»ö« | perl6.org/ | nopaste: paste.lisp.org/new/perl6 | evalbot usage: 'perl6: say 3;' or rakudo:, alpha:, pugs:, std:, or /msg p6eval perl6: ... | irclog: irc.pugscode.org/ | UTF-8 is our friend!
Set by lichtkind on 5 March 2010.
TimToady the ordering of rules can only be partial 00:00
and some rules are unorderable wrt each other
oh, another thing about the trie extension notion is that we can quit as soon as we're down to one possible fate 00:01
we don't actualy have to follow the \w+ out to the end if it won't change anything 00:02
since the actual token rule will still decide whether it matches or not
the lexers job is really just to order the real rules, not actually find the ends of the tokens 00:04
*'s
the major thrust of S05 is to find some kind of grand compromise between deterministic and non-deterministic matching, and manage the transition between them on behalf of the user 00:05
00:06 f00li5h joined
TimToady that's going to feel awkward at times, for sure 00:06
hmm, no pun intended
00:06 whiteknight joined
TimToady but awk matching is deterministic, and not very powerful, while p5 matching is non-deterministic, and not very fast 00:07
my research is in getting these two relatively immiscible ideas to mix nicely, so I need a certain amount of "detergent"
and as with most of the design of Perl over the years, I try to let the design flow primarily from practical considerations of what I really need at the moment, and less from those clean academic sources that are nevertheless fundamental 00:10
sorear Why is remember_alts separate from walk in the gimme5 regex compiler? 00:11
TimToady p6 turns that around a bit and tries to sneak in as much of the fundamental goodness while hiding it from mere mortals, but it's still a grand compromise underneath, and cheating happens to get it all to mesh
there was a reason, lemme think
jnthn The nice thing is that the cheats seem to be increasingly cleanly expressed in terms of the fundemantal goodness. :-) 00:12
lue Hello!
jnthn Hellue! 00:13
lue Fedora 13 is out, and PPC is now a "secondary architecture"
One reason I picked Fedora was because they treated PPC so well :(
jnthn Ooh, I didn't know you were running on PPC. Nice that somebody is running Rakudo there. :-) 00:14
clkao win 28 00:16
lue I'm not developing on it. It crashes every time I try to compile [I ssh]
But fedora does provide Rakudo :)
jnthn Oh :-(
TimToady sorear: I don't remember offhand from glancing at it; it may have been a result of some fossil external requirement, perhaps forcing certain cross-linkages that are no longer evident, or that I've forgotten
sorear ok, I won't copy it 00:17
the propagation of @_ also appears to be a fossil
TimToady if you try to do it in the same pass, and run into a problem, you may have found the reason :) 00:18
sorear the only real way to understand something is to reinvent it
TimToady it was probably always just there as a just-in-case
I don't think I ever used @_ for anything
sorear viv culture uses locals in %main:: for that stuff 00:19
TimToady the separate pass might have been required under a different model of LTM generation that did more at compile time
maybe it will come to me later
yes, but contextuals hadn't even been invented yet when that all started 00:20
sorear ...invented?
00:20 synth joined
TimToady invented into p6 00:21
when they were, it was a conscious decision to emulate them with locals and give up multi-threading the parser
or more precisely, give up on the notion of modelling a cleanly re-entrant parser as long as gimme5 semantics were involved 00:22
sorear gimme5 is older than I had realized 00:23
TimToady in the early design of p6, we just carried locals over as "temp", but eventually I figgered out that temporizing globals was just going to make the same threading headaches that p5 has, so that's when I started de-emphasizing 'temp' 00:25
and started emphasizing $*FOO semantics, even though we can't actually have them in gimme5-world
well, we could, but it would be expensive
Pad::Walker, and such
one of the reasons the old 'v6' engine was slow was because it did try to emulate a lot of that stuff in p5 00:26
and as it is, STD.pm5 is only tolerably slow
fsdo tolerable... 00:27
tedv Question about committing updates to the pugs unit tests. I gave someone my email and they said they gave me authorization, but when I commit, it asks for a user name and password. Where do I set those up? 00:30
phenny tedv: 24 May 03:07Z <jnthn> tell tedv I just use the command line svn client, e.g. from subversion.apache.org/packages.html#windows
diakopter tedv: you should've gotten an email from the svn auth system 00:31
if you didn't, someone can rerequest one
tedv I should check my spam filters
who should the sender be?
diakopter hrm
[email@hidden.address] maybe?
TimToady commitbit?
tedv definitely nothing in my inbox 00:32
00:32 yinyin joined
diakopter msg your email address to TimToady :) 00:32
tedv I kind of figured he was busy with important stuff!
TimToady there's no tedv in there, was it under a different name? 00:33
tedv but alright, I guess it's like calling shotgun, but in reverse
tedvessenes at gmail dot com
TimToady okay
diakopter my childseat isn't allowed in the front row
tedv because gmail doesn't let me have 4 letter email addresses. :(
TimToady what do you want for the svn id?
tedv?
tedv tedv sounds good
namespace there isn't in quite as much demand as gmail addresses 00:34
TimToady okay, sent
tedv Oh hey, I just got email from audreyt. Not sure if I quite believe that, but it looks legit
TimToady it's customary to add yourself to AUTHORS as a test commit
yes, she set it up
so that's probably it
tedv Did she just fall off the face of the earth or something? 00:35
TimToady it might have some pugs-vm specific things in there that are no longer operative, but the svn bits should be right
tedv I had found an extremely simple thing to fix for a test commit, but AUTHORS sounds even simpler.
diakopter twitter.com/audreyt
a nit: the pugscode tests aren't unit tests; they're functional tests 00:38
tedv Is there any difference between the pugs repository that rakudo checks out in rakudo/t/spec and the official pugs repository?
TimToady t/spec is a sub-branch of the official repo 00:39
if you check out the whole repo you can also fix other things like the specs
[Coke] if you're *only* dealing with spec tests, I find it easier to work out of rakudo's t/spec
tedv I suspect it will be a little while before I'm changing the specs
TimToady and AUTHORS is outside of t/spec too :) 00:40
tedv start small and all that
TimToady go for broke!
there's always a contradictory saying :)
look before you leap! he who hesitates is lost!
lue Do not make fun of a penguin. 00:41
TimToady Make fun of a penguin.
there's another
lue Ignorance Is Bliss <-> Knowledge is power 00:42
tedv those don't contradict
since bliss isn't the same thing as power
TimToady
.oO(Oranges, poranges--there's another one)
tedv "Be innocent as doves and shrewd as serpents"
Jesus knew how to do it in the SAME SENTENCE 00:43
that's pretty hardcore
diakopter Prov 26:4, Prov 26:5
TimToady welcome aboard! and remember that we work on forgiveness rather than permission, at least for the first 70*7 mistakes. :)
lue [I thought it was 70**7]
TimToady the other reading is 77
tedv aramaic is not a good language for mathematical precision
diakopter or accuracy!
tedv the options are 490 or 77 though. I don't think powers were in their understanding of math 00:44
TimToady anyway, I'll go for the more forgiving view, at least until you start abusing it. :)
lue "I deleted the t/spec folder and commited it. Is that bad?" <== don't do thta 00:45
*that
TimToady there's also: "If anyone has a sword, he should bring it." "We have two." "Er, that's enough."
that was two too many, as far as Malchus was concerned :)
let's see, if I have to forgive you for each file in t/spec, that adds up pretty rapidly... 00:46
I think you've just exceeded your 490 in one fell swoop.
sorear TimToady: Is the setting of $PURE used for anything? 00:47
sorear thinks not
lue Oh, but it was replaced with a huge t.p6 file. Does that make up for it? :)
TimToady I don't believe it is anymore. might have been related to the separate pass and a different LTM regime 00:48
00:48 snarkyboojum left
lue needs to learn to type 'yum upgrade' more often O.o 00:49
TimToady just types "yup"
and svn update is just "sup"
lue gup 00:52
Git PUll 00:53
TimToady whoops, gotta go to a meeting to find a new "shepherd"
since our old one retired
back in three hours ±
& 00:54
lue That explains why I saw you three hours ago, you time-traveller 00:55
[oh wait, maybe that's not what the ± means....]
01:02 snarkyboojum joined
lue afk whilst upgrading 01:09
pugssvn r30795 | tedv++ | Added self to authors file; also the standard commit test. 01:18
snarkyboojum looks forward to a working REPL :) 01:23
lue: how's your dna hacking going? :) 01:28
01:37 am0c left 01:45 whiteknight left 01:56 meppl left
pugssvn r30796 | sorear++ | [viv] New two-pass regex compiler based much more closely on gimme5 logic 02:04
02:05 tylerni7 joined 02:16 colomon_phone joined
sorear TimToady: Is there a tracker for STD.pm6? I just found another bizarre misparse 02:25
regex foo { <bar(1)> } # The string 'bar' does not appear anywhere in the AST. <bar> works fine and is parsed using *the same rule*
02:26 cls_bsd joined 02:27 cls_bsd left
pugssvn r30797 | sorear++ | [viv] Fix normal name assertion undef error 02:27
02:31 cls_bsd joined 02:32 colomon_phone left 02:33 masonkramer joined 02:42 snarkyboojum left, cls_bsd left, cls_bsd joined 02:43 cls_bsd left 02:45 plobsing joined 02:47 cls_bsd joined 02:48 cls_bsd left 02:52 cls_bsd joined, cls_bsd_ joined
pugssvn r30798 | sorear++ | [viv] Reintroduce | and || 02:53
02:55 snarkyboojum joined, cls_bsd left, cls_bsd_ left, cls_bsd joined 02:56 cls_bsd left 03:00 cls_bsd joined
pugssvn r30799 | sorear++ | [viv] Implement literal atoms and simple quotes 03:21
03:26 masonkramer left 03:33 mberends joined 03:35 agentzh joined 03:44 gfx joined
sorear TimToady: What does <regex_block>:!s mean? 03:45
03:45 jaldhar joined 03:47 stephenlb joined 03:55 mimcpher joined, snarkyboojum left 03:56 revdiablo joined 04:02 snarkyboojum joined 04:09 TiMBuS joined 04:13 stephenlb left, dju joined 04:14 [mark] joined 04:23 dju left 04:24 dju joined 04:30 dju left 04:31 dju joined, dju left 04:34 JimmyZ joined
pugssvn r30800 | sorear++ | [STD] Correctly parse sigspace in rules 04:40
r30800 | [viv] Generate .ws calls
04:40 snarkyboojum left
TimToady sorear: :s turns on :sigspace, so :!s turns it off 04:52
sorear oh, that simple 04:55
I figured it was an adverb to the assertion
TimToady no, just left the space out because it would have been significant :) 04:56
and all those internal :s, :i, etc are lexically scoped, for some definition of 'lexical'
sorear does %*RX predate slangs? 04:57
it seems weird that the regex engine tweaks entirely differently from quotes
TimToady well, those aren't exactly slangs, since they don't exactly change the parsing 04:58
they're more like pragmas
sorear %*RX<s> changes what <.ws> matches
there were 2 bugs
1. rule did nothing to actually set %*RX<s>
2. someone tested 's' using $*RX<s> in a fit of p5think 04:59
both fixed
TimToady well, kinda, but really it's only supposed to change the meaning of non-null whitespace
2. gee, who could've that been
TimToady looks around for someone to blame
04:59 snarkyboojum joined
TimToady as for STD trackers, I keep a private todo file 05:01
so your message is already in it :)
sorear and yes, now that I've gotten viv to the state where I can just copy code from gimme5, things are going much faster :D 05:03
TimToady \o/ 05:04
looks like we're throwing out the longname with a language change that is not subsumed, maybe
05:13 kaare joined, kaare is now known as Guest65941 05:20 kel__ joined 05:27 slavik joined 05:46 XaeroOne joined 05:56 uniejo joined 05:59 TSa joined
moritz_ good morning 06:00
spinclad gutn morgn, moritz_ 06:01
TSa HaloO
dalek p-rx: 814727d | pmichaud++ | (7 files):
Enable outer_ctx handling in NQP programs (e.g., via NQP::Compiler.eval).

to do the same.
p-rx: d905d5b | pmichaud++ | src/cheats/parrot-callcontext.pir:
Skip any null lex_pad entries in the outer_ctx chain.
p-rx: 9125c3c | pmichaud++ | src/HLL/Actions.pm:
Add HLL::Actions.CTXSAVE to make it easy for compilers to save contexts.
p-rx: c5659bb | pmichaud++ | src/HLL/Compiler.pm:
Clean up exception printing when catching exceptions in interactive mode.
sorear good morning pmichaud 06:05
was my efforts helpful?
06:20 jhuni joined 06:21 Su-Shee joined 06:22 jakk joined 06:23 jakk left 06:26 Eevee joined
JimmyZ std: my @foo = <a c d>; for @foo (sub bar() { say $^a; })(); 06:27
p6eval std 30800: OUTPUT«===SORRY!===␤Missing block at /tmp/SnAnZP2bFv line 1:␤------> my @foo = <a c d>; for @foo ⏏(sub bar() { say $^a; })();␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤Parse failed␤FAILED 00:01 114m␤»
JimmyZ rakudo: my @foo = <a c d>; for @foo (sub bar() { say $^a; })();
p6eval rakudo 240400: OUTPUT«Missing block at line 11, near "(sub bar()"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
TimToady you have to use map if you want that kind of indirection 06:32
JimmyZ I just got a known bug :) 06:41
aesop whats $^a 06:45
sorear a self-declared positional parameter
like a named $_
aesop ahh
sorear if you have $^a in a block, it's automatically added as a parameter
06:46 BrowserUk joined
sorear { $^a leg $^b } # this is a 2-ary block usable as a sort callback 06:46
moritz_ rakudo: say {$^a leg $^b} 06:55
p6eval rakudo 240400: OUTPUT«_block46␤»
moritz_ rakudo: say {$^a leg $^b}.signature.perl 06:56
p6eval rakudo 240400: OUTPUT«:(;; Mu $a, Mu $b)␤»
sorear std: /a ** 2 - 3/ 07:00
p6eval std 30800: OUTPUT«===SORRY!===␤Unrecognized regex metacharacter (must be quoted to match literally) at /tmp/jIjYCOpZW3 line 1:␤------> /a ** 2 - ⏏3/␤Unable to parse regex; couldn't find final '/' at /tmp/jIjYCOpZW3 line 1:␤------> /a ** 2 -
..⏏3/…
sorear TimToady: potential for awesomeness 07:01
TimToady: What does <foo>:? mean? 07:02
STD doesn't parse it, but it's implied to be legal in S05 07:03
TimToady S05:2089
sorear yes... 07:04
that's where it's implied to be legal
but I can't figure out what the semantics are
07:06 viklund joined
TimToady well, eager vs greedy is only meaningful when applied to a quantifier, so <foo>:? merely means it can be backtracked into, and :! would likely have the same effect 07:10
whereas <foo>: may not backtrack into <foo>
sorear I see.
What is :+ ?
TimToady same as :! 07:11
or is it the same as :
sorear it doesn't seem to be specced 07:12
except in gimme5
TimToady I think *+ and ++ are ratcheting forms in some regex languages 07:13
sorear p5 ;)
07:14 masak joined
masak oh hai, #perl6! 07:14
m6locks o hai guise
TimToady it was added after p6 had specced :!, so there's some thought of adding it or switching to it 07:15
sorear STD parses <foo>:? as two atoms, the latter of which is quantified
is this correct?
