»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by masak on 12 May 2015.
timotimo ugexe: why the F would you do that? :) 00:01
ugexe i wouldnt. why would others? no idea. but they do 00:02
in the case of Archive::Zip it would create a zip file in one test, test the extraction in another, and so forth 00:03
flussence
.oO( let's just give up and throw a makefile in the test directory! )
ugexe the 'green' test module solved it without adding a .prove type file
timotimo ugexe: also, i was talking about doing the tests inside a single file in parallel 00:05
ugexe ah, thats how green did it i think :) 00:06
00:07 larion joined
ugexe it requires you to be specific about it though, so you dont have say, 2 servers trying to listen on the same port 00:08
leont ugexe: are you aware of my TAP::Harness module? 00:12
00:12 telex left 00:13 telex joined, tokuhiro_ joined 00:14 leont left 00:16 colomon left 00:17 cognominal left 00:20 telex left, telex joined 00:24 colomon joined
ShimmerFairy Hm, I'm quite surprised ≤ and ≥ don't exist in core, seems like it'd be nice (esp. since <= looks too much like a fatarrow for my taste :P) 00:28
ugexe also fwiw... s22: "All .t files in this directory, will be tested in alphabetical order, possibly in parallel." 00:39
colomon ShimmerFairy: they’re trivial to add as a module… 00:56
00:58 aborazmeh joined, aborazmeh left, aborazmeh joined 01:00 khw left
ShimmerFairy colomon: while true, I think a module for two quite obvious Unicode characters is a pretty ridiculous concept :) 01:01
skids ShimmerFairy: A module that adds lots of them, though... 01:06
dalek kudo/nom: 55c1d9d | coke++ | docs/ (3 files):
Prep for 2015.09 - Zürich
 01:12
kudo/nom: 36ac769 | ShimmerFairy++ | docs/ChangeLog:
Add line about val() in ChangeLog

I always meant to do this, but managed to forget :) .		 01:16
01:21 laouji joined 01:22 aborazmeh left
labster ShimmerFairy: actually ≤ and ≥ are present in Latin-1. Even Parrot wouldn't hate it. 01:26
ShimmerFairy not that it matters, but that is an interesting point :) 01:27
(we used to use strings like "\x????" to work around Parrot's issues on that, and I at least would staunchly refuse reinstating that if Parrot support is ever resurrected) 01:28
01:28 khw joined 01:34 yqt left
labster I think we're past the whole stage of adapting the language to suit the compilers. And on a more important note, any user code was similarly slowed, and that's the kind of issue that would need to be solved before anyone adopted the VM anyway. 01:35
ShimmerFairy labster: oh, I didn't know user code was affected too, though I suppose it's the kind of thing only noticeable in CORE.setting, probably the largest P6 program by far :P 01:40
labster exactly. 01:41
m: my $a = 0.9999999999999999999999; say $a; my $b = $a.numerator/ $a.denominator; $b.say; $b.WHAT.say; 01:44
camelia rakudo-moar 86ddca: OUTPUT«0.10␤1␤(Num)␤»
ShimmerFairy And yes, I'd agree we're past adapting to compilers (except for a couple #? lines in Actions.nqp). Probably helps that our two current VMs are either more closely tied to P6 than Parrot was, or relatively far more mature compared to Moar and Parrot :P 01:45
01:45 Ben_Goldberg joined, aborazmeh joined, aborazmeh left, aborazmeh joined
labster JS is tied to P6? I always thought use strict; tied it to Perl 5. :P 01:46
01:47 Ben_Goldberg left, Ben_Goldberg joined
ShimmerFairy two current in-master/nom VMs, I meant :P 01:47
01:55 dayangkun joined
dalek c: 4693476 | coke++ | / (2 files):
Only write out needed docs.

When generating a very sparse copy of the site, this avoids trying to generate data that doesn't exist.		 02:05
c: c0667e2 | coke++ | htmlify.p6:
fix generated JS for search; track GLR change
 02:16
[Coke] ^^ that should unbust search.
02:32 abaugher left 02:33 abaugher joined
colomon is MAIN very very broken wrt slurpy arrays? 02:40
oh, nope, it’s my code that’s the problem.
02:42 AlexDaniel left
colomon ’s code for hacking @*ARGS was very broken under GLR 02:42
ssh: Could not resolve hostname github.com: nodename nor servname provided, or not known :( 02:47
02:52 noganex joined
[Coke] Just ran a very long --profile... get a runtime JS error opening the file. 02:54
Uncaught Error: [$injector:modulerr] errors.angularjs.org/1.2.21/$inject...A18%3A277) 02:55
02:55 noganex_ left
TimToady labster: what makes you think ≤ and ≥ are in Latin-1? 03:00
03:02 colomon left, kaare_ joined
ShimmerFairy A first try at adding ≤ and ≥ caused weird undeclared symbol errors (on a bunch of subs and other terms), so I'll have to try again later :) 03:07
03:09 raiph joined, orbusbak left, orbus joined 03:10 telex left 03:11 telex joined, telex left
ShimmerFairy TimToady: thinking about ≤ and ≥ got me thinking about aliasing operators again. My two thoughts atm are either sub infix:«<=»:«≤» (that is, stacking nameless colonpairs), or sub infix:«≤»:texas«<=» (having a named colonpair) 03:11
03:12 telex joined
[Coke] .tell dha looks like search is working again 03:13
yoleaux [Coke]: I'll pass your message to dha.
[Coke] zzz
03:15 colomon joined 03:16 eternaleye joined
ShimmerFairy Worth noting that the stacked nameless colonpair thing parses in rakudo, and even takes the first nameless as it usually does, it just doesn't do anything with subsequent ones :P 03:17
(the idea, for those who don't know, is that you can then define your own multis on ≤ or <= without worrying about which one is the "correct" one to overload) 03:20
TimToady would be better to have some way of adding on, so that modules can add aliases
ShimmerFairy true, and I wonder if that would be useful for subs in general 03:21
labster TimToady: I was looking at www.asciitable.com/ and saw ≤ ... apparently that isn't Latin-1, still trying to figure out what encoding that is.
ShimmerFairy
.oO( alias infix:«>=» of infix:<≥>; # perhaps an AliasHOW too, depending? )
 03:22
labster Apparently Mac OS Roman had ≤≥, but at different codepoints. Anyway, I think I just learned not to use that website. 03:25
ShimmerFairy Although that might be uncomfortably close to =alias (or alternately just fine, I'm not sure)
labster: I'd recommend only looking up ASCII for codepoints under 0x80; after that you're better off searching for the specific extra encoding instead :) 03:27
03:28 khw left, isacloud left, skarn left 03:30 isacloud joined
ShimmerFairy Could also be synonym infix:«>=» of infix:<≥> , which would keep with the somewhat-linguistic nature of Perl 6 :) (and thinking of it, I like 'synonym' better. Perhaps because 'alias' is a bit too generic to clearly state its purpose) 03:31
03:31 khw joined 03:41 prevost joined 03:42 khw left 03:46 thou joined 03:48 aindilis` joined 03:49 araujo left, aindilis left 03:54 Ben_Goldberg left, araujo joined 03:57 khw joined 04:06 skids left 04:12 skarn joined 04:16 lolisa joined
labster for 1..30 -> $i { my $x = ('0.' ~ '9' x $i).Rat; say $x.WHAT.gist eq ($x.numerator/$x.denominator).WHAT.gist } 04:19
m: for 1..30 -> $i { my $x = ('0.' ~ '9' x $i).Rat; say $x.WHAT.gist eq ($x.numerator/$x.denominator).WHAT.gist }
camelia rakudo-moar 36ac76: OUTPUT«True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤True␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤False␤»
labster Not sure why type is changing there. 04:20
ShimmerFairy Rats change to Num if they overflow
labster Why would it overflow if it's dividing numerator by denominator?
It fit in a Rat before. 04:21
04:26 rmgk is now known as Guest9550, rmgk_ joined, Guest9550 left, rmgk_ is now known as rmgk
ShimmerFairy labster: my guess is because Rat is still mis-defined as having an Int denominator instead of a UInt64 like it should, so a bare construction of a Rat lets you get away with bad values. (Just a guess, no clue if that's really the case) 04:27
04:34 prevost left
dalek kudo/nom: cbb8de3 | ShimmerFairy++ | src/Perl6/ (2 files):
Support comments in qqww lists

Also takes care of some smart quotes that were parsed for ww but weren't subsequently handled in the Actions (so «1 “2 3” 4» would drop the middle item). There don't currently seem to be any tests for this though.		 04:45
ast: 100f141 | ShimmerFairy++ | S02-literals/quoting.t:
Un-TODO tests on comments in quotewords.

Also fixes the syntax; these tests accidentally tried #' instead of #`.		 04:47
ast: 5692656 | ShimmerFairy++ | S02-literals/quoting.t:
Fix some wrong quoting.t tests

I'm not sure when those zen slices were ever supposed to be magically removed by quoting constructs that don't otherwise recognize those kinds of variables. And I can only hope that the qww test was supposed to be qqww, since nothing about ww implies variable interpolation.
With this, quoting.t is no longer fudged for rakudo :) .		 04:52
04:53 thou left
dalek ast: 4f3e528 | ShimmerFairy++ | S02-literals/quoting.t:
Minor test message correction.
 04:54
05:01 araujo left 05:13 khw left 05:44 [Sno] joined
dalek kudo/nom: bbf39d7 | skids++ | src/core/Any.pm:
Do not mention a specific postcircumfix in error. (fix for RT#125504)

Also, nag if someone makes an Associative that isn't associative.		 05:50
kudo/nom: d283300 | lizmat++ | src/core/Any.pm:
Merge pull request #531 from skids/lta_errors

Do not mention a specific postcircumfix in error. (fix for RT#125504)
05:52 aborazmeh left 05:55 domidumont joined 05:57 domidumont left 05:59 laouji left 06:00 domidumont joined 06:04 domidumont left 06:05 domidumont joined 06:06 rurban joined
dalek ast: 470cb7c | lizmat++ | S02-types/list.t:
X::Immutable on List.push/pop/shift/unshift
 06:10
06:11 diana_olhovik joined 06:21 FROGGS joined
dalek kudo/nom: 7931a2c | ShimmerFairy++ | src/core/allomorphs.pm:
Change back allomorphic .perl & .gist

After thinking about it for a while, I decided to change back the .gist and .perl methods for the allomorphic types to how they were originally. We really need .gist to show that you're dealing with an allomorphic type, and so it prints the use of val(), since that's the most succinct way I know of to do it.
The .perl method shows how to construct the object via .new, since I figure .perl is more about the object itself than whatever uses it. And considering the allomorphic types are meant to be user-facing (esp. if you use something like MAIN), I don't think it's so terrible to show it for .perl.
For neither method is using quotewords syntax (<> and «») appropriate; not all uses of val() are through those quoting syntaxes, and those uses of val() with leading/trailing whitespace would lead to an incorrect
  .perl previously. If we wanted to use quoting syntax for either or both
of these methods, it'd have to be a representation using q:v[] or similar.		 06:23
06:24 espadrine joined, raiph left 06:25 laouji joined
FROGGS .tell patrickz github.com/MoarVM/dyncall/pull/2 06:26
yoleaux FROGGS: I'll pass your message to patrickz.
moritz ShimmerFairy: fwiw there's an entry for val() in docs/ROADMAP
ShimmerFairy ah, I should probably take care of that too :)
moritz ShimmerFairy: when you're done with allomorphs, please remoe that
or even now, when you think it's done enough :-) 06:27
ShimmerFairy moritz: the only big things left are Version literals (which shouldn't be too hard to make a VersionStr), and enums (which... how? o_o), as alluded to in one part of S02 06:28
S02:3862 to be precise 06:30
moritz well, the idea behind val() was that it taps into the compiler 06:33
parsing a specific subset of Perl 6
and that could include constants and enums 06:34
lizmat m: say (a => 42).perl # ShimmerFairy
camelia rakudo-moar d28330: OUTPUT«:a(42)␤»
dalek kudo/nom: 5daff4c | ShimmerFairy++ | docs/ROADMAP:
Remove val() from ROADMAP

