Perl 6 spec! 22:48
22:50 eternaleye joined
DanielC jnthn: I managed to make the RPN calculator (several hours ago). It's very nice. In the end I completely rewrote the program. 22:50
masak jnthn: nice!
DanielC jnthn: I have a very nice grammar for it, and it can handle complex expressions like '5 4 + 3 / 5 3 - *'
jnthn DanielC: Awesome!
22:50 alester left
DanielC jnthn: I'll put it somewhere in perl6-examples. 22:50
22:53 ashizawa left
masak DanielC++ 22:53
DanielC :-)
masak ok, heading home for the night.
DanielC commits
masak see y'all tomorrow. 22:54
22:54 masak left
DanielC night 22:54
22:55 ashizawa joined, Whiteknight left
jnthn is tired 22:59
So, rest time
night all 23:00
DanielC night
23:02 presh is now known as bigpresh 23:03 nsh joined 23:05 justatheory joined 23:06 bigpresh is now known as bigpresh_, Muixirt left 23:08 frew|work left 23:17 skids_ joined 23:26 bigpresh_ left, bigpresh joined, bigpresh left 23:27 hanekomu left, bigpresh joined, bigpresh left 23:28 bigpresh joined, bigpresh left, bigpresh_ joined, bigpresh_ left 23:29 bigpresh_ joined, bigpresh_ left 23:30 bigpresh_ joined 23:32 kate21de1 joined, Lectus joined 23:34 Limbic_Region joined 23:35 unitxt left 23:41 hercynium joined
sjohnson the camel book shuold have on the front page 23:44
"Review the core modules at perldoc.perl.org/5.8.8/index-modules-A.html"
i can't believe how many times i've reinvented the wheel
23:48 kate21de left 23:50 rewt joined 23:53 amoc joined
Khisanth sjohnson: no, search CPAN :) 23:58
23:59 DemoFreak left

Powered by Cro and the Raku® Programming Language.

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

Thank you!