»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:08 noam_ joined 00:12 noam left 00:14 whiteknight left 00:18 whiteknight joined 00:22 packetknife left 00:24 awoodland left 00:36 benabik joined 00:54 havenn joined 01:05 scott__ joined 01:15 havenn left, felipe joined 01:18 PacoAir left 01:22 bbkr left 01:23 bbkr joined
sorear good * #perl6 01:37
araujo sorear, o/
01:38 sivoais joined 01:40 autarch joined
autarch hi all ... I'm working on an article for LWN on the state of Perl 5, and I want to talk a bit about the relationship between 5 and 6 01:41
is there an official statement on this anywhere I can quote?
sorear hello autarch
autarch hi
sorear TimToady: ping
autarch yeah, a quote from Larry would be ideal
01:41 gv left 01:44 isBEKaml left 01:51 odoacre left 01:52 odoacre joined 01:56 whiteknight left
autarch paste.scsys.co.uk/184822 02:04
how does that seem?
paste.scsys.co.uk/184822?tx=on&...rmat+it%21 - more readable
sorear I approve 02:08
autarch I wouldn't mind getting TimToady's blessing too
sorear I want to get more than just a blessing from him... 02:09
sorear waits. 02:10
02:13 scott__ left 02:16 scott__ joined 02:22 wolfman2000 joined 02:24 ribayr joined
sjn autarch: nice start on a text 02:27
autarch thanks
sjn not sure if "
not sure if "The Perl 5 of today is largely similar to the Perl 5 of ten years ago" is the best way to put it 02:28
some of the most active people in the perl community have started saying that "This is not your father's Perl" (albeit cheekily :) 02:29
autarch I'm more talking about the language itself
maybe I should just say "The Perl 5 language of today ..."
sjn Perl 5 has some really major differences today as compared to 5 years ago 02:30
and much comes from the "Modern Perl" "movement"
(maybe it's too much to call it a "movement" but there are some iteresting diversions from the good-old ways of thinking Perl there) 02:31
autarch But the Perl 5 language as interpreted by perl is not all that different 02:32
sjn there are some new additions that have made Perl 5 a more expressive and easier to use language 02:33
if you compare with Perl 10 years ago, there's a LOT
autarch I still maintain that they are "largely" similar
sjn much comes from the experimentation that has been done on CPAN, but some are core changes
autarch as in, it's been an evolution, whereas Perl 6 is a revolution
sjn yeah 02:34
that's true
autarch I'm just trying to contrast the two languages
sjn but do keep in mind that you can get significant changes with evolution too :)
autarch sure
sjn it would be wrong to give an impression that Perl 5 hasn't changed much in 10 years
that's patently untrue
autarch I'll try to think of a better way to phrase it 02:35
sjn (and directly misleading, imo)
there's been a very useful co-existence between P5 and P6 for several years
autarch I do say that 02:36
sjn and several of the cooler P6 features have been attempted to "backport" to P5 in the form of new CPAN modules
of which the biggest and best know example is the Moose OO system
known* 02:37
autarch I'm going to cover some of that later in the article
sjn talking about Perl 5 today without mentioning the newer modules and frameworks would be rather sad :)
autarch yeah, like I said, later 02:38
that's a small excerpt of a larger article
(which is mostly unwritten)
sjn but yes, Perl 6 is a project where the goal is to make a wonderful Perl-like language with all kinds of awesome features, but without the straightjackets that come from having to care about backward compatibility 02:39
and frankly, some of the stuff that's ready today can easily be called "mind-blowing" :) 02:40
</biased_opinion> :)
autarch: re: language, you might want to find more ways of saying "Perl 5" and "Perl 6" though :) 02:42
there's quite a lot of repetition
autarch I think Perl 6 has lots of cool ideas and features myself 02:43
well, I could write P5 and P6, but that seems to piss some people off
sjn "the language", "version 4", "the #perl6 community's product", "the different P6 implentations" 02:46
plenty of ways to vary the language ;) 02:47
(I'll leave it to you to do the rest ;) 02:48
02:52 noam_ left 03:02 NamelessTee left 03:11 alester joined 03:15 orafu left 03:16 orafu joined
sorear sjn: I think it's safe to say autarch is familiar with modern perl; ey's one of the Moose bigshots ;) 03:46
03:47 marmalade left
sjn he is? :) 03:51
sjn had no idea
autarch++ for keep it cool among n00bs :)
ooh 03:52
Dave Rolsky :)
autarch yeah, that's me 03:53
sjn autarch: sorry about my sillyness then :)
autarch no, no problem
sjn had no idea <_<
sjn should really learn to stop being "helpful" at 3am :-P 03:55
Tene gist.github.com/1955491 -- bug I found with current rakudo HEAD in grammars 03:59
That's as small as I was able to golf it.
04:00 am0c joined 04:04 NamelessTee joined 04:14 packetknife joined
alester FYI to everyone: I just transferred rakudo.org off of GoDaddy. 04:20
If there are any problems at all, please let me know.
Tene My company has some pretty bad history with godaddy. :( 04:22
alester petdance.com/2012/02/why-im-finally...g-godaddy/ 04:23
geekosaur godaddy needs to be gone,daddy,gone... 04:49
:p
05:05 machine2 joined 05:06 skids left, autarch left 05:25 woosley joined 05:29 xinming left, xinming joined 05:30 xinming left, xinming joined, birdwindupbird joined 05:33 machine2 left 05:38 machine2 joined 05:40 machine2 left 05:41 machine2 joined 06:19 lestrrat left 06:20 lestrrat joined 06:27 alester left 06:34 noam joined 06:37 woosley left, sivoais left 06:38 sivoais joined
moritz good morning 06:38
06:38 sivoais left 06:39 sivoais joined
TimToady howdy doo 06:39
06:41 packetknife left
sorear good mornign moritz. 06:43
hello TimToady, you are needed
06:49 alvis left
TimToady a lot of people seem to have that impression 06:52
06:54 wtw joined
sorear TimToady: irclog.perlgeek.de/perl6/2012-03-02#i_5232938 06:54
06:55 alvis joined
TimToady well, right now I have to go pick up future daughter-in-law from the airport 06:59
I saw that earlier, but didn't think of anything particularly clever to say that hasn't already been said too many times
07:00 NamelessTee left
TimToady the only thing that will prove Perl 6 to the world is doing it 07:00
afk &
07:08 machine2 left 07:09 alvis left, machine2 joined 07:15 noam left, noam_ joined 07:22 machine2 left 07:36 mj41 joined
moritz nom: class A { method &[]($x) { say 'there' } }; A.new.[5] 07:43
p6eval nom d0a245: OUTPUT«===SORRY!===␤Missing block␤at /tmp/EgK4pqqBRv:1␤»
moritz nom: class A { method @[]($x) { say 'there' } }; A.new.[5] 07:44
p6eval nom d0a245: OUTPUT«===SORRY!===␤Missing block␤at /tmp/EtiYZS9e9k:1␤»
moritz nom: class A { method @.[]($x) { say 'there' } }; A.new.[5]
p6eval nom d0a245: OUTPUT«===SORRY!===␤Missing block␤at /tmp/3UtIBcOaf6:1␤»
moritz nom: class A { method @.[$x] { say 'there' } }; A.new.[5]
p6eval nom d0a245: OUTPUT«there␤»
07:44 noam_ left
geekosaur *blink* not postcircumfix:< [ ] > ? 07:47
(argh notwork, 5 minutes to type that...)
which means I won't see the answer... 07:48
moritz geekosaur: it's a shortcut for postcircumfix:<[ ]>
07:49 alvis joined 08:06 packetknife joined, PacoAir joined 08:09 PacoAir left 08:12 alvis left 08:13 bbkr_ joined 08:14 bbkr left, bbkr_ is now known as bbkr 08:18 pat_js joined 08:19 fhelmberger_ joined 08:21 noam joined
jnthn morning, #perl6 08:32
sorear hello, jnthn
08:34 aindilis left, aindilis joined
moritz woah, lots of rakudo progress 08:37
08:37 pat_js left
jnthn Wow, time between entering $dayjob building and having a sticky note in my hand about a bug was literally 15 seconds today 08:37
moritz phenny: "tack så mycket"? 08:39
phenny moritz: "thanks so much" (sv to en, translate.google.com)
moritz phenny: en no "much"?
phenny moritz: "mye" (en to no, translate.google.com)
moritz no wonder I didn't reckonize "mycket" :-) 08:40
sorear is sv easy to transcribe? 08:41
jnthn sorear: Not so bad 08:44
sorear: Especially if you're lucky enough to live in a bit of Sweden without odd, Danish-influenced pronunciation. 08:45
jnthn lives in such an area, and finds his Swedish comprehension generally goes up noticably when he travels to other regions
sorear I think you mean does *not* live in such an area... 08:46
TODO: learn Swedish 08:47
moritz sorear: knowing German and English, I found learning Norwegian rather easy
sorear jnthn: do you beleive in role Hash[::T = Any] { ... } ? 08:48
moritz sorear: though my aim was mostly understanding people, not speaking it myself (so that I can understand my wife's family :-)
and I haven't quite succeded yet, but the half year where I put effort into it was rewarded with astonishingly successful 08:49
sorear TODO: learn German :)
sorear is starting to think about S09 in earnest now 08:50
jnthn sorear: I tried doing things that way before and it got icky.
sorear I'm also wondering if there should be a type distinction between shapeful and shapeless aggregates, or if it should be hidden like Int/Bigint 08:51
niecza: say (10**12).REPR # can't remember if I finished this
jnthn I was wondeirng if it's possible to do it was a set of composable mixins
p6eval niecza v15-2-gd19c478: OUTPUT«Unhandled exception: Unable to resolve method REPR in class Int␤ at /tmp/fEK3vlKrat line 1 (mainline @ 2) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3838 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3839 (module-CORE @ 65) ␤ at /home/p6…
sorear jnthn: could you elaborate on that? 08:52
shape handling seems pretty monolithic to me 08:53
08:55 alvis joined
sorear -> sleep, will quiz jnthn more tomorrow 09:03
jnthn Yeah, $dayjob is being quite busy here at the moment... 09:04
09:15 fglock joined, thou left 09:17 gv joined 09:39 dakkar joined
fglock I think I'm getting closer to a usable data model for perl5-in-js; next step would be to port it to perl6-in-js 09:46
09:48 tarch joined
moritz fglock++ 09:56
10:08 scott__ left 10:11 gv left
fglock moritz: I've found your post www.perlmonks.org/?parent=689702;node_id=3333 10:19
local and typeglob kinda work
what happenned with sprixel? 10:20
moritz fglock: it died. 10:22
au it begat niecza though 10:23
moritz github.com/diakopter/sprixel
fglock ah, cool (niecza, not died)
moritz well, it got rewritten for several different backends
the current one seems to be C/6model 10:24
fglock re data model - perlito5 now recompiles itself faster under v8 than under perl5; this gives some performance margin to add missing features and still have a usable impl 10:29
au oooh 10:32
fglock perl5-in-perl6 still needs a lot of work on the data model side
au: 10s in v8, 15s in perl5 10:33
moritz perl6: say $_.*WHAT
p6eval pugs b927740: OUTPUT«Scalar␤»
..niecza v15-2-gd19c478: OUTPUT«===SORRY!===␤␤NYI dottyop form .* at /tmp/krt8ABVT0w line 1 (EOF):␤------> say $_.*WHAT⏏<EOL>␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 1362 (die @ 3) ␤ at /home/p6eval/niec…
..rakudo d0a245: OUTPUT«===SORRY!===␤Cannot use .* on a non-identifier method call at line 1, near ""␤»
au praises inline caching vms in general and v8 in particular
fglock (10s bootstrap time is really nice to have)
au fglock: btw, can perlito.js run in node as well as d8? 10:35
s/can/is there some settings that makes/ 10:36
I ask because it might be very useful to have the bootstraped compiler as a npm module.