TimToady no
07:20 snarkyboojum left
moritz_ rakudo: say samecase('FoO', 'bar') 07:25
p6eval rakudo 240400: OUTPUT«foo␤»
moritz_ rakudo: say samecase('FoO', 'bAr') 07:26
p6eval rakudo 240400: OUTPUT«fOo␤»
moritz_ rakudo: say samecase('FoObar', 'bAr')
p6eval rakudo 240400: OUTPUT«fOobar␤»
moritz_ rakudo: say samecase('FoObar', 'bAR')
p6eval rakudo 240400: OUTPUT«fOOBAR␤»
07:34 rgrau[1] joined 07:38 XaeroOne left
TimToady zzz & 07:39
07:40 snarkyboojum joined
mathw Morning 07:41
masak go' mo' 07:43
07:45 snarkyboojum left
pugssvn r30801 | sorear++ | [viv] Implement quantifiers 07:47
jnthn oh morning
masak :) 07:50
mmm. auto-print in the REPL.
jnthn masak: You likes it? :-)
masak it almost made me forget for a while that variables don't survive to the next line. almost. :) 07:51
moritz_ ... until you used a variable from the previous line, and it segfaulted?
masak actually, yes.
and I see from the backlog that others did the same.
moritz_ my plan for this week's contribution: implement :samecase in .subst 07:53
masak sounds good.
moritz_ should be about 5 lines of patch, if done right
+ tests
jnthn +1 07:54
mathw tests++
masak so... next week, the rakudosketch meeting will take place in... #phasers? :P 07:57
moritz_ yes
jnthn Yes
masak I leave you guys alone for five minutes... :P
07:57 yinyin left
jnthn We're all pretty fired up about the idea. 07:57
I'd bank on this name lasting a while. 07:58
masak oh no, now all the puns get triggered.
jnthn It's OK.
Jag har svenska klass nu. :-)
masak \o/
07:58 rgrau[1] left
JimmyZ who owns #rakudosketch? 07:59
moritz_ actually understood that :-)
JimmyZ: nobody. That's the problem :-)
masak JimmyZ: it's not registered.
JimmyZ nobody can unset it? 08:00
08:00 yinyin joined
JimmyZ bad 08:00
moritz_ we've beeen more careful with #phasers
masak .oO( especially the 'business end' of the phasers ) 08:02
JimmyZ two guys is there. sorear and PerlJam
*are
masak nice fellows. I like them. 08:03
they're six days early, though :P 08:04
moritz_ no, one day late
mathw They're camping out
like you do for tickets for Wimbledon
moritz_ there'll no more meetings in #rs
JimmyZ hey,there is another irc://irc.perl.org/#rakudosketch :) 08:05
moritz_ if in $str.subst($pattern, $replacement, :samecase) the $replacement is a closure, should the samecase() function be applied to its return value? 08:08
moritz_ assumes "yes" and moves along 08:13
masak seems to me it would, yes. 08:14
is there an alternative? 08:15
moritz_ ignore it in the case of a closure
08:16 ejs joined
moritz_ or apply it to the match object, and pass the result string to the closure 08:16
lolihazblogged: perlgeek.de/blog-en/perl-6/contribu....writeback 08:18
mathw isn't it week three? 08:19
moritz_ no, week 4 :-) 08:20
08:20 ejs left
mathw It says week 2 :) 08:20
moritz_ it says week 4... now :-) 08:21
mathw yay
masak 'ignore it in the case of a closure' breaks Least Surprise. 08:26
08:26 ejs joined
masak moritz_++ # blog post 08:27
moritz_ hopes that somebody else comes up (and writes up) with next week's challenge 08:32
mathw yeah I'd say apply it to the closure's return value 08:34
masak moritz_: detaching from a recurring task is nontrivial. not saying it can't be done, though. 08:35
people get to expect person X doing task Y, so they don't jump in and take over. 08:36
mathw quite
I'm not going to volunteer, because I don't have the knowledge to come up with these things
08:36 iblechbot joined
moritz_ masak: that's why I tried to hook people at the very beginning... diakopter++ and colomon++ have put their names into the README, and have come up with some ideas already 08:36
masak ah, good plan. 08:37
moritz_ if it matters I can also host challenges by other people on my blog
or just post a short notices to look at a different location
masak was just going to suggest that
if not the first solution, then the other.
08:40 dakkar joined
moritz_ masak: any chance Emmentaler is in need of a website next week? 08:42
masak moritz_: I don't feel in a position to promise anything. but I can share my dateless time table.
it looks like this.
(1) fix up GGE with Rakudo master. almost done with that. 08:43
08:43 BrowserUk left
masak (2) set up automatic builds of at least GGE, probably other projects as well. will be proto-based somehow. there's already a proof-of-concept for November, but it's b0rken. 08:43
(3) have the builds generate sensible data and save it somewhere on feather. 08:44
that's it. with (3), someone can do a weekly challenge.
but I'd drift towards "no" for next week, actually. there's only one of me, and we have a Yapsi release on Tuesday.
moritz_ ok
I hope you know if TAP::Formatter::HTML, btw :-) 08:45
masak I've used it for the November reporter thing. 08:47
it's functional, but I'm not a fan of the layout.
and it's a bit too detailed for what I have in mind.
08:47 BrowserUk joined
masak perhaps links from the main report page can link to such output, though. 08:47
hm. doing spectests, I got a non-zero exit status from t/spec/S12-methods/submethods.rakudo. unable to reproduce it outside of the harness. 08:49
08:49 gfx left
masak ditto t/spec/integration/99problems-41-to-50.rakudo. 08:49
three files contain TODO passed tests: t/spec/S14-roles/composition.rakudo t/spec/S32-hash/delete.rakudo t/spec/S32-str/uc.rakudo 08:50
moritz_ first and third are not reliable
pugssvn r30802 | moritz++ | [t/spec] remove a few #todo markers, masak++ 08:52
09:02 plobsing left 09:09 BrowserUk left 09:14 XaeroOne joined, iblechbot left 09:26 ejs left 09:32 ejs joined
jnthn back 09:39
09:41 clintongormley joined 09:47 finanalyst joined 10:14 yinyin left
mberends hello jnthn, here's a slightly harder zavolaj question: could you extend NativeCall.pm6 to create wrappers for native functions that return void? The crucial mysql_free_result() does that, and our wrapper presumably still pops something off the native stack, causing Parrot to segfault. 10:17
10:17 am0c joined, am0c left, am0c joined
jnthn mberends: If you don't write a "returns" it should already handle that. 10:21
mberends: Add a debugging say to see what it's making for the Parrot NCI signature string; it should start with a "v" 10:22
mberends jnthn: now why didn't I think of trying that... ;-/
jnthn :-)
moritz_ HOW IS CODE DEBUGGED? :-)
mberends google for the answres 10:23
10:23 ejs left
moritz_ goggle only tellz me how is babi formed 10:23
oh wait, that was yahoo answers
jnthn They should do way instain bugs!
10:28 am0c left 10:30 ejs joined
masak they need to do way instain developer, who kill their buggies. 10:30
jnthn closes his open RT tabs quickly, lest he be done away for looking like he might be about to kill buggies. 10:33
10:38 IllvilJa joined 10:39 JimmyZ left 10:42 fridim joined
pragma_ My name is pragma_ and I approve of this channel. 10:48
10:50 ejs left, IllvilJa left
masak \o/ we approve of you too. 10:51
10:52 colomon joined
masak jnthn: it's ok to kill buggies. it's just a bit sad because these buggy can't frigth back. 10:53
jnthn Aye, but they're lady to rest gently. 10:54
masak oh, good. 10:55
10:55 ejs joined 11:12 rgrau[1] joined
mberends jnthn: unfortunately the NativeCall still segfaults on void functions. It seems that sub make-mapper always gives a return-mapper data type. 11:16
jnthn mberends: What siggy does it make? 11:17
(the Parrot one) 11:18
mberends looks
jnthn mberends: It should start with v 11:20
(for void)
If not, we certainly have a zavolajbug
(if yes, we probably still have one) 11:21
11:21 [Coke] joined 11:28 [Coke] left, envi^home joined
moritz_ back from nom 11:35
colomon rakudo: my $a = "hello world"; say $a ~~ s/o/0/; 11:37
p6eval rakudo 240400: OUTPUT«hell0 world␤»
colomon rakudo: my $a = "hello world"; say $a ~~ sa/o/0/;
p6eval rakudo 240400: OUTPUT«Confused at line 11, near "say $a ~~ "␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
colomon rakudo: my $a = "hello world"; say $a ~~ ss/o/0/;
p6eval rakudo 240400: OUTPUT«Confused at line 11, near "say $a ~~ "␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
colomon rakudo: my $a = "hello world"; $a ~~ s[o] = '0'; 11:38
p6eval rakudo 240400: ( no output )
colomon rakudo: my $a = "hello world"; $a ~~ s[o] = '0'; say $a
p6eval rakudo 240400: OUTPUT«hell0 world␤»
jnthn
.oO( $stomach ~~ essess/empty/full/ )
 11:39
moritz_ std: s:g:g/a/b/ 11:40
p6eval std 30802: OUTPUT«ok 00:01 114m␤»
moritz_ std: s:g:foo/a/b/
p6eval std 30802: OUTPUT«===SORRY!===␤Unrecognized adverb :foo(1) at /tmp/xiwaLnki7A line 1:␤------> s⏏:g:foo/a/b/␤Check failed␤FAILED 00:01 111m␤»
11:40 ejs left
colomon moritz_: I'm just looking at subst.t, and noticing there seem to be a decent number of skipped tests that should work in Rakudo now. 11:40
and also, some of the tests are wonky. 11:41
moritz_ no surprise here :/
colomon and many of the skip messages are a bit off. :(
11:42 masonkramer joined
colomon "hello".subst(rx:g:i/L/,'p'), 'heppo' 11:42
should that really work the same way as "hello".subst(rx:i/L/,'p', :g), 'heppo' ?
moritz_ might work, might not
colomon it seems weirdly non-intuitive to me. 11:43
in fact, it really seems like rx:g shouldn't work at all.
moritz_ aye, :g, :x, :nth could be seen as "external" to the regex 11:44
colomon moritz_: exactly.
rakudp 11:45
std: rx:g/hello/
p6eval std 30802: OUTPUT«ok 00:01 112m␤»
11:45 ejs joined
colomon rakudo: my $a = "hello world"; say $a ~~ rx:g/o/ 11:45
p6eval rakudo 240400: OUTPUT«Confused at line 11, near "say $a ~~ "␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
colomon rakudo: my $a = "hello world"; rx:g/o/
p6eval rakudo 240400: OUTPUT«Confused at line 11, near "rx:g/o/"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
moritz_ rakudo: s/a/b/g 11:46
p6eval rakudo 240400: OUTPUT«Unsupported use of /g; in Perl 6 please use :g at line 11, near ""␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
colomon rakudo: 'a'.subst(/(.)/,"$1$0").say
p6eval rakudo 240400: OUTPUT«Method 'postcircumfix:<[ ]>' not found for invocant of class ''␤current instr.: '!postcircumfix:<[ ]>' pc 11646 (src/builtins/Cool.pir:15)␤»
moritz_ rakudo: $1 11:47
p6eval rakudo 240400: OUTPUT«Method 'postcircumfix:<[ ]>' not found for invocant of class ''␤current instr.: '!postcircumfix:<[ ]>' pc 11646 (src/builtins/Cool.pir:15)␤»
colomon that "$1$0" should never work correctly.
or at least, as a sane programmer probably intended it to work.
mberends jnthn: the signature is correct (vp) and the native function sometimes returns correctly, so the bug is elsewhere. I'll commit the signature printer, commented out. 11:48
colomon rakudo: 'a b c d'.subst(/\w/, 'x', :x(*)) 11:49
p6eval rakudo 240400: OUTPUT«Method 'Num' not found for invocant of class 'Block'␤current instr.: 'perl6;Mu;' pc -1 ((unknown file):-1)␤»
11:49 JimmyZ joined
colomon wonders why we want to support :x(*)... 11:50
rakudo: say 'a b c d'.subst(/\w/, 'x', :nth(2), :g)
p6eval rakudo 240400: OUTPUT«a x c d␤»
jnthn mberends: k, hm
moritz_ colomon: :x(*) is the same as :g 11:51
colomon: and while trying to implement it I found RT #75302
colomon moritz_: ah.
11:52 ejs left
colomon woah, that's a crazy bug. 11:52
moritz_ aye :-)
it took me a while to find out it wasn't my fault 11:53
colomon rakudo: sub f($x is copy) { $x = 'abc'; say $x.WHAT; say $x.perl }; f 'a' 11:54
p6eval rakudo 240400: OUTPUT«Str()␤"abc"␤»
colomon just checking. ;)
rakudo: say 'a b c d'.subst(/\w/, 'x', :nth(2), :g)
p6eval rakudo 240400: OUTPUT«a x c d␤»
11:54 pmurias joined
colomon moritz_: the test for that is 11:54
'a b c d'.subst(/\w/, 'x', :nth(2), :g), 'a x c x' 11:55
moritz_ that's wrong
colomon has the same problem that :nth and :x mixed has.
lisppaste3 pmichaud pasted "REPL in nqp" at paste.lisp.org/display/100504 11:57
colomon rakudo: say 'a b c d'.subst(/\w/, 'x', :nth({$_ % 2}))
p6eval rakudo 240400: OUTPUT«x b x d␤»
colomon rakudo: say 'a b c d'.subst(/\w/, 'x', :nth({$_ !% 2}))
p6eval rakudo 240400: OUTPUT«a x c x␤»
pmichaud good morning, #perl6
11:57 jhuni left
colomon o/ 11:57
mberends \o
pmichaud NQP now ha a REPL that remembers lexicals. :-) 11:58
moritz_ \o/
masak yay
moritz_ now port it to rakudo... and ship it!
jnthn morning, pmichaud++
masak good morning, pmichaud!
pmichaud port to rakudo should be straightforward (and should remove some existing code from rakudo)
colomon moritz_: has TimToady weighed in on the :nth and :x thing, or is that just your (extremely sensible) opinion? 11:59
moritz_ colomon: just my opinion
masak colomon: he has weighed in once.
it's somewhere in the IRC logs.
11:59 pmurias left
moritz_ guided by the principle that one construct should not magically mean different things under certain conditions 11:59
colomon It's easy enough to get the desired behavior using :nth and a closure, too. 12:00
moritz_: maybe we should change the Spec and see if anyone complains? ;)
takadonet morning all 12:01
moritz_ IMHO the spec is pretty clear
masak colomon: hah! found the discussion! :) irclog.perlgeek.de/perl6/2010-03-26#i_2158320
moritz_ :nth(2) means just one match, in 2nd position
masak (grepping the IRC logs for RT numbers)++
12:02 agentzh left
colomon moritz_: that's not how I read TimToady there at all. looks to me like he's suggesting the test is correct. :( 12:02
moritz_ colomon: he says he can argue for it, not that it must be that way 12:03
masak indeed.
12:04 masonkramer left
moritz_ TimToady++ is sometimes very careful to indicate that he's not convinced "his" option might not be the best. We should take these clues as we see fit :-) 12:05
s:2nd/not//
12:05 mimcpher left
moritz_ s:nth(2):x(1)/not// # to make my case :-) 12:06
colomon moritz_: did you do those :nth with :x tests?
moritz_ colomon: I did... before I understood that :nth(2) never means "ever second match"
colomon: I've since then corrected some tests, but not all :(
and I did fix the spec :-)
masak moritz_: that's a useless use of :x ... :) 12:07
moritz_ masak: so you'd think... :-)
masak doesn't know what to think now 12:10
moritz_ then think some GSOC
like, adding a Stringy role
masak I did.
that test passes now.
it didn't last time I ran it. 12:11
moritz_ oic
masak something must have miraculously fixed itself in the meantime.
moritz_ but Str ~~ Stringy is NYI, right? 12:12
masak it's I in my buf branch.
oh wait.
no, just Buf ~~ Stringy.
colomon rakudo: my $a = "hello world"; $a ~~ s[o] = '000'; say $a
p6eval rakudo 240400: OUTPUT«hell000 world␤»
colomon oh, bother.
12:13 ejs joined
colomon rakudo: for <hello world> { s[o] = '000'; say $_; } 12:13
p6eval rakudo 240400: OUTPUT«hello␤world␤»
moritz_ masak: colomon had some fun changing prefix:<+> to call .Numeric, not .Num; I think in the long run you'll need to change prefix:<~> to .Stringy too
colomon "fun" in the "not really -Ofun" sense. :) 12:14
masak -O-fun
moritz_: I'll keep that in mind.
colomon what's the protocol for false positive tests? skip them? 12:15
masak rakudo: my $a; say "I do not have ", $a, " shoes on"
p6eval rakudo 240400: OUTPUT«I do not have Any() shoes on␤»
masak that's spec'd behaviour nowadays, right?
colomon rakudo: for <hello world> { say s[o] = '000'; say $_; }
p6eval rakudo 240400: OUTPUT«Substitution()<0x3f33340>␤hello␤Substitution()<0x3cc5af0>␤world␤» 12:16
colomon wtf?
oh.
rakudo: for <hello world> { say (s[o] = '000').Bool; say $_; }
p6eval rakudo 240400: OUTPUT«1␤hello␤1␤world␤»
colomon rakudo: for <hello world> { say (s[w] = '000').Bool; say $_; }
p6eval rakudo 240400: OUTPUT«1␤hello␤1␤world␤»
masak I thought the 's[] = $rhs' form was purely syntactical and didn't need a type of its own. 12:17
moritz_ it should default to $_, and doesn't
known bug 12:18
colomon masak: I have no clue, I'm just looking at the "s[...] = ... returns true on success" test in subst.t
jnthn masak: To do that you'd have to also make the smartmatch recognize it as some kind of special form too
masak moritz_: I tried augmenting Str to do the Stringy role. it seems to have worked! spectesting.
jnthn: hm, true. 12:19
jnthn masak: Which felt messy.