val() is pretty much done now, with only a couple spots (versions and enums) left to consider. But the majority of it has been dealt with.
  (Also cleaned up an outdated reference to :T)
lizmat ShimmerFairy: disagree with your reasoning re 7931a2c
06:35 thou joined, _mg_ joined
lizmat note that (a=>42).perl doesn't show a fat comma either 06:35
ShimmerFairy lizmat: not every type shows an object.new in .perl of course, I just think that IntStr and friends are complex enough to warrant showing a .new like other not-simple objects.
lizmat weren't they made to make life simpler 06:36
anyway, I've said what I wanted to say, letting TimToady decide on the matter
ShimmerFairy Yes, but that doesn't make them simple themselves :) 06:37
lizmat but do we need to bother users about this lack of simplicity ? 06:38
ShimmerFairy moritz: parsing valid enums would be interesting, but the far more interesting part is what an EnumStr could be, since enums are of an entirely different knowhow altogether. 06:39
lizmat isn't Perl 6 about torturing core developers so that users of the language have it easier ?
06:39 thou left, CQ joined
ShimmerFairy lizmat: why not? I don't think .perl is the place to be so concerned about being 100% user-friendly, but that may be just me :) 06:39
After all, .perl is supposed to give you valid P6 code (as best as it can, in the case of Code objects and such), so I don't think there's any particular "too complex" standard there. 06:40
lizmat well, whatever, breakfast now& 06:41
06:42 brrt joined
ShimmerFairy (As an example, I pretty much always find Match.perl output unhelpful for seeing what's going on, but then again that's not what .perl is for, but rather .gist ☺) 06:42
06:47 lolisa left
moritz ideas for how to make Match.perl more friendly (while still round-tripping) are very welcome :-) 06:48
TimToady we go to some pains to simplify .perl output on things like :{} and such
06:56 brrt left 07:05 zakharyas joined 07:06 ShimmerFairy left 07:09 araujo joined
moritz the real trouble wit Match.perl is that it needs to contain the .orig string 07:10
which can be quite big
CQ What's the best way to get a post-glr perl6 to play with? I don't want bleeding edge, leading edge is fine
moritz and obscures the actual matched string
CQ: rakudobrew build nom
FROGGS moritz: we could always auto-gist huge strings and refer to the gist url :o)
CQ moritz: Building Rakudo with backend 'nom' is NYI. Well volunteered! ... suggestions? :) 07:16
07:17 jack_rabbit joined 07:20 ShimmerFairy joined 07:26 larion left
cdc CQ: there's no post GLR release yet, that means you have to play with the cutting edge. "rakudobrew build moar", then "rakudobrew rehash" (assuming you are using the latest revision of rakudobrew). 07:26
07:26 abraxxa joined
moritz ah yes, sorry, 'moar' not 'nom' 07:28
nine CQ: bleeding edge really is not that bad. There's no special process for stabilizing a release, so really the a typical release is mostly missing bugfixes that are already in git ;) 07:38
07:41 cdc is now known as jeanmi-apropos 07:42 jeanmi-apropos is now known as cdc
lizmat commute& 07:44
07:44 lizmat left 08:05 leont joined, pierrot left 08:07 tping left
CQ Rakudo has been built and installed.; Switching to moar-nom; Done, moar-nom built ...thanks! 08:10
08:11 dakkar joined 08:14 brrt joined
ShimmerFairy fwiw I wouldn't mind a different .perl on the allomorphic types, I'd just appreciate if it were different from .gist (which may mean using q:v somewhere). I will however stay firm on making .gist show it's an allomorphic type, and in not using <>/«» in those methods' strings :) 08:17
08:17 g4 joined 08:19 pmurias joined
pmurias hi 08:19
yoleaux 14 Sep 2015 16:56Z <hoelzro> pmurias: should I go ahead and complete that merge tonight?
pmurias hoelzro: yes
08:21 rindolf joined
jnthn .gist on the allomorphic types really should just give their Str representation like .gist on a Str does 08:21
sub MAIN($filename) { say $filename; } # suddenly, when you invoke it with a numeric filename, the output will look weird
08:22 patrickz joined 08:23 thou joined
jnthn I don't want those kind of gotchas. Allomorphs are there to help get dispatch decisions right, and so make things easier, not to make things unpredictable. 08:23
08:24 jack_rabbit left
patrickz .tell FROGGS: Thanks! 08:25
yoleaux 06:26Z <FROGGS> patrickz: github.com/MoarVM/dyncall/pull/2
patrickz: What kind of a name is "FROGGS:"?!
patrickz .tell FROGGS Thanks!
yoleaux patrickz: I'll pass your message to FROGGS.
jnthn yoleaux is the opposite of Larry: totally doesn't get the colon... :P
FROGGS :D 08:26
yoleaux 08:25Z <patrickz> FROGGS: Thanks!
patrickz FROGGS: Dyncall seems to have moved to HG. I suspect the SVN is out of sync. 08:27
08:27 thou left
FROGGS hmmm 08:28
patrickz If the MoarVM dyncall is meant to stay up to date with upstream that might be the reason for last commit in march...
FROGGS patrickz: are there any changes upstream?
patrickz Yes
quite some
last 8 days ago
FROGGS ohh cool
patrickz hg.dyncall.org/pub/dyncall/dyncall
FROGGS \o/ they're finally doing a release 08:29
jnthn \o/ 08:30
pmurias hoelzro: I'll try to do that myself, the only thing missing is a memory bug fix/workaround to get a test to run on moar
leont Hmmm, is seems actions are even called on <.rule>s, is this intentional? 08:31
ShimmerFairy jnthn: I can see your point, though I don't think purely-numerical filenames are that common :P . I still would prefer .gist being more informative (since I find an allomorphic posing as any other ordinary string in a .gist misleading)
FROGGS patrickz: can you please file a dyncall issue in our repo with the new url?
patrickz: perhaps JimmyZ can pull in the commitsfrom the new repository location 08:32
jnthn leont: Yes, . means "don't capture"
patrickz will do (after $dayjob)
jnthn And is a property of the caller, not the callee
ShimmerFairy jnthn: We _could_ get rid of the allomorphic say/note candidates, and go back to when I had to use .gist to actually get the proper gist, though I imagine creating a disparity between .gist and &say output like that isn't preferable :P 08:33
leont Hmmm, it doesn't seem useful to me, but sure
patrickz FROGGS: Did that build issue PR found its way to upstream dyncall? 08:34
jnthn leont: Action methods can have side-effects (and indeed have plenty in the Perl 6 parser)
jdv79 how is a gist of val("123") better than 123?
ShimmerFairy leont: I can easily imagine an uncaptured thing's action method setting a contextual that's needed later on, regardless of capture (since action methods aren't _just_ about 'make', it's not exactly easy to figure if calling it would be useless on an uncaptured thing ☺) 08:35
jnthn ShimmerFairy: I don't see why we need extra candidates
jdv79 if i want more than a gist that's what .perl is for.
ShimmerFairy jdv79: because posing as just a string when it's actually two things is misleading to me.
jnthn ShimmerFairy: The candidate for a Str (any subclass of) in say/note should just nqp::unbox_s(...)
dalek p/js-merge-wip: 5926ead | (Pawel Murias)++ | tools/build/gen-js-makefile-parrot.nqp:
Remove old script for cross-compiling the js backend on parrot.
p/js-merge-wip: bb10638 | (Pawel Murias)++ | HACKING:
move the HACKING file to docs, and add a -js suffix to it
leont Right, that sounds reasonable
jdv79 a bunch of things have "lossy" gists. that's part of the point.
nine It's a gist after all, not the director's cut. 08:36
ShimmerFairy jnthn: There are say/note candidates in allomorphs.pm so that they could get the .gists properly. (They're not with the others because it's too early then) 08:37
"lossy" ≠ misleading, though
jnthn m: say 42.4
camelia rakudo-moar 5daff4: OUTPUT«42.4␤»
jnthn That doesn't scream "OMG I AIN'T A STRING!" when we print it :P
Well, say it, but doesn't on print either :) 08:38
jdv79 taking a gist to be a source of authoratative truth would be a mistake
that's not a "gist";)
jnthn ShimmerFairy: Looking at the extra candidates in allomorphs.pm only makes me more sure I'm right. :)
Needing those is decidedly a design smell. 08:39
FROGGS patrickz: does not look like
ShimmerFairy jnthn: that's true (I had a feeling someone was going to pull that one out :P), I just think that the allomorphic types warrant a gist that clues you in on what's happening.
jnthn: I agree it's a design smell, but I think it's because there shouldn't be a .gist-avoiding 'say' candidate in the first place :P
I didn't put in a lot of things with explicit allomorphic candidates, I did it for 'say' and 'note' because I figured those printing functions warranted accurate representation of the allomorphic object itself, and not picking one of its two sides. 08:42
jnthn ShimmerFairy: As already noted, the point of say/note isn't accurate representation, but human-friendly representation 08:44
ShimmerFairy (also, again, to get around the fact that the Str candidates on say and note avoid .gist, which I would consider a bug (even if you disagree for allomorphs, it's still a bug waiting to happen))
jnthn Anyway, not going to spend any longer on this, for one 'cus it's TimToady++'s call, and for two if it's delegated back to me I already made my mind up. :) 08:45
pmurias: I can reproduce that SEGV you found yesterday locally
pmurias jnthn: it was a realloc got an invalid size for me 08:48
jnthn pmurias: Yowser, it explodes in the MAST -> bytecode compiler
That's a new one...
pmurias and for me it passes if I write the mbc to disk 08:49
jnthn Doing a debug build at the moment to try and get a better location out of valgrind :)
08:50 araujo left
jnthn pmurias: Seems we write past the end of a buffer, and that's the real issue; the realloc exploding is probably a consequence of the corruption 08:51
08:51 cognominal joined, araujo joined
patrickz FROGGS: Last time you just wrote them an email, right? 08:53
08:53 darutoko joined
patrickz I can do that... 08:53
FROGGS patrickz: I didn't
jnthn Ah, it looks like an accident when labeled exception handlers were added.
08:54 domm left
pmurias is glad it turned out a normal bug rather than some weird gc freak accident as he expected 08:55
patrickz FROGGS: There is a commit in upstream dyncall referencing to you. ;-)
08:56 domm joined, rindolf left
FROGGS patrickz: yes, but that's abuot something else, not the cast problem what the issue #2 was about 08:57
08:57 domm left, domm joined, domm left 08:58 domm joined, domm left
patrickz FROGGS: I'm asking because they don't give any hints on their webpage how to propose pull requests. So that past commit came to them via email? If yes I'll also just email them this current PR. 08:58
08:59 domm joined, domm left
FROGGS patrickz: yes, I just emailed them a diff 08:59
08:59 domm joined
patrickz ok, will do the same. 09:00
09:00 larion joined, domm left, domm joined 09:07 leont left 09:08 domidumont left 09:13 brrt left 09:14 rindolf joined
jnthn pmurias: SEGV fixed in HEAD; thanks for reporting. Will you include the test file in question in the merge, so we have this covered? 09:19
09:24 [Sno] left, ][Sno][ joined 09:25 Woodi joined
pmurias jnthn: yes 09:25
jnthn pmurias: OK, great. Please can you close the issue once that's done? 09:30
pmurias MOAR_REVISION should contain a revision sha and the date from where it's from? 09:31
dalek ast: b8bcb2c | jnthn++ | S17-lowlevel/lock.t:
Test covering RT #125705.
 09:32
jnthn pmurias: Just run "git describe" in the MoarVM repo 09:33
(after pulling)
And put the result into that file
.tell lizmat if you're reliably seeing S17-supply/start.t passing now, then maybe rt.perl.org/Ticket/Display.html?id=125460 can be closed... I've not seen it fail since my recent fixes in Moar. 09:36
yoleaux jnthn: I'll pass your message to lizmat.
pmurias jnthn: thanks
09:40 telex left, telex joined
jnthn pmurias: Thanks for reporting it; it could easily have cropped up in Rakudo as well as NQP :) 09:42
So very happy to have it fixed
Also happy that rt.perl.org/Ticket/Display.html?id=125161 is no longer explosive :) 09:43
dalek ast: 1b3acbf | jnthn++ | S17-promise/start.t:
Test covering RT #125161.
 09:44
09:46 itz joined
itz o/ 09:46
jnthn And now we're down to 1020
dalek Heuristic branch merge: pushed 511 commits to nqp by pmurias 09:48
pmurias hoelzro: the js backend has been merged in 09:49
jnthn pmurias++ hoelzro++
09:51 pmurias left 09:52 pmurias joined 09:56 JimmyZ joined
jdv79 what was this recent invocation data race? 09:59
10:02 mprelude joined, andreoss joined
jdv79 113c3b5? 10:02
dalek p: b5d7db9 | (Pawel Murias)++ | tools/build/process-qregex-tests:
Run passing rx_quantifiers test