fglock yes - it needs a different implementation of CORE.print and IO.slurp; everything else is the same
au "npm install perlito ; perlito hello.pl"
fglock au: can you make a patch? :) 10:37
au what, no commit bits for me? :)
ok, I'll try making a pull request.
fglock btw, I'm trying to make a cpan package
ah, sure - let me find out how it works
I'm not familiar with github 10:38
au Admin -> Collaborators -> Add
(afaicr)
fglock au: looking - do you have some time for me to explain the bootstrap & test process? (maybe in a priv channel) 10:39
moritz if it's nothing secret, feel free to do it in here
moritz would be curious
au yeah
and/or capture it into a document in the repo
10:40 hugme joined
au or open a pad in beta.etherpad.org/ 10:40
10:40 ChanServ sets mode: +v hugme
au (I have ~20mins before lunch) 10:40
timtowtdi, though typing it here may be easiest :)
fglock README-perlito5 has most of it 10:41
perl -Ilib5 perlito5.pl -Cjs src5/util/perlito5.pl > perlito5.js # builds for perlito5 for v8 10:42
~/bin/d8 perlito5.js -- -Cjs src5/util/perlito5.pl > perlito5-new.js # bootstraps
running the bootstrap 3 times ensure it works
nice time util-perl5/bootstrap-perlito5-perl5.sh # recompiles the whole lib5 10:43
au takes notes (and would probably add them as makefile targets)
fglock time nice util-js/make-perlito5-js.sh # recompiles the web gadget (html/perlito5.js) 10:44
(perlito6 is about the same, but it also needs to bootstrap in python) 10:46
recovering from a broken bootstrap is basically: git checkout lib5 perlito5.pl 10:47
au got it. 10:48
fglock you can't write to lib5 while using the perl5 backend, so you need a temp file: 10:49
perl -Ilib5 perlito5.pl -Cperl5 src5/lib/Perlito5/Javascript/Emitter.pm > x && cp x lib5/Perlito5/Javascript/Emitter.pm
or just use the js compiler:
~/bin/d8 perlito5.js -- -Cperl5 src5/lib/Perlito5/Expression.pm > lib5/Perlito5/Expression.pm
(which is kinda cool - using v8 to compile back to perl5) 10:50
au yup 10:51
10:51 daxim joined 10:52 bbkr left
au idly wonders if theres a bin/into utility somewhere so "cat foo |into foo" can work 10:52
i.e. consume all of stdin, wait for it to close, then write it to arg1
moritz and doesn't write at all if SIGPIPE occurs? 10:53
au yeah
I guess "tail -n 9999" is something like that
(if tail supports an -o option, that is) 10:54
fglock testing in the command line with v8 also requires a temp file:
~/bin/d8 perlito5.js -- -Cjs -e ' $main::a = 10; sub x { local $main::a; $main::a = 3; say $main::a; return 123 }; x(); say $main::a; ' > x && ~/bin/d8 x
though it could just run eval-string, thinking about it 10:55
au ("|tail -n 999 |tee 1 >/dev/null" is close)
au punts that particular train of though
*t
fglock: you mean " | xargs d8 -e " ? 10:56
fglock au: I don't understand what you mean 10:57
au d8 perlito5.js -- -Cjs -e ... | xargs d8 -e 10:58
fglock this gives me: Error reading 'Do' 10:59
au probably not then 11:00
fglock au: adding -Cjs-node param to util/perlito5.pl might be better than tweaking a generic runtime 11:04
au why? probing at startup is ~zero cost 11:05
fglock because node has a lot more features, and then you keep the browser runtime smaller
maybe
11:06 kst left
fglock you may want "use" in the browser do an ajax call 11:06
11:06 kst joined
au yeah. we can test for the free variable require 11:07
and use it in both node and AMD/CommonJS/curlJS/requireJS/whatever context
11:11 havenn joined
fglock au: you should have a commit bit now 11:11
au <- been hacking in coffeescript/node for almost a year now and would love to bring p5/p6 modules into the ecosystem
fglock++ thx!
fglock you will probably find the calling conventions a bit odd, but that's needed to support some perl5 features efficiently 11:13
older versions had a simpler, more js-ish look
au *nod*
masak morning, #perl6 11:14
au /o masak 11:15
fglock o/ 11:16
au: README-perlito5-js
au got the callconv 11:18
fglock if you have problems with the parser... I can *try* to explain
it's certainly not perfect 11:19
the plan for @ISA is to merge the methods in the class magically - this is not implemented 11:20
fglock waves hands
au ah, so not using the .prototype.lookup chains?
fglock probably not, because then you can have multiple inheritance 11:21
masak I think the truce between the Perl 5 and Perl 6 communities can be achieved over the wording "Perl 6 won't replace Perl 5 any time soon". Eternal fivers can read it as an understatement and think "...or never", and be happy. Poeple in the sixers militia can read it as a long-term revolutionary plan and think "...but just you wait", and be happy.
daxim you don't fool me!! 11:22
au still thinks Perl 5+6i is a nice version number
masak s/never/ever/
au or 5.14+6.28i, for that matter 11:23
perl6: (5.14 + 6.28i).WHAT
p6eval pugs b927740, rakudo d0a245, niecza v15-2-gd19c478: ( no output )
au perl6: (5.14 + 6.28i).WHAT.say
p6eval rakudo d0a245, niecza v15-2-gd19c478: OUTPUT«Complex()␤»
..pugs b927740: OUTPUT«Complex␤»
fglock au: but .prototype.lookup chains is much better than no inheritance at all, of course 11:24
yes, it's complex :P
au fglock: dunno, I kinda think we can serialize over C3 and arrange a linear .prototype .prototype.prototype chains.
masak au: so you're saying the situation is complex? ;)
au masak: yeah, it's easier to accept it as complex than discarding either part :) 11:25
masak aye. 11:26
despite what I wrote above, I'm not a militant sixer, nor do I plan to become one.
moritz too still uses large amounts of p5 11:27
masak I'm eager to create a future where there's room for both languages.
moritz and will, for the foreseeable future
au PHP used to have militant sixers... they are sadly in short supply these days
masak I probably write more Perl 5 than Perl 6 nowadays.
au Perl community is comparably fortunate in that regard :)
masak <TimToady> the only thing that will prove Perl 6 to the world is doing it 11:28
this seems to be the predominant slogan the past few years. 11:29
a Candide-like "let's tend our garden" attitude. :)
au that's very candid for sure :)
11:29 havenn left
masak and p6l is full of Panglosses... 11:29
au -> lunch, bbiab 11:30
masak in fact, now that I think about it, the recent criticism heard on Twitter and elsewhere -- "the traffic on p6l/spec commits have decreased, therefore Perl 6 is in decline" -- is *pathetic*. 11:32
the people who say it probably are simply misinformed and would benefit from being corrected, but... wow, if that's the best outsiders can come up with these days, I'd say we're doing pretty well! 11:33
au “First, you win. Then they may ignore you, laugh at you, fight you… But none of that matters — You've won already!”
moritz well, the traffic on the mailing lists has declined. And that's usually not a good sign. 11:34
though it really just has gone elsewhere, which is not apparent for the outsider
masak moritz: it has. I haven't seen that as a worrying sign.
elsewhere = IRC?
blog posts? 11:35
moritz IRC, gists, blog posts, conferences
remeber the "calling positional parameters by name must by" gist? that would have gone to p6l 3 years ago 11:36
masak aye.
moritz and spawned a discussion with 200+ messages, arguing over the virtues of empowering the user vs. avoiding accidents vs. transcental tibetian philosphy, or so 11:37
masak when I want bikeshedding and out-of-the-box thinking with lots of noise and missed targets in it, I go to p6l.
when I want bikeshedding and constructive discussion, I post a gist URL here :)
I have a feeling that if one were to plot daily #perl6 activity over the past six-or-so years, one would see a definite increase. 11:39
moritz looks forward to high-bandwith communication with jnthn++ about the sink business
masak looks forward to going to Erlangen, and meeting moritz and the others
fglock I'd rather discuss alternative perl5 implementations here than in other lists - unless it bothers you as off-topic 11:40
moritz because I have the feeling that my first assumptions of how PAST nodes reflect program structure were childishly naíve
au fglock: considering the said implementation is written in perl6, I doubt it'd be off-topic here :)
fglock it is no longer written in perl6, I've translated to perl5 after bootstrapping 11:41
but conceptually yes :P
au that makes it doubly on topic, as the same process would apply to 5->6 and 6->js :)
masak mmm, 6->js 11:42
11:44 M_o_C joined, kst left, kst joined
masak so, I have a Perl 6 day today. 11:50
I'd like to spend parts of it on working on macros.
it's probably high time for another branch rebase. :)
I need to think more about the "two OUTERs" problem. 11:51
maybe blog about it and about the recent progress.
then I need to spend some time working on my GPW talk.
moritz: [backlog] "mycket" and "mye" are practically the same spelling! :) 11:56
at least by the loose, blurry vision etymologists (or enthusiasts) need to apply. 11:57
moritz masak: right, the edit distance is just 3 :-)
11:57 M_o_C left
masak there you go. 11:58
1, if you consider that the 't' is silent, and the 'ck' is just one phoneme. 11:59
sorear: I would learn German before learning Swedish... and after learning German, I wouldn't learn Swedish :) 12:00
that's even if you plan to live here at some point.
it's not worth it from a ROI point-of-view.
moritz kinda thinks it's *always* worth trying to learn the language of the country you live in 12:01
though that's more of a cultural choice than ROI
masak I've met dozens of people at this point who learned Swedish, even though they spoke English well. out of all those, I consider *one* of them to have learned Swedish to a level where speaking it made more sense than speaking English. 12:03
she had some sort of immersive learning process which apparently turned out quite well.
