»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.
00:01 tokuhiro_ joined 00:05 tokuhiro_ left
dalek kudo/nom: a275685 | hoelzro++ | docs/announce/2015.09.md:
Fix release announcement typo
00:06
00:10 cognominal left 00:11 telex left, telex joined 00:13 telex left 00:14 telex joined 00:20 telex left 00:21 telex joined 00:23 aborazmeh left 00:24 araujo_ joined 00:25 araujo left 00:28 pochi joined
flussence has anyone tried to implement The Algorithm in p6 yet? html.spec.whatwg.org/multipage/syn...ml#parsing 00:30
(I've made a start on section 12.5...)
00:32 adu left 00:39 sufrostico joined 00:45 laouji joined 00:51 AlexDaniel left
tony-o its mostly in html parsee xml with the exception of script execution 00:52
00:53 aborazmeh joined, aborazmeh left, aborazmeh joined 00:59 tokuhiro_ joined 01:02 mr-foobar left 01:03 kernel joined
flussence this module code scares me, it looks eerily similar to one I wrote in PHP years ago, tag-closing stuff and all... 01:03
(probably works better than mine, mind.)
01:03 kernel is now known as Guest22461 01:05 lichtkind left 01:06 araujo__ joined
flussence I take that back, mine was way harder to read :( 01:08
01:09 araujo_ left 01:12 telex left, telex joined 01:13 prevost joined 01:28 vendethiel joined 01:42 adu joined 01:45 mr-foobar joined
psch gosh 01:51
finally
$ ./perl6-j --ll-exception -e'say grep { True }, %(a => 1, b => 2).pairs'
(b => 2 a => 1)
now just run a few test files and then commit and then sleep :P 01:52
01:52 vendethiel left, sufrostico left 01:55 maverick87 joined
psch aww shugs 01:56
it doesn't actually *really* work
the patch that is
psch shrugs and goes to bed
02:02 telex left
adu what's "The Algorithm"? 02:03
02:03 telex joined
adu if I were writing a browser, I would make document.write evaluate to "undefined" 02:06
[Coke] just realized he's never seen weird al and larry wall in the same room. 02:09
flussence /o\ 02:13
japhb [Coke]: For some reason, that thought makes me want to listen to Word Crimes
flussence and, adu: that megabyte-long subsection of the HTML spec is The Algorithm. 25 years of parsing mistakes codified in one massive document. 02:15
02:15 perlawhirl joined
adu pretty verbose 02:16
[Coke]: lolol
flussence I've seen one implementation of a draft version of it, a 6000 line file. I think it's grown significantly since then... 02:18
02:21 stanrifkin_ joined 02:22 stanrifkin_ left
ShimmerFairy how do you set stuff for the :auth<> field? There's nothing about META6.json that would set it. 02:26
flussence that's a good question... I just try to manually keep mine consistent with what goes in %.<support><source>. 02:30
ugexe design.perl6.org/S22.html#auth ? 02:37
ShimmerFairy ugexe: that part doesn't describe META6.json, though 02:38
(that's just a section describing terminology, not offering anything about how stuff works yet)
ugexe i see. i think the majority of modules simply do "auth" : "github:username" 02:39
ShimmerFairy that's not spec though, so I can't do that :)
ugexe probably half of s22 isnt up to spec 02:40
er, rakudo
ShimmerFairy Point is, I'll need to hear from the people who know about modules before I just blindly throw in an "auth" field :) 02:41
ugexe rakudo will set it for you spec or not for instance (from author and authority). no emulates/supercedes, .content, excludes, %*RESOURCE, hooks/ 02:42
well i would consider myself one of those people
ShimmerFairy well, AFAIK it's lizmat and FROGGS who've done a lot of work on modules in P6. 02:44
ugexe who do you think is 3rd after them? 02:45
i mean jeez 02:46
ShimmerFairy I didn't know, sorry.
In any case, I need a more definitive answer before I can do anything about :auth<>, since I've no idea how it's supposed to work (as S22 is silent on the matter) :) 02:47
02:50 noganex joined 02:52 noganex_ left, maverick87 left
flussence afaik it just boils down to a hash key so that you can have two same-named modules from different authors installed... though I don't think we've had that actually happen yet, so it's fine to omit for now. 02:53
ugexe i imagine you wouldnt set auth. request auth sure, but it would be constructed from the required content storage and author 02:58
03:02 kaare_ joined
tony-o i wouldn't call it blindly throwing it in if it's spec'd.. 03:09
adu flussence: upon first inspection it looks like something that coroutines would be useful in implementing 03:12
ShimmerFairy tony-o: sure, but the point is that an "auth" field for META6.json *isn't* spec'd :) 03:13
ugexe and it shouldnt be
if you hard code the content storage portion of the auth, then you are stuck maintaing a branch for each content storage
requesting an auth is a different story 03:14
ShimmerFairy ugexe: if "auth" is meant to be pulled from another source, it'd be nice if that was explained somewhere (either under S22/TERMINOLOGY/auth, or before describing the META6.json fields), preferably also explaining how auths are supposed to be constructed
Before the META6.json fields, the description would explain the question of "why can I specify name and ver here, but not auth?", to clarify. 03:15
03:19 colomon left 03:43 telex left 03:44 telex joined 03:45 prevost left
dalek kudo/nom: 2266f4a | TimToady++ | src/core/List.pm:
simplify +args binding code
04:05
ast: 12dc2eb | TimToady++ | S06-signature/slurpy-params.t:
add some tests for +args binding
tony-o i think the intent is implied that you dont want to maintain a hardcoded value in auth, even if you disagree at the implication it seems intuitive that youd not want to do that 04:09
04:20 adu left, rjb joined 04:21 rjb left
tokuhiro_ how do i use string-io like `open my $io, "<", \$buffer;`? 04:23
04:26 rmgk is now known as Guest44882, rmgk_ joined, Guest44882 left, rmgk_ is now known as rmgk
flussence tokuhiro_: there isn't a module for that yet :( 04:28
tokuhiro_ orz
flussence (there's an IO::String, but reading from the string is a todo)
04:30 perlawhirl left 04:37 cdc left 04:57 khw left 05:02 araujo__ left 05:03 araujo joined, araujo left, araujo joined
ugexe maybe IO::Capture::Simple can do what you want 05:04
05:06 thou left 05:24 skids left 05:38 vendethiel joined 05:46 Guest38636 left 05:47 bowtie joined, bowtie is now known as Guest72901 05:54 simcop2387 joined 05:58 domidumont joined, domidumont left 05:59 domidumont joined 06:02 domidumont left, vendethiel left 06:03 domidumont joined 06:05 diana_olhovik_ joined, aborazmeh left 06:07 aborazmeh joined, aborazmeh left, aborazmeh joined 06:09 telex left 06:10 telex joined 06:29 thou joined 06:33 thou left 06:34 chenryn joined 06:41 [Sno] joined
dalek kudo/nom: 63a883f | TimToady++ | src/core/Any-iterable-methods.pm:
start using +args to glrify common listops
06:44
ast: 9679da8 | TimToady++ | S32-list/reduce.t:
reduce won't flatten anymore
06:53 FROGGS joined
FROGGS o/ 06:53
ShimmerFairy / ugexe: I expect that a module installer will set the auth/author/authority for you when downloading dists from a location that provides this information 06:54
ShimmerFairy / ugexe: github and cpan surely do that
ShimmerFairy / ugexe: though, one can and perhaps should add that field to the META6.json, at least as a fallback 06:55
06:57 ShimmerFairy left 07:02 ShimmerFairy joined 07:06 abraxxa joined, Peter_R left 07:15 rurban joined
[Tux] o/ 07:18
yoleaux 16 Sep 2015 17:23Z <nine> [Tux]: I tried bisecting your performance regression but could not find it. If anything, Rakudo seems to have become faster
[Tux] nine, I get a fluctuation between 37 and 40.5 depending on the moonphase squared by the daytime in milliseconds 07:19
moritz which again shows that measurements without error bars are mostly worthless 07:20
07:22 chenryn left 07:25 chenryn joined, jack_rabbit_ joined
nine [Tux]: I've found that when I do "cpupower frequency-set -g performance" my benchmarks give much more reliable results. 07:25
[Tux] cpupower: Command not found. 07:26
nine CPU frequency scaling is really an extremely complicated matter nowadays
[Tux]: it's in the cpupower package on openSUSE. Originally part of www.kernel.org/pub/linux/utils/kern...utils.html 07:27
[Tux] just installed it
re-running
FROGGS [Tux]: you're doing your benchmarks still on that hp-ux box? :o) 07:29
[Tux] no, OpenSUSE 13.1 - Linux 3.11.10-29-desktop [openSUSE 13.1 (Bottle)] HP Z220CMT Xeon(R) CPU E3-1245 V2 @ 3.40GHz/3706(8) x86_64 11914 Mb 07:30
ShimmerFairy FROGGS: would it make sense to have META6.json ask for an "owner" field, so that the fallback auth could be constructed as "fallback:JRANDOM" or whatever? (Instead of letting someone pick whatever they want for the left side of the colon?)
FROGGS ShimmerFairy: hmmmmm 07:31
ShimmerFairy: it is hard to tell how that works in the real world
ugexe different content storage might not let you have the same author name as another 07:33
FROGGS true 07:38
ShimmerFairy that's why I'm unsure about letting you specify whatever content storage you want :) (and the section on the "authors" field mentions that only the "owner" should be used for whatever, but there is no current "owner" field) 07:39
[Tux] test 50000 38.048 37.967 07:41
test-t 50000 40.237 40.156
ugexe you arent supposed to set the content storage yourself. thats just what happens
[Tux] => $work
ugexe right now
FROGGS that's problematic either way 07:42
ShimmerFairy ugexe: of course, we're discussing a fallback auth (which I think should be a fallback "owner", so you can't specify whatever content storage), in the event that an auth can't be constructed.
ugexe that should be implied by the person installing
ShimmerFairy ? We're not talking about module users, but module authors 07:43
ugexe yes
FROGGS ugexe: how will that work out with the use statements in your large $work application that require a certain :auth?
ugexe i dont see how its a problem 07:44
ShimmerFairy ugexe: then I still don't understand "should be implied by the person installing"
FROGGS well, if the auth in the META6.json (or in the CURLI database) changes so much, you can't use :auth filters in your use statements at all
right now I think I'd better use: use Foo:md5<z2397846298437602364z203zo> 07:45
moritz md5 *shudder*
FROGGS moritz: was just an example 07:46
:o)
moritz good at bikeshuddering
FROGGS shudder up!
:P
moritz :-)
ShimmerFairy FROGGS: I think the issue may be that :auth<> is actually a pretty open thing. Something feels off about how it works, I can't place my finger on it. 07:47
FROGGS nods
BooK wouldn't using md5 prevent quick local patches to the source?
07:47 aborazmeh left
FROGGS BooK: aye 07:48
BooK: it is pain, whatever we do
ShimmerFairy FROGGS: It feels like :auth<> would work better if Perl 6 handled authorship (some sort of 'author registry'), instead of letting *everyone else* handle it. But, I don't think making people register their auth somewhere is ideal either :)
FROGGS ShimmerFairy: that occoured to me as well 07:49
ShimmerFairy Perhaps the mere fact that it doesn't easily make sense in META6.json is a clue that it doesn't currently work as a filter on modules :)
BooK what problem is that trying to solve? 07:50
ugexe ah i was speaking for the recommendation manager. for using the module there should be no problem with :auth
as in, if i want my recommendation manager to install from github content storage only, it shouldnt jump to a fallback (the recommendation manager) 07:51
ShimmerFairy BooK: the idea is that we don't require a globally-unique module name in Perl6-land
BooK I thought the idea was to be able to load multiple version of "almost the same thing" at once, and use different ones in different cases/other modules 07:56
so that one could be very picky about what they want, while others would use * 07:57
ShimmerFairy Well yeah, it helps with doing that when everyone can just name their module JSON::Fast, with different auth/vers, instead of JRANDOM::v0_1::JSON::Fast and such :) 07:58
mst so far, we've mostly done fine with requiring unique names
BooK mst: my thought exactly
07:59 aborazmeh joined, aborazmeh left, aborazmeh joined
mst I can see multiple completely different JSON::Fast dists being a complete disaster 07:59
essentially you'd then have to -talk- about them as JRANDOM::JSON::Fast
BooK but as soon as JRANDOM's version is installed and has a incompatible interface, every module saying "I'm fine with using whatever JSON::Fast" will blow up
mst which, well, PAUSE already supports
BooK well, mst and I are already talking about diffrent JRANDOM::JSON::Fast :-) 08:00
ShimmerFairy FROGGS: what if, instead of a registered list of auths, it was just a registry of content storages? Could it help if we had a well-defined list of things for the LHS of the auth colon? (With perhaps an internal: or something set aside for whenever some company has an internal storage system)
Using JSON::Fast wasn't the best example, but in general we need some way to neatly bring together different "universes" of modules, where any number of them from each universe have a name clash. 08:01
dalek ast: 48999d9 | TimToady++ | S32-list/reduce.t:
reduce doesn't automatically flatten now
08:02
mst ShimmerFairy: why?
actually, no, wrong question
why is this any different to JRANDOM::JSON::Fast
in theory, I understand the idea 08:03
ugexe i mostly found it nice for when joe shmoe puts his boiler plate code in an otherwise good root namespace
mst but in practice, trying to do this stuff involves more complexity than useful, IME
BooK ugexe: but isn't that better solved with non-technical means?