nqp: f19a35e | (Pawel Murias)++ | src/vm/js/QAST/Compiler.nqp:
nqp: Add a hack so that ifnull returns true on undefined.
10:02 dalek left
pmurias hoelzro: the js backend is now fully merged in 10:03
10:03 dalek joined, ChanServ sets mode: +v dalek
JimmyZ \o 10:03
andreoss m: say ([1,2,3] Z+^ ([1,2,3] xx *) 10:04
camelia rakudo-moar 5daff4: OUTPUT«5===SORRY!5=== Error while compiling /tmp/sB2AOGiyzB␤Unable to parse expression in parenthesized expression; couldn't find final ')' ␤at /tmp/sB2AOGiyzB:1␤------> 3say ([1,2,3] Z+^ ([1,2,3] xx *)7⏏5<EOL>␤ expecting any of:␤ st…»
andreoss m: say ([1,2,3] Z+^ ([1,2,3] xx *))
camelia rakudo-moar 5daff4: OUTPUT«(2 1 0)␤»
andreoss m: say ([1,2,3] Z+^ flat([1,2,3] xx *))
camelia rakudo-moar 5daff4: OUTPUT«(0 0 0)␤»
andreoss m: say ([1] ^+ [1]) 10:05
camelia rakudo-moar 5daff4: OUTPUT«one([1], 1)␤»
andreoss m: say ([1] +^ [1])
camelia rakudo-moar 5daff4: OUTPUT«0␤»
andreoss why should +^ work with Positionals? 10:06
ShimmerFairy nope, +^ is for numeric stuff, closest to what you want is ~^ for Bufs 10:07
psch m: &infix:<+^>.candidates>>.signature.say
camelia rakudo-moar 5daff4: OUTPUT«(() ($x) ($x, $y) (Int:D \a, Int:D \b) (int $a, int $b))␤»
psch the fitting cand is the 4th
m: say [1,2,3] ~~ Int
camelia rakudo-moar 5daff4: OUTPUT«False␤»
ShimmerFairy andreoss: oh, I didn't read the "should". +^ coerces to Numeric, and +@array -> @array.elems 10:08
psch m: say ([1,2,3]).^can('Int')
camelia rakudo-moar 5daff4: OUTPUT«(Int Int)␤»
psch also, g'morn #perl6 o/
10:11 _mg_ left, thou joined 10:15 thou left 10:29 Andrie joined 10:31 [Tux] left 10:32 [Tux] joined
andreoss m: ([1], -> @p { [0, @p Z+ @p, 0] } ... *)[^5].perl.say 10:32
camelia rakudo-moar 5daff4: OUTPUT«===SORRY!===␤Type check failed in binding @p; expected 'Positional' but got 'Int'␤»
timotimo o/ 10:35
psch o/ timotimo
10:38 pmurias left
andreoss m: ([1], [1], [1] ... *)[^3].say.perl 10:39
camelia rakudo-moar 5daff4: OUTPUT«(1 1 1)␤»
FROGGS \o/
js in master!
timotimo why .say.perl?
andreoss m: ([1], [1], [1] ... *)[^3].perl.say
camelia rakudo-moar 5daff4: OUTPUT«(1, 1, 1)␤»
psch m: ([1], -> *@p { [0, @p Z+ @p, 0] } ... *)[^5].perl.say # vOv 10:41
camelia rakudo-moar 5daff4: OUTPUT«(1, 1, 1, 3, 3)␤»
psch i *think* i found a pattern why that weird BagHash set op things throws the CME...
s/things/thing/
andreoss it should generate pascal's triangles
psch it seems to happen whenever not all elements are in the intersection... 10:42
10:42 xfix joined
timotimo m: ([1], -> *@p { $[0, @p Z+ @p, 0] } ... *)[^5].perl.say 10:42
camelia rakudo-moar 5daff4: OUTPUT«(1, $[1, 1], $[2, 2], $[3, 3], $[4, 4])␤»
timotimo oh 10:43
m: ([1], -> *@p { $[(0, |@p) Z+ (|@p, 0)] } ... *)[^5].perl.say
camelia rakudo-moar 5daff4: OUTPUT«(1, $[1, 1], $[1, 3, 2], $[1, 3, 5, 3], $[1, 3, 5, 7, 4])␤»
timotimo almost! :)
m: ([1], -> *@p { $[|(0, |@p) Z+ |(|@p, 0)] } ... *)[^5].perl.say
camelia rakudo-moar 5daff4: OUTPUT«(1, $[1, 1], $[1, 3, 2], $[1, 3, 5, 3], $[1, 3, 5, 7, 4])␤»
timotimo m(
m: ([1], -> *@p { @p.perl.say; $[|(0, |@p) Z+ |(|@p, 0)] } ... *)[^5].perl.say
camelia rakudo-moar 5daff4: OUTPUT«[1]␤[1, [1, 1]]␤[1, [1, 1], [1, 3, 2]]␤[1, [1, 1], [1, 3, 2], [1, 3, 5, 3]]␤(1, $[1, 1], $[1, 3, 2], $[1, 3, 5, 3], $[1, 3, 5, 7, 4])␤»
timotimo ah, i see
m: ([1], -> @p { $[|(0, |@p) Z+ |(|@p, 0)] } ... *)[^5].perl.say 10:44
camelia rakudo-moar 5daff4: OUTPUT«===SORRY!===␤Type check failed in binding @p; expected 'Positional' but got 'Int'␤»
timotimo m: ($[1], -> @p { $[|(0, |@p) Z+ |(|@p, 0)] } ... *)[^5].perl.say
camelia rakudo-moar 5daff4: OUTPUT«($[1], $[1, 1], $[1, 2, 1], $[1, 3, 3, 1], $[1, 4, 6, 4, 1])␤»
timotimo m: ($[1], -> @p { $[(0, |@p) Z+ (|@p, 0)] } ... *)[^5].perl.say
camelia rakudo-moar 5daff4: OUTPUT«($[1], $[1, 1], $[1, 2, 1], $[1, 3, 3, 1], $[1, 4, 6, 4, 1])␤»
timotimo m: ($[1], -> @p { $[(0, @p) Z+ (@p, 0)] } ... *)[^5].perl.say
camelia rakudo-moar 5daff4: OUTPUT«($[1], $[1, 1], $[2, 2], $[2, 2], $[2, 2])␤»
timotimo so that | is the only one needed
m: ($[1], -> @p { $[0, |@p Z+ |@p, 0] } ... *)[^5].perl.say
camelia rakudo-moar 5daff4: OUTPUT«($[1], $[1, 1], $[1, 2, 1], $[1, 3, 3, 1], $[1, 4, 6, 4, 1])␤»
timotimo andreoss, psch, does that help? 10:45
psch ...not with the BagHash stuff :P
still, timotimo++
:)
timotimo sorry, psch :| 10:46
were you able to get the exact instruction that throws the CME?
jnthn Hm, in RT can you not search the bodies of tickets? 10:49
10:49 nwc10 joined
nwc10 pmurias++ 10:49
hoelzro++
vendethiel pmurias++, hoelzro++ too :-)
timotimo jnthn: don't you get all RT's via email, too? :P just search through your mail client :P
psch timotimo: well, almost. i can see the reason, but not the cause 10:50
nwc10 so, NQP now self-hosts and bootstraps on JS ?
timotimo hm
10:50 yakudza joined
nwc10 or is it a MoarVM-based cross compiler? 10:52
vendethiel nwc10: the latter, I think
psch timotimo: gist.github.com/peschwa/0ee944afeb1e1c874902
that's the reason it throws
i have no clue how to find the modification on the HashMap that happened after the Iterator was instantiated...
timotimo i don't get it, though 10:53
jdv79 timotimo: the QJson "too large document" error, which btw - great grammar there... - is likely unfixable
need to port to sommething else i guess 10:54
its somehow hitting a 27 bit boundary in some spec'd bin format that's trouble to fix
TMI but meh
timotimo are you making fun of my grammar? :) 10:55
psch the line in CORE.setting is in Seq.pull-one in EnumMap.values...
jdv79 no, that's what QJsonParseError spits out ^H
timotimo oh?
i didn't know it'd even admit that
jdv79 remember moritz's json didn't parse
that's why 10:56
timotimo yeah
i've had that problem with a different json document in the past
... perhaps it was the core setting compilation profile?
11:02 cognominal left
FROGGS shudders when looking at 'for my $cell (@{$row->[$#head..$#$row]}) {', which he just wrote 11:02
11:03 cognominal joined
jnthn .tell lizmat I think the .push/.pop error improvements in List have busted make test (54-lib.t) 11:04
yoleaux jnthn: I'll pass your message to lizmat.
11:09 rurban left 11:10 TEttinger left, raiph joined, Sqirrel left 11:11 Sqirrel joined 11:12 Sqirrel_ joined, Sqirrel left
masak afternoon, #perl6 11:12
jnthn o/ masak 11:15
11:16 domidumont joined 11:17 Andrie left
nwc10 good UGT heresy, masak 11:20
11:21 ][Sno][ is now known as [Sno] 11:25 dayangkun left 11:27 raiph left, _mg_ joined, zacts left
masak nwc10: I'm not on UGT right now. :) 11:28
11:28 zacts joined, raiph joined 11:31 telex left, telex joined, telex left 11:32 telex joined, g4 left 11:36 g4 joined
itz DST for UGT? 11:41
11:42 laouji left
dalek pan style="color: #395be5">perl6-examples: b36acee | (Steve Mynott)++ | / (2 files):
[99probs] P10-scottp.pl don't flatten in result
 11:42
itz m: dd <4 a> 11:47
camelia rakudo-moar 5daff4: OUTPUT«(IntStr.new(4, "4"), "a")␤» 11:48
itz m: dd <a>
camelia rakudo-moar 5daff4: OUTPUT«"a"␤»
itz m: dd <4>
camelia rakudo-moar 5daff4: OUTPUT«IntStr.new(4, "4")␤»
itz m: dd <~4> 11:49
camelia rakudo-moar 5daff4: OUTPUT«"~4"␤»
11:53 rurban joined 11:54 laouji joined 11:55 pmurias joined
itz m: dd > dd ([<4 a>]) 11:55
camelia rakudo-moar 5daff4: OUTPUT«5===SORRY!5=== Error while compiling /tmp/H37Y_IagPE␤Preceding context expects a term, but found infix > instead␤at /tmp/H37Y_IagPE:1␤------> 3dd >7⏏5 dd ([<4 a>])␤»
itz m: dd <4 a>>>.Str 11:56
camelia rakudo-moar 5daff4: OUTPUT«("4", "a")␤»
itz any neater way?
andreoss m: my @x = gather { take $_ for 1..* } #'{greedy} ; say @x[^10]
pmurias nwc10: it's not self-hosted yet
andreoss m: my @x = lazy gather { take $_ for 1..* } ; say @x[^10]
camelia rakudo-moar 5daff4: OUTPUT«(timeout)» 11:57
rakudo-moar 5daff4: OUTPUT«(1 2 3 4 5 6 7 8 9 10)␤»
11:57 Skarsnik joined
andreoss is gather now greedy by default? 11:57
11:59 thou joined
pmurias nwc10: it's the next milestone 11:59
12:02 xfix left 12:04 thou left
psch huh 12:04
i think i have an idea where this comes from
12:04 vendethiel left
psch we're using $k.WHICH for BagHashes 12:04
as key, i mean 12:05
and when copying the lhs to iterate over the pairs, we copy the underlying HashMap, but assign the same key
...actually maybe that doesn't really make sense :/ 12:06
'cause we have one hashKeyIter per VMIterInstance
but removing a String from one of them shouldn't touch the underlying HashMap
1) we have two HashMaps 2) one of them gets an VMIterInstance for its $!storage 12:08
dalek pan style="color: #395be5">perl6-examples: 94be1fa | (Steve Mynott)++ | / (2 files):
[99probs] P12-unobe.pl post-glr quoting behaviour
psch 3) the .next() of this iter is removed from the other HashMap
4) stuff explodes
that's what i'm pretty sure off... o.o
assuming the different BagHashes have actual different underlying $!storage this shouldn't happen... 12:09
12:09 xfix joined 12:22 xfix left 12:25 cognominal left
jnthn andreoss: Assignment is eager unless it encounters something marked as lazy. gather alone doesn't imply lazy. 12:29
nwc10 pmurias: aha right. thanks for the confirmation/clarification
good luck
have fun
jnthn andreoss: The way you wrote it (with lazy) is the right thing to do. 12:30
pmurias nwc10: the point where the real fun beings is when we get proper Perl 6 to start compiling to js ;) 12:31
nwc10 I hope that day comes soon, but as I'm not able to help (other than say thanks and offer encouragement) I'm not going to keep on about it 12:32
CQ pmurias: what do you mean compile to, do you mean code generation? 12:33
dalek kudo-star-daily: 4ad48b4 | coke++ | log/ (2 files):
today (automated commit)
 12:34
12:34 zakharyas left
dalek rl6-roast-data: 9f6a6fe | coke++ | / (9 files):
today (automated commit)
 12:34
jnthn Generate JavaScript from Perl 6, much like we generate MoarVM and JVM bytecode from Perl 6 today.
pmurias CQ: yes 12:35
jnthn Code generation is just the last step of compilation, though :)
pmurias the rest will be handled by rakudo++ 12:36
maybe with a custom analysis pass (or two) thrown in 12:37
CQ how would it be done? something like github.com/drforr/Perl-Mogrify ? 12:39
pmurias CQ: something like github.com/perl6/nqp/blob/master/s...mpiler.nqp 12:41
pmurias should propably split this file up...
braintwist wow I just found something neat, my systems time is 2 hours ahead of what the time actually is, but then I ran the DateTime in perl, and perl got the time right, it is really awesome, but I have no idea how perl does it ?
[Coke] drinks a coffee. 12:42
pmurias CQ: the biggest difference is that the javascript I emit won't be idiomatic 12:43
so it should emit code that you can run instead of stuff that sort of means that same things but can be tweaked and used as editable code 12:45
masak braintwist: if you want to find out, maybe talk to a Perl 5 forum? :) 12:49
12:49 xfix joined 12:50 xfix left
itz braintwist: I would guess either you or computer might be confused about timezones and/or daylight saving :) 12:50
dalek pan style="color: #395be5">perl6-examples: fb32836 | andreoss++ | / (10 files):
[euler] more GLR fixes
[Coke] I have a --profiel out from a big run that dies - a run from "say 3" works fine. will see if I can figure out what happened. 12:51
[Coke] commutes
braintwist masak: I was doing it in perl6
itz: I would think daylight savings would only explain one hour ? 12:52
12:56 El_Che left 13:01 hernanGOA joined
psch i'm confused 13:03
m: role R { has $!foo }; class C does R { }; say nqp::getattr(C.new, C, '$!foo')
camelia rakudo-moar 5daff4: OUTPUT«(Any)␤»
psch m: my $a = BagHash.new(<a b>); say nqp::getattr($a, BagHash, '%!elems');
camelia rakudo-moar 5daff4: OUTPUT«P6opaque: no such attribute '%!elems'␤ in block <unit> at /tmp/uwOvJpdphe:1␤␤»
psch but BagHash does Baggy, and Baggy has %!elems
jnthn psch: my $a := ..., or nqp::decont($a)
psch jnthn: ugh. right, decont is a thing that still is a bit spooky to me apparently :) 13:04
psch makes a note, "if it's weird and there's a $ try deconet" 13:05
s/deconet/decont/
hm, probably better as "if there's a sigil" 13:06
13:07 g4 left
dalek pan style="color: #395be5">perl6-examples: 5d0d700 | (Steve Mynott)++ | t/categories/99-problems.t:
[99probs] fix P25-topo.pl post-glr
 13:08
pan style="color: #395be5">perl6-examples: fb325c8 | (Steve Mynott)++ | / (10 files):
Merge branch 'master' of github.com:perl6/perl6-examples
13:08 g4 joined
dalek p: 771071d | (Pawel Murias)++ | src/vm/js/QAST/Compiler.nqp:
Insert newline when emitting strings to make it easier to read the generated code.
 13:12
p: d4330a7 | (Pawel Murias)++ | src/vm/js/QAST/Compiler.nqp:
Refactor atpos op handling.
p: 2a97e34 | (Pawel Murias)++ | / (2 files):
Test negative indexes to nqp::atpos.