12:03 mj41 left
masak I know it's such a cliché, but "my native language is actually quite difficult to master". 12:04
it's not so much the grammar and the words, which are close enough to English and German and French.
it's the pronunciation. the exact sounds, the stressing of words, and the sentence melody. 12:05
moritz "made more sense than speaking English" is exactly the ROI-type thinking that I don't mean
masak nod.
moritz how can you ever understand a culture on the gut level if you don't get their puns? :-)
masak oh, I agree with that. 12:06
that's why I'm always interested in the structure of languages, even though I don't speak them well.
12:09 bluescreen10 joined 12:14 bluescreen10 left
masak here's the current "issue" with macros: 12:18
macro foo($x) { $x }; foo my $a = 42; say $a 12:19
12:19 mj41 joined
masak we'd like this program to output '42', rather than, say, not even compile because $a isn't visible. 12:19
for that to work, the AST corresponding to the 'my $a = 42' bit has to be more of a thunk than a block. 12:20
moritz correct, it shouldn't be a call frame or outer frame on its own 12:25
masak nom: my $x = 0; sub foo { True && my $a = $x++; say $a }; foo; foo; foo
p6eval nom d0a245: OUTPUT«0␤1␤2␤»
masak as TimToady points out in irclog.perlgeek.de/perl6/2012-02-24#i_5204285 , this case is analogous.
12:25 bluescreen10 joined
moritz isn't there a PAST::Block type that corresponds to that? 12:25
masak how would I find out? 12:26
investigate how infix:<&&> is wired up?
moritz it's in Actions.pm
masak ok.
masak looks
moritz but I think it'll disappoint you
masak me too :)
moritz because iirc it's just :pasttype<if>
masak right. 12:27
12:27 hugme left, hugme joined
masak also, I have more immediate problems. 12:27
12:27 ChanServ sets mode: +v hugme
masak macro foo($x) { $x }; foo say "OH HAI" # still doesn't work 12:27
for known reasons.
there are only three occurrences of :pasttype<if> in Actions.pm -- neither of them relating to infix:<&&>. 12:29
they relate, respectively, to 'if', 'when', and quotepairs. 12:30
moritz sorry, it's Grammar.pm 12:31
masak ok.
token infix:sym<&&> { <sym> <O('%tight_and, :pasttype<if>')> }
right.
well, if :pasttype<if> has thunking behavior, then maybe it means that I can use that. 12:32
moritz well, thinking is easy on the PAST level 12:33
*thunking
nice typo :-)
because it just means passing a PAST::something around
masak right.
the burning question, however, is till this: can I give a *thunk* a different OUTER? 12:34
fglock wonder how to parse here-docs (easily)
moritz .*?$delimiter maybe?
masak: doesn't it automatically get a different outer when you put it into a PAST tree somewhere? 12:35
masak moritz: hold on, I'll give you an example. 12:36
moritz ah, the code in Perl6::World seems to fiddle with $pad<outer>
masak macro foo($x) { quasi { say {{{$x}}} } }; foo (my $x = 42); say $x # expected '42\n42\n' 12:37
the quasi has an OUTER being the macro &foo, even after being implanted in the mainline in lieu of the macro call. 12:38
therefore, the thunk representing '(my $x = 42)' needs to re-OUTER itself back to the mainline somehow.
that is the core of my current challenge.
moritz macro foo($x) { quasi { say {{{$x}}} } }; foo (my $y = 42); say $y 12:39
also expects 42\n\42, right?
masak ah, thanks. better.
yes.
they are two distinct variables.
and it may not really be an OUTER I need to change. as TimToady said, a thunk is something that parasites on an existing lexical scope.
moritz right, so a simple re-outering isn'T enough
right, you need to merge it into some pad 12:40
masak aye.
moritz think of it as a role you mix into it
masak tries to think of it like that, and falls off the deep end
yes... I think I get it. a role has its own lexical environment too.
moritz never mind then :-)
masak but this is somewhat inside-out from that.
we want to mix the mainline into the quasi block. 12:41
and implant the quasi block into the mainline.
it's like a "quilting" operation. :)
moritz well, if we mix the quasi into the mainline, the other direction happens automatically, no?
masak sounds like you're thinking of closures, not ASTs. 12:42
though I dearly hope you're right, and that I've misunderstood something :)
moritz I'm currently trying to figure out how the generated code for the already-applied macro would look like 12:43
masak from my perspective, it looks like a block (coming from the quasi block), proudly declaring that it has the macro as an OUTER. 12:44
containing thunks that somehow assert that they are part of the mainline's lexpad.
tadzik good afternoon, #perl6
masak tadzik! \o/
tadzik \o/ 12:45
moritz masak: I think that OUTERs aren't the solution
masak: because it's really about hygiene 12:46
masak no, probably not.
moritz masak: and you can't find a general solution that fddles with OUTERS and comes up with the right answer
masak moritz: a secondary cause of concern is setting those thunks up so that they get "rebound", just like each 'my $a = $x++' gets rebound to each invocation of 'sub foo'.
en.wikipedia.org/wiki/Hygienic_macro#Strategies is informative in the various ways we can solve this, I guess. 12:47
moritz masak: hold on, let me think through tthe current issue before coming up with new ones
masak gladly.
but they're all really related.
moritz sub mysay($x) { say $x; return $x }; macro foo($x) { quasi { mysay {{{ $x }}} } }; foo(mysay '42') 12:49
12:49 colomon joined
moritz this executes the 'mysay 42' first, right? 12:49
masak "first"? 12:50
moritz well, the output is 42\n42\n
let me make this a bit clearer
masak yes. as far as I can see, it is.
moritz sub mysay($x) { say $x; return 2 * $x }; macro foo($x) { quasi { mysay {{{ $x }}} } }; foo(mysay 42)
the output is 42\n48\n, correct?
masak yes.
well, "42\n84\n" 12:51
but you meant that.
moritz right
ENOMATH :-)
masak the 'mysay' in the quasi gets called first.
or, hm. 12:52
no, it doesn't.
the code that gets spliced in is 'mysay mysay 42', but as ASTs.
from then on, there's nothing more funny going on that sub calls. 12:53
I don't really see what you're trying to show here.
12:53 fhelmberger_ left
moritz well, the act of showing answered my initial question already :-) 12:54
masak oh, good.
moritz ok, here are my thoughts
it should compile to something like
$MACRO::x = (my $y = 42);
say $MACRO::x 12:55
say $y
where 1) $MACRO::x is a uniq name per macro instantiation
masak achieve hygiene by shoving things from inside the quasi into a gensym'd namespace? 12:56
moritz and 2) the first say() in there is looked up from the perspective of the macro's scope
masak: yes, or an equivalent mechanism
masak right.
moritz the question is if there's a sane way to unify the lookup of the &say and the $x inside the macro
12:56 JimmyZ joined
masak corresponding to the "Packages" solution on Wikipedia. 12:57
moritz and that's the point you've been trying to make all along, isn't it?
masak sometime like that.
a macro always has to deal with two lookup chains.
and the user very much expects the macro not to mess them up.
so much that it's almost a bit hard to make precise :) 12:58
a closure always deals with only one lookup chain. what's magical about closures is that it's the "original" lookup chain, not the "spliced-in" lookup chain. 12:59
but the code that results from a macro application needs to relate to both "original" (quasi) and "spliced-in" (mainline) lookup chains.
moritz gist.github.com/1958280 13:03
masak looks 13:04
moritz nothing really new
masak it tells me you're really getting it. 13:05
I've only showed the problem with scalar variables so far. you're showing it with sub lookup.
moritz well, they really obey the same lookup semantics
I could have done that by sheer pattern matching, without understanding (wait, where's the difference? :-) 13:06
masak ;) 13:08
masak .oO( en.wikipedia.org/wiki/Chinese_room )
moritz I'm now trying to visualize how the generated AST for gist.github.com/1958280 might look like 13:15
the first possiblity being
$MACRO::x := (say "hello $a"); 13:16
sub MACRO::greet($a) { say "LOL IT'S $a.uc()" };
MACRO::greet MACRO::$x 13:17
sub greet($a) { say "hello $a" }
greet $y
the second involving some weirdly connected blocks that twist my head in ways that I fear might be unrecoverable 13:18
13:21 tokuhirom joined 13:26 am0c left, mtk joined 13:29 simcop2387 left 13:31 simcop2387 joined
moritz hugme: hug masak because he has to implement macros 13:31
hugme hugs masak
phenny hugme: 03 Feb 15:08Z <[Coke]> tell hugme to hug p5p.
moritz hugme: hug p5p
hugme hugs p5p
masak wow, they had to wait for that hug almost a month.
moritz yes, hugmes auto-rejoin didn't work 13:32
I hope I've fixed it now, mit literal++'s help
masak moritz: I think I've been heading towards your second possibility involving weirdly connected blocks that twist your head in ways that may or may not be unrecoverable... 13:33
I will now seriously consider your namespacing solution.
for quite a while, I've had the feeling that *something* will have to be rewritten at insertion-time; either the quasi lookups or the mainline lookups. 13:35
I've been going back and forth on which ones to rewrite.
moritz note that you can also gensym names, or rewrite when you detect conflicts (though that's harder, I think), it really doesn't matter which way you fudge the names 13:37
you can safely gensym names that aren't valid Perl 6 identifiers
like with a double sigil or so
masak the thing about gensymming is that you know it's gonna come back and bite you. 13:40
it's basically incompatible with predictable lexpad introspection, for example.
and I don't know if I believe in "when you detect conflicts" -- that presupposes that all lookups are statically known, which isn't so. Perl 6 has lots of runtime-only lookup mechanisms. 13:41
moritz right 13:42
which is why you need namespaces
thought it *can* be gensym'med micro-namespaces
masak I need something, for sure.
maybe namespaces.
in a way, tying up OUTERs in exactly the right way can be seen as a form of clever anonymous namespacing. 13:43
moritz at some point you have to fool the dynamic lookup not to take those symbols into account that are implementation details
yes, it sounds more elegant 13:44
JimmyZ can't build rakudo :( 13:45
Serialization Error: could not locate static code ref for closure 'c3_merge'
13:48 overrosy left
masak JimmyZ: did you do a realclean or a 'git clean -dfx'? 13:49
moritz and is that newest rakudo and nqp? 13:53
JimmyZ moritz: 13:54
moritz: yes
moritz: everything is the newest, and did 'git clean -xdf' 13:55
masak did you try building everything from a freshly cloned rakudo repo?
moritz that... shouldn't be necessary
masak no, it shouldn't.
but you never know.
JimmyZ I'm everything is fresh 13:56
I'm sure *
nom: constant @primes = 2, 3, -> $p { ($p+2, $p+4 ... -> $n { $n %% none @primes ... * >= sqrt $n })[*-1] } ... *; 13:58
p6eval nom d0a245: OUTPUT«===SORRY!===␤Variable @primes is not declared␤at /tmp/RPLkFWmwOe:1␤»
JimmyZ I thought it was fixed.
masak what made you think that? 13:59
as far as I know, it wasn't even RT'd.