ShimmerFairy mst: I don't know the reasoning for those who originally came up with it, but for me having a specified :auth<> is better for conveying an author than some sort of "top level name is author" convention 08:04
mst ok, but then, what happens when the releaser changes?
FROGGS mst: the problem we are actually trying to solve is: dist Foo is popular, gets forked, other one (also Foo) gets more popular, so people switch auths
ShimmerFairy mst: in a sense, I myself think having a "named" package name component is better than a "positional" one for this sort of thing :)
mst consider DBIx::Class, which in the early days often had eight releasers across eight releases 08:05
FROGGS mst: like, allowing evolution
mst how is an author name even sensical at that point?
FROGGS mst: can also be an authority, like TheDBIxClassGroup 08:06
dalek kudo/nom: 32ce653 | TimToady++ | src/core/allomorphs.pm:
allomorphic .gist will just print the string

Fears of user confusion are less well founded than fears of users rampaging because they just wanted to print out the values, and got extra crap. This includes me, trying to run a rosettacode entry on a <> list of integers.
TimToady --> zzz & 08:07
08:10 aborazmeh left 08:17 thou joined 08:20 dakkar joined, telex left, telex joined 08:21 thou left 08:22 frodwith left 08:40 anaeem1_ joined, laouji left 08:41 darutoko joined 08:43 espadrine joined 08:45 laouji joined 08:46 sergot joined 08:49 anaeem1_ left 08:53 jack_rabbit_ left 08:59 zakharyas joined 09:04 morjo joined 09:15 cognominal joined
timotimo m: my $s = Channel.new; $s.close; say $s.poll; 09:28
camelia rakudo-moar 32ce65: OUTPUT«Nil␤»
psch r: my $a = Bag.new("a", "b", "b"); say $a.roll(2)
camelia rakudo-jvm 32ce65: OUTPUT«Type check failed in binding @pairs; expected 'Positional' but got 'Seq'␤ in block <unit> at /tmp/tmpfile:1␤␤»
..rakudo-moar 32ce65: OUTPUT«(a b)␤»
psch i got past the type check failure, but now i'm getting "This Seq has already been consumed" :/ 09:29
it points at the for in the gather in ROLLPICKGRABN, which i find confusing
moritz psch: how did you get past it?
psch moritz: correcting the .cache invocation in the Binder 09:30
09:30 geekosaur joined
psch moritz: gist.github.com/peschwa/6fff3a7c51bc27fa67bd 09:30
moritz psch: huh, then you need to debug where it has been consumed
nine mst: I'm hoping for the :auth mechanism to fix the ambiguity of Perl 5 and Perl 6 modules sharing a name
psch moritz: right, i'm trying :) 09:32
moritz m: my \seq = gather { take 1; take 2 }; my @list = seq.cache; .say for seq 09:34
camelia rakudo-moar 32ce65: OUTPUT«This Seq has already been iterated, and its values consumed␤ in block <unit> at /tmp/LRByNouMYm:1␤␤»
nine m: my \seq = gather { take 1; take 2 }; my @list = seq.cache; .say for seq.cache 09:35
camelia rakudo-moar 32ce65: OUTPUT«1␤2␤»
moritz nine: I know that's possible; I just wonder if Seq.iterator should be smart enough to fall back to Seq.cache.iterator 09:36
timotimo m: try die "oh noes" orelse say "test" 09:37
camelia ( no output )
timotimo m: die "oh noes" orelse say "test"
camelia rakudo-moar 32ce65: OUTPUT«oh noes␤ in block <unit> at /tmp/wrbcRN2G11:1␤␤»
timotimo m: (try die "oh noes") orelse say "test"
camelia rakudo-moar 32ce65: OUTPUT«test␤»
timotimo m: try die "oh noes" // say "test"
camelia ( no output )
timotimo m: (try die "oh noes") // say "test"
camelia rakudo-moar 32ce65: OUTPUT«test␤»
timotimo movable type doesn't let me edit my comment :( 09:39
i should have preview'd
blogs.perl.org/users/aaron_baugher/...erl-6.html ← this is what i replied to 09:41
morjo Anyone know of an easy to understand guide to Rules? 09:45
Timbus_ not me; they're so hard to follow.
timotimo have you read doc.perl6.org/language/regexes ? 09:47
nine moritz: that would be just gist.github.com/niner/e82fd8f33ca4c465477c which works and passes spec tests. I wonder about the performance implications though. 09:49
morjo Not everyting. Guess I should just start from the top :) 09:50
timotimo morjo: it's not much, but i would try to answer any questions you have 09:52
morjo timotimo: Thanks, I'll read that doc and try to make a test case and see how that goes. If I cant manage, I'm sure to come back :) 09:55
timotimo the most common problem i encounter is "significant whitespace" as "rule" gives you being less than helpful 09:56
so if you put mostly "regex" into your grammar, you'll not be so surprised
09:56 telex left
moritz or 'token' if you don't want backtracking 09:56
timotimo right 09:57
09:57 telex joined, telex left
timotimo oh, look! 09:57
wordpress has made another post i've spent some time on disappear out of the blue!
09:58 telex joined
timotimo it's nice that wordpress' editor saves in-progress posts in the local storage of the browser, but it seems like they are using the session storage, rather than the general local storage? 09:59
i'm not happy right now 10:00
10:01 andreoss joined, chenryn left 10:05 thou joined 10:07 colomon joined 10:10 thou left
andreoss m: my $x; say "hi" with $x; 10:31
camelia ( no output )
10:31 cdc joined
cdc m: 1 »+» (2, 3) 10:31
camelia rakudo-moar 32ce65: OUTPUT«Lists on either side of non-dwimmy hyperop of infix:<+> are not of the same length␤left: 1 elements, right: 2 elements␤ in block <unit> at /tmp/jcNHIdXWtc:1␤␤»
cdc m: (1, 2) »+» 3 10:32
camelia ( no output )
cdc ^^ LTA error
timotimo huh, that message seems not-right
cdc patch: gist.github.com/anonymous/3758973094fe2a3a1fe3
:)
this is just s/same/expected/ 10:33
timotimo mhm mhm 10:34
andreoss m: say "hi" with False 10:37
camelia rakudo-moar 32ce65: OUTPUT«hi␤»
psch m: my $x; say $x.defined; say False.defined
camelia rakudo-moar 32ce65: OUTPUT«False␤True␤»
andreoss say Nil.defined; say Empty.defined; 10:38
m: say Nil.defined; say Empty.defined;
camelia rakudo-moar 32ce65: OUTPUT«False␤False␤»
10:40 Ven joined
ilmari m: say( ("hi" with False) ?? "True" !! "False" ) 10:42
camelia rakudo-moar 32ce65: OUTPUT«True␤»
moritz m: say( ("hi" does False) ?? "True" !! "False" ) 10:48
camelia rakudo-moar 32ce65: OUTPUT«Cannot mix in non-composable type Bool into object of type Str␤ in block <unit> at /tmp/mk64YT9eSt:1␤␤»
moritz m: say( ("hi" but False) ?? "True" !! "False" )
camelia rakudo-moar 32ce65: OUTPUT«False␤»
andreoss say False but True 10:50
m: say False but True
camelia rakudo-moar 32ce65: OUTPUT«True␤»
Ven o/, #perl6! 10:58
10:58 andreoss left
[ptc] Ven: o/ 11:06
timotimo o/ 11:07
11:14 rindolf joined 11:15 mprelude joined
ilmari m: say( (True but False) ?? "True" !! "False" ) 11:16
camelia rakudo-moar 32ce65: OUTPUT«False␤»
timotimo yeah, mixing in an instance of the class "Bool" will override the ".Bool" method to return that exact value 11:17
11:17 chenryn joined
timotimo so when the ?? !! operator tries to boolify the value it got, it calls .Bool and that returns False 11:17
^- that's my interpretation of the result
of course it's strange that we call .Bool on something that we know is already a Bool
11:18 TEttinger left
ilmari m: my $x = "42" but 37; say $x + 5 11:19
camelia rakudo-moar 32ce65: OUTPUT«47␤»
[ptc] m: my @moo = []; @moo.push(1); @moo.say; @moo.push([1]); @moo.say 11:20
camelia rakudo-moar 32ce65: OUTPUT«[1]␤[1 1]␤»
timotimo oh, huh?
[ptc] why does pushing an array onto an array not result in an array of arrays? What am I misunderstanding here?
11:21 andreoss joined
psch m: my @moo = []; @moo.push(1); @moo.say; @moo.push(([1],)); @moo.say 11:21
camelia rakudo-moar 32ce65: OUTPUT«[1]␤[1 [1]]␤»
[ptc] it seems that .push(1) and .push([1]) do the same thing
psch single arg rule i think?
[ptc] ah, ok, could be
nine yes
m: my @moo = []; @moo.push(1); @moo.say; @moo.push($[1]); @moo.say
camelia rakudo-moar 32ce65: OUTPUT«[1]␤[1 [1]]␤»
nine Use itemization to get around that
[ptc] nine: that looks better. 11:22
[ptc] wonders where we need to update the docs now...
nine It's the difference between @moo.push(@new_values) and @moo.push($@additional_array)
[ptc] ah, that's a nice way to think about it! Thanks :-) 11:24
one can see in that example why one would need the $ in the single arg case
[Coke] Ugh. Forgot there was a child school thing tonight. gonna be up laaate with the release. 11:29
11:33 chenryn left
lizmat [Coke]: fwiw, one more day wouldn't hurt much, I don't think :-) 11:34
[Coke] eh. releases are most easy. 11:40
and there's enough coffee. Maybe.
[ptc] do we have a pre-GLR camelia? 11:41
11:42 telex left
[Coke] there was a tag so one can be setup. 11:42
11:42 telex joined
[ptc] [Coke]: thanks. Just wondered if I could save myself some work :-) 11:42
ilmari m: my $a = Nil; say Nil; say $a; say $a eqv Nil; 11:43
camelia rakudo-moar 32ce65: OUTPUT«Nil␤(Any)␤False␤»
timotimo we have star-m:
but by now that's rather old
[ptc] timotimo: might be good enough. thanks 11:45
star-m: my @directions = (1, -1 X 2, -2), (2, -2 X 1, -1); @directions.say 11:48
camelia star-m 2015.03: OUTPUT«1 2 1 -2 -1 2 -1 -2 2 1 2 -1 -2 1 -2 -1␤»
[ptc] m: my @directions = (1, -1 X 2, -2), (2, -2 X 1, -1); @directions.say
camelia rakudo-moar 32ce65: OUTPUT«[((1 2) (1 -2) (-1 2) (-1 -2)) ((2 1) (2 -1) (-2 1) (-2 -1))]␤»
ilmari huh, the first hit on duckduckgo.com/?q=perl6+eqv is doc.perl6.org/routine/eqv, but that 404s
timotimo doc.perl6.org/language/operators#infix_eqv 11:51
[Coke] searching for eqv fails. (as does ==)
ilmari reorganised recently? 11:53
[Coke] addeed github.com/perl6/doc/issues/129 for a new 404 handler.
timotimo hm, potentially
11:53 thou joined
ilmari doc.perl6.org/language/operators#infix_=== links to doc.perl6.org/type/WHICH, which als o404s 11:54
[Coke] and 130, 131 for the missing eqv/== terms
[ptc] timotimo: I think it's now in language/operators
11:54 larion joined
timotimo huh, WHICH should not be in /type/ at all 11:54
[ptc] timotimo: doc.perl6.org/language/operators#infix_eqv
timotimo i pasted that link up there already :) 11:55
[ptc] is just too slow today... 11:57
11:58 thou left 12:00 Skarsnik joined
ilmari doc.perl6.org/language/operators#po...%AB_%C2%BB links to doc.perl6.org/routine/%C2%AB%20%C2%...%AB_%C2%BB 12:02
that's « » for those who don't read uri-encoded utf8 fluently
circumifx « » isn't mentioned in /language/operators at all 12:03
s/ifx/fix/ # pesky laggy mobile data
12:03 rurban left 12:05 ShimmerFairy left
dalek c: 26441e3 | paultcochrane++ | lib/Language/operators.pod:
Update cross product output post-GLR
12:12
c: 9da60a0 | paultcochrane++ | lib/Language/operators.pod:
Fix zip operator examples post GLR
12:15 abraxxa left
andreoss should rakudobrew exec run command for all installed compilers or just one? 12:17
tadzik imho just one, but there should be a way to run it on all of them 12:18
yoleaux 15 Sep 2015 18:26Z <patrickz> tadzik: you can merge the PR now. Did some more testing and fixing and am confident this is loads better than before.