Implement them on the js backend.
13:13 skids joined
Woodi hallo #perl6 :) 13:15
pmurias++ :)
moritz pmurias++ # for nqp-js 13:16
pmurias++ # for increasing NQP test coverage
Woodi pmurias: so how to use that new addon ? install something and then in nqp configuration add --backends=js ?
masak pmurias++ # making Perl 6 web scale 13:19
13:19 El_Che joined 13:21 raiph left
dalek pan style="color: #395be5">perl6-examples: 924d611 | (Steve Mynott)++ | categories/99-problems/P31-rhebus.pl:
[99probs] fix P31-rhebus.pl post-glr
 13:23
13:23 nwc10 left
pmurias Woodi: --backends=moar,js should do the trick 13:23
[Coke] In AngularJS 1.2.0 and later, ngRoute has been moved to its own module. If you are getting this error after upgrading to 1.2.x or later, be sure that you've installed ngRoute. 13:28
dalek pan style="color: #395be5">perl6-examples: dc6f486 | andreoss++ | categories/euler/prob054-andreoss.pl:
fix indentation
pan style="color: #395be5">perl6-examples: be8087d | andreoss++ | t/categories/99-problems.t:
Merge branch 'master' of github.com/perl6/perl6-examples
pan style="color: #395be5">perl6-examples: c0c8711 | andreoss++ | categories/99-problems/P31-rhebus.pl:
Merge branch 'master' of github.com/perl6/perl6-examples
Woodi pmurias: thanx
nine pmurias: great to follow your progress :)
Woodi btw. Node/libuv company was lastly bought by IBM, interesting what will happens now... 13:30
timotimo is that joyent? 13:31
masak [Coke]: is that the profiler? 13:33
Woodi timotimo: no. and looks not autors of Node.js as I thinked: www.informationweek.com/cloud/platf...id/1322144 13:34
timotimo "[...] Node.js, a technology that enables developers to craft APIs to suit any need in the Java universe. " 13:35
oh, is that so!
itz 1999 java != javscript, 2015 our java was really javascript 13:36
timotimo oh?
itz that probably needed a :)
13:38 El_Che left
psch huh 13:39
that surely makes the author look credible 13:40
13:40 El_Che joined
timotimo seems like that's more of a business news website 13:40
psch timotimo: still, that he quotes someone speaking about JavaScript but talks about Java himself..? 13:41
timotimo it's an easy mistake to make; probably comes down to a typo
psch .oO( but PERL vs Perl is reason for complaint emails... ) 13:42
moritz yes, Iscript accidentally insertscript the word "script" into my text allscript the time
common typo
itz javascript was originally called livescript which is confusing since there is another livescript now 13:43
timotimo and then microsoft built something called "JScript", which is confusing because there's also "J"
"Because a developer can "express things as APIs, new services can be built around them," Soto said." ← ?? 13:45
13:46 khw joined 13:47 thou joined
dalek pan style="color: #395be5">perl6-examples: 0079070 | andreoss++ | categories/euler/prob081-moritz.pl:
remove unnecessary comments
 13:49
13:50 cognominal joined, andreoss left 13:51 aborazmeh joined, aborazmeh left, aborazmeh joined 13:52 thou left
dalek p/travjs: 679ecd4 | hoelzro++ | .travis.yml:
Add JS to Travis build matrix
 13:55
pan style="color: #395be5">perl6-examples: c12b95d | (Steve Mynott)++ | categories/cookbook/04arrays/04-05iterating-over-an-array.pl:
[cookbox] .Str now needed for <1 2>
 13:58
13:58 zakharyas joined
abraxxa is there a debug interface for NativeCall so e.g. see the memory location of a Pointer and its size and contents? 13:59
13:59 FROGGS left
[Coke] masak: yes, that was the profiler error. 14:00
14:00 abaugher left 14:01 abaugher joined
[Coke] needs to update the daily nqp runs to get nqp-js from master... 14:02
14:03 Ovidius joined, travis-ci joined
travis-ci NQP build failed. Rob Hoelz 'Add JS to Travis build matrix' 14:03
travis-ci.org/perl6/nqp/builds/80437183 github.com/perl6/nqp/compare/travjs-smoke-me
14:03 travis-ci left
dalek p/travjs-smoke-me: c0033ed | hoelzro++ | .travis.yml:
Remove Travis notifications while I'm hooking up JS
 14:04
14:04 thou joined
dalek p: d5702b6 | (Pawel Murias)++ | tools/build/process-qregex-tests:
Renable the rx_subrules tests on the js backend
 14:04
p: 3a54939 | (Pawel Murias)++ | tools/build/process-qregex-tests:
Run rx_modifiers on make js-test
p: 2f3c2c8 | (Pawel Murias)++ | src/vm/js/nqp-runtime/runtime.js:
[js] converting strings to integers
p: 4055609 | (Pawel Murias)++ | tools/build/ (2 files):
[js] Avoid calling npm (and recompiling all the c++ modules) when only a .js file changes.
14:04 Ovidius is now known as CurtisPoe
pmurias hoelzro: make now should be faster on a mere js change 14:04
CurtisPoe Is there anyway I can declare an attribute in a class as rw and automatically have it's mutator return the invocant?
s/it's/its/ 14:05
pmurias CurtisPoe: you want to have something like that ($foo.bar = 123).baz = 456) set baz on $foo? 14:06
jnthn Not really, given that you don't get a mutator, but rather an l-value accessor
14:06 abaugher left, laouji left
CurtisPoe So I'd need to write method wrappers for all of them? 14:06
moritz CurtisPoe: rw attributes *are* attributes with method wrappers
PerlJam CurtisPoe: or a trait that did it for you.
jnthn I'd just write a trait that uses the MOP :)
moritz uhm 14:07
but
it doesn't work, does it?
I mean, if you call $foo.bar, it just returns writable container
14:07 abaugher joined
moritz and if you assign an integer to it, it can't return something else, no? 14:07
jnthn Well, yeah, you'd have to actually adopt a mutator syntax
$foo.bar(42).baz(69) etc 14:08
CurtisPoe jnthn: that's what I was looking for. I wanted to write a clean Perl 6 example for en.wikipedia.org/wiki/Fluent_interface 14:09
14:10 muraiki joined
nine CurtisPoe: those are probably the worst examples for fluent interfaces one could think of :) 14:10
CurtisPoe I'm don't want to argue with a wikipedia page, I just want to see if we can have a nicer example :) 14:11
dalek p/travjs-smoke-me: 9cec7d3 | hoelzro++ | .travis.yml:
Add nodejs matrix for Travis tests