13:59 pernatiy left
moritz the commit message was Fix 'constant fib := 0, 1, *+* ... *; say fib[100]'. 13:59
JimmyZ github.com/rakudo/rakudo/commit/d4...d6d8d18ef7
masak that looks like a fix for something slightly different. 14:00
moritz aye
jnthn Yes, there were two issues, I fixed one out of two.
masak jnthn+
jnthn *sigh* four hours to find/fix a one-char bug...
JimmyZ still can't build rakudo
jnthn ...over 2 to reproduce the environment the bug was found in, mind :/
JimmyZ: Are you really sure you have latest Rakudo? 14:01
14:01 colomon left
jnthn I recognize that error. It was the first one I got when I built old Rakudo on nqp/bs before I started the Rakudo updates. :) 14:01
JimmyZ $ git pullAlready up-to-date. 14:02
both nqp and rakudo is Already up-to-date
14:02 jaldhar left 14:03 jaldhar joined
moritz JimmyZ: what's the output of git ref-parse HEAD ? 14:03
masak maybe something else in the system is old.
JimmyZ a one-char bug are always hard
$ git rev-parse head 14:04
c314198aeb77953bb15ad8d053381242828a3f52
moritz that's nqp, right? 14:05
looks correct
masak walk /
&
JimmyZ oh no, .. I 'm in (cont_reuse) branch ... 14:06
moritz JimmyZ: you should be in rakudo/nom and nqp/master 14:07
jnthn ah, that explains it.. :)
14:08 jaldhar left, NamelessTee joined
JimmyZ works for me now 14:16
jnthn \o/
JimmyZ compile CORE.Setting takes 980Mb memory here 14:22
moritz time to add more stuff to src/core/ :-)
jnthn :P :P 14:23
tadzik++ already did yesterday :)
felher perl6: sub cc($a, $b) { $a ~ $b }; <a b c> X&cc <1 2 3>;
p6eval rakudo d0a245: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in sub cc at /tmp/p32LlRTSf2:1␤ in block <anon> at /tmp/p32LlRTSf2:1␤␤»
..pugs b927740: OUTPUT«*** ␤ Unexpected "2"␤ expecting operator␤ at /tmp/xndbFkYbNN line 1, column 45␤»
..niecza v15-2-gd19c478: OUTPUT«Unhandled exception: No value for parameter '$b' in 'cc'␤ at /tmp/j1MrY3SgKd line 0 (cc @ 1) ␤ at /tmp/j1MrY3SgKd line 1 (mainline @ 2) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3838 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3839 (m…
moritz nom: class A { ... }; class B is A { }; class A { }; say B.new ~~ A 14:24
p6eval nom d0a245: OUTPUT«Method 'new' not found for invocant of class 'B'␤ in block <anon> at /tmp/xvg9wpPrT7:1␤␤»
jnthn felher: Declare a sub infix:<cc>($a, $b) { ...
and use Xcc
moritz: I know about that one.
moritz jnthn: is that supposed to work? 14:25
jnthn Not sure. I know that it doesn't though :)
moritz I mean, I could totally understand if not
jnthn I think I can make it fall out naturally from other things.
moritz so we don't need to know about the parent classes parents and methods at compose time? 14:26
felher jnthn: i though, since one could write infix:<foo>('a', 'b') (using operator foo as a binary function), maybe X works on binary functions (using binary-function as operator), too :) 14:27
*thought 14:28
jnthn moritz: Well, it's more that parents re-composing should announce the change to the hierarchy.
felher: No :)
felher jnthn: k, thnx :)
14:29 havenn joined 14:33 grondilu joined
fsergot o/ 14:34
14:36 grondilu_ joined
grondilu_ perl6: say constant X = 10**5 14:37
p6eval pugs b927740: OUTPUT«*** ␤ Unexpected "="␤ expecting ":" or "("␤ at /tmp/BdHKeJXtdZ line 1, column 16␤»
..rakudo d0a245, niecza v15-2-gd19c478: OUTPUT«100000␤»
grondilu_ perl6: say constant X = 10**50
p6eval rakudo d0a245, niecza v15-2-gd19c478: OUTPUT«100000000000000000000000000000000000000000000000000␤»
..pugs b927740: OUTPUT«*** ␤ Unexpected "="␤ expecting ":" or "("␤ at /tmp/8_G_jHFBcx line 1, column 16␤»
grondilu_ perl6: say constant X = 0x10000000000000000000000000000000000000000000000000000000;
p6eval rakudo d0a245: OUTPUT«===SORRY!===␤Object of type 'Int' cannot be referenced without having been assigned a serialization context␤»
..pugs b927740: OUTPUT«*** ␤ Unexpected "="␤ expecting ":" or "("␤ at /tmp/7CxndOyIpq line 1, column 16␤»
..niecza v15-2-gd19c478: OUTPUT«1684996666696914987166688442938726917102321526408785780068975640576␤»
grondilu_ rakudo--
14:38 grondilu left
moritz grondilu_: don't -- it, write a bug report and a test :-) 14:38
grondilu_ is there an Howto somewhere? 14:39
moritz I guess some of the number parsing code needs updates after the bs merge
grondilu_: bug reports are simply emails to [email@hidden.address]
with an expressive subject and a small bit of code and example output that demonstrates the problem 14:40
grondilu_ ok
do I have to make sure it hasn't been reported yet?
moritz and tests... I think I've written about that too, but I think it's quite obvious from looking at the existing tests
14:40 skids joined
moritz grondilu_: I don't think that one has been reported yet 14:40
grondilu_ Oh I see instructions in rakudo's README. Sorry. 14:41
felher so, if one has a binary function and want to use X on it, he make a new block and declares the function as an binary operator? like paste.pocoo.org/show/559762/? 14:42
moritz I guess we should really provide a crosswith() function or so 14:46
grondilu_ felher: I think if you define cc as an operator you'll be able to right directly $a X[cc] $b 14:47
moritz maybe zip() should sub zip(*@@lol, :with=&infix:<,>)
and then we should have a cross(*@@lol, :with=&infix:<,>)
then you could write that cross &cc, @list1, @list2 14:48
perl6: role A { method x() { } }; role B { method x() { } }; class C does A does B { }
p6eval niecza v15-2-gd19c478: OUTPUT«===SORRY!===␤␤Method 'x' must be resolved by class 'C' because it exists in roles 'A[curried][...]' and 'B[curried][...]' at /tmp/FYgypaJUFw line 1 (EOF):␤------> hod x() { } }; class C does A does B { }⏏<EOL>␤␤Unhandled…
..pugs b927740: ( no output )
..rakudo d0a245: OUTPUT«===SORRY!===␤Method 'x' must be resolved by class C because it exists in multiple roles (B, A)␤»
felher grondilu_: in my paste i did define it as an operator and then used it like '<a b c> Xcc <1 2 3>'. Or did i mistunderstand you? 14:51
moritz perl6: role A { method x() { ... } }; class B does A { } 14:53
felher Whats the reason for meta-operators to not be able to use functions as in X&foo?
p6eval niecza v15-2-gd19c478: OUTPUT«===SORRY!===␤␤Method 'x' must be implemented by 'B' because it is required by role 'A[curried][...]' at /tmp/AJnwecnwCT line 1 (EOF):␤------> method x() { ... } }; class B does A { }⏏<EOL>␤␤Unhandled exception: Check fai…
..rakudo d0a245: OUTPUT«===SORRY!===␤Method 'x' must be implemented by B because it is required by a role␤»
..pugs b927740: ( no output )
moritz felher: they are called meta *operators* because they act on *operators*
felher: I can understand that you want something that acts on non-operators. I want it too 14:54
felher: but then it should simply be a higher-order function, much like map or grep
though X&cc might work out. Maybe. Need to ask our parsing gurus (TimToady, sorear, jnthn) 14:55
felher moritz: I thought there is no difference between operators and subs save their names. So its primarily a naming-thing?
jnthn Well, the names imply a parser augmentation.
moritz felher: well, names and the syntax they imply
you can do say(1, 2, 3), you can't do 1 &say 2 &say 3 14:56
operators also have "fun" like precedence and associativity
felher Ah, i see. Thanks :) 14:57
moritz for example for [op], the associativity tells the meta-reduce thingy whether to reduce left-to-right or right-toleft
jnthn in some sense, operators are verbs whereass &foo is a noun.
moritz nom: say ([**] 2, 3, 4) == 2 ** 3 ** 4
p6eval nom d0a245: OUTPUT«True␤»
jnthn And maybe it's nice to keep things straight in that sense.
moritz note that ** is right-assoc
felher yeah, i understand the problem with the associativity. ty :) 14:58
moritz jnthn: but it might not hurt to be able to verb a noun :-)
jnthn True
14:59 kst left 15:00 kst joined
grondilu_ felher: (about earlier) Indeed. My bad. 15:02
felher hm, over all that meta-operator stuff i forgot on what i was working on in the first place... 15:03
grondilu_: kk :)
Oh, great, bs is merged to nom \o/ 15:06
jnthn aye :) 15:08
15:09 tokuhirom left 15:11 Trashlord left 15:12 GlitchMr joined
felher Great! (make clean && perl Configure.pl --gen-nqp --gen-parrot && gmake -j10 && gmake install) is no one minute faster :) 15:15
*now 15:16
15:16 marmalade joined 15:25 sisar joined
fglock moritz++ the ".*?$delimiter" suggestion was obvious, but actually helped - now I have something I can refine later 15:27
moritz felher: you could save much time by building parrot in parallel too 15:29
15:29 Trashlord joined
moritz felher: though it either needs a custom build script, or fiddling with the configure system 15:29
or maybe there's an env options
felher moritz: thanks for the hint. I'll look into that :) 15:31
15:38 noam left 15:39 MayDaniel joined 15:40 noam joined 15:41 kst left
moritz I wouldn't object to a patch that adds a --parrot-make-option option to rakudo's Configure.pl 15:41
15:42 kst joined 15:44 Psyche^ joined
jnthn decommute & 15:48
15:48 Patterner left, Psyche^ is now known as Patterner 15:53 mj41 left 15:54 havenn left 15:56 pernatiy joined 16:02 wtw left 16:04 havenn joined 16:06 thou joined 16:07 thou left 16:10 grondilu_ left 16:14 kaare_ joined 16:19 JimmyZ left 16:20 colomon joined
jnthn home o/ 16:25
16:27 havenn left, am0c joined 16:30 alester joined
[Coke] wonders what's on jnthn's rakudo-genda now that bs has landed. 16:31
16:34 fglock left 16:40 packetknife left
[Coke] pugs: say 1...2 16:41
p6eval pugs b927740: OUTPUT«*** ␤ Unexpected "2"␤ expecting operator, ":", "," or term postfix␤ at /tmp/YxTfhvRjaB line 1, column 9␤»
[Coke] pugs: say 1..2 16:42
p6eval pugs b927740: OUTPUT«12␤»
TimToady pugs: say 1 ... 2 16:46
p6eval pugs b927740: OUTPUT«*** ␤ Unexpected "2"␤ expecting operator, ":" or ","␤ at /tmp/YJ_Ray8xlA line 1, column 11␤»
TimToady I wonder how that's pars(ec)ing... 16:47
au pugs: my @x := (1...); say @x[1000] 16:50
p6eval pugs b927740: OUTPUT«1001␤»
au (postfix:<...>)
TimToady that's why I tried a space
au pugs: my @x := (1 ...); say @x[1000]
p6eval pugs b927740: OUTPUT«1001␤»
jnthn [Coke]: Try and unblock at least of the moritz++ branches, some feature hacking, and work on qregex bootstrap. 16:51
au (it pre-dates the "Postfix found where infix expected is an error" rule)
TimToady overagressive parsec space rule maybe
jnthn *at lesat one
au pugs: &infix:<...>(1,2) 16:52
p6eval pugs b927740: OUTPUT«*** No such subroutine: "&infix:..."␤ at /tmp/MowGquYYrD line 1, column 1 - line 2, column 1␤»
16:53 att left
moritz jnthn: sorting out the operator multi stuff in match-refactor branch might be an overall win too 16:58
jnthn: the sink branch needs some discussion
jnthn moritz: Is it best left until GPW when we'll have more bandwidth?