12:18 ShimmerFairy joined
tadzik .tell patrickz ok, I'll look into it :) 12:18
yoleaux tadzik: I'll pass your message to patrickz.
12:18 abraxxa joined
andreoss why run it with just one compiler? what's the point of exec then if one could just type perl6 instead? 12:22
moritz which perl6? 12:24
12:24 ZoffixWork joined
moritz what if it's not in $PATH? 12:25
perl6-m vs. perl6-j ?
and I hope we'll have a perl6-js (or perl6-q) one day
ZoffixWork Hey, I've just came across this ancient commit of mine and I think it's wrong. People shouldn't change "The Perl Foundation" text since the name refers to the copyright owner of the *license itself* and not the module, right? github.com/perl6/doc/commit/485645...e78a3fcca5
12:33 carlin joined
psch heh, trying to fix the Binder *and* adjusting jvm to not need the workaround in ROLLPICKGRABN at the same time wasn't particularly great an idea... 12:33
also grrr: 12:34
# expected: 'foofoofoobarbarfoobarbar'
# got: 'barbarbarfoofoobarfoofoo'
FROGGS :/
psch there's probably a dozend or so tests that fail exclusively due to hash elem ordering...
carlin m: my $module = "Test"; require $module; say 'alive';
camelia rakudo-moar 32ce65: OUTPUT«Could not find file 'Test' for module Test␤ in block <unit> at /tmp/ll_qx9VedB:1␤␤»
carlin m" require Test; say 'alive';
m: require Test; say 'alive'; 12:35
camelia rakudo-moar 32ce65: OUTPUT«alive␤»
carlin I'm sure that used to work ... :/
until quite recently?
psch star-m: my $m = "Test"; require $m; say "alive"
camelia star-m 2015.03: OUTPUT«alive␤»
psch that's not particularly recent, i guess
but it is a point for "it used to work" :) 12:36
might be cu_load changes?
12:39 Ven left
ZoffixWork Larry wrote the original 1.0 license... cool. I guess he'd know best the answer to my question. 12:39
TimToady, no idea whether this commit is right or wrong :) You be the judge :) github.com/perl6/doc/commit/485645...e78a3fcca5
12:39 ZoffixWork left 12:40 Ven joined 12:41 JimmyZ joined
ilmari m: my $foo = foo; my @bar = <bar blat>; say «$foo @bar baz» 12:41
camelia rakudo-moar 32ce65: OUTPUT«5===SORRY!5=== Error while compiling /tmp/L990QZYSAy␤Undeclared routine:␤ foo used at line 1␤␤»
ilmari m: my $foo = 'foo'; my @bar = <bar blat>; say «$foo @bar baz»
camelia rakudo-moar 32ce65: OUTPUT«(foo @bar baz)␤»
psch m: my $foo = 'foo'; my @bar = <bar blat>; say «$foo @bar[] baz» 12:43
camelia rakudo-moar 32ce65: OUTPUT«(foo bar blat baz)␤»
12:43 abraxxa left 12:47 tokuhiro_ left 12:48 laouji left
moritz m: my $foo = 'foo'; my @bar = <bar blat>; say «$foo @bar[] baz».perl 12:52
camelia rakudo-moar 32ce65: OUTPUT«("foo", "bar", "blat", "baz")␤»
12:52 Loren joined 13:01 abraxxa joined 13:10 skids joined 13:12 rurban joined 13:18 Ven left
dalek kudo/nom: 7a2b4a8 | peschwa++ | src/vm/jvm/runtime/org/perl6/rakudo/Binder.java:
Fix PositionalBindFailover binding code.
13:25
psch progress \o/ 13:26
13:26 larion left
moritz \o/ 13:26
psch++
psch there's two more things that are probably behind a lot of test failures 13:27
13:27 hernanGOA joined
psch one is the mystery StackOverflowError, the other is the allomorph mishandling 13:28
(and then there's of course the written-with-assumed-hash-element-ordering tests...)
13:28 Ven joined 13:29 Ven left
moritz those need fixing, yes 13:31
13:31 larion joined 13:34 sufrostico joined 13:39 aborazmeh joined, aborazmeh left, aborazmeh joined
dalek kudo-star-daily: 904d162 | coke++ | log/ (2 files):
today (automated commit)
13:40
[Coke] moritz: is docs.perl6.org a mojo5 app at the moment? 13:41
13:41 thou joined
[Coke] or is it a static site? 13:42
wondering how we'd do the 404.
[ptc] [Coke]: I believe it's a mojo5 app
[Coke]: no, sorry, it's not. The mojo5 app is for local use 13:43
[Coke]: the files get copied to the server via rsync and thus it's just a static site
moritz [Coke]: it's a static page 13:45
[Coke]: and 404 pages can be just a 404.html, and I can tell Apache to use that as the error document
13:46 thou left 13:47 tokuhiro_ joined
moritz idle wonders if he should name his facade classes "ThingFaçade", just to annoy cow-orkers :-) 13:48
psch r: say [max]() 13:49
camelia rakudo-moar 32ce65: OUTPUT«-Inf␤»
..rakudo-jvm 32ce65: OUTPUT«Too few positionals passed; expected 1 arguments but got 0␤ in block <unit> at /tmp/tmpfile:1␤␤»
psch r: my @a; say @a.max 13:50
13:50 andreoss left
camelia rakudo-{moar,jvm} 32ce65: OUTPUT«-Inf␤» 13:50
psch r: say ().WHAT 13:51
camelia rakudo-{moar,jvm} 32ce65: OUTPUT«(List)␤»
psch r: sub f(@a) { say @a.WHAT }; f ()
camelia rakudo-{moar,jvm} 32ce65: OUTPUT«(List)␤»
13:52 tokuhiro_ left, Ven joined
[Coke] moritz: ok. I'll add a 404.html at some point today, and you can do the apache magic? 13:54
moritz [Coke]: sure 13:55
dalek c: ec82f33 | (David H. Adler)++ | lib/Language/control.pod:
Added docs for with/without

  =head3 under the =head2 for C<if>
13:55 dalek left 13:56 dalek joined, ChanServ sets mode: +v dalek
dalek kudo/nom: 61022e7 | FROGGS++ | / (3 files):
make C++ NativeCall tests work on darwin

Sadly clang and gcc cannot export symbols of purely virtual functions when the implementation of the function (or rather method) is inside of the class body.
13:57
13:58 PerlJam joined, nucp6 joined
psch oh 13:59
that's the new param thingy, +args
can't even catch up before new bugs get introduced /o\
nucp6 hi perl6
psch hi nucp6 o/
14:03 telex left, telex joined
Ven FROGGS: do you need me to test some stuff:)? 14:04
14:05 aborazmeh left
Ven heh, I guess I can abstract "if %a<b>:exists { $_ = %a<b>; }" into `with %a<b>`, if I'm sure the key isn't null 14:06
14:06 rurban left
moritz s/abstract/simplify/ 14:07
14:09 carlin left, cognominal left
lizmat m: my %h = a => Any; say h<a>:exists; .say with %h<a> 14:10
camelia rakudo-moar 7a2b4a: OUTPUT«5===SORRY!5=== Error while compiling /tmp/CuRF8wGxVc␤Undeclared routine:␤ h used at line 1␤␤»
lizmat m: my %h = a => Any; say %h<a>:exists; .say with %h<a>
camelia rakudo-moar 7a2b4a: OUTPUT«True␤»
lizmat Ven ^^^ *and* the value is defined :-
)
FROGGS Ven: would be nice if you could pull rakudo and run 'make test', the two C++ tests should now pass
moritz lizmat: I guess that's what Ven meant with "if the kye isn't null"
Ven moritz: yar. not sure why I put that word there
also yes, "null" was meant for Nil/general undefinedness 14:11
FROGGS: am doing it :)
14:12 Sqirrel_ left, pyrimidi_ joined 14:15 telex left 14:16 telex joined
nucp6 racudo: 29*1 14:16
.
14:16 Sqirrel joined
moritz m: say 29 * 1; # for nucp6 14:17
camelia rakudo-moar 7a2b4a: OUTPUT«29␤»
14:18 thou joined
nucp6 thx moritz :) 14:18
dalek kudo/nom: b94b49d | peschwa++ | src/vm/jvm/runtime/org/perl6/rakudo/Binder.java:
Implement one-arg slurpies for the jvm.
14:19
psch r: say [max]() # fixes this, soonish probably
camelia rakudo-jvm 7a2b4a: OUTPUT«Too few positionals passed; expected 1 arguments but got 0␤ in block <unit> at /tmp/tmpfile:1␤␤»
..rakudo-moar 7a2b4a: OUTPUT«-Inf␤»
psch and when the spectest is clean i get to fix the interop list handling \o/ 14:20
FROGGS spectest clean?
psch eventually, hopefully :P 14:21
FROGGS psch: how many spectest files are left?
psch FROGGS: something between 55 and 70 i guess 14:22
14:22 telex left
psch FROGGS: i'm running spectest now after the onearg slurpies 14:22
FROGGS k
14:22 telex joined
psch FROGGS: eh, upper bound is actually 68, that's from before the onearg slurpies 14:22
14:22 morjo left
psch FROGGS: i'm pretty confident at least 4 or 5 files is hash ordering, i.e. bad tests 14:23
FROGGS psch: can you gist the summary then?
bbiab &
psch FROGGS: sure, i'll gist the new one after this run
14:23 FROGGS left 14:25 pyrimidi_ is now known as pyrimidine 14:26 MilkmanDan joined
Ven .tell FROGGS t/04-nativecall/13-cpp-mangling.t .... ok 14:34
yoleaux Ven: I'll pass your message to FROGGS.
14:46 telex left 14:47 telex joined 14:49 tokuhiro_ joined 14:51 FROGGS joined 14:53 khw joined
FROGGS \o/ 14:53
yoleaux 14:34Z <Ven> FROGGS: t/04-nativecall/13-cpp-mangling.t .... ok
14:53 tokuhiro_ left
[Coke] anyone else "the regex guy" at work? 14:53
Ven FROGGS++
[Coke] release today! Make sure Changelog keeps up to date. Don't sneak in any last minute big changes. 14:54
psch held a 2h regex workshop at one job
FROGGS [Coke]: customers even call and say: I just need to talk to somebody sane about regexes...
geekosaur implied contradiction? 14:55
PerlJam "sane" and "regexes" are often mutually exclusive (See Damian ;)
TimToady just snuck in a last-minute big change, and hasn't been updating the Changelog :)
lizmat looking forward to that entry :-)
geekosaur (and yes, I am very much "the regex guy" at work and elsewhere) 14:56
PerlJam TimToady: I think 63a883fccec95c692b1fd4b3f4f6e5b641dab5a7 broke panda (But I don't understand the difference between +args and *args yet)
[Coke] TimToady: I think everyone is used to you cheating. :P
TimToady: tell you what, let me know when you think it's safe to release with your latest stuff. :) 14:57
TimToady you might need to throw a flat in somewhere on one of the affected listops
PerlJam aye, that's what I'm thinking.
TimToady less likely, itemize a one-arg argument
jdv79 what big change? 14:58
14:58 cognominal joined
PerlJam So, *foo is a "flattening slurpy" and +foo does no flattening? 14:59
[Coke] tries to get a key for the release. (PGP will only build on PowerPC Macs.) wow, that's showing its age.
TimToady +foo is one-arg semantics
so it's more like **, but if there's a single arg, expands it to be multiple (unless itemized)
PerlJam ok 15:00
FROGGS I guess that's what I only can understand through examples
TimToady or looking at it the other way, it always takes the rest of the argument list as a single thing to be iterated, and sometimes the top-level iteration is commas, and sometimes it's just some object 15:01
15:01 nucp6 left
TimToady m: sub foo (+@args) { say @args.elems }; foo [1..4]; foo [1..4],2 15:02
camelia rakudo-moar b94b49: OUTPUT«4␤2␤»
TimToady there's some examples
that's the caching form; here's the non-caching form: 15:03
m: sub foo (+args) { say args.elems }; foo [1..4]; foo [1..4],2
camelia rakudo-moar b94b49: OUTPUT«4␤2␤»
PerlJam m: sub foo (*@args) { say @args.elems }; foo [1..4]; foo [1..4],2; # a contrasting example for me (should be 4,5 I think) 15:04
camelia rakudo-moar b94b49: OUTPUT«4␤5␤»
PerlJam cool
TimToady at least, it's supposed to be non-caching, haven't actually tested that :)
skids FROGGS: The +@args replaces the first box of code currently at doc.perl6.org/language/functions#Sl...onventions and the +args does the same thing but gives you a one-shot Seq so you can handle massive things with low memory.
FROGGS .tell jnthn How hard would it be to allow rw attributes on CStructs?
yoleaux FROGGS: I'll pass your message to jnthn.