I don't know if this will actually work, but here goes
masak nine: I wouldn't say it that strongly. that seems like a good API for a factory object. 14:12
14:12 Sqirrel_ left
nine CurtisPoe: my point being: fluent interfaces shine, when you _don't_ mutate the state. DBIx::Class resultset chaining and custom resultset methods are where it really starts to become useful. 14:13
masak unless -- is there some way to assign a lot of stuff to a factory object in a `given` block, using the usual mutator syntax, but then get back the fully constructed (immutable) object at the end?
maybe we could play around with a special `construct` keyword/slang for that? :)
masak .oO( sounds like a job for a macro! ) :P
it would be a nice party trick. the object as such is immutable, but using the `construct` keyword, you can use normal imperative code and assignment syntax to gradually build it up. 14:14
something like this: my $obj = construct MyImmutableThing { .foo = 42; .bar = "OH HAI"; .baz = computed() }; 14:15
muraiki the one time I used a fluent interface was in an openstack library for java, where it was used to mutate state and eventually call a constructor. 14:16
jnthn CurtisPoe: Quickly hacked this up: gist.github.com/jnthn/b291be5a7a18398849d4
moritz jnthn++ 14:17
masak jnthn++ 14:18
muraiki all I recall is that it was better than using the other java openstack library, but that might not have been the result of the fluent interface at all as I was ultimately using it from scala.
CurtisPoe jnthn++ :)
p6: class A { method foo() { return self } }; method bar() { return self }; A.new.foo.bar
camelia rakudo-moar 5daff4: OUTPUT«Potential difficulties:␤ Useless declaration of a has-scoped method in mainline (did you mean 'my method bar'?)␤ at /tmp/tmpfile:1␤ ------> 3{ method foo() { return self } }; method7⏏5 bar() { return self }; A.new.foo.bar␤Method 'bar…»
moritz ok, how can I profile htmlify?
CurtisPoe I don't understand that error message. 14:19
14:19 hernanGOA left
moritz I can reduce its runtime (by processing far fewer files) to 25s 14:19
nine In a language with real named arguments daisy chaining mutators for object construction doesn't buy you anything, unless you can cut the chain at any point and re-use the intermediary result to construct multiple objects from that common base.
muraiki but I agree with nine in that what is really the best is when you can chain things, not when fluent interfaces are just an alternate syntax for object construction
[Coke] moritz: heh. that was the large profile I had that broke profile. :)
moritz but the generated profile is a 145MB html file, and it's basically unusable in any browser
muraiki I guess java doesn't normally use / has support for named arguments? 14:20
moritz muraiki: it doesn't
nine muraiki: exactly, that's why people like this in Java.
muraiki ok, that explains it. maybe it's another case of "this design pattern exists because of the inflexibility of the language" not "because it's a good pattern" ;)
14:20 tokuhiro_ left
jnthn CurtisPoe: Do you understand what a method declaration and the mainline of a program are? :) 14:20
14:21 aborazmeh left
masak CurtisPoe: did you notice you put `bar` outside of the class? 14:21
14:21 hernanGOA joined
moritz fwiw mojolicious also uses it 14:21
muraiki I'm trying to reference that AI guy's critique of the GoF design patterns and how most of them don't need to exist in lisp
[Coke] moritz: did you see my patch that allows a pathological --sparse to work?
masak that is, you got the curlies right, but also wrong, at least if you want `bar` to belong to `A`.
muraiki sebastian thrun I think
[Coke] (I was doing --sparse=100 to try generating the search.js in a reasonable amount of time.)
jnthn wondres how we could actually make that one any clearer... 14:22
*wonders
moritz [Coke]: me too
jnthn Maybe "Useless declaration of has-scoped method outside of a class or role" would be better
muraiki I think where p6 shines is not in the java type of fluent interface, but in the type of chaining you can do with things like supplies, where it reads in a very declarative fashion. and where you can also branch one supply off into two different directions
jnthn But class/role are only two places you coulda declared it
muraiki which, once again, is what nine basically said. sorry fi I'm not adding anything here :)
PerlJam jnthn: highlighting that it's outside of a class/role might have helped though 14:23
14:23 _mg_ left
PerlJam maybe 14:23
jnthn PerlJam: Possibly
PerlJam initially thought that CurtisPoe put it outside of the class on purpose and really meant that the message was confusing. 14:24
jnthn Oh, as soon as I saw the message I figured "oh, I bet mis-nested curlies" :)
I guess that's experience... :)
masak I wanted to confirm whether CurtisPoe did it deliberately or not.
CurtisPoe Bah. I'm an idiot :) 14:25
masak we could maybe tailor the error message to say which class the method was outside of :P
jnthn I...what? :P
masak but may not be worth the extra trouble
dalek p: 4ab6d1f | (Pawel Murias)++ | src/vm/js/nqp-runtime/core.js:
[js] Make nqp::istype work on undefined
p: 0963c2d | (Pawel Murias)++ | src/vm/js/nqp-runtime/core.js:
[js] Make nqp::istype work on arrays
p: f895360 | (Pawel Murias)++ | tools/build/process-qregex-tests:
[js] add rx_captures and rx_qcaps to js-test
moritz masak: it's outside *any* class
masak: so it could list 'em all
jnthn I just worry about errors that try to enumerate class/role/grammar and so on
masak well, it's likely to come right after a particular class.
jnthn Because it's an open set, not a clsoed one
pmurias does adding '[js]' to commit msgs make sense?
jnthn *closed
PerlJam moritz: perhaps it could provide a list ordered by proximity to the method
masak I meant if it comes after some class. then we could give a better error, maybe. 14:26
no need to get more fancy than that.
so it *only* triggers if the class just closed.
even that may be too heavyweight.
PerlJam sounds highly specific. 14:27
masak but I'm pretty sure it would be the kind of awesome error message that Perl 6 is known for.
PerlJam "oops, the hint didn't trigger because I had an intervening statement"
masak people'd go "*whoa*, how does it know which class it's outside of!?"
PerlJam: well, think about the circumstances that led to CurtisPoe's error.
there wouldn't be a statement in between, because you wouldn't put one between methods.
this was solely the case of a premature closing brace. 14:28
PerlJam sorry, I can only imagine the circumstances and my imagination is quite big sometimes :)
masak which admittedly is much more likely in one-liners, I guess.
but still, it happens. I've done it, too.
itz Damn braces: Bless relaxes.
masak saying "it's outside any class" is premature generalization in this case. 14:29
PerlJam yeah, if you just listed the classes declared in the compilation unit that would be a huge hint. 14:30
PerlJam waits patiently for masak to implement his idea ;)
masak heh. 14:31
CurtisPoe I get a parsing error if I try to chain those method calls over multiple lines: gist.github.com/Ovid/8df7c963ead6d4f6f20e
14:31 telex left
masak CurtisPoe: without looking, unspace. 14:31
14:32 vendethiel joined
masak m: say 42 .abs 14:32
camelia rakudo-moar 5daff4: OUTPUT«5===SORRY!5=== Error while compiling /tmp/t8coPmq3aZ␤Two terms in a row␤at /tmp/t8coPmq3aZ:1␤------> 3say 427⏏5 .abs␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ state…»
14:32 telex joined
masak m: say 42\ .abs 14:32
camelia rakudo-moar 5daff4: OUTPUT«42␤»
CurtisPoe masak: that's what I thought, but it's nice when chaining methods to put one method per line.
Is that not possible due to whitespace significance?
masak there's a slang, Slang::Tuxic.
but no, not possible in the main slang due to TTIAR. 14:33
you need that backslash at least.
jnthn given open('file') { .print: 'foo'; .close; } # note that .foo means $_.foo
[Coke] notes that the youtube video is progressing. sooooon.
CurtisPoe Ah, yes.
itz TTIAR? 14:34
oh S299
arnsholt Two Terms In A Row
PerlJam using given rather than chaingin would make the code clearer IMHO too
cdc what breaks the Perl 6 clocking mechanism ;)
itz S99 I mean
shouldn't a bot do glossary lookups? :) 14:35
masak S99:TTIAR
hm.
14:36 synbot6 joined
PerlJam S99:TTIAR 14:36
synbot6 Link: design.perl6.org/S99.html#TTIAR
PerlJam :-)
one of these days someone will have to make a less fiddly bot.
geekosaur that may require something less fiddly-inducing than irc... 14:37
masak hey, IRC is perfect, OK? :P
it was invented in 1899. hasn't needed to change since. 14:38
arnsholt Just like X11 and Unix =)
geekosaur hands masak BRC
_sri <3 fluent interfaces
itz yeah things like X11 and init never change :)
masak _sri: five trillion jQuery users can't be wrong. :) 14:39
CurtisPoe If anyone wants to improve this, it would be great: en.wikipedia.org/wiki/Fluent_interface#Perl_6 14:44
Also, we should start getting more examples on Wikipedia, if possible.
PerlJam CurtisPoe++
moritz and do a green-field rewrite of the wikipedia page about Perl 6 14:45
masak what moritz said. 14:46
masak .oO( second system wikipedia page done right! )
14:48 hernanGOA left
cdc CurtisPoe: I think "given $employee { .name('Bob'); .salary(200) }" is more Perl 6 idiomatic, and this meets the aims of fluent interface (readability) 14:48
PerlJam
.oO( deja vu )
 14:49
CurtisPoe cdc: that's true. Skips the MOP, also.
timotimo but since you return self for those things
you can my $constructed = do given $employee { ... }
14:49 hernanGOA joined
moritz wow, I just had a profile with >30% time spent in dynamic optimization 14:50
jnthn moritz: New record!
timotimo oh wow
was the program running for longer than 10ms? :) 14:51
14:51 abraxxa left
jnthn moritz: That *can* happen if you're really unlucky (like, you run just for long enough for it to kick in), but more likely it warrants attention. 14:51
CurtisPoe That's giving me the error: Too many positionals passed; expected 1 argument but got 2
moritz timotimo: 170ms or so
jnthn: fwiw it was perl6-m --profile -Ilib t/pod-htmlify.t in perl6/doc 14:52
jnthn Hmm, a test file :)
moritz might be all fine; I was just surprised
jnthn is a little too 14:53
moritz: If you look at the Routines tab, do you see that it performed OSR anywhere?
moritz jnthn: profile is here: hack.p6c.org/~moritz/profile-144232...81512.html 14:54
cdc CurtisPoe: it's because the chainymutable trait adds one method with the same name (for instance foo() for the attribute $foo) 14:56
CurtisPoe: one doesn't need methods when using given :)
tony-o Stage parse : 34.599
cdc m: my $a = class { has ($.b, $.c, $.d) is rw }.new; given $a { .b = 1; .c = 2; .d = 3 }
camelia ( no output )
moritz jnthn: no OSR; two global deopts
cdc ^^ this could be indented à la fluent interface 14:57
I mean the body of given
CurtisPoe So, better? gist.github.com/Ovid/97d7bea032c00a8ccd44 14:58
Should I replace the wikipedia example with that?
moritz 6 local deoptimizations in reify-at-least
CurtisPoe (though I find it a shame that I can't have .salary = 200 (note the lack of decimal point)
cdc CurtisPoe: yes :) This is way more accessible for newd comers 14:59
moritz CurtisPoe: use Real as a type constraint
CurtisPoe: that can be Int too
CurtisPoe Yes, and I get to enjoy floating point errors then :) (but at least everyone will understand it) 15:00
cdc CurtisPoe: you can also add .say in the "given" block
or .Str.say, maybe 15:01
nine John the Iterator. He's a sequence operator. All he ever gives us is pain... 15:02
moritz CurtisPoe: it's much harder to accidentally create a floating-point number in Perl 6 than in Perl 5
subset RatOrInt of Real where Int|Rat 15:03
cdc CurtisPoe: you could rename method Str into gist, then no need to stringify explicitely when using .say or say().
CurtisPoe cdc: thank you! I never knew that :) (er, plenty of things I never knew) 15:04
cdc you're welcome :) 15:05
moritz In unrelated news, "Let's Encrypt" is in beta stadium
should I apply for a perl6.org certificate? 15:06
15:06 rurban left
cdc moritz: it makes sense since some search engines rank up https websites 15:07
PerlJam moritz: might as well. 15:08
pink_mist couldn't hurt
it should be open to the public come november afaik, so if you can't get in the beta, it's not that long until it should be available anyway
15:10 domidumont left
itz startssl certs are free (and in browsers) if you have access to the WHOIS email addresses for the domain 15:13
timotimo i'm looking forward to let's encrypt
moritz itz: I don't; and startssl doesn't issue revocations for free
timotimo they just put their first site online with one of their certificates 15:14
moritz I mean, I'd spend money on it if necessary (there's enough left from the hack server donations), but I hope it won't be necessary
nine jnthn: there are 3 [GLR] tickets left. All above my pay grade I fear. 15:15
15:16 diana_olhovik left, tokuhiro_ joined 15:21 tokuhiro_ left
hoelzro good *, #perl6! 15:21
pmurias++ # js merge
tony-o perl6 --ll-exception -Ilib t/01_basic.t
No such method 'load' for invocant of type 'Hash'
oops
15:22 TimToady left 15:23 Sqirrel joined, TimToady joined
ugexe gotta call load on your pipe and pass in the hash 15:27
tony-o there is no hash or load method in any of the files that thing, hah 15:29
[Coke] nine++ ticket wrangling.
ugexe it is a hashish pun
15:29 zacts left
vendethiel wow, just saw rurban's cperl... :o) 15:30
itz sighs
15:31 zakharyas left 15:35 larion left, zacts joined, rurban joined
tony-o ugexe++ lol 15:37
m: my @a; @a.push({ a => "A", b => "B"}); @a.push({ a => "A", b => "B"}); @a.perl.say;
camelia rakudo-moar 5daff4: OUTPUT«[:a("A"), :b("B"), :a("A"), :b("B")]␤»
tony-o m: my @a; @a.push({ a => "A", b => "B"}); @a.push({ a => "A", b => "B"}); @a.map({.perl}).join("\n").say; 15:38
camelia rakudo-moar 5daff4: OUTPUT«:a("A")␤:b("B")␤:a("A")␤:b("B")␤»
tony-o that's unexpected
dakkar www.thenautilus.net/cgit/WebCoso-p.../LibXML.pm ← I'm loving NativeCall 15:39
www.thenautilus.net/cgit/WebCoso-p...ests/xml.t ← it even works! 15:40
ugexe m: my @a; @a.push($[ a => "A", b => "B"]); @a.push($[ a => "A", b => "B"]); @a.perl.say; 15:41
camelia rakudo-moar 5daff4: OUTPUT«[[:a("A"), :b("B")], [:a("A"), :b("B")]]␤»
15:41 AlexDaniel joined
ugexe m: my @a; my %x; %x<a> = "A"; %x<b> = "B"; @a.push($%x); @a.push($%x); @a.perl.say; 15:41
camelia rakudo-moar 5daff4: OUTPUT«[{:a("A"), :b("B")}, {:a("A"), :b("B")}]␤»
skids m: my $pipe of Hash = (role Pot { }).new; 15:42
camelia rakudo-moar 5daff4: OUTPUT«Type check failed in assignment to '$pipe'; expected 'Hash' but got 'Pot'␤ in block <unit> at /tmp/yZ6G2W3TV6:1␤␤»
psch r: my $bh = BagHash.new(<a b c>); $bh.DELETE-KEY($_) for $bh.keys # found it
finally
camelia ( no output )
..rakudo-jvm 5daff4: OUTPUT«java.util.ConcurrentModificationException␤ in block <unit> at /tmp/tmpfile:1␤␤»
psch o.o
tony-o $%() did , awesome 15:43
that got Green - the ultimate tool for parallel testing , working with current nom 15:44
[Coke] m: say 33.333333333333333333333333333333333333333333333333333+i 15:45
camelia rakudo-moar 5daff4: OUTPUT«33.3333333333333+1i␤»
[Coke] I just got linenoise tab autocompletion to work by accident. ahhhh. 15:46
so nice. 15:47
itz looks at recursive code and wishes he had paid more attention to the CS lecturer and less to the beer in the bar
psch r: my $bh = Hash.new(<a b>); $bh.DELETE-KEY($_) for $bh.keys # ...i wonder?
camelia ( no output )
psch see, *that's* just crazy talk 15:48
dalek rl6-roast-data: 70f79f4 | coke++ | / (9 files):
today (automated commit)
 15:49
psch especially 'cause BagHash.DELETE-KEY calls Hash.DELETE-KEY...
15:51 fasfsehwrh joined
psch (actually, Baggy.DELETE-KEY) 15:51
[Coke] Perl 6 : a sufficinently smart(ass) compiler. 15:53
jnthn itz: Recursion is like you start drinking a beer, but half way through drinking it you have to go to buy another one, and then you start drinking that, and so it goes on...and eventually you get fed up, and finish the beers in reverse order of purchase. It sucks 'cus by the end you're drinking warm beer, just like recursion sucks for the CPU 'cus then the data it needs is no longer in the L1 cache...
arnsholt Trust jnthn to create an apt beer-analogy =D 15:54
itz jnthn++
TimToady warm is good for a cache
pink_mist stale then :P
15:54 patrickz left
TimToady we need something to pull/skip elements from a Seq without relying on subscripting, which caches gratuitously 15:55
15:55 fasfsehwrh left, andreoss joined
jnthn TimToady: Something that gracefully degrades to Linq's .Take and .Skip would be nice :) 15:57
TimToady except take is, er, taken
jnthn Well, or is a generalization of them 15:58
PerlJam m: (gather {} but role {})[0]
camelia ( no output )
jnthn Well, yeah, I'm expecting you to come up with a brilliant single mechanism that subsumes both of them :P
TimToady well, .neck($n) that returns a head and a tail, cut at $n
maybe that's a bit too...picturesque... 15:59
[Coke] pmurias, hoelzro: github.com/coke/perl6-roast-data/b...y.out#L130 - error on JS build, not sure how long it's been going on. probably just on hack.p6c.org 16:00
hoelzro [Coke]: which version of node does hack have? 16:01
ah, 0.10.29
hoelzro guesses from github.com/coke/perl6-roast-data/b...y.out#L144
PerlJam m: my @a will begin { say $_.WHAT } 16:02
camelia rakudo-moar 5daff4: OUTPUT«(Array)␤»
16:02 pmurias left
dalek rl6-roast-data: 18ec5b8 | coke++ | bin/nqp.js.sh:
js compiler was merged to master
 16:03
16:05 [Sno] left 16:06 larion joined 16:07 araujo left
arnsholt itz: Anyways, recursion is one of those things that are simpler than they seem initially, once you squint at it from the correct angle 16:07
vendethiel arnsholt: "apt beer" hey, I'd love a apt-get beer :P 16:09
moritz apt-beer add $url_to_pub 16:10
dalek ecs: bab09aa | ugexe++ | S22-package-format.pod:
Update CompUnitRepo class specification seperator
 16:13
16:16 araujo joined
ugexe beer is a FIFO queue. liquor is often more a LIFO queue 16:16
16:18 tokuhiro_ joined, dwarring left
jnthn I didn't know it was possible to drink so much you forget LIFO is a stack, not a queue... :) 16:20
16:22 tokuhiro_ left
cognominal "to iterate is human, to recurse is divine". I infer that god has an infinite L1 cache. 16:23
moritz cognominal: or is very good at TCO 16:25
TimToady or just very patient
dalek pan style="color: #395be5">perl6-examples: bcc9506 | (Steve Mynott)++ | / (3 files):
[99probs] GLR fix for P13-viklund.pl
 16:27
[Coke] the Swiss Perl Workshop 2015 Q&A video with Larry has been released! Thanks to Roman esp. for getting this put together and posted: www.youtube.com/watch?v=zmNInkzaYL...e=youtu.be
PerlJam Coke++ 16:32
16:33 leont joined
[Coke] tweeted and FB'd it as well, please share and enjoy. 16:33
16:39 telex left
moritz how fast is 'last' in a while loop? 16:40
16:40 itz left
moritz I ask because in a profile of htmlify, push-exactly is the one routine using up most of the exclusive time 16:40
and it uses a 'last', and that could be optimized out
while $i < $n {
$pulled := self.pull-one();
last if $pulled =:= IterationEnd;
could become 16:41
16:41 telex joined, telex left, larion left
moritz while $i < $n && ! IterationEnd =:= ( $pulled := self.pull-one) { ... } 16:41
jnthn moritz: Not so fast, more because of `last` itself than the while loop 16:42
16:42 telex joined
moritz jnthn: ok, I'll try the optimization 16:42
16:44 [Sno] joined
timotimo [Coke]: you are wearing a LRR shirt! i didn't even notice when i was at the SPW! 16:46
damn, and right now i have to go 16:49
16:49 TimToady left
tony-o is require doing weird stuff for non-core modules? perl6 -e 'require Term::ANSIColor;' is resulting in gist.github.com/tony-o/41203dfaaabca48547af 16:50
16:50 CQ left
ugexe m: say "\b"..."\t" 16:51
camelia rakudo-moar 5daff4: OUTPUT«( )␤»
16:51 TimToady joined
ugexe ah, on my terminal its just ' )' 16:51
moritz huh, there's both Term::ANSIColor and Terminal::ANSIColor in the ecosystem
tony-o: looks like a bug to me 16:52
m: require Term::ANSIColor;
camelia rakudo-moar 5daff4: OUTPUT«Could not find Term::ANSIColor in any of:␤ file#/home/camelia/.perl6/2015.07.1-942-g5daff4c/lib␤ inst#/home/camelia/.perl6/2015.07.1-942-g5daff4c␤ file#/home/camelia/rakudo-inst-1/share/perl6/lib␤ file#/home/camelia/rakudo-inst-1/share/perl6/ve…»
moritz m: require Test;
camelia ( no output )
tony-o i can only reproduce on 3rd party modules 16:53
[Coke] timotimo: I love me some LRR.
ugexe must be the update to use CUR to load modules
[Coke] I am going to mail Paul and tell him. He might dig it.
tony-o i'll dig into the CUR 16:54
after this conference call
moritz just had the idea for the first time to use 'perl6 -c' on a core file
16:57 dakkar left
moritz did everybody else already use that? 16:57
ugexe .candidates returns a hash
17:06 Andrie joined
timotimo coke, don't install Linux on the toilet!! 17:10
17:11 diana_olhovik_ joined
psch r: my %h = a => a => 1, b => b => 2; %h.DELETE-KEY($_) for %h.values.map: { .keys } 17:13
17:13 leont left
camelia ( no output ) 17:13
..rakudo-jvm 5daff4: OUTPUT«java.util.ConcurrentModificationException␤ in block <unit> at /tmp/tmpfile:1␤␤»
psch *that*'s what it comes down to...
dalek kudo/nom: 7662c46 | moritz++ | src/core/Iterator.pm:
Try to optimize Iterator.push-exactly not to use "last"