moritz yes (for sink) 16:59
jnthn Yes, the sink one was what I was meaning.
moritz considering that GPW is only 3 days away...
jnthn Right :)
moritz and pre-GPW lasagna only 2 days away... :-)
tadzik wonders if we have more lags now due to cables, or we will have on GPW due to sound barriesr
jnthn \o/
moritz: Is there some summary of what "the operator multi stuff" is? 17:00
jnthn thinks it was discussed a bit more here once... 17:01
moritz: Anyway, updating match-refactor with latest changes...
17:02 havenn joined, kst left 17:03 kst joined, NamelessTee left
dalek Heuristic branch merge: pushed 98 commits to rakudo/match-refactor by jnthn 17:05
moritz jnthn: in essence multi infix:<foo>(SomeType, SomeType) { } in a lexical scope is called for some values of 'foo', but for others it falls back to (Any, Any) case
class A { }; { multi infix:<+>(A, A) { 'OH HAI' }; say A + A } 17:06
nom: class A { }; { multi infix:<+>(A, A) { 'OH HAI' }; say A + A }
p6eval nom d0a245: OUTPUT«OH HAI␤»
moritz nom: class A { }; { multi infix:<==>(A, A) { 'OH HAI' }; say A == A }
p6eval nom d0a245: OUTPUT«OH HAI␤»
moritz nom: class A { }; { multi infix:<eq>(A, A) { 'OH HAI' }; say A eq A }
p6eval nom d0a245: OUTPUT«use of uninitialized value of type A in string context␤use of uninitialized value of type A in string context␤True␤»
moritz see, works for + but not for eq 17:07
and the match-refactor branch shifts infix:<==> from working to not-working
17:07 Chillance joined
jnthn nom: class A { }; { multi infix:<->(A, A) { 'OH HAI' }; say A - A } 17:09
p6eval nom d0a245: OUTPUT«use of uninitialized value of type A in numeric context␤use of uninitialized value of type A in numeric context␤0␤»
17:09 icwiener joined
jnthn Heh. So + works, - does not. 17:09
moritz and git bisect shows it's a commit that only touches the setting in rather harmless ways
jnthn wtf. :)
moritz I forgot which on it was, but I can find it in the logs
jnthn which one, ooc?
yeah, may help
I'm gonna switch back to nom branch and debug it there.
moritz either 560633d964a78848f6a78e2366189803ce54cacd or 5b4468309e03b0d6ff3f01070997b6ebb9687103 I think
yeah, match-refactor can be cleanly merged into pre-bs nom 17:10
jnthn I just merged latest nom branch into match-refactor and it built fine. 17:11
moritz 5b4468309e03b0d6ff3f01070997b6ebb9687103 is the evil one
make spectest should show two failing test files
and several passing TODOs, which are the main aspect of that whole branch :-) 17:12
jnthn k
17:12 sisar left
jnthn OK, I got the +/- diff reproduced locally. 17:13
moritz it's also something that's reflected in the way the spectests are fudged
so seems quite determinstic, even if in weird ways 17:14
17:14 fglock joined, birdwindupbird left, wolfman2000 left
jnthn moritz: Well, this is telling. 17:14
$ perl6 z.p6 17:15
OH HAI
use of uninitialized value of type A in numeric context
0
17:15 NamelessTee joined
jnthn $ perl6 --optimize=off z.p6 17:15
OH HAI
moritz oh.
and now you try the different optimization levels 17:16
jnthn --optimize=0 (analysis only) and --optimize=1 are fine
Starts at --optimize=2 (the default)
moritz oh, and we only got back operator overloading after the optimizer was implemented
jnthn OK, that gives me a relatively small amount of code to consider as to blame. :) 17:17
moritz so that why we didn't notice any regressions -- I attributed all the changes in non-passing tests to the nom-vs.-b difference
jnthn I blame the guy who wrote the optimizer.
TimToady thinks that something that does required transformations shouldn't be called an optimizer... 17:18
jnthn TimToady: ?
moritz well, it's more optimal than an incomplete implementation :-)
17:18 gv joined
jnthn TimToady: It works fine with the optimizer off. 17:18
TimToady okay
jnthn TimToady: It's when it's on that it's doing the transformation wrong.
moritz jnthn: well, without the optimizer we don't get CHECK-time routine checking, which is specced
(iirc) 17:19
jnthn True.
TimToady that's probably what I was thinking of...
jnthn --optimize=0 actually doesn't optimize, mind.
TimToady is still in 6 impossible things mode
jnthn It only does the analysis part.
--optimize=off turns off that analysis too.
I'm happy to name them in better ways. :) 17:20
moritz --static-frob=0..3 17:21
:-)
17:23 sisar joined
jnthn Hmmm. 17:26
Think I see it.
17:26 isBEKaml joined
jnthn also wonders why he ended up with the multi-dispatch related bits on Code rather than Routine 17:28
TimToady the problem with allowing &foo as an operator is it prevents the terse foo&bar style of & junction; while we recommend whitespace around infixes, I'm not much into requiring it
moritz because you wanted to multi-dispatch anon blocks? :-)
TimToady we might possibly allow &foo inside of [&foo] though
sisar o.O (description of 'length' => "This word is banned in Perl 6. You must specify units." !) 17:29
'banned' ?
masak 'severely discouraged'? :) 17:30
moritz banned. We do not use it, except to rant about it
17:30 Vlavv` joined
TimToady it is generally considered a sign of Everything Is A List mentality, which is not P6-Think 17:30
moritz well, I can talk about the length of my car, but then the answer isn't a number, but a number + unit :-)
masak +1 on X[&foo], or at least I don't feel opposition to it. I don't like X&foo at all, for reasons stated above.
TimToady though a list could have cars in it... 17:31
masak cddrs
sisar on first reading i though that i can't even define a sub called length in P6
17:33 Vlavv_ left
TimToady P6 sometimes tries to discourage you from thinking too generically where precision is required; length is one of those place 17:33
*places
s/places/times/ :)
17:35 sisar left 17:39 slavik1 left, kst left, slavik1 joined, kst joined 17:40 pernatiy left, pernatiy joined 17:43 pernatiy_ joined 17:45 pernatiy_ left, mucker joined, pernatiy left 17:49 pernatiy joined
TimToady wonders what precedence [&foo] should default to 17:51
arguably should depend on foo's signature 17:53
skids awestruck by all the fixes/new stuff in just last week. jnthn++ , moritz++ , tadzik++ ! 17:55
Definitely looking forward to testing out PROXY/postcircumfix:<( )> fixes. 17:56
jnthn moritz: Fixed. 17:57
17:57 tarch left
dalek kudo/nom: 93104f1 | jnthn++ | src/ (5 files):
Only need multi handling stuff in Routine; don't make every other block pay the price of the few extra attributes.
 17:57
kudo/nom: 99bc8b3 | jnthn++ | src/Perl6/Metamodel/BOOTSTRAP.pm:
Fix over-sharing of proto thunks, which caused mis-dispatches in some cases.
17:58 MayDaniel left
moritz \o/ 17:58
jnthn Wasn't actually the optimizer's fault in the end...was a bug in dispatcher derevation. 17:59
18:04 havenn left
jnthn moritz: There were two tickets related to this issue; tagged one testneeded, closed the other. 18:07
18:08 havenn joined 18:10 skipper joined 18:12 kst left, kst joined
moritz jnthn: ok, I'll see what we can unfudge 18:12
... and merge match-refactor :-)
18:13 havenn left
jnthn \o/ 18:13
dalek kudo/nom: e46bfea | moritz++ | src/ (4 files):
Merge remote branch 'origin/match-refactor' into nom
jnthn
.oO( that branch needed exhaustive testing )
dalek Heuristic branch merge: pushed 150 commits to nqp/qbootstrap by jnthn 18:14
18:15 dakkar left
jnthn ah, it didn't dispaly the commit message for the merge commit. Summary: qbootstrap doesn't build at the moment (expected that, though) 18:15
dalek d: bd9fad8 | larry++ | STD.pm6:
add conjectural infix [&foo] parsing

Variants such as [&.foo], [&(...)] also allowed. Variants with $ sigil also admitted (but not @ or %; I suppose those could be allowed if keyed 2-dimensionally).		 18:17
moritz jnthn++ # the bs merge made the whole dev cycle (recompile, test) much snappier
jnthn moritz: Yes, it's more -Ofun to hack now :) 18:19
felher likes X[&foo] 18:23
[Coke] tries to fudge S03-junctions/misc.t for pugs and goes blind. 18:25
[Coke] adds a pugs todo to test 6. it turns into test 5, which then passes. 18:26
dalek d: 8fb2ea6 | larry++ | STD.pm6:
set $<O> correctlier
 18:28
18:28 noam left, wolfman2000 joined
TimToady evalbot rebuild std 18:28
p6eval OK (started asynchronously)
18:28 noam joined
dalek Rebuild of std complete. 18:31
moritz std: 1 18:32
p6eval std 8fb2ea6: OUTPUT«ok 00:01 107m␤»
moritz ok, right revision
dalek ast: ea3031a | moritz++ | S02-names-vars/ (2 files):
rakudo unfudge
 18:36
ast: d178924 | moritz++ | S06-operator-overloading/workout.t:
fix S06-operator-overloading/workout.t

the previous version assumed that &is_approx from Test.pm lifts infix:<->, which it does not (and the spec does not say so either) Also contains rakudo unfudges
ast: a66ef9c | moritz++ | S0 (5 files):
rakudo unfudges
TimToady moritz: well, it could be lying 18:37
dalek kudo/nom: c3b4d54 | moritz++ | t/spectest.data:
run operator overloading workout.t
 18:38
sorear good * #perl6 18:39
moritz are classes exported by default?
dalek d: 9c3ef6d | larry++ | STD.pm6:
don't forget to check the variable if [$foo]
jnthn moritz: No 18:40
moritz: But they're our-scoped by default
moritz: And then GLOBAL-merging.