At the time, nobody could give me an answer for how it should be done, so I just Did Something.
colomon rakudo: for <hello world> -> $a { say ($a ~~ s[w] = '000').Bool; say $a; }
masak Something++
p6eval rakudo 240400: OUTPUT«Cannot assign to readonly value␤current instr.: '&die' pc 18205 (src/builtins/Junction.pir:381)␤»
jnthn It's almost certainly not right though.
colomon rakudo: for <hello world> -> $b { my $a = $b; say ($a ~~ s[w] = '000').Bool; say $a; }
p6eval rakudo 240400: OUTPUT«1␤hello␤1␤000orld␤»
colomon aarrrrrgh! 12:20
masak jnthn: fwiw, I would have guessed the smartmatch to be part of the special form.
jnthn Perhaps so.
Just...ugly to do.
masak aye. might involve backtracking.
jnthn I think pmichaud++'s context PAST node might help us fix it though.
Becuase then we can at least say "oh, we're in sink context, clearly we should do it to $_" 12:21
colomon must remember not to be sad about writing tests that fail...
moritz_ I think s[...] should always default to $_
and smartmatching sets $_ for the RHS
pmichaud moritz_: +1
masak the problem with irclog.perlgeek.de/perl6/2010-03-26#i_2158369 is, how do you find out the upper limit? and if that's not a hard enough problem, how do you do it with a closure?
moritz_ and then .sink triggers execution of the substitution
jnthn moritz_: Feel free to try it.
.sink is, iirc, probably actually spelt .eager. 12:22
pmichaud no, I think it's .sink
moritz_ jnthn: types are free to re-interpret .sink as something else
pmichaud thinks he'd better read some backscroll 12:23
masak uhn. are for loops eager because they're in sink context? 12:25
moritz_ masak: yes
colomon for loops are eager IF they're in sink context 12:26
masak what about a for loop at the end of a sub-or-method? will it behave differently? :(
moritz_ yes
pmichaud depends on how the sub-or-method is called
masak true.
maybe it's not so bad, then.
pmichaud a for loop at the end of a sub isn't in sink context, no.
but if the sub invocation is in sink context, the for loop gets executed eagerly then 12:27
jnthn pmichaud: Hmm. I can't find the bit of irclog for the discussion I was thining of.
*thinking
pmichaud it always felt to me as though :x and :n argument handling should be done with smart matching. 12:28
moritz_ :nths is
masak heh -- laziness context is like the value context for Perl 6. :)
moritz_ in rakudo
colomon code is $i !~~ any |$nth 12:32
moritz_ anyway, I'm rather firmly convinced that what we have now for :x + :nth is the sanest semantics, until somebody comes up with a different, consistent interpretation and shows working code for it 12:33
takadonet moritz_: sounds like a challenge to me..
masak colomon: I'd classify that as one of the confusing uses of negation and junctions...
moritz_++ # pragmatism 12:34
colomon masak: hmmm, I think it could be rewritten in the positive sense without breaking anything...
mathw isn't sure what the prefix | does there 12:35
masak colomon: if I read it correctly, it means '$i ~~ all $nth'.
moritz_ takadonet: not quite a challenge; rather a try to end non-productive bikeshedding
colomon masak: oh, no no no no
takadonet moritz_: sounds good
moritz_ mathw: interpolation into argument lists
masak colomon: see? confusing! :)
colomon: now explain to me why it doesn't mean that.
moritz_ masak: prefix:<|> is interpolation into argument lists 12:36
masak mathw: what moritz_ said but mistabbed.
colomon: sorry, left out the | in |$nth.
12:37 elmex joined
moritz_ takadonet: also if somebody actually cares enough to write code for it, I'm willing to reconsider my position. Until then I just assume that I put the most thought into (hubris :-) 12:37
mathw oh yes of course
I should have remembered that
masak colomon: $i !~~ any(|$nth) is the same as any(map { $i !~~ $_ }, |$nth), isn't it? 12:38
colomon masak: darned if I know. these things are hard to explain.
pmichaud actually, $i !~~ any(|$nth) should be the same as !($i ~~ any(|$nth)) iiuc
moritz_ masak: yes
masak ok, pmichaud and moritz_ just contradicted each other. 12:39
this should be interesting.
colomon pmichaud: that's how I read it.
moritz_ wait
moritz_ not so sure anymore
pmichaud++ is right
mathw I read it as "$i is not any of the things in $nth"
masak dang. :)
moritz_ the negation comes at the end
pmichaud negations and junctions are problematic
(in general)
masak apparently so.
I accept what you're saying, but I can't see how it works out that way.
jnthn To the point that the spec in one place suggests a warning.
masak how does the ! end up outermost?
pmichaud it's a metaop 12:40
it always ends up "outermost"
jnthn masak: All ! meta-ops are like that.
12:40 [mark] left
masak the ! metaop ends up outside of junctions? cool! 12:40
jnthn $a !== $b is !($a == $b) really
masak jnthn: oh, indeed.
so this doesn't go for the infix:<ne> operator?
moritz_ nope
that's not meta-y 12:41
masak that's probably what bit me once, then.
mathw infix:<ne> would presumably be subtly different to !eq
since it does all its !-ness inside
masak after that, I've made large circles around negated comparison ops combined with junctions.
moritz_ so !eq is not quite the same as ne, when junctions are involved
masak someone should write a blog post about that.
and a u4x entry.
moritz_ masak: that's a good item for our best practise list
jnthn btw, Rakudo's ne is implemented in terms of !eq 12:42
masak indeed.
jnthn If you want to avoid the inconsistency, then change infix:<ne> to take two Mus.
masak jnthn: so maybe that's a potential bug in Rakudo? :)
jnthn masak: Only if we can decide what the spec is. ;-)
masak didn't we just do that?
pugssvn r30803 | colomon++ | [t/spec] Clean up fudging and tests a bit. 12:43
moritz_ masak: where's the bug you're talking about?
jnthn moritz_: Did you mean !eq is not the same as ne is current Rakudo reality, or an awkward but fixable issue, or?
Or intentional?
jnthn would kinda feel more comfortable if they were the same.
moritz_ rakudo: say 1 ne 1|2 12:44
p6eval rakudo 240400: OUTPUT«any(Bool::False, Bool::True)␤»
moritz_ rakudo: say 1 !eq 1|2
p6eval rakudo 240400: OUTPUT«0␤»
masak what happens in that first one?
moritz_ that's correct as I understand the current spec
masak: autothreading over ne
jnthn masak: We auto-thread over ne
masak right, and the junction shouldn't collapse, of course.
jnthn In the second we auto-thread voer eq, and then the ! collapses it.
masak: Yeah, I was asking if we really want that 12:45
masak so it's actually correct? hooray!
jnthn masak: Or if we should mark ne and != as having Mu parameters.
So they work out the same as !eq and !==
masak jnthn: I don't see why we should.
mathw I'm wondering how often it's going to bite people
And if we want to allow the distinction
Is it potentially useful?
pmichaud as I read the spec, infix:<ne> is supposed to be the same as !eq
i.e., it's just a shortcut. 12:46
masak so it is a bug?
moritz_ masak: only by pmichaud++'s interpretation
masak submits rakudobug
that's enough for me. :)
mathw I'm not convinced
pmichaud S03:1189
As in Perl 5, converts to C<Str> before comparison. C<ne> is short for C<!eq>.
moritz_ ok 12:47
that's enough for me :-)
12:47 plainhao joined
pmichaud also later at S03:3915 12:47
mberends just re-asking before committing: is it OK to put the directories specified in PERL6LIB before the defaults in @*INC instead of after, to be able to override system modules with local ones? (it works fine locally) 12:48
pmichaud that tells me that &infix:<ne> is more macro-ish than a normal function, in the long run.
moritz_ mberends: +1 from me
masak mberends: that sounds like the Right Thing to me, at least.
mathw I think any way this goes is going to confuse somebody at some point
if ne is just syntactic sugar for !eq, it's going to surprise someone 12:49
and the current behaviour is going to surprise someone else
colomon masak: code is now if !$nth.defined || ($i ~~ any |$nth)
mathw So my inner paper shredder says why do we need infix:<ne> :)
colomon hope you find that nicer. :)
(spectesting)
moritz_ mathw: because TIMTOWTDI :-) 12:50
12:50 M_o_C joined
colomon moritz_: what, to surprise people? ;) 12:51
jnthn Because we might get lynched if we don't have == :-)
moritz_ colomon: it would certainly p5 programmers if we had eq but not ne
jnthn Heh. Back when we had prefix:<=> defined, if we'd dropped ==, then $a == $b woulda parsed as $a = (=$b) :-D
12:51 fridim left
moritz_ colomon: and then we'd have to install infix:<ne> just to generate an error message 12:51
12:52 SmokeMachine joined
jnthn oh wait 12:52
it would be != we dropped
But same
Two prefixes :-)
moritz_ colomon: and then people will ask "why don't you just make infix:<ne> the same as infix:<!eq>"
dalek kudo: 39c38ca | (Martin Berends)++ | src/Perl6/Compiler.pir:
[Perl6/Compiler.pir] change the order of directories in @*INC so that the
jnthn But then two terms in a row. :-)
masak colomon: I do find it nicer. colomon++
colomon moritz_: I'm not actually against ne, just amused at invoking TIMTOWTDI in that context :)
12:52 alanhaggai joined
jnthn Anyway, we do ne and != often enough that's probably a huffman argument for 'em anyways. :) 12:53
pmichaud mberends: it just occurred to me... what does perl5 do in the case of @INC?
mathw Maybe
moritz_ colomon: you're right, it was not the best argument :-)
moritz_ knows that 'use lib $path;' prepends to @INC
mathw It just feels like an inconsistency to me if <ne> behaves with junctions like a meta-opped <eq>
Because one of them would return a bool, and one of them would return a junction of bools 12:54
12:54 perlygatekeeper joined
moritz_ mathw: if you take the radial approach, junctions are the problem, not the ne/!eq multi existence 12:54
mathw: so just abolish junctions, and everything is suddenly much easier :-) 12:55
mberends pmichaud: good point. PERLLIB also goes before, just checked empirically
mathw moritz_: just make all comparison operators produce a single bool when used on junctions :)
pmichaud mberends: if perl 5 does it that way, that's a strong argument for rakudo to do the same :)
mberends yes, they agree now :)
masak also, there's an argument by sanity. :) 12:56
jnthn mberends: Nice. :-)
mathw moritz_: well except the ones that don't return boolean values of course
12:56 SmokeMachine left
masak if the user actually *specifies* a path in the invocation, that's pretty prioritized. 12:56
pmichaud (junctions and negation) S03:3121 discusses this in some detail. ("Use of negative operators...")
I'm not sure if that needs to be updated to take into account !'s meta-ness. 12:57
moritz_ pmichaud: you're so boring, citing specs when we could bikeshed for ages... :-)
mathw pmichaud: yes can't you tell I'm trying to procrastinate here? 12:58
mathw reads S03
pmichaud because the way S03:3121 reads, it makes it sound as though infix:<!=> is its own operator, and not the same as !($a == $b)
12:59 snarkyboojum joined
pmichaud mathw: (procrastination) -- trying to resolve ambiguities/contradictions in the synopses is a great way to procrastinate. Unless you're trying to procrastinate resolving the ambiguities and contradictions in the synopses. :-) 12:59
mathw pmichaud: the best form of procrastination gets something else useful done
and yes, S03:3121 does seem to indicate that
pmichaud mathw: that sounds good... I'll need to think on it more a bit later.
mberends uh-oh, perl5 puts '.' at the end of @INC, perl6 has it immediately after PERL6LIB, which is also better for overriding system with local. Hmm, don't want to follow perl5 there. 13:01
pmichaud mberends: iiuc, having . anywhere but the end is a potential security risk 13:02
moritz_ '.' in INC is potentially harmful
we should disable that once wie have 'use lib' and Cwd and such nice modules
mberends ok, '.' can move to the end 13:03
pmichaud "overriding system with local" is also another way of saying "silently replace the standard system module with a hacker-compromised version" 13:04
mberends it sounded like a good idea for testing, but indeed it's bad for production 13:05
pmichaud while it makes things easier for developers, it .... right
developers should probably have the burden of extra work here. :-)
mberends heh, yes
13:05 moritz_ sets mode: +o mberends 13:08 Guest65941 left
mathw yes doing that should be possible, but very explicit 13:11
13:14 wasmahen joined
wasmahen hey 13:14
is this perl6
mberends wasmahen: oh yeah!
masak it is.
welcome!
wasmahen i have been looking at this wikipedia article on perl6 en.wikipedia.org/wiki/Perl6
are we regressing?
masak a little, sometimes.
wasmahen perl 5: my $i = "25" + 10; # $i is 35
masak aye. 13:15
wasmahen perl6: my Str $s = "Hello, world";
why
p6eval elf 30803, pugs, rakudo 240400: ( no output )
moritz_ rakudo: my $i = "25" + 10; say $i
p6eval rakudo 240400: OUTPUT«35␤»
masak wasmahen: you don't have to type your variables if you don't want to.
moritz_ still works.
wasmahen do i have to declare my variable type
masak no.
pmichaud wasmahen: no.
masak that's what I meant by 'you don't have to'.
pmichaud wasmahen: if you do, it's a type constrait. 13:16
wasmahen so $str = "hello world" works
masak yup.
wasmahen great
pmichaud well, you still have to declare $str
moritz_ perl6: my $str = "hello world"; say $str
p6eval elf 30803, pugs, rakudo 240400: OUTPUT«hello world␤»
pmichaud you just don't have to type it
wasmahen great, i was afraid for a moment there 13:17
am expecting a lot from perl6
masak wasmahen: does the Wikipedia article give the impression that you have to provide types for all variable declarations?
wasmahen masak: ya
masak: it says: In Perl 6, the dynamic type system of Perl 5 has been augmented by the addition of static types
snarkyboojum and then "However, as with Perl 5, programmers can do most things without any explicit typing at all:" 13:18
13:18 nodie joined
masak snarkyboojum! \o/ 13:18
wasmahen ya, as if telling me perl5 could do this but in perl6, they made is more complicated
it*
PerlJam greets
snarkyboojum masak: o/
pmichaud PerlJam: /part #rakudosketch 13:19
masak wasmahen: 'Perl 6 offers a hybrid typing system whereby the programmer may choose to use Static Typing, Dynamic Typing or mix the two.'
wasmahen: seems to me to be pretty explicitly telling you that it's optional.
pmichaud I'd move the option description a bit higher
snarkyboojum I guess it points out static types first because that's new in Perl 6 13:20
wasmahen masak: all this is too complicated words for me
PerlJam pmichaud: Is someone going to register the channel, etc.?
wasmahen english is not my maternal language
masak wasmahen: ok. thanks for letting me know.
pmichaud "However, static typing remains optional, so that ..."
instead of "However, as with Perl 5, ..."
moritz_ PerlJam: we moved to #phasers because nobody had registered #rs
PerlJam: so we couldn't even set the topic
pmichaud only sorear++ is left in the channel when PerlJam++ parts 13:21
masak pmichaud: how do you know? :)
wasmahen: the Wikipedia page was clear, but maybe not perfectly organized. thanks for pointing it out.
jnthn
.oO( nobody a/part from ... )
 13:22
wasmahen ok goodnite
13:22 wasmahen left
pmichaud masak: by surreptitiously peeking my head inside the channel from time to time :-) 13:22
PerlJam 13:22 -!- PerlJam [duff@feather.perl6.nl] has left #rakudosketch [] 13:23
masak pmichaud: you can't measure an IRC system without also affecting it :P
pmichaud masak: I'm willing to claim that my statement was/is correct, however. :) 13:24
masak: and I'm not sure anyone can disprove it. :)
masak pmichaud: lol
moritz_ :-)
pmichaud when PerlJam++ parted, only sorear++ was left. That seems the most likely situation.