in my benchmarks, it is mostly drowned out in noise though		 17:20
TimToady m: for $[1,2,3] { .say }; .say for $[1,2,3];
camelia rakudo-moar 5daff4: OUTPUT«[1 2 3]␤1␤2␤3␤»
17:20 patrickz joined
TimToady nine: ^^^ you missed one :) 17:21
moritz ok, this routine is the one with the second-highest exlusive (!) time in the htmlify.p6 profile: github.com/perl6/doc/blob/master/h...fy.p6#L651 17:23
but, all it does is calling other routines 17:24
it doesn't really do any work itself
how can it use up so much exclusive time?
vendethiel moritz: the grep, etc, are considered as taking their own time? 17:28
..maybe spurt is *really* slow on your system :P
moritz: "return if $name ~~ / '/' /;" <- what's wrong with eq? 17:30
that's stupid a question, disregard it
PerlJam vendethiel: how about .... What's wrong with index()? :) 17:33
psch r: my %h = a => 1, b => 2; for %h { %h{$_.key.uc} = $_.value ** 2 }; say %h.perl # i am starting to see how it happens \o/ 17:34
camelia rakudo-jvm 5daff4: OUTPUT«java.util.ConcurrentModificationException␤ in block <unit> at /tmp/tmpfile:1␤␤»
..rakudo-moar 5daff4: OUTPUT«{:A(1), :B(4), :a(1), :b(2)}␤»
vendethiel PerlJam: sure! but "defined $name.index('/');" is starting to get long
17:34 ShimmerFairy left
psch unfortunately i can only think of sticking the VMIterInstance into the VMHashInstance so they know they belong together 17:34
i guess i'll have to take a closer look at how moar handles that...
dalek ast: 3913af2 | TimToady++ | S0 (2 files):
check that for modifier respects item
 17:37
17:39 andreoss left 17:40 diana_olhovik_ left 17:41 CurtisPoe left 17:46 domidumont joined 17:47 ShimmerFairy joined 17:50 rurban left 17:51 spider-mario joined
vendethiel TimToady: that's one I hadn't seen 17:55
m: $[1, 2, 3].map: { .perl.say }
camelia rakudo-moar 5daff4: OUTPUT«1␤2␤3␤»
vendethiel TimToady: that breaks the one-arg rule, and the "map=for", no?
17:55 _mg_ joined 18:02 Sqirrel left, Andrie left, Sqirrel joined 18:03 Andrie joined, Andrie left 18:04 Andrie joined, Andrie left, Andrie joined 18:05 xinming left 18:06 telex left 18:07 telex joined, Andrie left, xinming joined
moritz vendethiel: spurt is its own routine too, and thus doesn't count for the exclusive time 18:13
vendethiel: and regex search for a constant substring mostly optimizes to an index() call anyway 18:14
TimToady vendethiel: in GLRR, methods don't pay attemtion to $
only functions 18:15
well, and 'for'...
skids Well, $ on invocants does not matter. On additional method args it does. 18:16
vendethiel mmh, ok. 18:18
so for = map, not .map
18:19 telex left, tokuhiro_ joined
skids (You can think of it as the $ being used up calling the method. And after all, $foo.a where foo has @.a is technically a violation of sigil invarience but we live with it.) 18:19
18:19 telex joined 18:24 tokuhiro_ left
patrickz .tell tadzik you can merge the PR now. Did some more testing and fixing and am confident this is loads better than before. 18:26
yoleaux patrickz: I'll pass your message to tadzik.
18:27 baest_ joined 18:28 baest left 18:39 meisl joined, meisl left 18:40 domidumont left
dalek c: 23f2a08 | moritz++ | lib/Perl6/Documentable/Registry.pm:
Htmlify: cache get-kinds
 18:45
18:50 cognominal left 18:53 _mg_ left, telex left, telex joined 18:56 domidumont joined 19:01 telex left 19:02 telex joined 19:23 domidumont left 19:28 xfix joined 19:34 _mg_ joined 19:35 FROGGS joined
FROGGS hello dear sir or madam 19:36
vendethiel \o FROGGS 19:38
hoelzro o/ FROGGS
FROGGS o/
19:40 Sqirrel_ joined, Sqirrel left 19:48 _mg_ left 19:50 diana_olhovik_ joined 19:51 CurtisPoe joined
CurtisPoe My client died. I don't know if my last question went through or not! Did I just ask a question about "is"? :) 19:51
timotimo o/ 19:53
CurtisPoe Did I just get warnocked on asking if I asked a question? :) 19:54
CurtisPoe guesses that most are in bed or at a pub, as they should be. 19:55
timotimo you did not ask a question about is before asking about asking a question about is
hoelzro o/ CurtisPoe
CurtisPoe In that case: I know that using "isa" for inheritance is a contentious issue, so rather than discuss a sore subject, is there some place I can read up on the reasoning for this?
masak can I try to make your question a bit more precise? 19:56
timotimo i'm not sure what you mean exactly?
masak are you asking "why do people dislike class inheritance more and more nowadays?" ?
timotimo you mean using the keyword "isa" to say "this class derives from that class"?
CurtisPoe Basically, we use "is" for inheritance and "is" for traits. Traditionally, many languages use "isa" for inheritance. I find overloading the meaning to be confusing (is that a base class or a trait?), so I wanted to understand why this was done. 19:57
masak aha. 19:58
timotimo ah, yes, i remember at least two discussions about that thing
masak basically it's done because the advantages of deliberately confusing those two things into `is` makes for a language that's easier to learn, at the expense of a little bit of flexibility/clarity later. 19:59
uhm. that sentence almost made sense. :)
what I meant to say was: confusing traits and inheritance into `is` makes for an easier-to-grok Perl 6.
20:00 telex left, telex joined
masak you don't have to worry about the difference when you're starting out 20:02
20:03 AlexDaniel left
timotimo also, having classes start with upper case and traits with lower case makes it less dangerous in actual usage 20:03
dalek kudo-star-daily: 45a250a | coke++ | log/ (2 files):
today (automated commit)
 20:04
timotimo i wonder if - at use-time of a module - we should check for name collisions between traits and classes you're importing
masak it could even be (post-)rationalized by saying that in this case, class inheritance gets to ride on the trait mechanics.
timotimo: yes, that might be a good idea.
timotimo that's how i imagine it, too, masak
[Coke] timotimo: Paul says he's not got a lot of perl experience, but that it was awesome to see the LRR shirt in the wild. :)
timotimo will watch the Q&A to the end now
sweet, [Coke]! :) 20:05
PerlJam This has come up enough (is v isa), that maybe there should be an entry in the FAQ for it
timotimo and of course you can write a slang to "fix" that :)
masak timotimo: not that for that to happen in the first place, either a trait must be upper-cased, or a class must be lower-cased. which is already a nonstandard choice by the programmer.
timotimo right 20:06
masak the fact that class inheritance is riding on the traits mechanism can even be motivated by saying that as a language, we care a lot more about traits than we do about class inheritance. 20:08
20:08 patrickz left, telex left 20:09 telex joined 20:10 xfix left
gfldex r: say sprintf('%f', 1, 2); # shouldn't there be a line number with that error message? 20:11
camelia rakudo-{moar,jvm} 7662c4: OUTPUT«Directives specify 1 argument, but 2 arguments were supplied␤␤»
PerlJam huh 20:13
looks like that's an NQP message
ugexe sprintf is nqp 20:15
TimToady Not Quite Presentable
jnthn CurtisPoe: It's probably worth noting that in Perl 6, a trait is a compile-time call to a multi sub, and the effects of that are whatever it fancies. I have the impression Moose et al has them mean something other than that (more tied up with roles, perhaps).
20:15 telex left 20:16 telex joined 20:20 tokuhiro_ joined 20:22 telex left, telex joined 20:25 tokuhiro_ left
timotimo ohey jnthn :) 20:26
20:26 muraiki left, darutoko left
CurtisPoe In perl5, a pragma is a package (sort of a class) and is lower-cased. Will that happen in Perl 6 and clash with traits? And will case-insensitive file systems be an issue? 20:28
timotimo for one, most installed modules get registered in a database 20:29
so that even if the filesystem is fat16 and only allows 8.3 we can have a module called ψεπψαιστοσ
it'll just get saved into "12345_hephaistos.pm6" or something
20:29 Andrie joined 20:30 telex left 20:31 araujo left
CurtisPoe So at trait can't be applied to a class? 20:31
timotimo traits can be applied to classes 20:32
m: class Foo is rw { has $.a; has $.b }; my $a = Foo.new; $a.a = 10; $a.b = 99; say $a.perl
camelia rakudo-moar 7662c4: OUTPUT«Foo.new(a => 10, b => 99)␤»
CurtisPoe So what's the difference between traits and roles?
20:32 telex joined, nowan_ is now known as nowan
timotimo when you "is Role", you'll just mix in the role 20:32
20:32 Alina-malina left
timotimo when you "is some-trait", the trait can do whatever it wants to the class 20:32
CurtisPoe So a trait applied to a class merely applies that trait to all attributes?
timotimo no, that's only a thing "is rw" does
PerlJam CurtisPoe: after you get all of your answers, you need to blog about it :) 20:33
TimToady you don't generally 'is' roles
timotimo give me a second
oh!
right, of course
you generally "is SomeClass"
20:33 Andrie left
timotimo so it's not "a trait can't be applied to classes", but "a role is usually not combined with a class using the 'is' keyword" 20:33
CurtisPoe PerlJam: I feel very uncomfortable blogging about something in Perl 6 when it looks *wrong* to me. I love the language and right now, I don't want to scare people.
timotimo best forget about "is SomeRole" altogether :) 20:34
PerlJam CurtisPoe: good point. I retract my suggestion to blog about it ;)
20:34 Peter_R joined
TimToady roles are a very specific and controlled composition mechanism; traits are for cheating in any other way 20:35
we just treat inheritance as a form of cheating :)
timotimo many traits that are already built into perl6 so far are actually implemented by mixing a role into a class or into an attribute or something like that 20:36
CurtisPoe Experienced programmers generally know that it's OK to overload behavior, but overloading semantics is a big, fat, red flag. It seems like the semantics of "is" are very heavily overloaded, so many experienced programmers are going to have an issue with this.
Consider Perl 5 pseudo-hashes: years of overloading the semantics of $array[0] if it contained a hashref. Many mysterious error messages.
Or consider how many C programmers would have an account_total() subroutine return a negative value on error, until negative values were suddenly allowed. Again, overloading the *semantics* shot them in the foot. 20:37
20:38 telex left
CurtisPoe Thus, this looks like overloading the semantics of "is" and I am very, very confused about why this is the way it is. That's why I was hoping someone could point me to docs rather than possibly risk irritating people again :) 20:38
20:38 telex joined
TimToady the issue really almost never comes up in practice because traits tend to be lowercase adjectives, and classes tend to be uppercase nouns 20:39
timotimo to give us a common base again: "is" is used both for applying traits and for deriving from classes; "does" is for applying roles
CurtisPoe (n.b. as mentioned previously, overloading behavior is fine. That's why we have polymorphism)
20:39 _mg_ joined
CurtisPoe So what if you're writing your code in a language which doesn't distinguish between upper and lower case? 20:39
20:40 _mg_ left
timotimo oh, you mean a natural language with a script that doesn't have lower vs upper case? 20:40
i imagine japanese, chinese writing systems don't have something like that, and neither does hangul? 20:41
TimToady then you're still not as bad as all the commonly used languages that throw all words into the same namespace 20:42
20:42 espadrine left
CurtisPoe p6: sub ラン { say "Hi!" }; ラン() 20:42
camelia rakudo-moar 7662c4: OUTPUT«Hi!␤»
timotimo turned on the auto-generated captions for the Q&A, chuckles a bit, then turns it off again 20:43
CurtisPoe I don't mean to be rude, but is the official position of Perl 6 "if you use a language that doesn't have a concept of case, too bad"? 20:44
timotimo definitely not
CurtisPoe s/language/writing system/
timotimo you are not forced to use upper vs lower case for classes vs traits at all
m: class lowercaseclass { }; my $a = lowercaseclass.new; say $a.perl
camelia rakudo-moar 7662c4: OUTPUT«lowercaseclass.new␤»
TimToady maybe if you use a language that doesn't distinguish adjectives from nouns 20:45
20:45 telex left
timotimo if your language is Oog!, you'll run into a tiny bit of trouble 20:45
CurtisPoe But that means one developer writes a trait named ラン and another one writes a class named ラン and their code breaks.
timotimo because we don't allow whitespace in these names
PerlJam TimToady: and you know programmers are a strange lot ... they may adjectivify nouns or noun adjectives or something too.
20:45 telex joined
timotimo well, that one developer would have to import that trait into the same piece of code that also uses the class 20:45
CurtisPoe Or a language which has adjectives which have the same meaning as nouns which have the same meaning as verbs :) 20:46
timotimo based on your insights, i've suggested we make a compile-time warning or error available that complains when traits and classes collide like that
TimToady it's all declaration time, anyway, and you have to know the names that have been declared, so it's not a run-time confusion
timotimo so the code won't break in a surprising way
CurtisPoe timotimo++
timotimo we should RT that
i'm not going to build that right now
CurtisPoe I'm OK if it's a warning and not a compile time failure. 20:47
It's rare enough that it seems like a reasonable compromise.
CurtisPoe is sorry for beating a half-dead horse.
timotimo no worries :) 20:48
TimToady well, it might be a half-dead unicorn...
20:48 diana_olhovik_ left 20:50 liztormato joined
CurtisPoe Or a one-winged butterfly ... :) 20:50
20:50 zacts left 20:51 diana_olhovik joined
TimToady m: class Foo :rw { has $.foo }; 20:51
camelia rakudo-moar 7662c4: OUTPUT«===SORRY!===␤Cannot invoke this object␤»
PerlJam Also note that you could create a slang to distinguish trait application and class inheritance.
TimToady we also have this syntax in reserve for disambiguation 20:52
since types take precedence over traits
20:55 [Sno] left
timotimo that looks like how you declare inheritance in cpp 20:55
20:57 kaare_ left 20:58 diana_olhovik left, zacts joined 21:03 skids left
jdv79 i'm lost but cool. CurtisPoe++ for entertainment value. 21:04
21:05 lizmat joined, liztormato left 21:08 cognominal joined 21:15 dha joined
dalek ast: e788818 | usev6++ | S14-roles/mixin.t:
Add tests for RT #119925
 21:15
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=119925
lizmat good *, #perl6!
yoleaux 09:36Z <jnthn> lizmat: if you're reliably seeing S17-supply/start.t passing now, then maybe rt.perl.org/Ticket/Display.html?id=125460 can be closed... I've not seen it fail since my recent fixes in Moar.
11:04Z <jnthn> lizmat: I think the .push/.pop error improvements in List have busted make test (54-lib.t)
bartolin o/ lizmat
lizmat hmmm... I did spectest the changes to List.push etc. not the "make test" :-( 21:16
bartolin o/
jdv79 lizmat: o/
home yet?
lizmat jdv79 o/
nope... in Erlangen at the moment
tomorrow we'll be driving home... 21:17
bartolin lizmat: I created a PR to fix 54-use-lib.t
21:17 rindolf left
dalek kudo/nom: 1e58fc6 | usev6++ | t/01-sanity/54-use-lib.t:
Adjust tests to improved error reporting