TimToady evalbot rebuild std
p6eval OK (started asynchronously)
jnthn dinner time & 18:41
moritz oh, I see what's wrong
TimToady std: 42 [&atan2] 43
p6eval std 8fb2ea6: OUTPUT«ok 00:01 113m␤»
moritz t/spec/S06-operator-overloading/imported-subs.t starts as
use Exportops;
and t/spec/packages/Exportops goes like 18:42
module Exportops;
18:42 MayDaniel joined
moritz class NotANumber { stuff here } 18:42
dalek Rebuild of std complete.
18:42 MayDaniel left, MayDaniel joined
jnthn Right, that'll be available as Exportops::NotANumber 18:42
moritz and then imported-subs.t tries to uses that NotANumber (unqualified)
can I make NotANumber also 'is export' and have the short name available? 18:43
jnthn Yes
ooh, I should clean up the is export trait now we have bs too :)
18:45 kst left
jnthn First though, really dinner. 18:45
18:45 kst joined
dalek ast: 6551484 | moritz++ | / (2 files):
fix Exportops.pm, and unfudge S06-operator-overloading/imported-subs.t
 18:45
18:45 thou joined
TimToady std: 42 [$foo] 43 18:46
p6eval std 9c3ef6d: OUTPUT«===SORRY!===␤Variable $foo is not predeclared at /tmp/r49AhTSI7A line 1:␤------> 42 [⏏$foo] 43␤Check failed␤FAILED 00:01 110m␤»
moritz \o/
TimToady std: 42 [&foo] 43; # assumes post-declared
p6eval std 9c3ef6d: OUTPUT«ok 00:01 114m␤»
TimToady std: &foo
p6eval std 9c3ef6d: OUTPUT«ok 00:01 111m␤»
TimToady that too 18:47
niecza: &foo
p6eval niecza v15-2-gd19c478: OUTPUT«===SORRY!===␤␤Undeclared routine:␤ 'foo' used at line 1␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 1362 (die @ 3) ␤ at /home/p6eval/niecza/src/STD.pm6 line 1147 (P6.comp_unit @ 33) ␤ at /home/…
TimToady arguably should work like that though
dalek kudo/nom: b6c09b4 | moritz++ | t/spectest.data:
run S06-operator-overloading/imported-subs.t
TimToady sorear: where does niecza add_mystery for a bare &foo ? 18:49
18:49 daxim left 18:52 localhost left 18:54 localhost joined
moritz std: sub foo; foo() 19:00
p6eval std 9c3ef6d: OUTPUT«===SORRY!===␤Malformed block at /tmp/tJ4MC1rBfe line 1:␤------> sub foo⏏; foo()␤ expecting any of:␤ new name to be defined␤ routine_def␤ trait␤Parse failed␤FAILED 00:01 107m␤»
moritz rakudo: try eval 'sub foo; foo()'; say $!.perl 19:01
p6eval rakudo b6c09b: OUTPUT«X::AdHoc.new(payload => "Method 'ast' not found for invocant of class 'Undef'")␤»
moritz nom: sub foo; foo()
p6eval nom b6c09b: OUTPUT«===SORRY!===␤Method 'ast' not found for invocant of class 'Undef'␤»
moritz nom: sub foo //; 19:03
p6eval nom b6c09b: OUTPUT«===SORRY!===␤Missing block␤at /tmp/stu6NjbORV:1␤»
masak moritz: can I submit a rakudobug for the LTA error message on 'sub foo;'? 19:06
moritz masak: sure 19:07
masak submits rakudobug
moritz and when we fix it, I'll test it properly (ie by type of exception)
oh, it might even be my fault :/ 19:09
19:10 mj41 joined
felher moritz: would be so kind to take a look at github.com/felher/rakudo/commit/f9...2128ab5676 -- Maybe thats a patch that you wouldn't object to :) 19:11
+you
19:11 tokuhirom joined
moritz felher: I'll take a look, but it might take after lunch 19:11
19:11 sisar joined
felher moritz: thanks, thats fine with me :) 19:11
19:12 PacoAir joined 19:13 Teratogen left
Tene masak: can you also make a rakudobug out of this? gist.github.com/1955491 19:13
19:15 cbk1090 joined
sorear TimToady: (niecza's) STD.pm6 line 5465 19:16
masak Tene: I will if you golf it a bit further :) 19:17
Tene bah; that's already golfed down from my original by more than half.
masak (I already saw it in the backlog and decided not to submit it until it's shorter)
[Coke] phenny: "Wat is en Dampfmaschin? Da stelle mer uns janz dumm." 19:20
masak [Coke]: question mark
[Coke] phenny: "Wat is en Dampfmaschin? Da stelle mer uns janz dumm."?
phenny [Coke]: "Wat is a steam engine ? Then ask us janz stupid number ." (de to en, translate.google.com)
[Coke] That's not helpful. (subject of a message to p5p) 19:21
masak has the structure of a pun, but ich don't get it.
moritz it's a famous line from a German movie 19:22
and it's dialect, which is why Google Translate doesn't like it 19:23
"WHat is a steam engine? Let's act very stupid... " (ie, start at the basics)
Tene masak: what about gist.github.com/1955491 ?
19:23 birdwindupbird joined 19:24 kst left
flussence
.oO( we should use test suite progress to determine when Perl 6 Is Ready - all we need to do is fail less tests than PHP :)
 19:24
19:24 icwiener left, icwiener joined
cbk1090 I'm having problems with user input, is there a way to change a Str and make it an Int? 19:24
dalek ecs: f538b17 | larry++ | S01-overview.pod:
default to P5 only when invoked as 'perl'
moritz cbk1090: sure, call .Int on it
nom: say '42'.Int 19:25
p6eval nom b6c09b: OUTPUT«42␤»
moritz nom: say '42'.Int.WHAT
p6eval nom b6c09b: OUTPUT«Int()␤»
cbk1090 ok
19:26 pernatiy left, PacoAir left
cbk1090 It's not working... 19:26
if $data ~~ m/ (\d+ ) \s (\d+) / { $num1.Int = $0; $num2.Int = $1; say "Data: " ~ $num1 ~ ", " ~ $num2; return(1); }
moritz you probably want 19:27
$num1 = $0.Int
cbk1090 ok
19:28 kst joined
cbk1090 moritz, Thank you so much! 19:31
sisar why does comb have a default while split does not ? 19:35
moritz sisar: what would you default it to?
19:36 att joined
sisar is not falling for that ;-) 19:37
moritz sisar: but it's the heart of the answer
sisar: in p5, split() is rather magical
dalek d: 137d5f5 | larry++ | STD.pm6:
backport niecza's undefined &foo detector
moritz sisar: like, removing leading and trailing empty items
19:37 bluescreen10 left
moritz sisar: in p6, we don't do that, and instead have .comb and .words 19:38
19:38 kst left
moritz sisar: so if you want a split that defaults to /\s+/ or so, you should really be using .words instead 19:38
19:39 kst joined
sisar so when the spec says "The default is no longer ' ' since that would be interpreted as a constant. ", it is saying "no longer" to contrast with P5? 19:39
moritz right
Tene rakudo: grammar Foo { token element { | 'a' <element> { make 'a' ~ $<element>.ast } | ';' { make ';' } } }; say Foo.parse("aaaa;", :rule<element>).ast;
p6eval rakudo b6c09b: OUTPUT«Method 'match' not found for invocant of class 'Any'␤» 19:40
TimToady evalbot rebuild std 19:43
p6eval OK (started asynchronously)
Tene That parses fine if you remove the actions.
19:43 Tedd1 joined
dalek Rebuild of std complete. 19:44
masak sisar: even if Perl 6 is not the successor or "next version of" Perl 5, it still has its roots in it. 19:45
TimToady std: 1 [&atan2] 3; 1 [&btan2] 3
p6eval std 137d5f5: OUTPUT«===SORRY!===␤Undeclared routine:␤ 'btan2' used at line 1␤Check failed␤FAILED 00:01 114m␤»
fglock yay - heredocs work: perlcabal.org/~fglock/perlito5.html
masak fglock++
TimToady I hope you parsed them the right way and not the Perl 5 way :) 19:46
sisar moritz: "String delimiters must not be treated as rules but as constants. " Can you explain?
19:47 wolfman2000 left, kst left
moritz sisar: $str.split('.+') doesn't interpret the '.+' as a regex 19:47
sisar: but it actually searches for the substring '.+'
19:47 kst joined
sisar so it simply forbids regexes ? 19:48
sorear hi fglock! 19:49
moritz sisar: no, you *can* pass a regex
$str.split(rx/.+/)
sisar: the spec simply says that a string shouldn't be compiled automagically to a regex, like p5 does it
fglock base/lex.t is *very* hard
sorear: o/
dalek kudo/nom: ced04f7 | (Felix Herrmann)++ | / (2 files):
enable Configure.pl to pass options to Parrot's make
 19:50
kudo/nom: d4dc7d7 | moritz++ | src/Perl6/Grammar.pm:
fix regression introduced in 6588dd39, fixes bug #111492
sisar moritz: hmm. I'll have a look at P5 split, hopefully it won't confuse me :
moritz sisar: it simply means "don't do magic" :-) 19:51
sisar thinks magic is an overused word in perl lingo 19:52
[Coke] would update "no longer foo" to something like "contrasted with perl5's default, foo"
dalek ast: fcd1668 | moritz++ | S32-exceptions/misc.t:
Test for proper error of "sub foo;", RT #68710 and RT #111492
moritz [Coke]: feel free :-)
[Coke] (because I get the feeling some of those are referring to old p6 behavior) 19:53
moritz: I will eventually give the spec a front to back readthrough.
sisar [Coke]: i got the same exact feeling
moritz evalbot rebuild nom
p6eval NOT OK (maybe a rebuild is already in progress?)
[Coke] perhaps during the enormous layovers on the next business trip 19:54
evalbot rebuild THE UNIVERSE
p6eval OK (started asynchronously)
dalek Rebuild of THE complete.
masak haha
tadzik :)
[Coke] ... that was astonishingly fast.
sisar ;)
hehe
tadzik evalbot rebuild my house
p6eval OK (started asynchronously)
dalek Rebuild of my complete.
masak evalbot rebuild me into a cyborg
p6eval OK (started asynchronously)
dalek Rebuild of me complete.
tadzik :)
TimToady evalbot rebuild THE COMPLETE MULTIVERSE
p6eval OK (started asynchronously)
dalek Rebuild of THE complete.
masak it only gets to THE 19:55
that's not impressive.
I could do that.
TimToady ooh, white did it split on unbreakable space?