[Coke] when creating a key to sign releases with, should I pick a key with an expiration date? 15:07
lizmat is looking at #126073 trying to get it fixed before release 15:08
[Coke] lizmat++ 15:10
Ven reads about + 15:13
Ven doesn't get the difference between +a and +@a yet :) 15:14
is + a ... pregil?
psch FROGGS: 54 files show up in the summary, but at least one is a TODO PASS
TimToady it's a slurpy
psch gisting it now...
15:14 diana_olhovik_ left
Ven right. * would have to be considered a pregil for + to even be. 15:14
it's a funny word, though :P.
psch gist.github.com/peschwa/90c8dcc516ea934a17be # latest j-spectest, for the curious 15:15
Ven so, *@a is non-flattening, **@a is total-flattening, and +@a/+a is one-arg-rule-style flattening..?
FROGGS psch: thanks
TimToady you got * and ** backwards
Ven whoops. thanks. 15:16
psch not happy about all the "Bad plan"s :/
cause that's usually leaking jvm-level stuff
Ven I wonder why you didn't go with *** and label 6.0 "Dancing with the stars" *g*
dalek kudo/nom: 0fca4dc | coke++ | docs/release_guide.pod:
fix typo
kudo/nom: bcde489 | coke++ | docs/release_guide.pod:
use moar Unicode
skids m: sub f(+a) { a.WHAT.say }; f((1,2,3).grep({$_})); # not Seq? 15:19
camelia rakudo-moar b94b49: OUTPUT«(List)␤»
TimToady patches welcome :) 15:20
skids :-) 15:21
15:22 Loren left
psch r: my ::foo $x; 15:24
camelia ( no output )
psch r: my ::foo $x, say $x
camelia rakudo-jvm b94b49: OUTPUT«java.lang.RuntimeException: java.lang.StackOverflowError␤␤»
..rakudo-moar b94b49: OUTPUT«Method 'gist' not found for invocant of class 'foo'␤ in block <unit> at /tmp/tmpfile:1␤␤»
psch r: my $x, say $x 15:25
camelia rakudo-{moar,jvm} b94b49: OUTPUT«(Any)␤»
rindolf Hi all. paste.debian.net/312188/ - this code gives me Method 'count' not found for invocant of class 'Range' 15:27
in block <unit> at prob329-shlomif.p6:28
- what should I do?
Anyone? 15:32
15:33 telex left, telex joined
[Coke] First, patience. 15:35
PerlJam rindolf: are you using a recent version of rakudo?
rindolf PerlJam: possibly not.
PerlJam: let me upgrade.
[Coke] I get a different error.
Cannot call map(Block: Range); none of these signatures match: 15:36
m: map { 1.say }: 1..2 15:37
camelia rakudo-moar b94b49: OUTPUT«Cannot call map(Block: Range); none of these signatures match:␤ (\SELF: &block;; :$label, :$item, *%_)␤ (HyperIterable:D $: &block;; :$label, *%_)␤ in block <unit> at /tmp/UkqDfSZuT3:1␤␤»
rindolf m: map { 1.say }: (1,2,3)
camelia rakudo-moar b94b49: OUTPUT«Cannot call map(Block: List); none of these signatures match:␤ (\SELF: &block;; :$label, :$item, *%_)␤ (HyperIterable:D $: &block;; :$label, *%_)␤ in block <unit> at /tmp/9ldDRUOstN:1␤␤»
[Coke] m: map { 1.say }, 1..2
camelia rakudo-moar b94b49: OUTPUT«1␤1␤»
[Coke] ^^ you'll need to change that, also. 15:38
line 34
15:38 larion left
PerlJam or use the indirect object notation correctly :) 15:38
m: (1..2).map: { .say }; 15:39
camelia rakudo-moar b94b49: OUTPUT«1␤2␤»
lizmat m: Supply.new.grep(1).grep(1).unique.tap: *.say # shortest golf for #126073 so far
camelia rakudo-moar b94b49: OUTPUT«Memory allocation failed; could not allocate 58608 bytes␤»
[Coke] ... and then there's a ton of uninitialized value warnings, and it dies with an another error.
lizmat it needs two greps, 1 unique and a tap to go into never never land
even *without* any value needing to be emitted 15:40
dalek c: 591a8a5 | TimToady++ | lib/Language/modules.pod:
Revert "Added note that raw license copy/paste needs modification of copyright owner"

This reverts commit 48564569710f175d8baa61b4ef1038e78a3fcca5.
As Zoffix++ suspected, the copyright is on the license itself, not on what is being licensed.
rindolf [Coke]: done. 15:42
FROGGS finally starts to work on XML::Compile 15:43
dakkar FROGGS: in perl6? are you going to bind libxml2?
(because I started looking into libxml and libxslt bindings)
|Tux| FROGGS, markov's suite?
FROGGS |Tux|: aye 15:44
|Tux| FROGGS++; # brave guy
FROGGS dakkar: have already started ages ago
rindolf [Coke], PerlJam : thanks! I mixed up ":" and "," - now can you please tell me how to get a list/array/etc. of characters from a string ? .chars() only does a count, and .split('') adds an empty string at beginning and end.
PerlJam m: "string".comb.perl; 15:45
lizmat m: Supply.new.grep(1).grep(1).unique.tap # even shorter, jnthn any ideas on what might going on here ?
camelia ( no output )
rakudo-moar b94b49: OUTPUT«Memory allocation failed; could not allocate 640 bytes␤»
FROGGS dakkar: well, months, the basics work... now I want to start to work on XML::Compile to see what is needed from XML::LibXML and how it is used
dakkar FROGGS: I'm clearly not paying enough attention :) link?
PerlJam m: "string".comb.say.perl;
camelia rakudo-moar b94b49: OUTPUT«(s t r i n g)␤»
FROGGS ... to get the interface straight
PerlJam blah
rindolf: anyway ... use comb
dakkar I was going to mirror Shlomy's work for the perl5 bindings
FROGGS dakkar: github.com/FROGGS/p6-XML-LibXML # though I've got unpushed changes....
dakkar Shlomi, even
FROGGS dakkar: if you intend to look more closely, I'll push more often 15:46
rindolf dakkar: I am Shlomi Fish.
dakkar rindolf: hi!
rindolf dakkar: hi.
dakkar sorry for mistyping your name
FROGGS: www.thenautilus.net/cgit/WebCoso-p.../LibXML.pm ← that's my exploratory attempt 15:47
dalek pan style="color: #395be5">perl6-examples: f1f8dc8 | (Shlomi Fish)++ | categories/euler/prob329-shlomif.p6:
Add a solution to Euler #329
15:48
FROGGS dakkar: nice
dakkar I was aiming to get the same API in perl5 and perl6 (eventually) 15:49
dalek p: d96dcc1 | (Pawel Murias)++ | src/vm/js/nqp-runtime/ (3 files):
Split cclass ops into a separate file.
15:50
p: 84386f3 | (Pawel Murias)++ | src/vm/js/nqp-runtime/nfa.js:
[js] fix nfa bug
p: d3b3e62 | (Pawel Murias)++ | src/vm/js/QAST/Compiler.nqp:
[js] Fix regex bugs.
rindolf OK, it's working now and even runs under a minute here.
15:50 domidumont left
rindolf But my machine is pretty fast. 15:50
TimToady skids: actually, I think I know how to fix the Seq thing
PerlJam rindolf: maybe make youre #! line say "/usr/bin/perl6" or "/usr/bin/env perl6" instead of "/usr/bin/perl" :-) 15:51
15:52 frodwith joined
rindolf PerlJam: sure. 15:52
FROGGS dakkar: I hope to not end up with the same api as in P5, because for one it is not very Perl 6ish and some stuff is just weird
15:52 pmurias joined
rindolf paste.debian.net/312199/ - seems like Perl 5 and Perl 6 execute it in roughly the same time. 15:52
tony-o PerlJam++ on the env thing 15:53
FROGGS dakkar: like I think that parse_string, parse_file, parse_fh etc etc do not make much sense in Perl 6 15:54
dakkar: I can give you a commit bit if you want...
dinner &
15:55 travis-ci joined
travis-ci NQP build passed. Pawel Murias '[js] Fix regex bugs.' 15:55
travis-ci.org/perl6/nqp/builds/80855483 github.com/perl6/nqp/compare/47408...b3e62e41b2
15:55 travis-ci left
dalek pan style="color: #395be5">perl6-examples: 3974007 | (Shlomi Fish)++ | categories/euler/prob329-shlomif.p6:
Correct she-bang and use v6.
15:55
rindolf PerlJam: thanks!
dakkar FROGGS: agree, we have multi-methods now ☺ I was mostly thinking about all the other methods and classes 15:56
nine lizmat: endless recursion in live 16:01
lizmat: called from add_source 16:02
16:03 perl6_newbee joined, pmurias left, andreoss joined
lizmat nine L823 ? 16:04
16:05 Ven left
nine lizmat: yes, looks like it 16:05
andreoss perlbrew's exec runs command for each of perls, rakudbrews's exec runs it only for currently selected compiler
i believe it's wrong
can rakudobrew run command for all compilers? 16:06
lizmat nine: think I found t 16:10
nine cool :)
lizmat SupplyOperations, L48: I think it should refer to the source tap, not to the source 16:12
L148 rather
nine Sounds reasonable 16:13
dalek ast: c2c13fb | TimToady++ | S06-signature/slurpy-params.t:
make sure +args passes through Seq directly
lizmat hmmm.. wrong :-(
dalek kudo/nom: 25c8d24 | TimToady++ | src/core/List.pm:
+args should bind a Seq directly, not via List
16:15
lizmat m: Supply.new.flat.flat.squish.tap # weirder still, then maybe not 16:17
camelia rakudo-moar bcde48: OUTPUT«Memory allocation failed; could not allocate 58608 bytes␤»
16:17 larion joined
lizmat if we have 2 identical methods out of SupplyOperations, followed by one from Supply, and tap that, this happens 16:17
feels like a scoping issue on the SupplyOperations methods 16:18
moritz does SupplyOperations uses roles that reference outer lexicals?
lizmat yes to roles, but I don't see any outer lexical references (yet) 16:19
16:19 telex left 16:20 Sqirrel left, telex joined
nine lizmat: method live in SupplyOperations does indeed access the outer lexical scope 16:21
timotimo are they supposed to scope over any lexicals?
don't we usually have to push values in through the parameters?
nine $source is the argument of grep, not available in the anonymous class
timotimo i've recently been taught classes don't do closure semantics
nine exactly that's the problem here 16:22
lizmat ah, duh...
nine Should be $!source, not $source
lizmat is testing 16:23
nine Confirmed, bug fixed :)
timotimo nice 16:24
good job
timotimo afk for a bit
nine moritz++ # excellent hint!
lizmat thanks all: will commit fix (to other methods that suffer the same issue)
dalek rl6-roast-data: f791d0f | coke++ | / (9 files):
today (automated commit)
16:30
[Coke] tools/contributors.pl6 is broken. 16:32
needs some post-GLR love.
lizmat [Coke]: will do
16:33 vendethiel joined
thou hi, i'm wanting to do something like $match = Perl6::Grammar.parse($src); then use the parse to read out a list of identifiers, method names, pod chunks, etc. for making an indexed code browser. I've looked at Rakudo::Perl6::Parsing module but it's not working and seems to be a tough way to go (mucking with nqp-level stuff). Is there a better way to go about this? 16:33
dalek kudo/nom: d25a07a | lizmat++ | src/core/SupplyOperations.pm:
Fix for RT #126073

Problem was caused by thinko: when checking for a live source, the anonymous class should check the attribute, not the original parameter being passed.
  nine++ moritz++ timotimo++ for support in fixing this bug
16:35
timotimo the grammar class is unable to function without the actions and a world object
and then you will have something that also runs begin blocks and such
dalek kudo/nom: fb9f788 | lizmat++ | src/core/IO/Notification.pm:
Add convenience .IO to .IO.watch generated objects
16:36
timotimo thou, i have no good suggestion sadly 16:37
but if you take care to properly handle quoting, you can basically just do a simple search through the code i imagine 16:38
nine Perl cannot be properly parsed without running Perl code. That's been true for Perl 5 and is still true with Perl 6. It's where much of its power comes from. 16:40
16:41 larion left
thou Yeah, I was thinking of making a custom :actions() that would generate something more accessible, but I'm not sure how to proceed w/ that. 16:41
Or run STD from Inline::Perl5 and get something from there? :-D 16:42
16:44 telex left 16:45 telex joined, espadrine left
nine thou: that would probably have been the very last use case I could have come up with for Inline::Perl5 ;) 16:45
16:48 spider-mario joined
vendethiel timotimo: another way would be to wait until runtim e and collect info there 16:49
16:50 tokuhiro_ joined, dakkar left 16:52 abraxxa left
dalek kudo/nom: 3dfa856 | TimToady++ | src/core/List.pm:
call .cache on Seq if +@args used

Also, avoid returns in hot code.