After commit 86ddca2003 the tests in question die with X::Immutable		 21:18
kudo/nom: 38c94bd | lizmat++ | t/01-sanity/54-use-lib.t:
Merge pull request #533 from usev6/x_immutable

Adjust tests to improved error reporting
lizmat hmmm... essentially the tests I added to roast :-)
bartolin :-)
jdv79 mmm.. franconia. not far from Shlenkerla and the like 21:19
vendethiel CurtisPoe++ as well. That was a fun ride ('cept not a unicorn one)
dha Do while loops still allow for a LABEL, and, if so, do we want to make a point of documenting that?
yoleaux 03:13Z <[Coke]> dha: looks like search is working again
dha \o/
Also, should the section in control.pod on return/while and return/until be after C<while> and C<until>'s documentation? 21:20
jdv79 i'll be in munich tomorrow for the fest 21:23
timotimo the münchenfest?
jdv79 yes, i've never been 21:26
timotimo oktoberfest*
jdv79 w/e
timotimo i heard there's lots and lots and lots of passed out drunks on all the streets 21:27
jdv79 i hope there's enough room to step between the then 21:28
*them
dalek ast: 5f01fa1 | usev6++ | S02-literals/listquote.t:
Fix test which dies with X:Immutable now
timotimo i hope you'll have a proper place to sleep instead
jdv79 i'll take pictures
i have booked a hostel 21:29
21:29 [Sno] joined
masak dha: yes, `while` loops allow for a label. and `next`/`redo`/`last` work in them, with and without labels. 21:35
m: L1: while False {}; while True { last L1 }; say "alive"
camelia rakudo-moar 7662c4: OUTPUT«No exception handler located for last_label␤␤»
masak that error message could probably be improved :) 21:36
21:36 colomon left
dha Great. In the meantime I'm going to write general while/until docs, and leave a TODO for labels. 21:36
masak I know it's a lexotic thing, but we can see L1 from that `last`
21:37 colomon joined, telex left, telex joined
dha ok, basic while/until docs. gist.github.com/dha/0a0736b9da74a8e61e49 21:40
tony-o just wrote an article discussing some of the higher level problems with parallel testing that Green addresses, and how it addresses them for anyone interested. j.mp/parallel-testing 21:47
dalek ast: 6a475d4 | lizmat++ | S02-types/hash.t:
Adapt error message checking to new message
 21:49
21:51 rurban joined
dalek osystem: 9cc7454 | tony-o++ | META.list:
Green, a parallel testing option

Check it out.		 21:53
21:54 camelia left 21:55 camelia joined 21:57 ChanServ sets mode: +v camelia
grondilu m: multi infix:<[+]>(Range $r) { .Int given ($r.min + $r.max)*($r.max - $r.min + 1) }; say [+] 1 .. 10; 21:58
camelia rakudo-moar 38c94b: OUTPUT«55␤»
grondilu oops, meant prefix:<[+]>, but that can't work apparently. 21:59
m: multi prefix:<[+]>(Range $r) { .Int given ($r.min + $r.max)*($r.max - $r.min + 1) }; say [+] 1 .. 10;
camelia rakudo-moar 38c94b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/6_XUTILSxu␤Calling prefix:<[+]>(int) will never work with any of these multi signatures:␤ (Range $r)␤at /tmp/6_XUTILSxu:1␤------> 3 + $r.max)*($r.max - $r.min + 1) }; say 7⏏5[+] 1 .. 10;␤»
grondilu (also, forgot to divide by two. Nevermind) 22:00
flussence m: multi prefix:<[+]>(Range $r) is looser(&infix:<..>) { .Int given ($r.min + $r.max)*($r.max - $r.min + 1) }; say [+] 1 .. 10 22:01
camelia rakudo-moar 38c94b: OUTPUT«110␤»
grondilu oh cool
flussence wait, that doesn't seem right...
grondilu as I wrote, I forgot to divide by two
m: multi prefix:<[+]>(Range $r) is looser(&infix:<..>) { .Int given ($r.min + $r.max)*($r.max - $r.min + 1) div 2 }; say [+] 1 .. 10 22:02
camelia rakudo-moar 38c94b: OUTPUT«55␤»
flussence there we go :)
grondilu I was wondering if that's the kind of optimization mentionned in ROADMAP:
3 ** Optimizing multis for `[+] 1..10` and `[<=] 1..10` etc.
TimToady well, that's gonna be a bit of a problem
dalek kudo/nom: 425845d | lizmat++ | src/core/Whatever.pm:
Make Whetever.new == Inf True