13:24 nodie left 13:27 takadonet left
dalek kudo: 1728e99 | (Solomon Foster)++ | src/core/Cool-str.pm:
Clarify logic at masak++'s suggestion.
 13:28
kudo: 51b61c5 | (Martin Berends)++ | src/Perl6/Compiler.pir:
[Perl6/Compiler.pir] re-order @*INC again, '.' goes to the end, pmichaud++
13:28 jhuni joined
dalek kudo: 6f0e67d | (Martin Berends)++ | src/core/Cool-str.pm:
Merge branch 'master' of github.com:rakudo/rakudo
 13:28
kudo: fd769c8 | pmichaud++ | (2 files):
Update Perl6::Actions to grab dynamic outer compilation via HLL::Actions
kudo: 20068fa | pmichaud++ | src/ (2 files):
Merge branch 'master' of github.com:rakudo/rakudo
pugssvn r30804 | colomon++ | [t/spec] Add tests for .subst with :p and :c options.
colomon woah
13:30 pmurias joined 13:31 zostay left 13:34 isBEKaml joined 13:39 ejs left 13:40 pmurias left, SmokeMachine joined
isBEKaml hi all! :) 13:40
13:41 JimmyZ left
moritz_ colomon: do those tests pass? 13:41
13:42 kaare joined
isBEKaml oh, I was wondering if my network went into latency or this channel was _really_ silent.. ;) 13:43
13:43 kaare is now known as Guest74734
pmichaud Perhaps the people in the channel went into latency. 13:43
moritz_ isBEKaml: you can always check the website with the IRC logs :-)
isBEKaml :) 13:44
PerlJam isBEKaml: We're all here to varying degrees :)
13:44 ejs joined
isBEKaml #perl6++ # I wish I could +1 everyone. :) 13:45
pmichaud I bet it could be done with a Perl script. :)
moritz_ irssi directly supports Perl plugins 13:46
isBEKaml yes, and I'm yet to learn how to do those. But I use XChat since I don't have an unicode ready console, yet..
masak hi, isBEKaml! :) 13:47
isBEKaml yo, masak! :)
moritz_ the documentation of the plugin API is in a rather sad state
it basically consists of a short intro (Juerd++), the source code and many example scripts 13:48
isBEKaml moritz_: XChat or irssi?
moritz_ isBEKaml: irssi
masak Perl 6 doesn't have a way to combine hexadecimal literals and scientific notation, does it?
isBEKaml I have used both, but I knida like irssi better since i don't have to go into X for that. :)
moritz_ masak: sure it does :/ 13:49
search S02 for :16< or so
13:49 patspam joined
moritz_ rakudo: say :16<a0*10**2> 13:50
p6eval rakudo 240400: OUTPUT«16000␤»
masak moritz_: I know about that, but it doesn't elimiate the fact that you can't use E to mean 'scientific notation' in a hexadecimal literal.
colomon moritz_: yes, the :p and :c tests all pass, no prob.
masak but I suppose ** works just as well. 13:51
moritz_ urk, t/spec/integration/advent2009-day02.t test 8 failed
fixing... 13:53
pugssvn r30805 | moritz++ | [t/spec] fix typo
dalek kudo: 4569863 | masak++ | src/Perl6/Module/Loader.pm:
[Module::Loader] removed redundant period

without a full stop after them.		 13:59
13:59 gbacon joined
moritz_ masak++ # cleaning up my mess :-) 13:59
masak it was a very minor mess. :)
but it's a great feeling to have the power, when bothered by a minor detail, to just commit the change, and it's fixed! \o/ 14:00
moritz_
.oO( masak admits his power hunger :-)
 14:01
masak om nom nom.
isBEKaml rakudo: say 'The foo and the bar'.subst(/:i the/, 'that', :g, :samecase); 14:02
p6eval rakudo 240400: OUTPUT«that foo and that bar␤»
isBEKaml moritz_: I'm confused by your blog challenge. What was it about? hooking :samecase to Cool-str ? 14:03
moritz_ yes
and still is
if it worked, the first 't' in the output would have been upper case
isBEKaml ah 14:04
rakudo: say 'The foo and the bar'.subst(/:i the/, 'that', :g);
p6eval rakudo 240400: OUTPUT«that foo and that bar␤»
moritz_ it just ignores options it doesn't understand 14:05
isBEKaml so, :samecase is redundant as of now. But challenge is to add weights to it. :)
moritz_ right
masak isBEKaml++ # I can see you're thinking about it... :) 14:06
isBEKaml masak: lol, I don't know if I can come up with something. I'm only git pulling.... :) 14:07
PerlJam 'THE'.subst(/:i the/, 'that', :samecase); # what should this do? :)
moritz_ rakudo: say samecase('that', "THE")
p6eval rakudo 240400: OUTPUT«THAT␤»
moritz_ exactly THAT :-)
14:07 M_o_C left
PerlJam excellent. 14:08
snarkyboojum yay, I just did an HTTP request using Rakudo :P
lol
masak PerlJam: S05:225.
PerlJam masak: yeah, just got there :)
isBEKaml snarkyboojum++ ## that could turn into something of a tryrakudo.org ? :)
moritz_ we really need a specbot that answers FAQs with links to synopsis :-) 14:09
snarkyboojum isBEKaml: :)
masak snarkyboojum: you're about two years behind. :P november-wiki.org/
PerlJam moritz_: that would make a good not-necessarily-perl6 challenge :)
snarkyboojum masak: I have no clever comeback for that :)
moritz_ masak: doing a HTTP request != answering a HTTP request 14:10
14:10 BrowserUk joined 14:11 am0c joined
masak moritz_: oops. 14:12
actually, I don't think I've ever *done* an HTTP request from Rakudo. snarkyboojum++
snarkyboojum :)
jnthn wonders if IRC::Client works on current Rakudo 14:13
14:18 M_o_C joined
pmichaud jnthn: ping 14:19
14:19 viklund left
moritz_ received a submission for this week's challenge o/ 14:19
and from a first look it's pretty good 14:20
isBEKaml moritz_: so soon? \o/
14:20 ruoso_ joined
jnthn pmichaud: pong 14:20
pmichaud: Gimme some relief from Java! :-)
pmichaud I need some guidance with the YOU_ARE_HERE stuff that's been added to Rakudo.
jnthn Out of the fat and into the fryer... :-) :-) 14:21
pmichaud iiuc, YOU_ARE_HERE is intended to make it appear that code is executed lexically inside of the setting
jnthn pmichaud: It mostly works just by twiddling the outer of the outermost lexical scope of the program to actually be the setting. 14:22
pmichaud I don't like the $*IN_EVAL contextual -- it's fragile.
jnthn As in, user-space twiddlable? 14:23
In that sense yes.
pmichaud no, in the sense that there may be code that gets evaled (e.g. interactive mode) but doesn't set $*IN_EVAL
jnthn Ah, and then we don't want the outer scope of the thing to be the setting...
Yes.
Maybe it's the wrong flag. 14:24
And we need to look at it the other way
$*IN_MAINLINE # this is the mainline code
pmichaud dynamic vars as flags tend to be code smells to me
jnthn Many things relating to the current bit of parsing/compilation we're doing are done with dynamic vars. 14:25
Guess this could be an over-use of them.
What would you suggest instead?
pmichaud yes.... it feels like the use of $past<something> as flags as well -- they should be used minimally
jnthn Aye, I agree on those. 14:26
pmichaud I don't know -- I just needed to make sure I understood what YOU_ARE_HERE is doing
jnthn The action for YOU_ARE_HERE just shoves a tiny bit of code in the setting that twiddles the outer lexical scope of the bit of the Parrot sub it is passed.
Since that code is in the lexical scope of the setting, the chain ends up having the setting in it. 14:27
pmichaud looks to me like it actually generates a sub in the setting
jnthn Right.
pmichaud any particular reason it needs to be !YOU_ARE_HERE?
i.e., couldn't it also be &YOU_ARE_HERE just as easily?
jnthn No, I just needed something to call it. :-)
Well
I didn't really think it should be user-space exposed.
Folks can't call a thing from Perl 6 land with a ! at the start. :-) 14:28
moritz_ S06 (iirc) mentions YOU_ARE_HERE, and doesn't say it's hidden
jnthn That was pretty much the entire reason though.
pmichaud YOU_ARE_HERE shows up in the spec, so I'm guessing it's userspace exposed
jnthn As something you can pass a chunk of code to and have it run in the current setting?
If so, then yes, by all means let's rename it.
pmichaud Haven't read yet -- I just know that it's in the spec
okay, I'll work on any refactors there. Currently that $*IN_EVAL appears to be destroying my outer context in interactive REPL 14:29
jnthn tbh, the majority of the effort in making this work was fixing other areas we got lexical-ish bits not quite right.
14:29 uniejo left
jnthn Not the actual YOU_ARE_HERE thing that did set_outer 14:29
14:29 isBEKaml left
jnthn Yes, I can imagine it would cause problems there. 14:30
Feel free to change it as you see fit.
pmichaud will do
In some sense I'm thinking of changing YOU_ARE_HERE.pm to have a full sub declaration
our sub YOU_ARE_HERE(...) {...} 14:31
jnthn That would also be workable
pmichaud okay, I'll read up on the spec and see what comes out
jnthn I was more attempting to match STD and svn.pugscode.org/pugs/src/perl6/CORE.setting
pmichaud yes, I noticed that also :-)
jnthn (see YOU_ARE_HERE token at end)
pmichaud I can see that it's following STD. I'm wondering if this is a place wehre Rakudo should simply so that STD.pm6 can follow what it does instead :) 14:32
*simplify
jnthn :-)
mathw jnthn: I'm right with you on Java generics 14:34
jnthn: the only thing that's worse is using the collection classes without them
pmichaud yeah, term:YOU_ARE_HERE appears to be more of a "invoke the mainline code" term rather than a "define the YOU_ARE_HERE sub at this point"
(according to S02)
14:35 XaeroOne left
TimToady for 14:36
for VMs that can't do 'set outer' it's more of a freeze/thaw point
pmichaud TimToady: makes sense to me
jnthn mathw: I know just fine what type erasure is, but I'm quite sure Joe Programmer is in for a shock when the compiler starts whining that he can't define these two methods with different signatures because the type-erase to the same thing. :-/ 14:37
*they
oh hey, I think I fixed $nasty-work-bug
pmichaud jnthn: I wonder if we can dynamically detect "mainline" code as being code that doesn't have an outer context set. 14:38
(at which point we force its outer to be the setting)
eval'd code would already have an outer context set
jnthn pmichaud: Maybe. 14:39
TimToady in a sense, so does the mainline code
jnthn pmichaud: May get an off-by-one if there is some outer scope which is the thingy that actually gives the mainline to YOU_ARE_HERE (or whatever we call the code that does the twiddling) 14:40
TimToady YOU_ARE_HERE should be more or less indistinguishable from a 'my package UNIT { ... }' at that point
pmichaud TimToady: yes, I was thinking of something like that too :-) 14:41
jnthn pmichaud: Then, I guess that's solvable by .lexical(0) on the mainline PAST::Block.
TimToady but anything you can do to fake that is fine
pmichaud jnthn: right.
jnthn So yeah, sounds feasible.
TimToady as long as it happens soon enough that the compilation of UNIT knows it
jnthn "if get_outer returns NULL..."
pmichaud or the outer scope is the thing that does the checking for "do I have an outer scope?" before calling UNIT_START
as opposed to checking outer of the "inner mainline" 14:42
14:42 M_o_C left
pmichaud (sorry, that was totally confusing) 14:42
(rephrasing)
the off-by-one wrapper scope can check itself for "do I have an outer?" to decide if it's really the mainline execution or no 14:43
it can then decide what to do with the "mainline" that it's wrapping.
jnthn pmichaud: Ah, get it. 14:44
pmichaud: That also works.
pmichaud if the wrapper already has an outer scope set, then it can just invoke the mainline and be done.
jnthn *nod*
pmichaud and we then reserve UNIT_START for actual mainline execution. 14:45
jnthn That feels righter than making the code that does the set_outer have to do that check.
snarkyboojum masak, moritz_: ok, I just posted a dent from rakudo :P
masak cool!
jnthn oh noes...you dented it?!
.oO( It'll buff out )
pmichaud and in fact!
snarkyboojum command-line identica'ing from rakudo :) 14:46
pmichaud we can make sure our mainline block is always .lexical(0)
if the off-by-one wrapper has an outer scope set, it passes that scope to mainline before executing it
otherwise it calls UNIT_START or YOU_ARE_HERE or whatever, which sets the mainline's outer to the setting before executing it 14:47
14:47 am0c left
masak found it! identi.ca/notice/33651989 14:47
pmichaud either way, the off-by-one wrapper is out of the outer context chain by the time it passes control to mainline
jnthn *nod*
pmichaud then mainline's OUTER would indeed be the setting directly
jnthn Even better. 14:48
snarkyboojum hehe that's it ;) now I just have to make it say "from rakudo" instead of "from api"
pmichaud okay, I'll work on those refactors.
jnthn pmichaud++
pmichaud I'm much more comfortable with this now.
jnthn Me too. ;-)
pmichaud it feels much less workaround-y and more "do exactly what is supposed to happen here" 14:49
it also stays consistent with other aspects of PCT/NQP compilation
(i.e., the evalpmc[0] is the one that determines outer context)
pugssvn r30806 | moritz++ | [t/spec] tests .subst() with :samecase
r30806 |
r30806 | Patch courtesy of Hongwen Qiu
masak snarkyboojum: make it say 'from perl6awesomedent' :) 14:50
snarkyboojum masak: hehe
pmichaud okay, need a short break here -- will be back later
masak on the list of Things We Have Plan 9 to Thank For: UTF-8. 14:51
moritz_ I'm curious: what else is on that list? 14:54
[particle] the plan 9 mascot.
masak I bet one could put something having to do with C compilers on that list.
the Plan 9 people actually share much of the hubris of the Perl 6 crowd. they know they got a lot of things right, and deplore that it's not more widely known. 14:55
pmichaud s/Thank/Blame/ # perhaps? 14:56
moritz_ kinda likes UTF-8
14:56 finanalyst left
moritz_ it's not ideal for implementing regex engines, but I don't do that very often :-) 14:56
TimToady UTF-8 is pretty okay as long as you don't have to random access it 14:57
arnsholt I'd argue that it's also a matter of foisting complexity off to the implementors
PerlJam moritz_: Plan 9 is also relatively unknown as far as OSes go. Hopefully Perl 6 doesn't share the analogous trait wrt languages. 14:58
masak this presentation especially kinda reeks of "here's what we know we did wrong and want to do better": www.cs.unm.edu/~fastos/05meeting/PL...EADYET.pdf
PerlJam: Perl 6 is already famous. unfortunately not yet for all the right reasons :/
& 14:59
TimToady if you squint, Perl 6 looks like a dyslexic Plan 9
moritz_ that's a line for my fortune file :-)
arnsholt masak: Perhaps infamous is even better than famous. Especially when Perl 6 suddenly turns out to be awesome =) 15:00
jnthn -> beer shopping, bbiab
15:02 am0c joined 15:03 justatheory joined
moritz_ masak: I like it. "Don't believe me? Type 'stty' at an xterm 15:03
and tell me why a window has a baud rate"
PerlJam heh, I just read that slide too :)
moritz_ they do have a point 15:08
15:09 justatheory left
snarkyboojum masak: there ya go - identi.ca/notice/33654228 :P 15:09
moritz_ :-)
snarkyboojum 'from perl6awesomedent' 15:10
ok.. enough spam from me :)
15:13 orafu joined
TimToady
.oO(Perl 5 : Perl 6 :: Ma Bell : Bell Labs)
 15:13
15:19 isBEKaml joined
pugssvn r30807 | pmurias++ | [mildew] renamed Mildew::Parser::* to Mildew::Frontend::* 15:20
ajs rakudo: use Test; "abc" ~~ /a[ $<x> = (b) ]?c/; ok($/<x> eq "b", "aliased match within optional non-capturing group") 15:22
p6eval rakudo 456986: OUTPUT«No applicable candidates found to dispatch to for 'infix:<eq>'. Available candidates are:␤:()␤:(Any $a, Any $b)␤␤current instr.: '_block14' pc 29 (EVAL_1:0)␤» 15:23
pmichaud jnthn: ping
ajs rakudo: use Test; "abc" ~~ /a[ $<x> = (b) ]c/; ok($/<x> eq "b", "aliased match within non-capturing group")
p6eval rakudo 456986: OUTPUT«ok 1 - aliased match within non-capturing group␤»
ajs One of these things is not like the other....
moritz_ ajs: the problem with the first test is that $/ isn't a a Perl 6 object at the moment
so it's not compatiable with Any :( 15:24
ajs So how do you do aliasing in that case?
moritz_ rakudo: use Test; "abc" ~~ /a[ $<x> = (b) ]?c/; is("$/<x>", "b""aliased match within optional non-capturing group") 15:25
p6eval rakudo 456986: OUTPUT«Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤current instr.: 'perl6;Regex;Cursor;FAILGOAL' pc 1971 (ext/nqp-rx/src/stage0/Regex-s0.pir:1050)␤»
moritz_ rakudo: use Test; "abc" ~~ /a[ $<x> = (b) ]?c/; is("$/<x>", "b", "aliased match within optional non-capturing group")
p6eval rakudo 456986: OUTPUT«not ok 1 - aliased match within optional non-capturing group␤# got: "1"␤# expected: "b"␤»
pmichaud stringify it first
try ~$<x> eq 'b'
moritz_ rakudo: use Test; "abc" ~~ /a[ $<x> = (b) ]?c/; is("$/<x>[0]", "b", "aliased match within optional non-capturing group")
p6eval rakudo 456986: OUTPUT«ok 1 - aliased match within optional non-capturing group␤»
moritz_ ah, that's the problem 15:26
the array in $<x> stringifies incorrectly
(again it's a parrot thing, not a proper Perl 6 array)
pmichaud ....because it's a parrot array. Good catch.