16:52
ast: a78f857 | TimToady++ | S06-signature/slurpy-params.t:
test +args vs +@args distinctions

  +@args must be repeatable, while +args must not be
16:53
16:54 tokuhiro_ left 16:55 patrickz joined
masak evening, #perl6 16:58
dalek ast: 3165b8c | lizmat++ | S17-supply/on.t:
Add test for #126073
17:00
patrickz Hi. I'm currently trying to build MoarVM on Windows using MSVC. It builds successfully, but the resulting moar.exe fails with exitcode -1073741819 when doing `MoarVM\moar.exe --libpath="src/vm/moar/stage0" "src/vm/moar/stage0/nqp.moarvm" -e "print('hi')"`
yoleaux 12:18Z <tadzik> patrickz: ok, I'll look into it :)
patrickz When running without arguments it does give the usage.
17:01 tokuhiro_ joined
patrickz Giving wrong paths for libpath or the nqp.moarvm file gives error messages. 17:01
I suspect it's failing on some higher level but I have no idea how to look deeper. Any hints/ideas? 17:02
tokuhiro_: plenv is quite an awesome tool. Thanks for building it!
lizmat [Coke]: will look at contributors.pl after dinner& 17:03
patrickz tokuhiro_: I'm currently trying to mimic some of it for Perl6 in rakudobrew
17:06 tokuhiro_ left
[Coke] lizmat++ 17:08
17:13 zakharyas left
cdc  /pos 17:16
17:16 muraiki joined 17:20 espadrine joined 17:24 thou left, thou joined 17:25 espadrine left 17:30 lichtkind joined 17:34 Sqirrel joined 17:35 thou left 17:36 domidumont joined
dalek kudo/nom: 25fed44 | TimToady++ | docs/ChangeLog:
added TimToady changes since last release
17:36
TimToady has been busy... 17:37
muraiki "New `supply`, `whenever` and `react` blocks for easy reactive programming" neat 17:38
where can I find more info on those :D
flussence just noticed in the backlog a mention about `cpupower frequency-set -g performance`... I just have a shell alias that does `echo performance | sudo tee /sys/devices/system/cpu/*/cpufreq/scaling_governor` for the same result, if that helps anyone :)
(on gentoo, installing cpupower downloads an entire kernel tarball...)
moritz m: say 42 17:39
camelia rakudo-moar 3dfa85: OUTPUT«42␤»
moritz m: say 42␤say 43
camelia rakudo-moar 3dfa85: OUTPUT«5===SORRY!5=== Error while compiling /tmp/sQcATCcOCM␤Bogus postfix␤at /tmp/sQcATCcOCM:2␤------> 3say 427⏏5<EOL>␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ statement mo…»
moritz TimToady: ^^ have you seen that error message?
17:39 Ven joined
moritz TimToady: I'm pretty sure it used to say something like "two terms in a row" and hint that there might be a forgotten semicolon 17:39
TimToady yeah, I remember you mentioned it before 17:40
moritz ok, then I'll stop bugging you :-)
TimToady it's in my personal todo 17:41
17:41 patrickz left
moritz are +arg and +@arg in a signature the same? 17:42
TimToady the latter caches 17:43
moritz ok
TimToady there's even tests for it now :)
moritz m: sub f(+@arg) { say @arg }; f()
camelia rakudo-moar 3dfa85: OUTPUT«[]␤»
moritz had hoped for a "slurpy but at least one argument" syntax, and that would naturally have been +@arg
17:44 Ven left
TimToady I decided this was worth stealing the + for 17:44
there wasn't really any other good character left on the keyboard, and I got tired of typing *| 17:45
and it just looks a lot cleaner
FROGGS TimToady: do you also have the feeling that current rakudo+roast (including GMR) feels pretty RC0ish? 17:46
17:46 patrickz joined 17:48 perl6_newbee left, mullagainn joined
mullagainn so DBI has not been ported to perl6 yet? 17:48
FROGGS mullagainn: only DBIish 17:49
mullagainn ok
ilmari or you can use DBI via Inline::Perl5
mullagainn I know this question is asked a lot, but when will rakudo be stable? 17:50
jnthn Stability (e.g. "won't break my code a few months down the line") is more a property of the language than the implementation 17:51
yoleaux 15:04Z <FROGGS> jnthn: How hard would it be to allow rw attributes on CStructs?
17:51 telex left
FROGGS that too, yes 17:51
17:51 telex joined
jnthn So "when the language design is", and that's what we're aiming to commit to by December this year. 17:51
mullagainn well let me re-phrase. When will it be production ready?
jnthn Given a bunch of people already seem to be using it in production, "now"? :) 17:52
mullagainn ok, so finalization of the language hopes to be done by December
Oh, I did not know that
flussence
.oO( has DBDI gone the way of parrot? )
FROGGS mullagainn: given that we are preparing "The Release™", i expect that all will settle down in the next three months 17:54
jnthn mullagainn: Trouble is that people's requirements for "production ready" vary a good bit. But are people using it to do stuff useful to their $dayjob? Sure.
colomon has actually been using p6 to do useful $dayjob work for about six years now… 17:55
mullagainn I dont see a lot of XS modules out there. 17:56
muraiki yeah, "production ready varies." I've found a few bugs but am essentially using p6 in production (well, I will be once this product launches)
TimToady XS modules may only be accessed via Inline::Perl5 17:57
Perl 6 has a different way to get to C
mullagainn oh ok
TimToady much, much cleaner
17:58 Sqirrel_ joined
muraiki yeah, the perl6 C stuff is hardcore 17:58
17:58 Sqirrel_ left
muraiki a beauty to behold 17:58
mullagainn yeah. I use a lot of it though
17:58 Sqirrel_ joined
mullagainn SNMP vs Net::SNMP is about a 300% difference in speed 17:58
arnsholt NativeCall (the C interaction stuff) is similar to Python's cffi, if you're familiar
17:59 Sqirrel left
_itz or perl5 NCI 17:59
muraiki mullagainn: github.com/jnthn/zavolaj#getting-started 18:00
skids ergh now I'm reminded that in addition to multidim and the new async stuff I also wanted to check out the new C++ thing. Much indecision.
mullagainn So are there any major performance upgrades in p6, specifically in the realm of network io? 18:01
FROGGS mullagainn: we currently focus more on getting the semantics right... I expect that we care more about performance in a few months, not much earlier 18:02
mullagainn That is actually a really bad question lol.
poorly worded i guess.
skids More potential for performance than actually-optimized so far. But the async code is likely to work out well in that area.
mullagainn I have been using async for years in p5. 18:03
raiph mullagainn: have you tried using Inline::Perl5 with one or more of the XS modules you regularly use with Perl 5?
mullagainn But for almost everything I have to handle everything myself.
No. I am really just trying to learn more before I get started. 18:04
Kind of assess my challenges.
muraiki coming from perl5, the async stuff in perl6 is, in my opinion, the best part :)
mullagainn and at least we get real threads now :) 18:05
raiph mullagainn: it would be very helpful to the P6 effort, even if not to yourself, if you simply hammered on being able to use your existing P5 mainstays with P6
mullagainn Oh like re-writing all those modules I know and love? 18:06
muraiki I should try loading our p5 stack via Inline::Perl5 again. I made it fairly far a few months ago... but something was causing it to die
mullagainn: no, you can simply use them via Inline::Perl5
skids mullagin: no just trying to use them with Inline::Perl5 and keeping nine++ very busy fixing anything that's broke :-)
muraiki I was using Perl 5's YAML in Perl 6 without a problem
raiph mullagainn: and in particular seeing if you can wrap P6 PAC (parallelism/async/concurrency) around your existing P5 codebase
muraiki raiph: I think that's where my things started exploding
mullagainn hmm, that is a good idea 18:07
any good reads on p6 PAC? 18:08
raiph www.reddit.com/r/perl6/comments/3i...in_perl_6/
mullagainn What is all this in the Rakudo Readme about Java? 18:10
18:11 Peter_R joined
nine muraiki: I still don't know if I need to adapt Inline::Perl5 better to the post GLR world. I works fine, but Perl 5 APIs may have become a bit more difficult to use due to the new (non-)flattening rules. 18:11
muraiki: feedback would be greatly appreciated
raiph mullagainn: in the near term the Rakudo Perl 6 compiler is focused on running on a VM designed for P6 called MoarVM 18:12
mullagainn what on earth is that? 18:13
raiph mullagainn: you could treat that as an unimportant implementation detail if you prefer
moarvm.org 18:14
jnthn mullagainn: Rakudo can target different VMs; one of them is the JVM.
If you don't want to do that, you can ignore those bits of the README :) 18:15
raiph mullagainn: what are some examples of the P5 XS modules you regularly use? 18:17
mullagainn SNMP (which is packaged in the net-snmp toolset) 18:18
JSON::XS
[Coke] (christmas) we need to decide what platforms we are officially targeting, also, as we don't have parity on OS X at the moment.
mullagainn Redis::Fast
jnthn [Coke]: In what areas, ooc? 18:19
raiph mullagainn: Could you try, say, JSON::XS with your Rakudo and confirm it works fine for you?
mullagainn yes, I am compiling Rakudo now 18:20
want me to pm you or group?
tony-o i'd be interested to know
mullagainn k 18:21
18:23 leont joined
FROGGS [Coke]: I'd think we target windows, linux and osx (64bits, as usual) mainly 18:24
[Coke] jnthn: many nativecall failures in make test.
raiph mullagainn: to this channel at least at first
mullagainn getting issues using panda
Use of uninitialized value of type Str in string context in method update at
[Coke] rt.perl.org/Ticket/Display.html?id=124351
FROGGS [Coke]: isnt there just one about an rt...? 18:25
[Coke]: weird... that did work on lizmat++'s box
raiph mullagainn: do you know about rakudobrew?
[Coke] adds that one to the christmas ticket...
mullagainn I do not
[Coke] FROGGS: happy to give you more diagnostic info from my box.
mullagainn like perlbrew but p6? 18:26
[Coke] I am on 10.10.5. Just updated xcode today, will double check that.
FROGGS [Coke]: ahh nvm, the original output is outdated...
mullagainn This is perl6 version 2015.07.2 built on MoarVM version 2015.07
FROGGS [Coke]: t/04-nativecall/16-rt125408.t fails everywhere
[Coke]: and the C++ tests do pass now
raiph mullagainn: sorta, perl6advent.wordpress.com/2014/12/...akudobrew/
[Coke] double checks. 18:27
FROGGS [Coke]: I'm going to look into t/04-nativecall/16-rt125408.t now
[Coke]++
[Coke] FROGGS++!
muraiki nine: sounds good, I'll hopefully have some time to test post-GLR Inline::Perl5 stuff soon
[Coke] FROGGS: closed 124351. 18:29
FROGGS cool :o) 18:30
[Coke] RT: 1,015 tickets; 3 [GLR] tickets, 3 Christmas blockers, 8 testneeded
is there a consensus that RT should be for bugs, NYIs that are spec'd, but not for feature requests? 18:33
moritz FROGGS: I've notices there aren't any module loader changes in docs/ChangeLog
FROGGS moritz: ups
moritz FROGGS: is there anything user-visible worthy of a changelog entry? 18:34
[Coke] (basically asking for RT #61602)
FROGGS I'll add a line
[ptc] m: my (1, -1 X 2, -2)
camelia rakudo-moar 25fed4: OUTPUT«5===SORRY!5=== Error while compiling /tmp/b1e8BilGbf␤Malformed parameter␤at /tmp/b1e8BilGbf:1␤------> 3my (1,7⏏5 -1 X 2, -2)␤ expecting any of:␤ formal parameter␤»
moritz [Coke]: feature requests should be issues on perl6/specs on github
[ptc] m: my say(1, -1 X 2, -2)
camelia rakudo-moar 25fed4: OUTPUT«5===SORRY!5===␤Type 'say' is not declared␤at /tmp/gAo_eL__sh:1␤------> 3my say7⏏5(1, -1 X 2, -2)␤Malformed my␤at /tmp/gAo_eL__sh:1␤------> 3my7⏏5 say(1, -1 X 2, -2)␤␤»
[ptc] m: say(1, -1 X 2, -2)
camelia rakudo-moar 25fed4: OUTPUT«((1 2) (1 -2) (-1 2) (-1 -2))␤»
FROGGS moritz: a user can now CompUnit.new(...).load
[ptc] m: say (1, -1 X 2, -2).flat 18:35
camelia rakudo-moar 25fed4: OUTPUT«(1 2 1 -2 -1 2 -1 -2)␤»
[Coke] ok. I'm going to close 61602 with a final note pointing back to the already extant spec ticket.