It occurred to me that we have many places in the core setting where we are checking for whether a value is either Whatever or Inf. It also occurred to me that adding a Numeric method for Whatever would make such a check much easier, because it would just allow you to say:
   sub a($value) { say "foo" if $value == Inf }
and pass either a(*) or a(Inf) would give the desired effect. Before applying this new capability, I thought I'd get it out in the world first.
TimToady m: multi prefix:<[+]>(Range $r) is looser(&infix:<..>) { .Int given ($r.min + $r.max)*($r.max - $r.min + 1) div 2 }; say [+] 1,2,3,4,5,6,7,8,9,10
camelia rakudo-moar 38c94b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/wBvmcAgLkG␤Calling prefix:<[+]>(int) will never work with any of these multi signatures:␤ (Range $r)␤at /tmp/wBvmcAgLkG:1␤------> 3max)*($r.max - $r.min + 1) div 2 }; say 7⏏5[+] 1,2,3,4,5,6,7,8,…»
tony-o TimToady: you in the city on october?
s/on/in/
TimToady yup
grondilu m: say (1, 2, 3).WHAT
camelia rakudo-moar 38c94b: OUTPUT«(List)␤» 22:04
grondilu ^it's not a RAnge
tony-o i may drive down for that, lots of good riding plus i can heckle you
TimToady mebbe I should figure out what a "Release Talk" should say
tony-o how long is that session planned for? 22:05
TimToady actual talk is 1 hour, but social stuff around it
lizmat wonders what city this city is 22:06
grondilu m: multi prefix:<[+]>(Range $r) is looser(&infix:<..>) { .Int given ($r.min + $r.max)*($r.max - $r.min + 1) div 2 }; multi prefix:<[+]>($l) { reduce * + *, $l }; say [+] 1,2,3,4,5,6,7,8,9,10
camelia rakudo-moar 38c94b: OUTPUT«12345678910␤»
tony-o san francisco lizmat
lizmat ah, that city :-)
22:06 hernanGOA left
grondilu m: multi prefix:<[+]>(Range $r) is looser(&infix:<..>) { .Int given ($r.min + $r.max)*($r.max - $r.min + 1) div 2 }; multi prefix:<[+]>($l) is looser(&infix:<,>) { reduce * + *, $l }; say [+] 1,2,3,4,5,6,7,8,9,10 22:06
camelia rakudo-moar 38c94b: OUTPUT«12345678910␤»
cognominal tales of the city... 22:07
grondilu m: multi prefix:<[+]>(Range $r) is looser(&infix:<..>) { .Int given ($r.min + $r.max)*($r.max - $r.min + 1) div 2 }; multi prefix:<[+]>(*@l) is looser(&infix:<,>) { reduce * + *, @l }; say [+] 1,2,3,4,5,6,7,8,9,10
camelia rakudo-moar 38c94b: OUTPUT«12345678910␤»
grondilu gives up
tony-o i've lived in this area too long, rarely ever hear it called san francisco
cognominal en.wikipedia.org/wiki/Tales_of_the_City
tony-o cognominal: nice 22:08
cognominal I am not from SF but I have read them all.
well not all, apparently there are new ones. 22:09
grondilu m: say .elems given 1, [\*] 1..*
camelia rakudo-moar 38c94b: OUTPUT«2␤»
grondilu hum, the usual expression for factorial is not GLR-friendly 22:10
m: say .elems given 1, |[\*] 1..*
camelia rakudo-moar 38c94b: OUTPUT«Cannot .elems a lazy list␤ in block <unit> at /tmp/OHZv3bAwT3:1␤␤Actually thrown at:␤ in block <unit> at /tmp/OHZv3bAwT3:1␤␤»
grondilu m: say .elems given my @ = 1, |[\*] 1..*
camelia rakudo-moar 38c94b: OUTPUT«Cannot .elems a lazy list␤ in block <unit> at /tmp/uV4mpu4rPt:1␤␤Actually thrown at:␤ in block <unit> at /tmp/uV4mpu4rPt:1␤␤»
22:12 vendethiel left 22:15 bbkr joined, bbkr_ left
dalek kudo/nom: b3a7722 | lizmat++ | src/core/IO/Handle.pm:
:eager is a noop for IO::Handle.lines now
 22:16
22:16 dha left 22:18 Skarsnik left
grondilu m: constant fact = 1, [\*] 1 .. *; say fact[6] 22:19
camelia rakudo-moar 38c94b: OUTPUT«Nil␤»
grondilu m: constant fact = 1, |[\*] 1 .. *; say fact[6]
camelia rakudo-moar 38c94b: OUTPUT«720␤»
masak 'night, #perl6 22:20
TimToady m: constant fact = [\*] 1, 1, 2, 3 ... *; say fact[6] 22:24
camelia rakudo-moar 38c94b: OUTPUT«720␤»
22:26 TEttinger joined, FROGGS left
grondilu TimToady: I'm surprised infix:<...> finds a logical sequence here 22:28
22:29 Alina-malina joined
grondilu m: constant fact = [\*] 1, 1, 2 ... *; say fact[6] 22:29
camelia rakudo-moar 425845: OUTPUT«5===SORRY!5=== Error while compiling /tmp/VjhafoIf7j␤An exception occurred while evaluating a constant␤at /tmp/VjhafoIf7j:1␤Exception details:␤ Unable to deduce arithmetic or geometric sequence from 1,1,2 (or did you really mean '..'?)␤ i…»
grondilu has no idea how infix:<...> works 22:30
TimToady it considers only the final 3
grondilu ok
it's a better expression then, but it looks a bit magical
TimToady probably slower too
tony-o have you ever been heckled TimToady ? 22:33
TimToady certainly
though usually not maliciously 22:34
timotimo so only by accident?
TimToady "Never attribute to malice..." 22:35
timotimo This repr (malice) doesn't support attribute access 22:38
22:42 skids joined 22:44 cognominal left
lizmat sleep& 22:49
22:52 LordVorp joined
psch hrm 22:54
22:54 telex left
psch after a few hour break and some looking at moarvm it seems i have to understand the uthash code to figure out how they prevent breaking on changing-while-iterating 22:55
i don't *really* feel up for that, neither in the short- nor mid-term :P
i mean, i'm focusing on the jvm backend for a reason...
22:55 telex joined
timotimo because it's so simple 22:56
22:56 kanishka joined
psch ...is that sarcasm? 22:56
timotimo i think so
psch :P
22:57 bbkr_ joined
psch i'm a bit weary of just stuffing bits onto VMHashInstance that (probably) wouldn't conform with any sixmodel interface 22:57
but that's the only way i can think of to make this work out
as in, in the end it could be as easy as telling every VMHashInstance about all VMIterInstances that exist that access it 22:58
22:58 LordVorp left
psch but 1) that's *probably* not particularly 6model-y 2) it might be noticeable overhead 22:58
i'm guessing here, but uthash probably was chosen because it allows (let's call it) async access to hash elements 22:59
and the jvm doesn't allow that, which is why we can't iterate over a hash while changing it
or rather, the other way around - change it while iterating over it
...maybe moar is wrong here? (one can hope...) 23:00
timotimo doesn't know uthash does anything to prevent trouble when accessing concurrently
23:00 bbkr left
timotimo i expect it'd break spectacularly if you use it non-well 23:00
psch r: my %h = a => 1, b => 2; for %h { %h{$_.key.uc} = $.value * 2 }; say %h.perl
camelia rakudo-{moar,jvm} b3a772: OUTPUT«5===SORRY!5=== Error while compiling /tmp/tmpfile␤Variable $.value used where no 'self' is available␤at /tmp/tmpfile:1␤------> 3b => 2; for %h { %h{$_.key.uc} = $.value7⏏5 * 2 }; say %h.perl␤ expecting any of:␤ term␤»
psch r: my %h = a => 1, b => 2; for %h { %h{$_.key.uc} = $_.value * 2 }; say %h.perl 23:01
camelia rakudo-moar b3a772: OUTPUT«{:A(2), :B(4), :a(1), :b(2)}␤»
..rakudo-jvm b3a772: OUTPUT«java.util.ConcurrentModificationException␤ in block <unit> at /tmp/tmpfile:1␤␤»
timotimo oh, modification while iterating
not modification from multiple threads
psch yeah
timotimo yeah, no clue. but that can certainly be done
psch i agree 23:02
timotimo it's just that you may end up skipping things
like you skipped A and B there :P
23:02 telex left
psch but on jvm it means every VMHashInstance has to know all the VMIterInstances that iterate over it 23:02
timotimo otherwise it would have been :A(4) :B(8)
psch ...or we copy
huh
maybe that works 23:03
instead of creating an iterator for the HashMap we actually iterate over, we create it for a clone of that HashMap
which should give the same semantics
23:04 telex joined
psch well, it seems to build nqp fine at least :) 23:06
if this works i might fix 3+ test files with 7 chars :P
which would somewhere between 50 and 400 tests vOv 23:07
timotimo ".clone" is 6 chars
hmm 23:08
psch oh, 8 chars then
'cause java, which means explicit ()
timotimo ah
of course
psch the reason for the range of tests is that failures in e.g. S02-types/bag.t abort the rest of the tests
but not all tests that are aborted suffer from this bug 23:09
timotimo right, that's often problematic
i wonder why this is A Thing now and wasn't before?
psch probably because the Seq code changed something
timotimo did we grab the full list of keys or pairs before?
psch star-j: say "hi"
camelia star-j 2015.03: OUTPUT«Error occurred during initialization of VM␤Could not reserve enough space for object heap␤Error: Could not create the Java Virtual Machine.␤Error: A fatal exception has occurred. Program will exit.␤»
psch :/
yeah, glr code boils down to this: 23:10
r: my %h = a => a => 1, b => b => 2; %h.DELETE-KEY{$_} for %h.values.map: { .keys }
camelia rakudo-{moar,jvm} b3a772: OUTPUT«Cannot call DELETE-KEY(Hash); none of these signatures match:␤ (Any:U $:, *%_)␤ (Any:D $:, *%_)␤ (Hash:U $: *%_)␤ (Hash $: \key, *%_)␤ (Hash $: \key, :$SINK!, *%_)␤ in block <unit> at /tmp/tmpfile:1␤␤»
psch err
r: my %h = a => a => 1, b => b => 2; %h.DELETE-KEY($_) for %h.values.map: { .keys } # this
camelia ( no output )
..rakudo-jvm b3a772: OUTPUT«java.util.ConcurrentModificationException␤ in block <unit> at /tmp/tmpfile:1␤␤»
23:10 Alina-malina left
psch i'd guess before we had Seq that map got all the keys first, cached them and then iterated and deleted 23:10
timotimo 'k 23:11
psch didn't check history for that though, so that's really just a guess
but now the .map builds a Seq
and yeah, somehow that gets in its own way
i still don't completely get where and why
23:12 spider-mario left
psch interestingly the tests for <BagHash (&) Array[Str]> is what brought this up :P 23:12
which maybe means "can be change hashes (or other collection-y types) while iterating over them" is under-tested 23:13
s/ be / we /
23:15 thou left
ShimmerFairy I thought I'd look up the COMPOSE phaser (since I know it's NYI), but the spec only says "when a role is composed into a class" on it. That doesn't help much :( (are you supposed to put it in the role, class, or either/both? What if it's in the wrong spot? And what's it meant to help with?) 23:18
psch ShimmerFairy: from what i remember, i thought it to be put into the role, useful to check for what the class already .^can, for example 23:19
ShimmerFairy: that's just my interpretation of course, and i suspect TimToady probably has a more descriptive concept :)
psch runs a spectest 23:20
ShimmerFairy weird, because some of rakudo's core types have a "# XXX should be COMPOSE" comment in their classes, where they implement ^parameterize_role
(which would at least suggest that it's not _quite_ as simple as "in a role only" ☺) 23:21
psch shrugs 23:22
23:22 telex left
psch you said yourself the design docs aren't particularly detailed :) 23:22
23:22 telex joined
psch (maybe s/detailed/helpful/, to be pedantic :P ) 23:22
ShimmerFairy there's more documentation on a separate-seeming COMPOSE thingy for module exports :P
23:24 Peter_R left 23:27 AlexDaniel joined
timotimo ShimmerFairy: a bit further up in the backlog, jnthn showed how to have mutator methods for attributes that return self so they can be chained 23:46
it uses a method compose that gets overriden
(or is that method ^compose?)
timotimo goes to bed
psch progress \o/ 23:47
only 88/200 tests in bag.t failed 23:48
instead of 156/200
ugexe modules installed with CURLI cause an error when use'd/require'd now :( 23:49
psch i'll leave the rest of spectest to hack and also go to bed o/ 23:50
23:55 cognominal joined 23:57 tokuhiro_ joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!