I should be able to look at those today.
moritz_ we have an old ticket for that in RT
ajs freaky.... so why is it an array when I say [$<x> = (b)]? and a string when I say [$<x> = (b)] 15:27
pmichaud ajs: because of the quantifier
moritz_ [...]?
pmichaud consier: [$<x> = (b)]*
*consider
ajs Ah, I see
OK, thanks
pmichaud $<x> would be an array of captures
s/captures/matches/
ajs That's going to really trip some people up, becuase they have to look outside of the alias to determine what kind of match they're going to get. 15:28
But I can work with that now
pmichaud I've been wondering if quantified captures should act the same as the :global flag, though
i.e., they return a match object with the individual captures in .[] 15:29
instead of returning an array of match objects
moritz_ that could also be pretty confusing
because match objects usually represent continuous matches
and those wouldn't
pmichaud ...except in the case of :global :-)
it just seems like a potential semantic mismatch there 15:30
ajs Unrelated: in perl 5, I would return undef to indicate "no result". In P6, is that Mu or should I be returning a more specific type or fail()? 15:31
pmichaud ajs: fail(), usually.
ajs ok,thanks
moritz_ or maybe Nil
pmichaud yes, Nil if you really mean "no result"
moritz_ that' like the perl 5 return(); # empty list in list context 15:32
pmichaud moritz_: yes, I agree that using match instead of array could end up being confusing. someone could then do $<x>.to and be totally surprised at the result
whereas if it's an array, $<x>.to at least gives an error about no method 'to' 15:33
so, I guess we leave it as an array
still feels a little weird that :global is different :-)
moritz_ +1 to leaving as-is
or even better... 15:34
pmichaud okay, I'm grabbing lunch, then will come back to work on YOU_ARE_HERE refactors
moritz_ don't we have some listy objects that behave just like the stored object if there's only one inside?
pmichaud sure -- they're called Parcels :-)
moritz_ then why don't we use a Parcel instead of an Array?
that way matches quantified with ? will DWIM as much as possible 15:35
pmichaud well, at the parrot level that would be problematic.
moritz_ Perl 6 will construct its own Match objects
pmichaud i.e., nqp-rx would never be able to match that.
(not that nqp-rx needs to be driving perl 6 design)
isBEKaml rakudo: my %hash =<a 1 b 2>; say %hash.keys; say %hash{$_} for %hash.keys; 15:36
p6eval rakudo 456986: OUTPUT«ba␤2␤1␤»
15:36 pmurias joined
pmichaud there might also be the problem that single-element captures might cease to act like an array when we need them to 15:38
moritz_ that might indeed be a problem
pmichaud (i.e., the array-ness would get lost somewhere)
15:38 Ross joined
moritz_ anyway, I don't want to stop you from grabbing lunch :-) 15:38
pmichaud I'm more comfortable with things as they re now.
*are
yes, lunch
pmichaud _ISNT_HERE
pugssvn r30808 | pmurias++ | [mildew] add missing file 15:39
r30809 | pmurias++ | [mildew] add -Fm0ld frontend
pmurias moritz_: will there be a gsoc irc meeting?
moritz_ pmurias: we've had masak report as part of the #rakudosketch meeting yesterday 15:40
pmurias: you'd certainly be welcome there next week (Tue 1900 UTC), or you could arrange something yourself
masak snarkyboojum++ # perl6awesomedent! \o/ 15:41
snarkyboojum masak o/ :)
jnthn lol...systemet unicode FAIL.
masak pmurias: but be aware that 'there' is in another place from next week on. #phasers
jnthn: what now? 15:42
pmurias #phasers is the new #rakudosketch?
moritz_ yes :-)
for Perl 6 and Rakudo phase transitions, or so
jnthn masak: Was just glancing my receipt...and turns out whatever font they print in it doesn't have ý in it. :-) 15:43
Well
masak jnthn: oh. receipts.
ajs Gotta go, but I leave you with this one:
jnthn I'm not sure what the glyph that appears there is. :-)
ajs rakudo: "abc" ~~ /[$<x> = (b)]?/; say @($/<x>)
p6eval rakudo 456986: OUTPUT«Method 'list' not found for invocant of class 'ResizablePMCArray'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
masak jnthn: I don't know anyone who gets receipts right.
jnthn :-)
masak: As long as the beer's right, it's fine. :-)
pmurias moritz_: i think i'll just write a blog post when my milestone is finished and show it to ruoso
15:44 am0c left
masak the ESR quote at en.wikipedia.org/wiki/Plan_9_from_B...abs#Impact is interesting from a Perl 6 perspective as well. Perl 5 may ultimately be a 'dead end', but it's a quite successful one. :) 15:46
pmurias if i convert m0ld -> AST -> m0ld it's called round-tripping?
masak pmurias: aye. 15:48
pugssvn r30810 | pmurias++ | [mildew] added a -Cm0ld backend to test roundtriping 15:51
diakopter pmurias: ih
15:54 constant joined, constant left, constant joined
diakopter lol at "ydlsxeci Plan 9" 15:54
pmurias diakopter: hi 15:56
15:57 masak left
diakopter pmurias: wuzzup 15:57
pmurias hacking on rewriting m0ld from haskell to perl5 (part of my GSoC project) 15:58
15:59 synth left
diakopter pmurias: cool 15:59
16:00 IllvilJa joined, synth joined
pmurias it's more a thing that needs to be done, rather then a particularly cool/fun piece of hacking 16:00
16:01 jaldhar left
diakopter oh 16:01
what, you're not creating a haskell-esque runtime in Perl? :) 16:02
that was a joke 16:05
pmurias diakopter: see failing test i committed for perlesque 16:07
dalek meta: r297 | pawelmurias++ | trunk/Sprixel/t/parse_bug3.t:
failing test - dunno why
 16:08
pmurias please rename it when you find out the true orgin of the bug 16:09
when are classes better then roles? 16:11
TimToady when you might want to do monkey patching, since roles are immutable 16:12
when you want your class to have state 16:13
16:13 ejs left
TimToady when you want uneven mixin semantics rather than flat composition 16:14
mberends jnthn: I probably haz $work in Helsingborg June 20-23 (a M$ bootcamp). Will you be in Lund before/after those days?
pmurias TimToady: mixin semantics? 16:15
diakopter: the other thing i'm hacking on is a game in erlang
jnthn mberends: ooh, maybe, let me check travel/conf schedule.
mberends: Helsingborg is extremely near here.
mberends: I'm away from the 23rd. 16:16
mberends: I'm around the couple of days before.
mberends ok, maybe a visit from Friday 18th evening or Saturday morning is possible 16:17
isBEKaml moritz_: sorry I was away for some time, "power" problems. :(
moritz_: that challenge was unbelievably simple. :O
jnthn mberends: wfm.
You're welcome. :-)
mberends :-) self.invite(); 16:18
jnthn ;-)
mberends: Did you get any further on the void thingy?
mberends: If not, have you got a small test case I can try?
mberends nope. the examples file runs, but not the DBI stuff under construction. It's much bigger :-( 16:19
mberends gets back into FakeDBI mode 16:20
jnthn Ah
mberends: I wonder if the null return value is somehow causing an explosion.
mberends null != void of course
jnthn That is, I'd expect that doing $P0 = foo() where foo is an NCI with a signature starting "v" would just shove a null PMC in the register (or complain) 16:21
Yeah
16:21 pnate joined
jnthn wonders what a good fix may be 16:21
mberends: There isn't any chance you can get a backtrace?
16:22 colomon left
jnthn e.g. run it under gdb? 16:22
jnthn is really curious where it segfaults
mberends ok, will try gdb
PerlJam from #p5p ... 16:24
<@xdg> anyone want to opine on how hard it would be to let "when" work inside a map block?
slowly morphing perl 5 into perl 6 :)
mberends heehee
jnthn rakudo: (1..10).map({ when 3 { say "tre" }; when 5 { say "fem" }; }).eager 16:26
p6eval rakudo 456986: OUTPUT«tre␤fem␤»
jnthn Easy. ;-)
jnthn mostly wanted to make sure Rakudo handled it right :-)
16:27 justatheory joined, hercynium left
snarkyboojum isBEKaml: my "I did an HTTP request!" *did* turn into something useful - a little cli identi.ca client :) 16:27
moritz_ isBEKaml: never said it was hard
isBEKaml snarkyboojum: nice. link? 16:28
moritz_: :)
snarkyboojum isBEKaml: well, it's "as rough as guts" but this is the gist gist.github.com/414632 16:29
isBEKaml snarkyboojum++ ## Dented IO::Socket::INET!! :) 16:30
16:31 M_o_C joined
snarkyboojum isBEKaml: :P 16:31
pmichaud back again 16:33
jnthn: ping 16:34
jnthn pmichaud: pong
pmichaud jnthn: want to see a Truly Evil Cheat? ;-)
(found this last night while working on repl)
jnthn Uh-oh...
:-)
lisppaste3 pmichaud pasted "lexpad evil cheat (for jnthn++)" at paste.lisp.org/display/100517 16:35
16:35 synth left
jnthn copy pad, $P1 16:35
?!
16:35 XaeroOne joined
pmichaud I CAN HAZ DYNAMIC LEXPADZ 16:35
moritz_ epic
jnthn omg.
There is a proper way to do that. :P 16:36
pmichaud++
pmichaud there is?
jnthn pmichaud: LexPad and LexInfo are, iirc, HLLMappable.
pmichaud: But that's effort. ;-)
pmichaud I only need to do it in one instance, though.
jnthn Ah
Then hack away...and hope. :-)
pmichaud I don't want all of them to be dynamic... just the one that holds the outer context information for each interactive repl line
jnthn Where are you doing it, ooi?
ah, ok 16:37
Makes sense.
16:37 meppl joined
pmichaud essentially, invoke a line of code, capture its context while executing, when it's done executing collect all of the lexicals into the dynamic lexpad 16:38
jnthn Ah
pmichaud on the next line of code, use the context containing the dynamic lexpad as the outer context for the new line of code
jnthn Saves building up a long long chain of 'em. 16:39
pmichaud rinse, repeat.
actually, the long-long chains weren't working out.
Parrot doesn't provide sufficient introspective context getting/setting ability to really get the chains to work out
jnthn Oh.
:-|
pmichaud also, if a sub doesn't believe it's an :outer, it tends to want to throw away its context when it's done
16:39 synth joined
pmichaud i.e., the return continuation says "okay, we're done, we're not an outer, so we can just throw everything away now!" 16:39
and it clears out the registers 16:40
jnthn :|
Happy there's a solution.
pmichaud me too.
jnthn Even if it does win you disgusting hack on the month award. :-)
*of
pmichaud I thought it was pretty clever
nicer would've been if I could simply setattribute ctx, 'lexpad', pad instead of using the copy 16:41
but allas, that's not allowed either
jnthn Aye. :-)
pmichaud *alas
16:42 cdarroch joined, cdarroch left, cdarroch joined
jnthn wonders...what else we can use this cheat for. :-) :-) 16:42
pmichaud I knew you'd start thinking of that... :-)
I also added a method that allows us to get an entire lexical environment of a context into a single hash 16:43
jnthn Ah, flattened?
pmichaud where "entire lexical environment" means context+all outer contexts
yes, flattened
jnthn Handy.
[particle] easy to look up OUTER then, i suppose 16:44
OUTER::$* anyway
pmichaud somewhat... I think OUTER may be just one level
jnthn Yeah, same
[particle] oh, right. OUTER::OUTER::$*. forgot about that 16:45
jnthn CALLER:: OTOH may want a to-the-sub-boundary flat view.
pmichaud anyway, it's in PIR, so it's a good template for other stuff we might want/need.
jnthn would have to check the spec
That is
16:45 __2synth joined 16:46 __2synth left
jnthn sub foo() { my $a = 42; if 1 { bar() } }; sub bar() { say CALLER::<$a> } # does this work? 16:46
sub foo() { my $a = 42; if 1 { bar() } }; sub bar() { say CALLER::OUTER::<$a> } # or do you have to write this?
diakopter my class LexicalScope : PrototypeChain<string, LexicalScope> { } has a similar (well, PrototypeChain<TKey,TValue> does) all-keys-flattened method
moritz_ jnthn: nope, CALLER only works if the variable in the caller is declared as 'is context' 16:47
pmichaud jnthn: I think CALLER variables are hidden unless 'is dynamic'
16:47 __2synth joined
moritz_ wasn't CONTEXT:: used for caller + caller's lexical environment? 16:47
pmichaud it's now DYNAMIC:: 16:48
moritz_ ok
pmichaud and it only looks in callers, not their surrounding outers
16:48 synth left
jnthn Ah, OK. 16:48
16:50 dakkar left 16:52 __2synth left
pmurias diakopter: any idea what causes that bug? 16:54
that = one in the failing test 16:55
sorear good morning pmurias 16:56
pmichaud
pugssvn r30811 | lwall++ | [S05] attempt to clarify :nth and :x
pmurias sorear: hi 16:57
16:59 patrickas joined
patrickas ola! 16:59
isBEKaml hello, patrickas! :)
mberends perusing docs/test_summary.times, somewhere between May 16th and 20th, the execution time of the first test in every script increased dramatically (factor 2x to 10x). Subsequent tests unchanged. Any idea what caused that (not suggesting a bisect witchhunt)? 17:00
patrickas hey isBEKaml! what's up ? what have you been working on lately ?
moritz_ mberends: so.. startup cost? 17:01
mberends no
cost of calling the first Test.pm function
jnthn mberends: Doesn't include calling plan?
(that is, could plan have got more costly?) 17:02
isBEKaml patrickas: nothing much. Just hanging around, mostly. :)
mberends it might be plan(), yes. looking...
17:03 pmurias left
mberends the increase is between the end of plan() and the first execution of proclaim(), usually via ok(). 17:05
jnthn Curious.
diakopter phenny: tell pmurias no I don't know what's wrong with that bug; I recently slowed down my rate of deluding myself into thinking I have more round tuits than I actually have... 17:06
phenny diakopter: I'll pass that on when pmurias is around.