18:36 andreoss left
FROGGS uhh, github has issues 18:37
[Coke] m: say :10(':16<99>')
camelia rakudo-moar 25fed4: OUTPUT«153␤»
jnthn is glad to see TimToady++ decided the .gist behavior for allomorphs was just the string representation :) 18:38
[Coke] m: say :10('99'), :16('99'), :10('0x99'), :10(':16<99>') #RT #77624
camelia rakudo-moar 25fed4: OUTPUT«99153153153␤»
[Coke] m: say (:10('99'), :16('99'), :10('0x99'), :10(':16<99>')).join(":") #RT #77624
camelia rakudo-moar 25fed4: OUTPUT«99:153:153:153␤»
[ptc] m: (1, -1 X 2, -2), (2, -2 X 1, -1) 18:39
camelia ( no output )
[ptc] m: say (1, -1 X 2, -2), (2, -2 X 1, -1)
camelia rakudo-moar 25fed4: OUTPUT«((1 2) (1 -2) (-1 2) (-1 -2))((2 1) (2 -1) (-2 1) (-2 -1))␤»
[ptc] sorry for the mistakes; I must be tired... 18:40
any ideas how I could flatten that nicely?
I've tried ((1, -1 X 2, -2).flat, (2, -2 X 1, -1).flat).flat which works, but is horrible...
TimToady jnthn: well, ShimmerFairy++ for pushing the allomorph stuff through
moritz m: say flat (1, -1 X 2, -2), (2, -2 X 1, -1)
camelia rakudo-moar 25fed4: OUTPUT«(1 2 1 -2 -1 2 -1 -2 2 1 2 -1 -2 1 -2 -1)␤»
moritz [ptc]: ^^ like this?
[Coke] m: say <1 2 3 4 5 10>.max
camelia rakudo-moar 25fed4: OUTPUT«10␤»
jnthn TimToady: Oh, indeed. I'm generally happy with it, just wasn't with that part :) 18:41
[ptc] moritz: yes. I'm just blown away that I never stumbled across that option!
[Coke] m: <1 2 3>[2].WHAT.say
camelia rakudo-moar 25fed4: OUTPUT«(IntStr)␤»
[ptc] m: my @directions = (1, -1 X 2, -2), (2, -2 X 1, -1); say flat @directions 18:42
camelia rakudo-moar 25fed4: OUTPUT«(((1 2) (1 -2) (-1 2) (-1 -2)) ((2 1) (2 -1) (-2 1) (-2 -1)))␤»
[ptc] moritz: ^^ that's the original problem I'm trying to solve. Maybe the code doesn't need to be an array anymore? 18:43
dalek kudo/nom: 78c4805 | FROGGS++ | docs/ChangeLog:
mention CompUnit changes, moritz++
moritz m: my @directions = (1, -1 X 2, -2), (2, -2 X 1, -1); say flat |@directions 18:44
camelia rakudo-moar 25fed4: OUTPUT«(((1 2) (1 -2) (-1 2) (-1 -2)) ((2 1) (2 -1) (-2 1) (-2 -1)))␤»
moritz huh
m: my @directions = (1, -1 X 2, -2), (2, -2 X 1, -1); say @directions.flat
camelia rakudo-moar 25fed4: OUTPUT«(((1 2) (1 -2) (-1 2) (-1 -2)) ((2 1) (2 -1) (-2 1) (-2 -1)))␤»
moritz thorougly surprised
[ptc] another question: why does .flat seem to work differently to the flat function?
[Coke] m: say ("foo" & "a nice old foo") ~~ /foo/ 18:45
camelia rakudo-moar 25fed4: OUTPUT«「all(Failure.new(exception => X::OutOfRange.new(what => "Start argument to substr", got => "5", range => "0..3", comment => "use *5 if you want to index relative to the end"), backtrace => Backtrace.new), "e o")」␤»
[ptc] m: my $directions = (1, -1 X 2, -2), (2, -2 X 1, -1); say $directions.flat
camelia rakudo-moar 25fed4: OUTPUT«(1 2 1 -2 -1 2 -1 -2)␤»
[ptc] hrm, my rakudo doesn't give that output...
jnthn moritz: An Array already is flat, so it returns identity
[ptc] thinks an update of rakudo is probably in order
[Coke] r: say "foo"[1..*] 18:46
camelia rakudo-{moar,jvm} 25fed4: OUTPUT«()␤»
[Coke] r: say "foo"[2] 18:47
[ptc] s/update/local update/
camelia rakudo-{moar,jvm} 25fed4: OUTPUT«Index out of range. Is: 2, should be in 0..0␤ in block <unit> at /tmp/tmpfile:1␤␤Actually thrown at:␤ in block <unit> at /tmp/tmpfile:1␤␤»
[Coke] r: say "foo"[2..*]
camelia rakudo-{moar,jvm} 25fed4: OUTPUT«()␤»
raiph mullagainn: I have a panda error following a rakudobrew refresh; investigating
[Coke] (RT #112216)
m: use MONKEY_TYPING; augment class Any { method foo {say "FOO"}}; "barf".foo 18:48
camelia rakudo-moar 25fed4: OUTPUT«===SORRY!===␤Could not find MONKEY_TYPING in any of:␤ file#/home/camelia/.perl6/2015.07.1-973-g25fed44/lib␤ inst#/home/camelia/.perl6/2015.07.1-973-g25fed44␤ file#/home/camelia/rakudo-inst-2/share/perl6/lib␤ file#/home/camelia/rakudo-inst-2/s…»
[Coke] m: use MONKEY-TYPING; augment class Any { method foo {say "FOO"}}; "barf".foo
camelia rakudo-moar 25fed4: OUTPUT«Method 'foo' not found for invocant of class 'Str'␤ in block <unit> at /tmp/c5n5raKkz3:1␤␤»
[Coke] jnthn: ^^ should that work?
(if not, please gently reject RT #115618) 18:49
FROGGS it would have to clear the method cache somehow, right?
psch for sub-classes even
FROGGS m: use MONKEY-TYPING; augment class Any { method foo {say "FOO"}}; augment class Str { }; "barf".foo
camelia rakudo-moar 25fed4: OUTPUT«FOO␤»
psch ISTR some discussion that it will never cascade downwards automatically 18:50
jnthn [Coke]: Eventually, yes
But we're not so smart at the cache invalidation yet
psch probably misunderstood or misremembers the discussion then
dalek kudo/nom: f0044bc | (Stefan Seifert)++ | docs/ChangeLog:
Add ChangeLog entry for backwards incompatible lines() fix
18:51
FROGGS cache invalidation is probably the biggest problem in software engineering space :o)
dalek kudo/nom: 1e70188 | TimToady++ | src/Perl6/World.nqp:
notice TTIAR across line before Bogus postfix
TimToady moritz: ^^ 18:52
18:52 brrt joined
TimToady lunch & 18:53
18:53 clearkruti joined
[Coke] RT: 1,014 tickets; 4 [GLR] tickets, 3 Christmas blockers, 11 testneeded 18:53
18:53 brrt left, brrt joined 18:57 mullagainn left 19:02 tokuhiro_ joined
nine m: my $a = 1; my $b = 2; $a = (|$a, $b); say $a[0] 19:02
camelia rakudo-moar 25fed4: OUTPUT«1␤»
nine m: my $a = 1; my $b = 2; $a = ($a, $b); say $a[0] 19:03
camelia rakudo-moar 25fed4: OUTPUT«(\List_139651993582776 = (List_139651993582776 2))␤»
nine I think we can safely reject this ticket. The bug has never been in list assignment and always in .perl. And it's fixed even there. 19:04
[ptc] this is weird. I just rebuilt moar, nqp, rakudo, however this code works...
m: my $directions = (1, -1 X 2, -2), (2, -2 X 1, -1); say $directions.flat
camelia rakudo-moar 25fed4: OUTPUT«(1 2 1 -2 -1 2 -1 -2)␤»
19:04 FROGGS left
[ptc] whereas in my rakudo I get This Seq has already been iterated, and its values consumed (within the repl) 19:04
19:05 diana_olhovik_ joined
[ptc] m: my $directions = (1, -1 X 2, -2), (2, -2 X 1, -1); my $blah = $directions.flat; 19:06
camelia ( no output )
nine [ptc]: I can see why: the repl prints the result of the last statement, thereby consuming the Seq returned by X
psch: when you then try to print them again, the Seq will rightfully complain
brrt .tell FROGGS i'm very interested to see what code has a JIT bug
yoleaux 14 Sep 2015 22:05Z <meisl> brrt: oh-wow...
brrt: I'll pass your message to FROGGS.
14 Sep 2015 22:05Z <meisl> brrt: never had such a nice welcome - and such an immediate one!
PerlJam someone mentioned earlier about p6 smartly calling .cache Maybe this is why
yoleaux 14 Sep 2015 22:06Z <meisl> brrt: I *am* interested in Aho-style tiling, will have a look.
[ptc] nine, psch: thanks. Was wondering what I was missing...
brrt .botsnak 19:07
19:07 tokuhiro_ left
brrt .botsnack 19:07
yoleaux :D
psch nine, [ptc]: ...i'm not sure how i got involved in that discussion :S
oh, probably inattentive tabbing! 19:08
muraiki hrm, I just updated rakudobrew, built perl6, and build panda. if I try to `panda install` anything, I get a type check failure. is anyone else running into this?
dalek pan style="color: #395be5">perl6-examples: 5963e85 | paultcochrane++ | categories/99-problems/P91-edpratomo.pl:
[99probs] fixed P91-edpratomo post GLR
19:08 brrt left 19:09 FROGGS joined
PerlJam [ptc]: That code would work in the repl too as long as you put them on one line :) 19:09
raiph muraiki: same here
muraiki raiph: :(
I'll open an issue on panda 19:10
raiph muraiki: just installed zef (for first time) to see how that does...
PerlJam bah!
[Coke] finds i.imgur.com/uW4Zhm7.png which may be relevant to this discussion.
19:10 brrt joined
PerlJam muraiki: I think I just fixed it, if someone would merge my PR 19:10
dalek kudo/nom: ecdf815 | skids++ | src/Perl6/ (2 files):
Do not shortcut anon capture param if it has subsig or post constraint

Fixes :(| ($f)) but due to a parse issue, not :(|($f))
kudo/nom: fb852ed | niner++ | src/Perl6/ (2 files):
Merge pull request #536 from skids/binder2

Do not shortcut anon capture param if it has subsig or post constraint
muraiki PerlJam: oh, thanks!
PerlJam or give me a commit bit on the panda repo
19:11 patrickz left, domidumont left
muraiki weird 19:11
psch so S03-metaops/hyper.t breaks on test 312 (l:907) but writing lines 904 to 908 to a file and running that files on the equivalent of l:908...
muraiki it says your commit was merged 3 hours ago
[ptc] PerlJam: thanks :-) I think I'm a bit mentally tired to think 100% straight atm.
PerlJam muraiki: that was a different one. I just realized that the same fix was needed in multiple places.
19:11 thou joined
[ptc] probably time to avoid the keyboard 19:11
muraiki perljam: oh, I get it. thanks again
psch s/that files on/that file breaks on/ 19:12
FROGGS PerlJam: done 19:13
yoleaux 19:06Z <brrt> FROGGS: i'm very interested to see what code has a JIT bug
PerlJam FROGGS++ danke 19:14
FROGGS PerlJam: gern geschehen :o)
psch gist.github.com/peschwa/149552262ee830d90fc5 # spooky state hiding somewhere in HYPER..? 19:17
19:18 [Sno] left, mullagainn joined
mullagainn back 19:18
still getting erros with panda
Type check failed in assignment to '$prefix'; expected 'Str' but got 'List' 19:19
muraiki mullagainn: it was just fixed, hehe
mullagainn lolol
muraiki muraiki: do `rakudobrew build-panda` assuming you are using rakudobrew
mullagainn ok
lol
SInce I just did, do I need to clear it out or anything? 19:21
PerlJam Maybe "rakudobrew nuke moar" to start from scratch, but I wouldn't think you'd need to.