*why
moritz it might just search for \w+
luke the source
sisar evalbot rebuild evalbot 19:56
p6eval OK (started asynchronously)
dalek Rebuild of evalbot complete.
sisar does it actually rebuild evalbot? 19:57
moritz no
sisar ( it should)
moritz no
sisar so who builds evalbot ? 19:58
moritz nobody; it doesn't need a build step
masak evalbot rebuild masak 19:59
p6eval OK (started asynchronously)
masak ahh.
dalek Rebuild of masak complete.
moritz do we have a masak clone now?
tadzik that could be useful 20:00
masak no, just a fresher build.
tadzik I could use a clone of myself too
moritz :-)
masak one day I'll write that blog post about having seven clones.
TimToady it has to dismantle the previous one for parts to build the new one
masak I have it mostly mapped out in my head.
moritz yeah, that's the big problem with hardware
masak it's not completely problem-free to have clones, I'll tell you that much. 20:01
moritz yes, all the hassle with the authorities
sisar what language would a TimToady's clone invent ? might be interesting ;-)
moritz who pays the tax?
sisar: Perl. Of course :-)
felher++ # configure patch 20:02
masak moritz: we're under the radar. all the clones except one always stay in the apartment.
moritz: no, it's more a psychological thing. I never realized how obnoxious I can be!
hard to reason with sometimes. especially since I know just about all my own buttons to push. 20:03
moritz masak: ah, that explains the supposedly good wifi connection in Swedish busses
it's the clones from the apartment that do the IRCing
masak dang, I should have been more careful! 20:04
moritz it all makes sense now :-)
masak I mean... hehe, yes. it's them.
20:05 masak2 joined
masak2 masak: get off, it's my turn 20:05
masak :P
wow, that's unexpectedly provocative.
masak2 masak: seriously, you've been here all day 20:06
masak rule 1 when actually getting clones: agree to never be several online on IRC at the same time.
20:06 wolfman2000 joined
masak masak2: you're a phony, and I'll tell you why. 20:06
I'd never name a clone "masak2".
masak2 dang
masak that's classical rookie mistake in clone naming.
masak2 it's that "masak" is taken on freenode!
masak every clone will consider itself "the original masak", and will have to have a name that reflects that.
20:07 masak2 is now known as kasam
masak including, to the extent the distinction can be made, the original. 20:07
you see, I've been giving this quite a bit of thought.
sorear why is masak2 connecting from Poland with tadzik's hostmask?
masak for, um, research into my story.
tadzik sorear: no idea :)
masak sorear: yeah, it's all very confusing.
sorear did you accidentally leave a clone behind during one of your tadzik hackathons?
flussence sorear: they get around a lot
tadzik masak: I thought that was real you! 20:08
moritz tadzik: they are *all* real
tadzik /o\
skids
.oO(tadzik pirated a copy?)
kasam cover blown!
20:08 kasam left
masak yes. we have slightly different memories after the point of cloning, but we're still mostly the same. 20:09
flussence
.oO( compared to most other places, hostmasks are completely useless as unique identifiers in here... )
moritz masak: have you heard of the "paranoia" pen&paper RPG? 20:12
masak moritz: yes, only recently. I can see why you're thinking of it.
moritz if you die, you're replaced almost immediately by a clone with full knowledge of your previous self 20:13
(which is a bug in the system, you shouldn't have those memories :-)
masak :P
moritz oh, and it's lots of fun to play
masak I'd love to try it sometime. 20:14
tadzik never played it 20:16
20:16 kst left 20:17 kst joined
moritz it's kind of a meta-RPG; it makes you do all the stuff that you usually don't do in a "real" RPG 20:18
like, distrust and repeatedly kill your fellow players 20:19
tadzik heh
20:19 pjcj left
tadzik on a last D&D session there appeared a gang of whale hunters, whose goal in life was powerleveling (whales give quite a plenty of XP) 20:19
talk about broken 4th wall
moritz well, there's an easy solution for that: whales dying out :-) 20:20
tadzik they were just NPCs, no need to try to stop them :)
diakopter I'll dye your whale 20:21
masak diakopter! \o/
diakopter: you're supposed to shave the whales, not dye them!
diakopter writes JIRA plugins this week 20:22
diakopter wrote an html5 app optimized for iPad last week
too much variety
masak moritz: my new favorite game could be called a micro connection game: www.cameronius.com/games/span/
diakopter: is this $dayjob, or something else? 20:23
tadzik woot, rakudo passes 23k tests 20:25
sorear diakopter! o/ 20:26
jnthn tadzik: :D
20:27 skipper left 20:31 birdwindupbird left 20:32 noteventime joined
masak tadzik: didn't we pass 20k tests only weeks ago? 20:35
tadzik something along this, yes
20:35 Trashlord left
masak crazy. 20:36
so not only have Niecza and Rakudo been locked in an even race, they've been running full speed at the same time? 20:37
we shoulda done this competition thing ages ago :P
sisar masak: strangelyconsistent.org/blog/june-2...11-hangman doen't work for me 20:38
fglock we need a 3rd team, just writing more tests
20:38 Trashlord joined
sisar i'm getting 'words' as the word to be guessed, everytime 20:39
jnthn bets a missing .IO
sisar but line 1 for "states" works 20:40
masak sisar: huh. I'll look into it. sounds odd. 20:41
sisar ok
20:42 parcs` joined
moritz my $WORD = lines("words").roll; 20:47
yep, that's outdated
lines() now just breaks a string into lines 20:48
masak aha, that simple.
sisar: fix: lines("words".IO).roll;
I'll update the post.
moritz can be fixed with an .IO, just as jnthn++ predicted
or open('words').lines
masak sisar++, jnthn++, moritz++
moritz which I personally prefer
sisar masak++
masak I never liked open.lines -- the point of .lines is to abstract away the open 20:49
sisar .IO treats "words" as a Input/Output object rather than a string? 20:50
slurp does not need .IO ? 20:51
20:52 pernatiy joined
moritz well, .IO treats "words" as path 20:52
and lines(IO) reads from the IO object passed to it 20:53
20:53 kst left
sisar slurp defaults to slurp(IO) ? 20:54
moritz slurp opens the file for you if you pass a string (and not an IO object) to it 20:56
20:58 kst joined
sisar zzz... 21:00
masak moritz: you should learn to explain that in a way that doesn't make your audience fall asleep :P 21:01
21:12 snearch joined 21:15 Khisanth left
moritz masak: maybe it was something you didn't say :-) 21:16
21:16 snearch left
masak I'm glad sisar is perusing my Perl 6 intro blog posts from this summer. I'll try to live up to the honor. 21:19
dalek ast: 1af5f7a | moritz++ | S32-exceptions/misc.t:
RT #71814, =begin without identifier
 21:21
moritz I *love* it when I can use typed exceptions to close tickets that complain about bad errors 21:22
there are many tickets in the 'testneeded' queue where the error is already a typed exception
gfldex it all falls into place 21:23
there must be some sort of master plan
moritz well, it was one of my major reasons for applying for that exception grant :-)
so it's not really unexpected 21:24
but stilly very gratifying
masak moritz++ 21:26
21:26 gv left, GlitchMr left 21:27 kst left 21:28 kst joined 21:29 Khisanth joined 21:32 MayDaniel left
sorear jnthn: how is your time now? 21:33
jnthn sorear: I'm aboutish. 21:39
As in, physically here. Mentally, questionable. :)
(Insufficient sleep...)
moritz sorear: it seems that niecza's "make" isn't safe for parallel make when obtaining a new bootstrap compiler 21:41
and, completely unrelated 21:42
in rt.perl.org/rt3/Ticket/Display.html?id=76486 it's always the first multi that's called. Is that right?
jnthn moritz: looking
moritz and if yes, why doesn't the case with [] give ambiguous dispatch? 21:43
jnthn argh, that ticket has a bunch of examples. Which are we looking at?
moritz the very last one
in the last reply
perl6: multi sub a(@a) { say 1 ~ @a.perl }; multi sub a([]) { say 2 ~ [].perl }; a [] 21:44
p6eval rakudo d4dc7d: OUTPUT«1Array.new()␤»
..pugs b927740: OUTPUT«*** ␤ Unexpected "[])"␤ expecting formal parameter or ")"␤ at /tmp/UwCYKvZkPH line 1, column 50␤»
..niecza v15-2-gd19c478: OUTPUT«2[]␤»
21:44 y3llow_ joined
moritz I kinda agree with niecza here 21:44
21:44 pothos_ joined
jnthn Same...I'm thought we had a passing test for that case too 21:45
moritz I was about to write on... and it didn't pass :/
21:45 havenn joined
jnthn ah 21:46
Something is off here.
21:46 pothos left, y3llow left 21:47 y3llow_ left, y3llow joined
dalek ast: a288e1f | moritz++ | S06-multi/unpackability.t:
RT #76486
 21:48
moritz I'll just roast it, and reply to the ticket
jnthn oh, I think I see...
21:48 pothos_ left, parcs` left, pothos joined
noteventime I just started learning a little Perl 6 (having basically no prior experience with any Perl), and I'm a little confused by named parameters. How can one tell the difference between passing a named parameter and passing a non-named pair? If it does it by inspecting the type of the left hand side of the pair, what is that type and where/how is it bound? 21:49
21:49 y3llow left
masak noteventime: it's a syntactical distinction. 21:50
jnthn moritz: Looks like it may have been a bug in building the signature object. Compiling an attempted fix at the moment.
moritz noteventime: it's syntactic
21:50 y3llow joined
masak is faster :) 21:50
noteventime Ok
moritz noteventime: the a => $value and :a($value) forms are named
whereas 'a' => $value (note the quotes) and $pair forms are positionals that happen to be pairs
noteventime So I can't do anything magical like passing named variables by a hash table?
moritz noteventime: you can, be you have to declare your magic 21:51
mysub(|%h) uses all the pairs from %h as named arguments
noteventime And what types do the keys have to be?
21:51 pothos left 21:52 wolfman2000 left
noteventime (Or I guess, how does it map the keys of the hash table to the named parameters?) 21:52
moritz nom: sub greet(:$whom) { say "welcome $whom" }; my %h = whom => 'noteventime'; greet |%h
masak whatever type you declare your parameter to be.
p6eval nom d4dc7d: OUTPUT«welcome noteventime␤»
21:52 y3llow left, pothos joined
moritz by name :-) 21:52
21:52 y3llow joined
moritz the keys of a standard hash are always of type Str 21:52
noteventime Interesting
masak noteventime: it maps the hash key to the name of the named param.