17:13 alester left
pugssvn r30812 | lwall++ | [S03] explain how not-raising works on != and ne 17:21
TimToady pmichaud, mathw: S03 no longer implies that != and ne are distinct operators 17:23
jnthn \o/
diakopter !!
pmichaud TimToady: excellent, thanks.
TimToady so the semantics of != should match !== 17:24
jnthn TimToady: Is negate the right thing to call the HOF that the ! meta-ooperator desugars to?
TimToady that's why I double-hedged it :)
jnthn TimToady: If so, will update Rakudo there too (or somebody can beat me to it; currently we call it something else).
TimToady what do you call it?
jnthn notresults iirc
I prefer negate.
rakudo: notresult 17:25
p6eval rakudo 456986: OUTPUT«Could not find sub &notresult␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
jnthn rakudo: notresults
p6eval rakudo 456986: OUTPUT«No applicable candidates found to dispatch to for 'notresults'. Available candidates are:␤:(&op, Mu $a, Mu $b)␤:(&op)␤␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
TimToady I like verbs like cross, reduce, etc
jnthn That
Same
'cus others are like
rakudo: crosswith
p6eval rakudo 456986: OUTPUT«No applicable candidates found to dispatch to for 'crosswith'. Available candidates are:␤:(&op, Iterable $a-iterable, Iterable $b-iterable)␤:(&op, Any $a, Any $b)␤␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
jnthn etc
snarkyboojum just realised you can exit the REPL with 'last' :) 17:33
17:33 colomon joined
moritz_ now that's the last command... :-) 17:33
sorear this is somewhat a bug and may be fixed, depending on how much of my code pmichaud used
snarkyboojum good that the REPL doesn't segfault any longer 17:34
moritz_ std: 1/0e0
p6eval std 30812: OUTPUT«ok 00:01 114m␤»
moritz_ rakudo: 1/0e0
p6eval rakudo 456986: OUTPUT«Confused at line 11, near "1/0e0"␤current instr.: 'perl6;HLL;Grammar;panic' pc 528 (ext/nqp-rx/src/stage0/HLL-s0.pir:348)␤» 17:35
snarkyboojum sorear: cool - I certainly didn't expect it to exit when I typed 'last'
moritz_ what's confusing about 1/0e0?
colomon maybe leftovers from the days when 1/0 was a single number instead of 1 / 0? 17:36
moritz_ maybe
rakudo: say 2/2**3
p6eval rakudo 456986: OUTPUT«1␤»
moritz_ yes
pugs: say 2/2**3
p6eval pugs: OUTPUT«0.25␤»
colomon I thought we'd gotten rid of that complete.... :\
*completely
moritz_ token number:sym<rational> { <nu=.integer>'/'<de=.integer> } 17:37
moritz_ tries to remove
colomon any idea why TimToady just nixed smartmatching in :nth() ?
moritz_ ETOOMUCHMAGIC? 17:38
colomon it seems the natural way to implement the rest of the specified features...
17:38 perlygatekeeper left
patrickas how can I add a multi method to Hashes ? 17:38
naively I would have expected "augment class Hash" or "%h.^add_method" to work but they didn't :-(
TimToady because smartmatching does the *wrong* thing with :nth(1,3,5)
moritz_ patrickas: currently rakudo implements Hash as a role... so not at all 17:39
sorear TimToady: Is there any difference between COMPILING::<$?FOO> and $*FOO ?
moritz_ rakudo: Hash.^add_method('foo', sub ($a, $b) { say $b } }); say {a => 2}.foo(3)
p6eval rakudo 456986: OUTPUT«Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤current instr.: 'perl6;Regex;Cursor;FAILGOAL' pc 1971 (ext/nqp-rx/src/stage0/Regex-s0.pir:1050)␤»
colomon TimToady: you mean if you do it $i ~~ 1, 3, 5 rather than $i ~~ any 1, 3, 5 ?
moritz_ rakudo: Hash.^add_method('foo', sub ($a, $b) { say $b }); say {a => 2}.foo(3) 17:40
p6eval rakudo 456986: OUTPUT«Method 'foo' not found for invocant of class ''␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
TimToady colomon: yes
colomon ah, okay.
are closures still allowed?
moritz_ nope
TimToady and you can still get the smartmatching via grep
patrickas moritz_ I can add a method foo, but not a multi method
moritz_ patrickas: no idea then :/ 17:41
TimToady sorear: yes, COMPILING could conceivably be many dynamic levels outward, and any number of dynamic scopes could have overridden $*FOO inside that
so they are not the same
and in any case, $?FOO only defaults to $*FOO if there isn't a predefined meaning 17:42
patrickas rakudo: my %h;%h.^add_method('foo', sub ($a, $b) { say $b }); say {a => 2}.foo(3)
p6eval rakudo 456986: OUTPUT«3␤1␤»
TimToady also, it's pretty easy to tell when a list is exhausted, and the matching can quit early 17:43
colomon TimToady: If I'm understanding correctly, I'm not sure I like that at all. Seems like we just went from two different easy ways to specify "match every other time" (either :nth(2) :x(*) or :nth({ %_ !% 2 })) to no easy way.
moritz_ :nth(1, *+2 ... *)
TimToady :nth(2,4,6...*)
:nth(1,3,5...*) 17:44
moritz_ colomon: :nth(2) :x(*) never did that
that would always only match the 2nd item, never the 4th
TimToady well, it did in the original formulation lo these many years ago
17:44 ajd joined
TimToady but a series is much more powerful and expressive 17:44
ajd is there anything out on the state of perl 6? 17:45
TimToady yes
moritz_ ajd: planetsix.perl.org has blogs about many Perl 6 things 17:46
ajd like a paper I could read talking about where is is now, what needs to be done, what has been done, etc
TimToady however, it's a rather large state vector, so it would help to ask more specific questions
moritz_ there's also rakudo.org/status 17:47
TimToady the state of the language as opposed to the various implementations is defined in the synopses
(and in the standard parser, STD) 17:48
17:48 patrickas left, perlygatekeeper joined 17:49 envi^home left
colomon TimToady++ # series 17:51
though that completely blows our current code. 17:52
17:53 sundar joined
colomon does that imply that if you have :nth and :g, hitting the end of the :nth list resets the counter to 1? 17:53
17:54 snarkyboojum left
moritz_ doesn't think so 17:55
sorear TimToady: How does using a series as a smartmatch work?
rakudo: say 2 ~~ 1, 3 ... * 17:56
p6eval rakudo 456986: ( no output )
moritz_ it's not a smartmatch
read the diff again :-)
pugssvn r30813 | moritz++ | [S03] typo 17:57
lisppaste3 sundar pasted "Why isn't a non-increasing list an error in :nth? (S05-regex.pod)" at paste.lisp.org/display/100524 17:58
sundar To clarify, such a non-increasing list will most probably be a mistake, so isn't it better to flag it as such and help the programmer? 17:59
colomon sundar: are you arguing it should be "such a list must be monotonically increasing" rather than "such a list should be monotically increasing"? 18:01
18:04 XaeroOne left
sundar colomon: I had to refer to the difference, but yes. :) 18:06
colomon sundar: if so, probably because detecting it is very unreasonable (maybe impossible) if the list extends beyond the number of matches.
TimToady sorear: you can't smartmatch against an infinite series, except maybe with some kind of prover that shows two expressions will produce the same series 18:08
list smartmatches do not imply any()
unlike in p5 and ancient p6 18:09
we might possibly define any() on monotonic infinite serieses; there may already be a conjecture in the spec about that 18:10
moritz_ a conjectural conjecture :-) 18:11
TimToady but there has to be some way to know when to stop looking
in the case of :nth, we have the benefit of knowing we're checking against 1...$n for finite $n 18:12
well, until we get infinite strings :)
colomon hmm? 18:14
moritz_ like .cat?
colomon I just figured every time we got a match, we looked at the next thing in the :nth list, and walked up the list as needed, which is why it has to be in monotonically increasing. 18:16
(which is an approach well-suited to being lazy in the case of infinite strings (ie Cat)) 18:17
moritz_ now we need to JFDI
colomon is still confused by :x 18:18
18:18 ajd left
diakopter Return of the Jfdi 18:19
colomon moritz_: you want to work on .match or on the tests? ;)
moritz_ colomon: will look at the tests first 18:21
18:21 iblechbot joined
colomon moritz_: okay, I'll see if I can quickly puzzle out the new algorithm 18:21
moritz_ so I guess ranges are still allowed? 18:22
(assuming they somhow interpolate into lists 18:23
rakudo: say (1, 2..3).perl
p6eval rakudo 456986: OUTPUT«(1, 2..3)␤»
colomon I'm assuming it's actually Iterable which is allowed, which includes Ranges
dalek kudo: 4e3d29d | (Hongwen Qiu)++ | src/core/Cool-str.pm:
add :samecase to Cool.subst
 18:25
kudo: bafb6d9 | moritz++ | src/Perl6/ (2 files):
remove rational number literal fossile
18:25 bluescreen joined 18:26 bluescreen is now known as Guest72640 18:27 charsbar left, charsbar joined
colomon say 4.Iterator 18:28
rakudo: say 4.Iterator
p6eval rakudo 456986: OUTPUT«Method 'Iterator' not found for invocant of class 'Integer'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
colomon rakudo: say 4.list
p6eval rakudo 456986: OUTPUT«4␤»
colomon rakudo: say 4.list.iterator
p6eval rakudo 456986: OUTPUT«4␤»
pugssvn r30814 | moritz++ | [t/spec] bring :nth tests up to current spec
colomon rakudo: my $a = 4.list.iterator; say $a.get; say $a.get
p6eval rakudo 456986: OUTPUT«4␤␤»
18:29 BrowserUk left
colomon rakudo: my $a = 4.list.iterator; say $a.get; say $a.get ~~ EMPTY 18:29
p6eval rakudo 456986: OUTPUT«4␤1␤»
18:30 TiMBuS left
pugssvn r30815 | moritz++ | [t/spec] remove more outdated usages of :nth 18:30
18:30 iblechbot left 18:31 uvtc joined
sorear interesting 18:32
TimToady: I think I broke something
TimToady so did I :)
sorear ./viv STD.pm6 --> Check failed; undeclared routines: invert, note
uvtc Hi p6. Does code written in Perl 6 use 2 different filename extensions for scripts and modules (like Perl 5: .pl and .pm)?
moritz_ uvtc: yes (though .pm6 is also OK) 18:33
TimToady sorear: maybe you need a clean remake
those should come in from CORE.setting via syml/CORE.syml or some such
sorear unless viv is using the wrong setting 18:34
colomon moritz_: have a trial implementation. now compiling.
pugssvn r30816 | moritz++ | [t/spec] remove old :nth usages from .subst tests
colomon nice thing about this version is it radically more efficient when looking for a few early matches out of many. 18:35
if it sees :nth(2) it stops processing the string for matches after the 2nd is found.
moritz_ colomon: I tend to prototype the implemention of such methods in "normal" (non-setting) code, because it's faster to compile
colomon moritz_: that's probably sensible. but too late... :)
my brain is still set to the days when compiling and testing ng was really fast. 18:36
moritz_ for example for .subst I just monkey-typed class Str, that way it was always tighter than the built-in
which is in Cool
colomon moritz_++ # I had no idea you could do that. 18:37
and say probably works then too, eh?
moritz_ it does
decadent luxury, I know 18:38
sorear Just how old is STD and gimme5?
colomon say "hello world".match(/./, :nth(1,2,*+*...*)) 18:39
duh
> say "hello world".match(/./, :nth(1,2,*+*...*))
heloo
18:40 slavik left
colomon > say "hello world".match(/./, :nth(1,3...*)) 18:40
hlowrd
18:40 slavik joined
colomon moritz_: are the tests ready to go? 18:40
moritz_ colomon: yes
not terribly many use :nth
colomon @match = $data.match(/fo+/, :nth({ $_ !% 2 })); still in counted-match.t 18:41
moritz_ oh
colomon should be 2, 4 ... *
?
moritz_ yes
colomon I've got it;. 18:42
you added a non-monotonic test, moritz_++
pugssvn r30817 | moritz++ | [t/spec] remove another :nth fossile, colomon++ 18:43
uvtc When building rakudo, `make install` just installs things within my rakudo-2010.05 directory -- and not in my /usr/local -- is that correct?
sorear yes 18:44
uvtc sorear: thanks
sorear make install installs things within your Parrot installation
if you're using a system-wide Parrot install, you can install Rakudo system-wide 18:45
pugssvn r30818 | colomon++ | [t/spec] Simple :nth test with infinite series.
uvtc I'm doing a `perl Configure.pl --gen-parrot` build.
moritz_ that will install into rakudo-2010.05/parrot_install/ 18:46
so that after 'make install' in Rakudo you'll have a parrot_install/bin/perl6 binary
jnthn The fibonaci hello world! 18:48
jnthn likes :nth
colomon moritz_: had to patch my fix to make your non-monotonic test work. compiling version 2 now... 18:49
ooo, infinite loop
uvtc If I then want to make that perl6 available to all users on the system, can I just copy the whole rakudo-2010.05 directory to /opt, create an /opt/rakudo symlink there, and then tell them to use this shebang in their Perl 6 scripts: `#!/opt/rakudo/parrot_install/bin/perl6`?
sorear no 18:50
if you want to make perl6 available to all users on the system, you'll need to do a global install of Parrot
uvtc (Sorry, I meant mv, not cp.)
sorear parrot_install is not relocatable 18:51
1. get Parrot
2. perl Configure.pl --optimize && make && sudo make install # this will put Parrot in /usr/local
uvtc sorear: but i mean moving the whole rakudo-2010.05 directory to /opt, not just the parrot_install dir.
sorear uvtc: parrot bakes paths into its files
moritz_ uvtc: perl Configure --gen-parrot --gen-parrot-prefix=/usr/local/
sorear 3. switch back to Rakudo
oh, that works too. 18:52
moritz_ uvtc: then run 'make install' in rakudo => profit
colomon just realized he shouldn't have committed that fix, as it will cause failing tests for everyone else 18:53
moritz_: counted-match.t works completely here now.
uvtc Oh, interesting. Sounds like somewhere it's using absolute paths instead of relative ones.
pugssvn r30819 | sorear++ | [viv] Allow saving of parsed code 18:54
sorear uvtc: s/somewhere/everywhere/
moritz_ colomon: \o/
uvtc moritz: would that be `--gen-parrot-prefix=/opt` or `--gen-parrot-prefix=/opt/rakudo-2010.05`?
sorear uvtc: why not /usr/local?
colomon where were the other :nth tests? 18:55
moritz_ colomon: subst.t
sorear whatever you use, Rakudo will be installed as PREFIX/bin/perl6
colomon where there some in a match.t or no?
moritz_ there too
uvtc Either would be fine. I guess I just got into the habit of using /opt for "larger" self-contained software (like java).
moritz_ colomon: but those are not runnable by rakudo, iirc
no wait, in counted.t 18:56
$ ack -l :nth S05-*
S05-modifier/counted-match.t
S05-modifier/counted.t
S05-substitution/subst.t
colomon moritz_: apparently I took stupid pills this morning, latest version handles weird lists but breaks non-nth matches. :(
moritz_ colomon: nopaste patch? 18:57
colomon already fixed it, I think.
uvtc sorear: in that case, shouldn't PREFIX be `/usr/local/rakudo-2010.05/parrot_install` ?
colomon (compiling / testing now)
sorear uvtc: that's the default PREFIX, but you just said you wanted to move it to /opt 18:58
moritz_ uvtc: where do you want the binary to be?
uvtc sorear: Sorry, I don't think I understand ... do I need the whole rakudo-2010.05 directory and all its contents to be in /usr/local?