raiph hmm. panda still not working for me; doing rakudobrew nuke moar and trying over
dalek kudo/nom: 9363e00 | (Steve Mynott)++ | src/core/Deprecations.pm:
remove Sept part of deprecated message
19:22
kudo/nom: 0b47248 | (Will Coleda)++ | src/core/Deprecations.pm:
Merge pull request #537 from stmuk/nom

remove Sept part of deprecated message
19:23 brrt left 19:24 mullagainn left
nine psch: what do you mean? 19:26
psch nine: i think "< nine> psch: when you then try to print them again, the Seq will rightfully complain" was mis-addressed to me :) 19:27
(the gist is pretty clear, but that's probably not what you're asking about...) 19:28
raiph mullagainn, muraiki, PerlJam: needed a nuke for me, panda now seems to be working again 19:29
nine psch: oh the first message was a tab completion screw up. The second however was about the spooky state thing
muraiki raiph: yeah, FROGGS merged PerlJam's fix. :) 19:30
psch nine: oh, okay. the first three evals in the gist show <<+>>, <<+<< and >>+>> not-failing when run alone
raiph muraiki: just sayin' I had to do a nuke to get it to work
PerlJam doesn't "rakudobrew build-panda" fetch the latest one? 19:31
psch nine: the last two show different orders of those three hypers fail
muraiki PerlJam: raiph: Yeah, `rakudobrew build-panda` fixed it for me without a nuke
assuming nuke means erase everything and start over from scratch
nine psch: I cannot reproduce those failures here
psch nine: are you running r-j? :)
nine psch: ah, that's it! 19:32
psch j: my @a = 1, 2, 3; @a >>+>> (); @a <<+>> (); @a <<+<< ()
camelia rakudo-jvm 1e7018: OUTPUT«Cannot call HYPER(Sub+{<anon>}+{Precedence}, Int, Int, Int); none of these signatures match:␤ (&op, \left, \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \left, Associative:D \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \l…»
nine j: my @a = 1, 2, 3; @a >>+>> (); @a <<+>> (); 19:34
camelia ( no output )
nine j: my @a = 1, 2, 3; @a >>+>> (); @a <<+<< ()
camelia ( no output )
nine j: my @a = 1, 2, 3; @a <<+>> (); @a <<+<< ()
camelia ( no output )
nine weird
psch yeah 19:37
it's the third one that triggers it
but like, not really i guess
nine j: my @a = 1, 2, 3; @a >>+>> (); @a >>+>> (); @a >>+>> ();
camelia rakudo-jvm 1e7018: OUTPUT«Cannot call HYPER(Sub+{<anon>}+{Precedence}, Int, Any, Int); none of these signatures match:␤ (&op, \left, \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \left, Associative:D \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \l…»
psch 'cause in hyper.t there's a lot more than three
nine j: my @a = 1, 2, 3; @a >>+>> (); @a >>+>> ();
camelia rakudo-jvm 1e7018: OUTPUT«Cannot call HYPER(Sub+{<anon>}+{Precedence}, Int, Any, Int); none of these signatures match:␤ (&op, \left, \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \left, Associative:D \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \l…»
nine twice in the same direction?
psch oh 19:38
nine j: my @a = 1, 2, 3; @a <<+<< (); @a <<+<< ();
psch i hadn't looked for that yet!
camelia ( no output )
dalek c: b0a6ed3 | (David H. Adler)++ | lib/Language/variables.pod:
Fixed code example for C<state>
psch j: my @a = 1, 2, 3; @a <<+>> (); @a <<+>> ();
nine j: my @a = 1, 2, 3; @a >>+<< @a; @a >>+<< @a;
camelia rakudo-jvm 1e7018: OUTPUT«Cannot call HYPER(Sub+{<anon>}+{Precedence}, Int, Int, Int); none of these signatures match:␤ (&op, \left, \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \left, Associative:D \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \l…»
( no output )
psch uhh
oh
raiph mullagainn: PerlJam++'s fix plus muraiki++'s `rakudobrew build-panda` worked for me; are you too now up and running with panda? 19:39
psch j: my @a = 1, 2, 3; @a <<+<< (); @a <<+>> ();
camelia ( no output )
psch j: my @a = 1, 2, 3; @a >>+>> (); @a <<+>> ();
camelia rakudo-jvm 1e7018: OUTPUT«Cannot call HYPER(Sub+{<anon>}+{Precedence}, Int, Int, Int); none of these signatures match:␤ (&op, \left, \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \left, Associative:D \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \l…»
TimToady j: my @a = 1, 2, 3; @a >>+>> (); $@a <<+>> (); 19:40
camelia rakudo-jvm 1e7018: OUTPUT«Cannot call HYPER(Sub+{<anon>}+{Precedence}, Int, Int, Int); none of these signatures match:␤ (&op, \left, \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \left, Associative:D \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \l…»
psch that looks like it's about two dwimmy rhs..? (assuming "it points outward to dwim" is the right semantics)
TimToady j: my @a = 1, 2, 3; $@a >>+>> (); $@a <<+>> ();
camelia rakudo-jvm 1e7018: OUTPUT«Cannot call HYPER(Sub+{<anon>}+{Precedence}, Int, Int, Int); none of these signatures match:␤ (&op, \left, \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \left, Associative:D \right, :$dwim-left, :$dwim-right)␤ (&op, Associative:D \l…»
19:43 hlelo joined, mullagainn joined
mullagainn panda worked 19:43
19:43 telex left 19:44 telex joined
mullagainn use Inline::Perl5; 19:45
use JSON::XS::from<Perl5>;
my $json = JSON::XS.new();
TimToady m: say 42␤say 43
camelia rakudo-moar 1e7018: OUTPUT«5===SORRY!5=== Error while compiling /tmp/HfGJNIZa9Z␤Two terms in a row across lines (missing semicolon or comma?)␤at /tmp/HfGJNIZa9Z:2␤------> 3say 427⏏5<EOL>␤ expecting any of:␤ infix␤ infix stopper␤ postfi…»
mullagainn ===SORRY!=== 19:46
Could not find JSON::XS::from in any of:
raiph mullagainn: single colon before the 'from'
mullagainn aah
ok
thanks
PerlJam That could probably use a more awesome message as it's likely to be a commonish error. 19:47
mullagainn no errors
slow to start though
I tried to load SNMP.pm and I got a Do no know how to load code from perl5 19:48
grr
another typo
I am not used to using capital P in perl
19:49 colomon left, colomon_ joined
raiph mullagainn: fwiw, decent Rakudo module loading performance needs module precompilation but precomp was disabled a week or so ago pending changing where it happens 19:49
mullagainn oh, so it will be fixed eventaually 19:50
PerlJam all things will be "fixed" eventually :-)
19:51 hlelo left
mullagainn lol 19:51
skids Ohhh... I *thought* something seemed slower these last couple of days but couldn't put my finger on it.
dalek c: 2fda12d | paultcochrane++ | lib/Type/List.pod:
Fix example output for map post GLR
19:54
[Coke] wonders why the first part of the nqp build step is to tag and push. That seems premature.
er, nqp -release- steps
I'm going to remove parrot from the nqp release information. No one is currently maintaining that, and there are known test failures. 19:56
raiph mullagainn: do you think either or both of the ::from and <perl5> errors need a better error message?
[Coke] ... ugh. except there's stuff that exists only in the parrot Makefile. 19:57
19:57 telex left
mullagainn is there something wrong with this syntax? 19:57
y $sess = SNMP::Session.new( 'DestHost' => $ip, 'Community' => $community, 'Version' => 2 ); 19:58
my**
19:58 telex joined
mullagainn raiph: yes probably 19:59
I am getting this. 20:00
Could not find symbol '&Session'
in method AT-KEY at /opt/nms/.rakudobrew/moar-nom/install/share/perl6/runtime/CORE.setting.moarvm:1
in block <unit> at ./json:15
[ptc] m: say ((1, 2), <a b>).flatmap(&uc).join('|'); 20:01
camelia rakudo-moar 0b4724: OUTPUT«1 2|A B␤»
[ptc] m: say ((1, 2), <a b>).map(&uc).join('|');
camelia rakudo-moar 0b4724: OUTPUT«1 2|A B␤»
[ptc] hrm, according to the docs, flatmap should give 1|2|A|B as output 20:02
nine mullagainn: that's a nown limitation. Try use SNMP:from<Perl5>; use SNMP::Session:from<Perl5>; 20:03
mullagainn: I only create a Perl 6 class for the use'd package, not any sub-packages that may load.
mullagainn oooh 20:06
ok
nine Unfortunately it's not exactly straight forward to find out which packages are loaded in a use statement.
mullagainn Can't locate SNMP/Session.pm in @INC ( 20:07
dalek kudo/nom: 643eb6d | TimToady++ | src/Perl6/Grammar.nqp:
correctly detect malformed parameters
mullagainn SNMP::Session is within SNMP
nine mullagainn: ok, till that's fixed, you can always do; use Inline::Perl5; my $p5 = Inline::Perl5.new; my $session = $p5->invoke('SNMP::Session', 'new', DestHost => $ip, Community => $community, Version => 2); 20:08
mullagainn ok. 20:09
thanks. That will work ok
nine This worked for me: perl6 -e 'use Inline::Perl5; my $p5 = Inline::Perl5.new; $p5.use("SNMP"); my $session = $p5.invoke("SNMP::Session", "new", DestHost => "127.0.0.1", Community => "public", Version => 2)' 20:10
20:14 darutoko left
[Coke] raiph: when opening a ticket, if there's not that much text, inline the text so it's easy for someone to write a test or retry the bug without having to click through. 20:14
linking back for more discussion is always awesome, keep doing that.
raiph++
PerlJam nine: how do you create Perl 6 classes at all? I mean, someone could name a file Foo.pm that only has package Bar in it (no Foo package) 20:15
20:15 telex left 20:16 telex joined, telex left
nine PerlJam: currently I just create a Foo class. Once I implement support for actually trying to figure out which packages were loaded, this will be gone. 20:18
20:18 telex joined
muraiki nine: I got a "*** longjmp causes uninitialized stack frame ***" crash when trying to run the same perl5 function in two different start blocks :( 20:19
20:20 diakopter joined, kenhat joined
muraiki unfortunately this is proprietary code so I can't share it... I'm not really certain what I can give you to help diagnose what happened 20:20
dalek ast: a8f9dca | TimToady++ | S13-overloading/typecasting- (3 files):
remove obsolete overloading tests
20:21
kudo/nom: be0fdeb | TimToady++ | t/spectest.data:
remove obsolete overloading tests
20:22
nine muraiki: short answer: don't run Perl 5 code from different threads. Perl 5 code is not thread safe.
20:23 espadrine joined
muraiki ah, I thought somebody before said to try using perl5 code with the various async/concurrent features :( 20:23
that's what I figured, though
nine muraiki: you may create multiple Perl 5 objects (provided your Perl is compiled with -DMULTIPLICITY) and use them in different threads. I haven't tried that in almost a year though, so there may be bugs.
multiple Inline::Perl5 objects that is 20:24
muraiki ah yeah, I think I had done that about 6 months ago
leont Yeah, using one interpreter in multiple threads would be "fun"
PerlJam nine: you could inspect (keys %::) recursively to see what symbols show up when you use a P5 module. I don't know if you'll have enough information to decide which you should create as P6 classes though.
20:24 pyrimidine left
muraiki it makes sense now; I assumed I'd automatically get one interpreter per p6 thread 20:24
nine PerlJam: certainly worth a try
20:24 pyrimidine joined
nine Ok, off to bed now. Good night #perl6! 20:24
muraiki night
PerlJam nine: good night 20:25
TimToady o/
dalek ast: c702610 | usev6++ | S02-literals/radix.t:
Mention RT #77624 for existing test
20:25 telex left, kaare_ left
masak good nine, night 20:26
20:26 telex joined
muraiki night nine good 20:26
20:26 lizmat left 20:27 lizmat joined
btyler_ muraiki: one thing I was playing around with a while ago was to use OO::Monitors to manage a perl5 object worker pool, which is nice if you don't want to have an unbounded number of perl5 interpreters in your process :) 20:28
maybe I can find the gist, moment
muraiki btyler_: oh, that sounds really nice
masak m: use Test; done
camelia ( no output )
dalek kudo/nom: 47b4c1a | lizmat++ | tools/contributors.pl6:
Fix contributers script, it needed a slip
masak ...why doesn't `done` output "1..0" in the above case?
lizmat [Coke] ^^^ 20:29
masak m: use Test; ok 1; done;
camelia rakudo-moar 643eb6: OUTPUT«ok 1 - ␤»
masak this seems to be a recent regression.
lizmat masak: because it's called "done-testing" nowadays
m: done
camelia ( no output )
masak lizmat: :( :( :(
lizmat done is part of the supply sugar
masak lizmat: back several years ago we managed to rename it from `done_testing` to `done` :(
lizmat m: use Test; done-testing 20:30
camelia rakudo-moar 643eb6: OUTPUT«1..0␤»
lizmat yes, I know
but it was felt that the syntactic sugar for supplies was more needed than "done" from Test.pm
masak ok.
lizmat since you probably shouldn't be using "done", but plan your tests anyway :-)
jnthn Yes, I think the suply sugar deserved the keyword 20:31
[Coke] lizmat++ #contribs
20:31 diakopter left
masak lizmat: whether to plan one's tests or to use `done` still seems like a matter of taste to me. 20:31
lizmat I disagree: a plan will make sure that all tests are actually run
masak so will `done`. 20:32
lizmat and give you a fail if they didn't
dalek kudo/nom: 1fdf9df | (Steve Mynott)++ | docs/ChangeLog:
ChangeLog now has `dd` now displays variable type
kudo/nom: 48c03ad | (Will Coleda)++ | docs/ChangeLog:
Merge pull request #538 from stmuk/nom

ChangeLog now has `dd` now displays variable type
[ptc] masak: isn't it done-testing now?
masak [ptc]: yes. I still haven't quite updated my brain.