21:52 kst left 21:53 fglock left
moritz and the value can be of type Mu, that is to say any possible value 21:53
sorear noteventime: you can also say fun((a => 2)) to pass a Pair value
21:53 kst joined
moritz ah right, that trick also exists. I usually forget about it. 21:53
sorear noteventime: there seems to be a semi-general rule in Perl 6 that parens disable syntactic magic
noteventime Ok, I didn't realise that hash table keys were strings by default :)
moritz sorear: except when they don't :-)
noteventime Nice, thanks for the explanation 21:54
21:54 y3llow left
moritz nom: my %h{Any}; class A { method Str() { 'foo' } }; %h{A.new} = 1; sub f(*%h) { say %h.perl }; f |%h 21:54
p6eval nom d4dc7d: OUTPUT«("A|7279588910590874249" => 1).hash␤»
moritz heh
it uses .WHICH
21:54 y3llow joined, pothos left
moritz which is... dubious, but somehow not surprising 21:55
sorear Hash table keys in Perl have historically[1] always been Str. It was only a week or two ago that Rakudo added support for non-Str hash keys, and Niecza still doesn't have them
[1] it's possible to plug in modules that override this
21:55 pothos joined
sorear rakudo: my $r1 = role { method foo() { 5 } }; my $r2 = role { method foo() { 7 } }; my %hash{Any}; %hash{"quux" but $r1} = 9; %hash{"quux" but $r2} = 11; say %hash.keys>>.foo 21:57
p6eval rakudo d4dc7d: OUTPUT«5␤»
moritz ouch
21:58 pothos left
masak huh. 21:58
21:58 pothos joined
moritz rakudo: my $r1 = role { method foo() { 5 } }; my $r2 = role { method foo() { 7 } }; say ("quux" but $r1).WHICH; say ("quux" but $r2).WHICH 21:58
p6eval rakudo d4dc7d: OUTPUT«Str+{<anon>}|quux␤Str+{<anon>}|quux␤»
moritz seems we are insufficiently paranoid in ClassHOW.gist, or we should be using something else in .WHICH 21:59
masak those look more like .WHAT than .WHICH to me.
moritz masak: feel free to rakudobug
jnthn Er, we shoudln't be using the typename. :)
21:59 kst left
jnthn Just use nqp::where($the_type.HOW) to get a unique identifier for the type. 21:59
moritz ok
22:00 kst joined
araujo wonders if someone is already using perl6 in production environments out there 22:00
jnthn I guess it doesn't matter what we use so long as it's type-unique.
sorear araujo: yes, it's done
araujo sorear, though it is still under development, I guess ... small scripts 100~200 LOC would be ok to go right?...
I mean for such a tasks
22:01 pothos left
moritz yes 22:01
sorear araujo: even large scripts are ok! it's more a question of performance and longevity
22:01 pothos joined
araujo sorear, I care about the latter mainly, yes 22:01
moritz people use such scripts to generate test data for "production" tasks, for example
22:01 havenn left
sorear you *will* have to make small changes to your scripts every few months to keep them running. 22:01
araujo sorear, right, that is why I guess small scripts shouldn't be a big deal to maintain 22:02
moritz, interesting
good to know
sorear araujo: use common sense and try not to write excessively fancy code
araujo nods 22:03
sorear if you write a torture test for the junction engine, it's much more likely to need tweaks
22:04 pothos left
araujo I guess using .. basic constructs and such... not getting too fancy as you say ... would keep the code easy to maintain 22:04
22:04 pothos joined
[Coke] moritz: I have fond memories of paranoia. 22:05
moritz too
time for sleep here
22:05 rodolo joined
[Coke] tadzik: I think you might be counting skips in your 23k. 22:05
moritz waves to the cheering crowd :-)
22:05 tokuhirom left
masak good night, moritz. 22:06
sorear jnthn: I'm wondering if now would be a good time to ask about your "shapes as mixins" comment
jnthn 'night moritz
masak dream of cheering crowds and butterflies.
22:07 pothos left
masak submits rakudobug of the 'but $r1' case above. 22:07
22:07 rodolo left, pothos joined
jnthn sorear: Today I implement typed hashes as Hash + a mixin. 22:07
sorear: Mostly, I was wondering if other such bits of S09 can be handled in that way. 22:08
sorear: I haven't thought it all the way through yet.
sorear: Seems a natural way of only making types with extra constraints pay for them, or on the flip side getting the information into the type system for optimizers to care about. 22:09
22:09 skids left
tadzik [Coke]: I probably am 22:09
jnthn sorear: There may well be drawbacks to that approahc. 22:12
*approach
masak araujo: my blog engine is written in Perl 6. it has been building HTML to be served to the world since September 2010. 22:13
araujo masak, hehehe nice :D 22:15
masak, how easy has been to maintain during this time? 22:16
many changes along the way?
masak practically none. 22:17
I don't remember anything that made me tear out my hair. 22:18
compared to maintaining, say, November back in 2008/2009, this has been a breeze.
flussence I have to say, I was really impressed after the announcement of forum.dlang.org's existence a few days ago 22:19
their site loads as fast as your blog does!
sorear meh 22:20
araujo masak, good to know :)
sorear anyone who's really serious about running a web site should use a PHP wiki engine
that's what mikemol does
dalek kudo/macros3: fca3907 | masak++ | / (7 files):
implemented quasi quotes and macros

This works:
  - Macro declarations
  - Calling a macro (using `macro()` and `macro` and operators)
  - Quasi quotes
  - Variable lookup from within the quasi quote
This doesn't, yet:
  - Returning a macro parameter instead of a quasi
  - Variable lookup from within a macro parameter
masak macros2 is dead -- long live macros3. 22:21
flussence third time lucky? :)
jnthn wonders how high we'll get :)
.oO( We probably need to get pretty high to work out macros :P )
mikemol sorear: >.> 22:22
masak hugs sorear
[Coke] writes a very very very small amount of haskell. 22:23
tadzik masak: you didn't name your clone masak2, but your branches are named macros, macros2, macros3... ;) 22:24
sorear If I have a clone I'll name em `uuidgen`
masak tadzik: the branches don't risk having their sense of primacy hurt. 22:25
tadzik :)
that's what _you_ think!
masak yes, it is.
flussence sorear: I think that's how a few sci-fi stories of clones going insane start...
dalek kudo/nom: dc6b0ef | jnthn++ | src/Perl6/Actions.pm:
For sub-signatures using [...] we should constrain to Positional.
 22:27
kudo/nom: 79c8e19 | jnthn++ | src/Perl6/Actions.pm:
Fix the do-we-have-a-subsig test so it doesn't drop 0-arity subsignatures.
ast: 2228433 | jnthn++ | S06-multi/unpackability.t:
Untodo a couple of multi tests.
22:30 kst left, kst joined
masak earlier this evening, I realized that I would probably have lots of long, silent psychological struggles with my clones. like, moving back and forth between a cupboard and a cup hanging from the wall. 22:40
no-one would say anything, but it would be like a tense situation. it would go on form months, or years.
for*
dalek kudo/nom: 4f62718 | jnthn++ | src/core/traits.pm:
Eliminate old hack from the bad old days when we ran traits again at startup.
22:41 kst left, kst joined
tadzik jnthn: how far is require from working? 22:41
masak s/moving back and forth/moving a toothbrush back and forth/ 22:42
tadzik or maybe import, not require 22:44
22:44 mucker left
jnthn tadzik: import as in...? 22:44
tadzik as in require 'Foo::Bar'; import 'Foo::Bar' 22:45
or such
jnthn That can't work as import operates at compile time
But require can probably be made to work pretty soon 22:46
I plan to look over the whole import/export thing a bit more soon. It gets a bit easier now.
Do you have a use case in mind? :)
tadzik hmm, require seems to Just WOrk
jnthn Oh. :)
tadzik looks for a use case he had in mind 22:47
jnthn Oh, it's marked as working but no import lists in features.
Which is the state I thought it was in.
tadzik what was the syntax for getting a class by name? 22:48
like ::("Foo") or something
jnthn nom: class A { }; say ::("A") 22:49
p6eval nom d4dc7d: OUTPUT«A()␤»
tadzik huh
jnthn Apparently, that :)
tadzik gist.github.com/1962048 this doesn't work for some reason 22:50
22:51 kst left
jnthn Try 22:52
::('MyModule')::('&foo')();
tadzik no change
22:55 kst joined
tadzik but then: my $a = &MyModule::foo; $a.() works 22:56
::('&MyModule')::('foo')(); does not
[Coke] seen au?
aloha au was last seen in #perl6 6 hours 4 mins ago saying "pugs: &infix:<...>(1,2)".
jnthn tadzik: INDIRECT_NAME_LOOKUP doesn't fall back to looking in GLOBAL, it seems 22:58
[Coke] phenny, ask au: given feather.perl6.nl/~coke/pugs.diff, all the method seem to work: I can do "3.Int.Num.Rat" -> 3/1 - however, when I say "Int" with this patch, I get an error about multis. pre-patch it returns "::Int"; any pointers?
phenny [Coke]: I'll pass that on when au is around.
22:59 att left
[Coke] (rough guess, this is worth at least 1000 tests. ;) 22:59
23:00 kaare_ left 23:02 glass left 23:03 havenn joined
jnthn tadzik: Got a patch locally 23:04
tadzik yay!
jnthn tadzik: 'twas a one-liner
Running spectest justin case 23:05
(only takes 3 mins or so here...)
tadzik :)
jnthn It *was* 3 mins before tadzik++ went and got us passing more tests :)
tadzik if a minute was 300 seconds it'd also take about 3 minutes in here :)
:P
23:07 mj41 left, kst left
jnthn tadzik: If you have a moment to add a test, that'd be awesome ;) 23:08
dalek kudo/nom: 715aed6 | jnthn++ | src/core/operators.pm:
Fix ::('blah') lookups to also fall back to GLOBAL (tadzik++ for noting the bug).
jnthn Note it needs to be ::('MyModule')::('&foo')(); 23:09
tadzik jnthn: I'll find one, sure :)
jnthn++
23:09 mj41 joined 23:10 kst joined 23:15 mucker joined
dalek ast: dfd1074 | tadzik++ | S11-modules/require.t:
Add a test for calling subroutines by name in a module that was require()d
 23:19
jnthn \o/
tadzik let's see if that works for classes... 23:20
\o/ 23:21
now zby should be able to port WebNano to Perl 6 :)
jnthn woo 23:22
masak \o/
tadzik at least until I get to know about another blocker :) 23:23
23:24 aindilis left 23:26 icwiener left 23:34 whiteknight joined
masak 'night, #perl6 23:36
jnthn 'najt, masak
23:38 skids joined
mathw lo 23:38
jnthn hi, mathw 23:39
mathw I seem to have broken my sleep schedule 23:40
it's 23:40 and I just woke up feeling like it's time to start the day
jnthn Oops. 23:42
:)
'tis done far too easily.
Though normally I do it in the summer, when there's not much darkness to remind me that night exists :) 23:43
mathw ah we don't really have that problem so much
my problem was that this morning (which I'm now thinking of as 'yesterday') I woke up at about 3am thanks to the pain in my stupid leg, went for a walk to loosen it out, and kind of never went back to sleep... 23:44
so it seems I went to sleep again at about 6pm
maybe
jnthn Sorry to hear about the leg :( 23:45
Hope it improves.
mathw I have my second visit to the physiotherapist in 10 hours
so far he's got it to loosen up in exchange for near-constant pain. I'm assured this is normal. 23:46
23:53 colomon left 23:55 mj41 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!