18:59 arthur-_ joined, arthur-_ left
sorear uvtc: NO 18:59
er
no.
colomon moritz_: that's better.
sorear do not move rakudo-2010.05
it will not work
clean everything, change the prefix, and rebuild
also, where do you want the binary to be?
uvtc sorear: Ok. Just finished a make that I'd started before we began chatting. 19:00
sorear whee. ./viv -5 --thaw STD.store in 4.5sec
19:00 M_o_C left
uvtc sorear: Where is the customary place for the binary? I suppose `/usr/local/parrot/bin`? 19:01
afk
sorear uvtc: just /usr/local/bin
colomon moritz_: prepping for spectest
TimToady sorear: I think I finally have a fix for your <foo()> dropping the longname 19:03
uvtc sorear: {back} thanks.
sorear: Sorry, but I'm still confused. Doesn't the perl6 binary need some accompanying parrot directory to run? 19:04
moritz_ uvtc: it needs an installed parrot + tools 19:05
uvtc sorear: if I download rakudo and build in my ~/temp directory, I'd like to be able to delete that temporary build stuff after installing what's necessary somewhere into /usr/local.
moritz_ so you configure it iwth --gen-parrot-prefix=/usr/local 19:06
but we had that already, no?
uvtc moritz: right. So, it seems easiest to have the perl6 binary and the parrot + tools all living together, in per-fect har-mon-ey. :)
moritz_ correct.
sorear uvtc: you should install the parrot stuff in /usr/local/bin/parrot, /usr/local/lib/parrot/2.4.0/runtime, etc
this can be accomplished by a default install of standalone parrot or a --gen-parrot --gen-parrot-prefix=/usr/local build of rakduo 19:07
TimToady: Excelllent.
uvtc sorear: I'd rather not let parrot install itself throughout my /usr/local. Seems tidier to have it in its own /usr/local/parrot directory. 19:08
19:08 awwaiid joined
moritz_ then use /usr/local/parrot as prefix 19:08
and rakudo will be in /usr/local/parrot/bin/perl6 19:09
uvtc moritz: beautiful. Will try that, but gotta run at the moment. Thanks very much!
19:12 dju_ joined, uvtc left 19:18 Guest72640 left, Guest72640 joined
pugssvn r30820 | sorear++ | [viv] Store generated class information in compiled files 19:21
19:21 ShaneC joined
TimToady sorear: did you fix the :sigspace problem in viv? if not, I think my <foo()> fix might've done so...at least it shows up now 19:21
19:21 synth joined
sorear yes, I fixed :sigspace 19:22
$*RX<s>, remember?
TimToady 'kay
right
19:22 ShaneC left
TimToady (just testing my patch) 19:22
19:23 rhr left
sorear is running viv on live STD.pm6 now :) 19:29
pugssvn r30821 | colomon++ | [spec] Add Real.Int, Real.Rat, Real.Num, and Real.Complex to the spec. 19:30
r30822 | sorear++ | [viv] Implement token foo:bar syntax 19:31
TimToady sorear++ 19:32
sorear now I've gotten far enough down STD.pm6 to hit what appears to be the <foo()> crash
dalek kudo: 475d1c7 | (Solomon Foster)++ | src/core/Cool-str.pm:
Change .match to support the new list approach for :nth.
 19:33
sorear hmm, I don't have anything for protoregex traits
TimToady about to check in fix, but it breaks $<foo> = <bar> for some reason 19:37
19:40 rhr joined 19:41 nadim joined
TimToady well $<foo> = (@x), more specifically 19:42
19:44 plainhao left
pugssvn r30823 | colomon++ | [t/spec] Unfudge tests that now work. 19:46
moritz_ colomon++ 19:48
19:49 Ross^ joined, Ross left
sjohnson rakudo: my $x = 19:53
p6eval rakudo bafb6d: OUTPUT«Confused at line 11, near "my $x ="␤current instr.: 'perl6;HLL;Grammar;panic' pc 528 (ext/nqp-rx/src/stage0/HLL-s0.pir:348)␤»
sjohnson oops... enter key fail
... anyone else ever hit the enter key often, when trying to hit the single quote?
moritz_ I hit enter when I want to hit backspace
jnthn My keyboard has a #~ key in the way. :-) 19:54
19:54 justatheory left
jnthn (between ' and enter) 19:54
moritz_ not a good habit when you're typing rm $somthing on the command line, and want to change the wrong last character
pugssvn r30824 | lwall++ | [STD] don't lose existing bindings when we go to a sublanguage 19:58
r30824 | [STD] refactor $*GOAL to only be informative, never used as <stopper>
r30824 | instead, we create a <stopper> rule for $*GOAL if necessary
r30824 | [gimme5,Cursor] various tweaks in support of the above
19:59 Ross^ left
sorear TimToady++ 19:59
TimToady phone
20:00 Ross^ joined
colomon rakudo: say 1 20:02
p6eval rakudo bafb6d: OUTPUT«1␤»
moritz_ recompilation in 13min
20:17 justatheory joined
colomon rakudo: say 1 20:17
p6eval rakudo bafb6d: OUTPUT«1␤»
colomon recompile starts in 13min, I'm guessing that should have said?
colomon has no idea why he wants to play with p6eval when he has an up-to-date REPL at home. 20:18
rakudo: say 1 20:26
p6eval rakudo bafb6d: OUTPUT«1␤»
pmichaud hugme: add chromatic to nqpbook 20:29
hugme hugs chromatic. Welcome to nqpbook!
20:31 TSa left 20:32 rhr left 20:39 clintongormley left 20:47 pmurias joined 20:52 kel__ left 21:01 patspam1 joined, patspam left 21:02 patspam1 is now known as patspam
cognominal searched a long time why parse actions where not triggered. I did parse(...., :actions<ActionClass>) instead of ... :actions(ActionClass) 21:05
21:06 Guest74734 left
cognominal I hate sllent fails... 21:06
m6locks fale 21:07
21:08 rhr joined 21:09 SmokeMachine left
pmichaud cognominal: I think we can remove the silent fail. 21:16
oh, actually, in this case it might be tricky. Hrm.
I'll think about it.
cognominal I don't whine much because I suppose someday, once bootstrapped, you (or someone) will rewrite that in Perl 6 and make that more solid. 21:17
You have so much on your plate...
jnthn The thing is that methods in general take a *%_ 21:20
Which swallows up the typo'd names.
21:20 whiteknight joined
cognominal my error was more a braino than a typo, passing a string instead as a class as a pair value 21:21
s/as/of/ 21:22
jnthn Ah 21:23
That's harder to catch, yes
Well, maybe a :U on actions so you gotta pass some type object.
Oh, no, 'cus you may pass an instance... 21:24
Hm
Yeah, hard to catch. :)
cognominal what a :U?
why not testing if a string and if so searching the corresponding class and passing that instead, Is that so difficult? 21:25
anyway I am happy I caught my mistake 21:26
21:26 ruoso_ left 21:27 Su-Shee left 21:28 Guest72640 left 21:31 pmurias left 21:43 tedv joined
jnthn pmichaud: ping 21:44
pmichaud jnthn: pong
pugssvn r30825 | tedv++ | Fixed incorrect test descriptions.
jnthn pmichaud: I'm working on auto-currying.
pmichaud: What you rather we construct the PAST for a closure, or instead just call some PIR thingy that manufatures them?
pmichaud you mean at runtime? 21:45
jnthn For the PIR case, yes.
Well
We could write it in Perl 6 actually :-)
pmichaud heh
jnthn I can go either wa
*way
colomon in p6 would be pretty straightforward, wouldn't it? 21:46
pmichaud you don't have a preference or bias?
jnthn It occurs to me that we can also for * op * just return the op itself, but maybe that'd be cheating.
pmichaud: Well, given it can be done in Perl 6, that eliminates my preference not to write more stuff in PIR. :-)
pmichaud do you need an answer soon, like, tonight?
jnthn pmichaud: I can go either way. The Perl 6 way is a little more runtime cost, but a little less PAST.
colomon still have to handle the * op (* op2 1) case, right?
(etc)
jnthn colomon: oh, yes
colomon: That...shouldn't be a problem with either approach. 21:47
pmichaud: Am just hacking on it now and was about to JFDI one way or the other, and wondered if you had a big preference, that's all.
pmichaud jnthn: my gut reaction is that the PIR case might be _really_ nice.
as in, cleaner overall for dealing with closures in general. 21:48
jnthn pmichaud: It'd be short but...well...then we end up with annoying problems that make it less nice.
colomon jnthn: I can sort of see how to do it, but was wondering about * op * just being op and how it would interact with that.
pmichaud jnthn: thus my question about how soon do you need it -- I think I need an evening (but just one) to let it stew in my head a bit
jnthn Like, we want to return a Perl 6 Block, not a Parrot Sub. And we want it to have a proper signature...
pmichaud: It's, like, what I've dug into doing tonight. :-)
colomon pmichaud: will that depend on fixing closures?
pmichaud colomon: I think it might simplify fixing closures 21:49
colomon oooooo
pmichaud jnthn: I'm thinking it might simplify the Perl 6 Block issue a bunch
jnthn ...hadn't seen those as related.
pmichaud oh, I think I might've misunderstood what you said earlier, then. 21:50
jnthn pmichaud: I'm just working on the
* + * => { $^a + $^b }
thingy
pmichaud jnthn: right
I'm wondering if it could be generalized slightly to clean up Perl 6 block construction in general throughout the actions 21:51
right now it's a bit... messy
jnthn Ah, OK
Now I see what you're getting at a bit more.
sorear pmichaud: How is the REPL stuff faring? 21:53
pmichaud: Is any of my research helping?
pmichaud sorear: yes, it helped quite a bit.
I _really_ appreciated your conversion of HLLCompiler stuff into nqp. Much nicer to work with.
I have REPL in NQP implemented now. Adding it to other HLLs will be really easy.
Adding it to Rakudo is a bit tricky because the YOU_ARE_HERE stuff needs a bit of a refactor (and because of the way Perl 6 in general deals with lexical setting) 21:54
sorear Yes. You're running into a problem which I did. 21:56
My solution was to change YOU_ARE_HERE so that it only called set_outer if the old outer was null
pmichaud that's the solution we cam up with today
*came
so looks like you were ahead of us a bit
but YOU_ARE_HERE still needs some other cleanups 21:57
sorear your handling of the REPL as outers is a bit different from mine though
pmichaud actually, outer singular now
sorear this custom lexpad stuff is pretty crazy ;)
pmichaud there's one outer
and its lexpad continues to be replaced with an updated lexpad
currently I have it as a package-scoped object, I'm likely to let it be created local to any given interactive loop 21:58
sorear where I went with this was, since eval always sets the outer, it's no longer needed to keep two separate code paths in src/builtins/eval.pm
then $*IN_EVAL can die, !UNIT_START can be simplified a tiny bit
pmichaud correct 21:59
again, you're doing the same as what I plan
sorear also, you're about to break spectests
pmichaud oh?
sorear the spectests assume that eval "return" will drop a non-null value into $! 22:00
in current Rakudo, this works entirely by accident
return throws a control exception, which gets caught by the compile-exception-catcher in eval
once eval is made symmetric, that won't work
pmichaud oh, that's probably not too big an issue 22:01
I think I can work around it, at any rate.
sorear so you'll need to port a patch from topic/lexical-persistence-2 which adds a CONTROL { die "unexpected control exception" }
jnthn Hmm.
* + foo(42)
Means { $_ + foo(42) }
Or
$temp = foo(42); { $_ + $temp } 22:02
pmichaud I'd be okay if &eval isn't exactly the same as HLL::Compiler.eval
jnthn: my expectation would be the latter. 22:03
jnthn Ah, S02 suggests the first.
pmichaud doh!
jnthn Well
It's making a subtler and trickier point though 22:04
Eevee latter seems potentially surprising: foo(*) + foo(42)
jnthn oh hmm
Eevee: That won't get curried, fwiw
Eevee where, e.g., foo == print
oh well then nevermind
jnthn Needs to be a literal * directly on one side of the operator.
pmichaud: The bit of the synopsis is:
This rewrite happens after variables are looked up in their lexical scope,
and after declarator install any variables into the lexical scope,
with the result that
* + (state $s = 0)
is effectively curried into:
-> $x { $x + (state $OUTER::s = 0) }
Eevee how much does * gobble? only infix operators? 22:05
jnthn Eww. :-/
lue Ohai!
jnthn Oh, wait, it's maybe not nasty
Because we have already shoved the decl in the right place.
It seems to imply that the assignment happens inside the closure though.
Eevee: Postfixes and prefixes too, it appears. 22:06
For any unary or binary operator (specifically, any prefix, postfix,
and infix operator), if the operator has not specifically requested
to handle C<*> itself, the compiler is required to translate directly
to an appropriately curried closure at compile time.
Compiler translates...directly.
jnthn sees the evidence starting to stack up one way :-)
Eevee hm. not list? 22:07
pmichaud jnthn: yeah, but I just see weird issues with something like
my $x = * + foo($y);
$y = 7
say $x(3)
jnthn 10? :-)
pmichaud because at the time we evaluate the closure, $y is changed
jnthn pmichaud: Aye - I'm not sure if I see that as weird or not. :-) 22:08
pmichaud I suppose if we think of * as always meaning "construct a closure", then it makes okay sense
jnthn pmichaud: Yeah, that's how I'd been seeing it. 22:09
22:09 IllvilJa left
pmichaud and I'm guessing that *-currying is now always syntactic? 22:09
since it's a compile-time feature?
jnthn pmichaud: That's very much how the synopsis reads to me.
arnsholt It sounds like a potential feature, but one that'll take a bit of getting used to
pmichaud yeah, me too.
jnthn OK, I'll go that way.
sorear It's actually a quite critical feature.
pmichaud how does something like * + foo(*) work? 22:10
jnthn Nastily, I suspect.
pmichaud or even * + (3 - *) ?
sorear It's what makes @foo[*-1] work
jnthn That second one works OK
sorear pmichaud: like { $^a + foo($^b) }
pmichaud: I asked earlier
arnsholt The first example would be the same as *+foo(3), no?
colomon I would assume foo(*) is called, or at least that call is compiled in.
jnthn Ah, or is it..
colomon: Yeah
sorear er, actually it wouild be { $_ + foo(*) }
Eevee wouldn't the first be { $^a + { foo($^a) } }
colomon sorear: right
Eevee and thus sort of useless
jnthn Eevee: No 22:11
It'd end up as { $^a + foo(*) }
sorear Eevee: no. WhateverCode closures curry like Whatever itself
22:11 sundar left
jnthn sorear: WhateverCode is what we can rid of by doing it syntactic, btw. :-) 22:11
pmichaud but foo(*) isn't a WhateverCode
jnthn pmichaud: Correct.
It's just a call that passes *.
pmichaud so what happens with 22:12
* + foo(*.xyz)
Eevee it might be time to read S02
this is getting increasingly contrived
sorear It's never not time to read S..
jnthn { $^a + foo({ $^a.xyz }) }
pmichaud: The two auto-curryings are independent there.
afaiu. 22:13
pmichaud jnthn: right
and foo() probably won't know how to handle a WhateverCode argument?
or....
jnthn pmichaud: There is no WhateverCode
pmichaud sorry, closure
jnthn pmichaud: It's just a Block now?
jnthn had been thinking the switch to syntactic was so we could do away with all the various types and multi candidates.
cognominal the game is to read a S* in one sitting without missing anything. I have lost eacht time so far 22:14
jnthn And then map will Just Work 'cus it knows what to do with a Callable
pmichaud jnthn: well, I think that is the thinking, but in my experience it's a waterbed model
i.e., we think we're doing away with it but we've really just pushed the problems somewhere else :-)
kind of like lists and arrays, or the original interpretations of Whatever :)
jnthn pmichaud: It feels to me like we're trying to take a little bit of sugar that's handy for simple things and trying to make it handle every case we could ever come up with.
pmichaud jnthn: well, that was the original problem with Whatever, also. :) 22:15
Eevee well, the language is perl 6... :P
jnthn I'd kinda rather it didn't, and people can write out exactly what they want if the simple thing isn't it.
pmichaud okay, so we only curry on unary and binary operators 22:16
jnthn Otherwise we get * + 3 * * - 42
And it's like...huh wtf.
pmichaud I can live with that.
jnthn I can live with people being disuaded to write things like that. :-) 22:17
pmichaud I meant I can live with the currying happening at compile time for the binary operators
jnthn pmichaud: Ah, OK. :-)
pmichaud (and unaries)
jnthn OK.
pmichaud I'm guessing that would have to patch into the expression actions somewhere?
i.e., we need nodes that can say "I'm a currying argument?" 22:18
jnthn pmichaud: I was thinking just a check if $key is INFIX, POSTFIX or PREFIX and if so call whatever_curry or something that goes checking for the whatevers, and also that it's not one of the operators we shouldn't auto-curry.
pmichaud: I just added :returns('Whatever') to the PAST node that * generates. 22:19
pmichaud jnthn: sounds like a very reasonable starting point
jnthn Then it's a very easy check
pmichaud go with that :)
jnthn No $pastnode<whatever> flags. ;-)
pmichaud: OK, I'll fetch a beer and dig in. :-)
Eevee (*, *, $x) = (1, 2, 3); <= appears in S02. wouldn't this require * to eat listops too
jnthn Eevee: No, that works for Other Reasons. :-) 22:20
Eevee oh dear. Other Reasons are frightening
jnthn Eevee: It's just a list assignment.
pmichaud but it does have infix:<,> in it :-)
jnthn And it so happens that if you assign to * it throws away the value.
Eevee in that case I guess that's a good argument for not eating listops
jnthn pmichaud: oh ouch...
Eevee oh I thought , was list
jnthn pmichaud: I think I need a check for binary. ;-)
*binaryness 22:21
binarity?
Meh, I should learn English some day.
pmichaud just check that it's not :assoc<list>, probably.
jnthn pmichaud: That works too.
rakudo: * = 'lol thrown away';
p6eval rakudo 475d1c: OUTPUT«Cannot assign to readonly value␤current instr.: '&die' pc 18373 (src/builtins/Junction.pir:373)␤»
jnthn ...oh.
pmichaud either that or make sure that infix:<,> doesn't curry. 22:22
which, it probably doesn't since it takes Object as arguments.
jnthn rakudo: my $x; (*, *, $x) = (1, 2, 3);
p6eval rakudo 475d1c: ( no output )
jnthn pmichaud: Aye.
pmichaud sorry, Mu
jnthn :-)
pmichaud so, does an operator have to explicitly declare a Whatever formin order to be not curried? or is declaring a Mu parameter sufficient? 22:23
i.e., if I have sub infix:<abc>(Mu $a, Mu $b) { ... }
does "* abc 3" auto-curry?
Eevee rakudo: (*) = (1); 22:24
p6eval rakudo 475d1c: OUTPUT«Cannot assign to readonly value␤current instr.: '&die' pc 18373 (src/builtins/Junction.pir:373)␤»
jnthn pmichaud: Well, TimToady mentioned as a first cut we just black-list some we know. 22:25
pmichaud or would it have to be sub infix<abc>(Mu $a, Mu $b) is like-Whatever-and-stuff { ... } ?
jnthn pmichaud: afaik you'd need a signature that somewhere mentioned Whatever
pmichaud (note that "like-Whatever-and-stuff" is what is written in S02)
jnthn omg
22:25 Khisanth joined
pmichaud oh, we'd have to proto it. 22:25
jnthn n operator can declare that it wants to 22:26
handle C<*> either by declaring one or more of its arguments for at
least one of its candidates with an argument of type C<Whatever>, or
by marking the proto sub with the trait, C<is like-Whatever-and-stuff>.
pmichaud Right.
that's the section. :)
jnthn :-)
Yeah
Things get easy on that on the (*sigh*) post-R* multi refactor. :-)
Or at least, I'll do the multi compile time introspection enhancements at the same time I work on getting the new meaning of proto in place. 22:27
pmichaud anyway, I'll let you work on it
jnthn :-)
I has selected a beer and am ready to hack. :-)
Eevee wow, parrot's in ubuntu's universe repo 22:28
lue So nice of them to invite us :) 22:30
22:30 gbacon left
Eevee maybe star will get rakudo in there :( 22:34
diakopter Ubuntu Silver Star ? 22:35
pmichaud probably. rakudo isn't quite yet stable enough
22:35 elmex_ joined
diakopter Ubuntu Spry Star ? 22:35
lue Which one is that?
diakopter the S one?
22:35 elmex left
Eevee haha, no, rakudo star 22:35
22:35 elmex_ is now known as elmex
Eevee rakudo★ 22:36
lue ahh, no I see what you're saying.
s/no/know
s/know/now/
gah!
diakopter .u hexa 22:37
phenny U+4DCA HEXAGRAM FOR PEACE (䷊)
diakopter .u spl
phenny U+4DD6 HEXAGRAM FOR SPLITTING APART (䷖)
diakopter my font doesn't have those
Eevee mine does but they overflow slightly 22:38
lue Mine fit just fine within the parens
arnsholt On my OS X they overflow a bit on the right
lue .u  22:39
phenny U+F8FF (No name found)
lue Why does my keyboard let me type that? What is it?
Eevee an apple
it's actually the last glyph in the private use block
which apple deemed should be a tiny apple logo
lue Well that figures, it's the Apple fourth level I got that from :)
.u E000 22:40
phenny U+E000 (No name found)
lue rakudo: say "\xE000"
p6eval rakudo 475d1c: OUTPUT«␤»
Eevee and that's the beginning
yeah I don't have a glyph
lue Anyone that has Linux Libertine can see uE000 :)
Eevee hm, I seem to have a bunch of misc junk all throughout private use
 U+E033 BUTTERFLY 22:41
lue That's what it's for :) [0xE000 in Linux Libertine is Tux, the penguin]
Eevee ha
lue .u butter
phenny lue: Sorry, no results for 'butter'.
lue rakudo: say "\xE033" 22:42
p6eval rakudo 475d1c: OUTPUT«␤»
Eevee at this size it looks like an airplane with the nose missing
lue I get an ffj glyph, just like fi (fi) and fl (fl)
[for E033]
rakudo: say "\x[$_]" for 0xE000..0xE010 22:43
p6eval rakudo 475d1c: OUTPUT«Unrecognized backslash sequence: '\x' at line 11, near "[$_]\" for "␤current instr.: 'perl6;HLL;Grammar;panic' pc 528 (ext/nqp-rx/src/stage0/HLL-s0.pir:348)␤»
lue darn :) 22:44
Eevee ha I don't think escapes work like that
rakudo: say chr($_) for 0xe000..0xe010 22:45
p6eval rakudo 475d1c: OUTPUT«␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤␤»
jnthn rakudo: say "$_: &chr($_)" for 0xe000..0xe010; 22:46
p6eval rakudo 475d1c: OUTPUT«57344: ␤57345: ␤57346: ␤57347: ␤57348: ␤57349: ␤57350: ␤57351: ␤57352: ␤57353: ␤57354: ␤57355: ␤57356: ␤57357: ␤57358: ␤57359: ␤57360: ␤»
lue Apparently, Linux Libertine and Orbitron have provided some nice characters in the Private Use Area
rakudo: say "{$_.fmt('%x')}: &chr($_)" for 0xe000..0xe010;
p6eval rakudo 475d1c: OUTPUT«e000: ␤e001: ␤e002: ␤e003: ␤e004: ␤e005: ␤e006: ␤e007: ␤e008: ␤e009: ␤e00a: ␤e00b: ␤e00c: ␤e00d: ␤e00e: ␤e00f: ␤e010: ␤»
Eevee I just have a bunch of diamonds, an arrow, a checkmark, and circled digits at the end
lue E009 and E00A are cool W's from Orbitron 22:47
sorry, E008 and E009
rakudo: say "«{$_.fmt('%x')}: &chr($_)»" for 0xe000..0xe010; 22:48
p6eval rakudo 475d1c: OUTPUT««e000: »␤«e001: »␤«e002: »␤«e003: »␤«e004: »␤«e005: »␤«e006: »␤«e007: »␤«e008: »␤«e009: »␤«e00a: »␤«e00b: »␤«e00c: »␤«e00d: »␤«e00e: »␤«e00f: »␤«e010: »␤»
lue Is Pretty Print [colors, font weight, etc.] possible thru perl6? At least to the STDOUT in a console. 22:50
Eevee rakudo: say [~] map 0xe000..0xe010: { "« {$_.fmt('%x')}: &chr($_) »" }; 22:51
p6eval rakudo 475d1c: OUTPUT«Confused at line 11, near "say [~] ma"␤current instr.: 'perl6;HLL;Grammar;panic' pc 528 (ext/nqp-rx/src/stage0/HLL-s0.pir:348)␤»
Eevee rakudo hurt itself in its confusion!
rakudo: say [~] 0xe000..0xe010.map: { "« {$_.fmt('%x')}: &chr($_) »" }; 22:52
p6eval rakudo 475d1c: OUTPUT«␤»
23:00 Psyche^ joined
cognominal rakudo: say (class A {} ).defined 23:00
p6eval rakudo 475d1c: OUTPUT«1␤»
cognominal rakudo: class A {} say ::A.defined 23:01
p6eval rakudo 475d1c: OUTPUT«Confused at line 11, near "class A {}"␤current instr.: 'perl6;HLL;Grammar;panic' pc 528 (ext/nqp-rx/src/stage0/HLL-s0.pir:348)␤»
cognominal rakudo: class A {}; say ::A.defined
p6eval rakudo 475d1c: OUTPUT«0␤»
cognominal beats me.
jnthn rakudo: class A {} say A.defined 23:02
lue rakudo: class ::A{}; say ::A.defined
p6eval rakudo 475d1c: OUTPUT«Confused at line 11, near "class A {}"␤current instr.: 'perl6;HLL;Grammar;panic' pc 528 (ext/nqp-rx/src/stage0/HLL-s0.pir:348)␤»
rakudo 475d1c: OUTPUT«0␤»
jnthn std: class A {} say ::A.defined
p6eval std 30825: OUTPUT«===SORRY!===␤Strange text after block (missing comma, semicolon, comment marker?) at /tmp/2opAbO3d0N line 1:␤------> class A {}⏏ say ::A.defined␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤ statement modifier loop␤Parse
..failed␤…
jnthn Oddness. 23:03
23:03 Patterner left, Psyche^ is now known as Patterner
jnthn std: class A {} say A.defined 23:03
p6eval std 30825: OUTPUT«===SORRY!===␤Strange text after block (missing comma, semicolon, comment marker?) at /tmp/NpoVdMhlmE line 1:␤------> class A {}⏏ say A.defined␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤ statement modifier loop␤Parse
..failed␤FA…
jnthn std: class A {}; say A.defined
p6eval std 30825: OUTPUT«ok 00:01 111m␤»
jnthn std: class A {}; say ::A.defined
p6eval std 30825: OUTPUT«ok 00:01 111m␤»
jnthn rakudo: class A {}; say ::A.defined
p6eval rakudo 475d1c: OUTPUT«0␤»
jnthn ETOOLIBERAL
yay 23:04
> say (1 + *).WHAT
Block()
> say (1 + *).(2)
3
> say (* + *).(2, 4)
6
> say (1..*).WHAT
Range()
pmichaud wow 23:06
afk, dinner 23:07
lue rakudo: my $a = 3; say 1+$a; say (1+*).($a)
p6eval rakudo 475d1c: OUTPUT«4␤4␤»
cognominal rakduo: say (* * *).(*,*) 23:08
rakudo: say (* * *).(*,*)
p6eval rakudo 475d1c: OUTPUT«Method 'Num' not found for invocant of class 'Block'␤current instr.: 'perl6;Mu;' pc -1 ((unknown file):-1)␤»
lue rakudo: say (* *).(3,++) 23:10
p6eval rakudo 475d1c: OUTPUT«Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤current instr.: 'perl6;Regex;Cursor;FAILGOAL' pc 1971 (ext/nqp-rx/src/stage0/Regex-s0.pir:1050)␤»
jnthn rakudo: say (* * *).(2,4) 23:11
p6eval rakudo 475d1c: OUTPUT«Method 'Num' not found for invocant of class 'Block'␤current instr.: 'perl6;Mu;' pc -1 ((unknown file):-1)␤»
jnthn That one works now.
> say (* * *).(2,4)
8
Eevee rakudo: say (*.*).(3, postfix:sym<!>) 23:12
p6eval rakudo 475d1c: OUTPUT«Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤current instr.: 'perl6;Regex;Cursor;FAILGOAL' pc 1971 (ext/nqp-rx/src/stage0/Regex-s0.pir:1050)␤»
23:12 jhuni left
jnthn The parser won't accept *all* line noise. ;-) 23:12
cognominal :)
23:14 jhuni joined
lue rakudo: say (*,*).(3, postfix:sym<!>) 23:17
p6eval rakudo 475d1c: OUTPUT«Could not find sub &postfix:sym<!>␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
lue rakudo: say (*,*).(3, !)
p6eval rakudo 475d1c: OUTPUT«Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤current instr.: 'perl6;Regex;Cursor;FAILGOAL' pc 1971 (ext/nqp-rx/src/stage0/Regex-s0.pir:1050)␤»
lue rakudo: say (*,*).(3, "&!")
p6eval rakudo 475d1c: OUTPUT«invoke() not implemented in class 'ResizablePMCArray'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
lue std: say (*,*).(3, "&!") 23:18
p6eval std 30825: OUTPUT«ok 00:01 114m␤»
jnthn Works on user defined opps too :-)
> our multi postfix:<!>(0) { 1 }; our multi postfix:<!>($n) { $n * ($n - 1)! }
> my $fact = *!; say $fact.WHAT; say $fact(10);
Block()
3628800
lue rakudo: our multi infix:<+>($a is rw,$b){$a++ for ^$b;}; our multi postfix:<++>($a is rw){$a + 1;}; my $a = 3; say $a++; say $a+3 23:22
p6eval rakudo 475d1c: OUTPUT«Method 'set_candidates' not found for invocant of class 'MultiSub'␤current instr.: '' pc -1 ((unknown file):-1)␤»
lue ō.o Last time it was infinite recursion
23:22 Ross^ left
jnthn rakudo: our multi infix:<+>($a is rw,$b){$a++ for ^$b;}; 23:23
p6eval rakudo 475d1c: ( no output )
jnthn rakudo: our multi postfix:<++>($a is rw){$a + 1;};
p6eval rakudo 475d1c: OUTPUT«Method 'set_candidates' not found for invocant of class 'MultiSub'␤current instr.: '' pc -1 ((unknown file):-1)␤»
jnthn Evidentaly postfix:<++> didn't make it into the setting yet. 23:24
lue rakudo: our multi infix:<+>($a is rw,$b){$a++ for ^$b;}; our multi sub postfix:<++>($a is rw){$a + 1;}; my $a = 3; say $a++; say $a+3
p6eval rakudo 475d1c: OUTPUT«Method 'set_candidates' not found for invocant of class 'MultiSub'␤current instr.: '' pc -1 ((unknown file):-1)␤» 23:25
23:25 sorear joined
lue It worked last time... 23:25
pugssvn r30826 | sorear++ | [viv] Refactor emit_p5 using an augment/inner paradigm. -2500 characters.
r30826 | ret and $lvl are no more.
lue rakudo: say "\xE020" 23:26
p6eval rakudo 475d1c: OUTPUT«␤»
lue Why do we use ICU? What does it give us that we use? [just curious] 23:28
jnthn lue: A bunch of unicode stuff.
lue: Like character properties, etc.
lue I thought we really only used it for character tables. 23:29
I was thinking of whether it would be feasible to "remove" that dependancy [I've heard about doing that before, IIRC] 23:30
pugssvn r30827 | sorear++ | [viv] Finish translation of <foo(bar)>
sorear lue: Why?
Repeat after me.
Dependencies are Good.
lue Not like I was going to do it, I was just wondering. 23:31
23:31 demond joined
diakopter sorear: r30826 !!?!??! :D 23:34
just from reading the commit message, I suspect this will make me willing/able to work on it now!
23:47 snarkyboojum joined
diakopter phenny: tell pmurias I looked at paste.lisp.org/display/100307 and it doesn't work just b/c of the non-working expression parser - my $poss = self.positionals; $poss.Add($pos); # works 23:53
phenny diakopter: I'll pass that on when pmurias is around.
23:55 patspam left 23:56 justatheory left
sorear std: method foo ($x) ($y) is rw ($z) { } 23:56
p6eval std 30826: OUTPUT«ok 00:01 115m␤»
sorear TimToady: what does that mean? 23:57
jnthn
.oO( 2 sigs, 1 sub )
 23:58
Eevee whoa
sorear 3 sigs, mixed in with a trait
jnthn Stop spoiling my meme references with accuracy. :P
sorear: Somewhere in S12 there's some conjectural syntax relating to being able to give a single multi candidate multiple signatures. 23:59
sorear: I suspect it's that.

Powered by Cro and the Raku® Programming Language.

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

Thank you!