[Coke] jnthn: github.com/rakudo/rakudo/pull/306 may be rejectable as pre-GLR.
lizmat masak: no, it doesn't, it only means you don't expect any tests after 20:33
20:33 telex left
lizmat if for some reason a loop decides to run fewer tests than you thought, done-testing will *not* catch that 20:33
jnthn masak: I have ran into cases where the number of tests run actually changed due to a bug
masak: And the explicit plan in the file caught it.
masak I agree that an explicit `plan` makes sense in the spectests.
PerlJam (explicit plan)++
masak for my module tests, I've tended to prefer `done`. 20:34
jnthn I think done-testing is useful when developing a test file
Then when it's "stable" drop a plan in
PerlJam jnthn++ exactly
20:34 telex joined
lizmat jnthn: problem with that approach, is that by the time it's stable, you forget to put a plan in 20:34
FROGGS also true 20:35
raiph Thanks [Coke]++ # for RT guidance
lizmat so personally, I *always* put a plan in from the start, and ignore the wrong number of tests message in the beginning
jnthn lizmat: Do I? :P :P
lizmat: Sure, people should do what works for them :)
leont One other advantage of done is that it prevents merging errors
skids ok 0, "you will always fail 1 test until you put a plan in"; done-testing();
lizmat jnthn: well, *I* tend to forget, so I taught myself to use a plan from the very beginning :-)
leont You can't get a conflict on the number if there is no number
arnsholt I usually put in a plan at the top, and just keep updating the numbers when the test harness reminds me to =) 20:36
That's what TDD is for, no? =)
lizmat arnsholt++ :-)
jnthn Anyways, if we didn't steal "done" for the supply stuff, we'd have had to go rename *that*, which would also have cauesd breakage...
(we were already using the name "done", just didn't have a function with that name)
skids remembers Array.plan is still XX GLR. 20:37
masak so what does `done` actually do outside of a `supply` block?
dalek p: b3d6658 | coke++ | VERSION:
bump VERSION to 2015.09
p: f50929a | coke++ | docs/release_guide.pod:
Explicitly test the 2 backends we expect to pass
p: 885986b | coke++ | docs/release_guide.pod:
Fix typo
p: 67653b8 | coke++ | tools/build/MOAR_REVISION:
Use latest MOAR release
jnthn masak: Nothing, just as "last" doesn't do anything outside of a loop
PerlJam masak: throws a CONTROL_DONE
masak m: last
camelia rakudo-moar 643eb6: OUTPUT«last without loop construct␤ in block <unit> at /tmp/boWVx0Xrit:1␤␤»
jnthn Well, yeah, it whines you used it in the wrong place
masak jnthn: at least it warns.
`done` does nothing.
jnthn Doesn't do anything *useful*
m: done
camelia ( no output )
jnthn Hm
raiph mullagainn: I've added RTs #126087 and #126088 to cover the ::from and :from<perl5> surprises you had 20:38
lizmat m: emit
camelia rakudo-moar 643eb6: OUTPUT«5===SORRY!5=== Error while compiling /tmp/8lnsfK4qIr␤Calling emit() will never work with declared signature (\value)␤at /tmp/8lnsfK4qIr:1␤------> 3<BOL>7⏏5emit␤»
jnthn m: emit 1;
camelia ( no output )
lizmat m: emit 42
camelia ( no output )
lizmat :-)
masak jnthn: which is unfortunate, because I suddenly had a lot of failing tests but no reason why.
jnthn Wonder why that gets dropped
masak well, confused test files, rather.
jnthn It *is* throwing a control exception
masak if you say so :P
jnthn Read the bloody code if you don't belive me 20:39
masak anyway, it seems you agree with me it should do something ;)
lizmat
.oO( stop stabbing the code! )
20:39 erxeto joined
jnthn Aye 20:39
masak submits rakudobug 20:40
20:41 TimToady left
jnthn oh, damn 20:42
print_control just drops unknown control exceptions on the floor
[Coke] nqp release done, including gpg signing.
20:43 diakopter joined, TimToady joined, telex left
[Coke] don't see any difference in the signed tag on github. 20:43
dalek kudo/nom: 3091699 | FROGGS++ | t/04-nativecall/16-rt125408.t:
skip test, the fix is not in the 2015.09 release
20:44
20:44 telex joined
FROGGS [Coke]: it is not visible there 20:44
jnthn masak: I'll try and sneak a patch in for this ahead of the release, if [Coke] permits :)
[Coke] jnthn: JEZU, I just did the nqp release. 20:45
FROGGS okay, that was the last thing I committed today
jnthn [Coke]: This is in Rakudo
[Coke] that isf ine.
I ONLY did the nqp release. :)
jnthn What does JEZU mean? :P
FROGGS [Coke]++ # :o)
masak [Coke]++ # release
[Coke] and I can't start on rakudo until about 4 hours out. 20:46
PerlJam [Coke]: did you end up keeping the parrot stuff?
[Coke] some scifi author used that instead of "jesus" when swearing.
20:46 [Sno] joined
masak jnthn: rt.perl.org/Ticket/Display.html?id=126089 20:46
jnthn Ah
masak detrains &
[Coke] PerlJam: had to, we need to build the parrot makefile to release nqp
I just removed the bit that tried to run parrot's tests for now. 20:47
FROGGS btw, on my box is all clean... nqp, rakudo and roast
jnthn Well, darn...the change to print_control shows we don't actually make it into there :S 20:48
mullagainn BTW JSON::XS worked using Inline::Perl5 20:49
SNMP is half working
dalek kudo/nom: bc9ec3a | coke++ | tools/build/NQP_REVISION:
[release] bump NQP revision
FROGGS half?
raiph mullagainn: thanks for that update, was just typing a message to ask :)
vendethiel is the beta release happening? :o)
[Coke] vendethiel: a release of the compiler is happening. :P 20:50
FROGGS vendethiel: of the compiler, yes :o)
jnthn m: say 0xFFE
camelia rakudo-moar 643eb6: OUTPUT«4094␤»
vendethiel when is TimToady++'s bday?
20:50 erxeto left
_itz is this a pre-alpha beta? 20:50
FROGGS I still think that I've got a good feeling about roast... and would consider its state worthy a RC0 20:51
jnthn m: say 0xFFFE
camelia rakudo-moar 643eb6: OUTPUT«65534␤»
pink_mist vendethiel: 27th according to wikipedia
FROGGS BOM
[Coke] finds another bug in tools/contributors.pl6
FROGGS [Coke]: fwiw, panda and Inline::Perl5 are doing well as well 20:52
jnthn Bad news, it seems that there *is* an NQP level bug wrt the disappearing done control exception
FROGGS :S
how come it appears now?
it is not like the GLR changed the way our mainline works
PerlJam bad luck
FROGGS or how exceptions are handled outside of loops...
[Coke] jnthn: bad news in that it won't make it to 2015.09, or bad news in that I have to cut another release in an hour? ;)
jnthn It's that I added extra control exceptions and they were outside the mask 20:53
lizmat [Coke]: what's the pb in contributors.pl ?
[Coke] lizmat: it assumes we release every month.
jnthn [Coke]: Depends how seroius you consider the issue
[Coke] I will work around it with a hardcoded run instead of letting it pick a date, and then pretend we release every month again.
lizmat yea, you can specify a date with YYYY-MM-DD as the parameter 20:54
[Coke] jnthn: what's the symptom? does done work, except in masak's case? That seems like something we can live with for a month.
jnthn [Coke]: Basically, we stole "done" from Test.pm for something else
[Coke] I have to head out shortly - I have no problem doing another nqp release today if this is fixable. 20:55
FROGGS [Coke]: things like last or done do not explode when used outside of their blocks/loops
skids
.oO(irony of a feature named "done" holding up a release)
_itz :)
FROGGS *g*
jnthn [Coke]: And the incorrect use of "done" doesn't explode, so we silently do nothing
FROGGS: No, last is fine
FROGGS: It's only for emit and done that we have an issue
[Coke] travis says that I broken nqp by updating to moar 2015.09
FROGGS jnthn: ahh, that's fine then isnt it? 20:56
20:56 clearkruti left
lizmat m: "emit".flip.say # :-) 20:56
camelia rakudo-moar 48c03a: OUTPUT«time␤»
[Coke] if the failure is failure to explode in the bad path, rather than failure to work in the happy path, I think we can live with that.
jnthn lizmat: I noticed that and thought it was wonderfully appropriate for async :)
[Coke]: It may just cause some confusion for people is all 20:57
bartolin lizmat: would you mind to take a look at github.com/usev6/rakudo/commit/bf1537daa9 ? i tried to fix RT #124205 but my patch feels smelly. i couldn't think of a clean way to execute special case for the first element of a lazy list (and so i added a my hack within .map) ...
FROGGS [Coke]: travis just reports that the javascript backend failed a test 20:58
20:58 skids left
[Coke] FROGGS: I can't get travis to tell me what failed. 20:58
looks like it's thinking about it, then just ... . . . ...
FROGGS [Coke]: travis-ci.org/perl6/nqp/jobs/80904368#L924 20:59
jnthn bartolin: Could use a state var instead
FROGGS err, now I clicked on the redo button and travis starts over again...
but anyway, was just a problem with js 21:00
lizmat bartolin: looking at it now
bartolin jnthn, lizmat: thanks! 21:01
dalek p: ac856c1 | jnthn++ | src/vm/moar/QAST/QASTOperationsMAST.nqp:
Fix control exception mask for new additions.

Now CONTROL also handles AWAIT/EMIT/DONE.
21:02
timotimo how do people feel about building speculative tests for UDP sockets?
timotimo goes to bed :P 21:03
FROGGS gnight timotimo
bartolin jnthn: i'm unhappy with the extra conditional within the map block. could a state var help to avoid that block?
bartolin has to think about that some more 21:04
21:04 tokuhiro_ joined 21:05 sufrostico left
lizmat bartolin: looking at the ticket, I think it should be rejected 21:05
bartolin lizmat: oh!
lizmat if you provide a :with that is stupid enough to always return 1, then you get what you deserve ?
now, maybe the initialization of $last should be a bit smarter 21:06
[Coke] jnthn: that was quick.
bartolin m: my @rt124205_b = "", "", "b", "b", "B", "B"; say @rt124205_b.squish(:with(*.Str eq *.Str)).perl
camelia rakudo-moar 48c03a: OUTPUT«("b", "B").Seq␤»
21:07 telex left
bartolin m: my @rt124205_b = "", "", "b", "b", "B", "B"; say @rt124205_b.unique(:with(*.Str eq *.Str)).perl 21:07
camelia rakudo-moar 48c03a: OUTPUT«("", "b", "B").Seq␤»
[Coke] leaving now, but I guess I'll cut 2015.09.01 for nqp when I get back. toodles.
jnthn [Coke]: OK. I'm getting a Rakudo patch in with further improvement too
lizmat bartolin: this was about squish ?
ah, ok
bartolin lizmat: that's from S32-list/squish.t -- so it's not only a :with that always returns 1 21:08
21:08 tokuhiro_ left, telex joined
bartolin yeah, i think squish should return the same thing as unique here 21:08
lizmat bartolin: ok, clear now, lemme try something
21:10 Peter_R left
FROGGS good night #perl6 21:11
21:11 FROGGS left
jnthn 'night, FROGGS 21:11
.oO( he really shouldn't spill his beer on his connection like that...it's a waste of beer! )
21:12
21:12 Peter_R joined
PerlJam bartolin: why wouldn't you use $last = self[0]; and gather self[1..*].map: { ... } ? 21:12
jnthn PerlJam: If self is a Seq that won't go well 21:13
PerlJam I suppose that would still return the empty list
dalek kudo/nom: 6afd267 | lizmat++ | src/core/Any-iterable-methods.pm:
Fix for #RT 124205, usev6++ bartolin++
bartolin PerlJam: I think I tried that and it didn't work out. 21:14
lizmat hmm.... not so good after all :-( 21:15
21:18 rurban joined 21:19 rurban left, colomon_ left 21:20 pmurias joined, kenhat left 21:21 colomon joined, rindolf left
lizmat hmmm.. 20,21 now passing, 22-23 not yet 21:22
pmurias arnsholt: if you don't count the tests manually but just make the test harness do the counting doesn't it defeat the purpose of having a plan?
jnthn pmurias: Not the anti-regression one, know 21:24
bartolin lizmat: 22 and 23 don't have their itemization right, do they? looks GLR related to me
lizmat yeah, fixed that now as well, all tests passing 21:25
pmurias in the spec tests where weird stuff can happen due to a buggy compiler plans make a lot more sense
bartolin \o/ lizmat++
dalek kudo/nom: 0921fa4 | lizmat++ | src/core/Any-iterable-methods.pm:
Better fix for #RT 124205, usev6++ bartolin++
kudo/nom: e25cbf4 | jnthn++ | src/core/Exception.pm:
Don't leak unexpected control exceptions.
21:26
kudo/nom: f435959 | jnthn++ | src/core/Exception.pm:
Good errors for unhandled emit and done.
kudo/nom: e95a7f3 | jnthn++ | src/core/Exception.pm:
Remove superstitious parens.
jnthn There we goes
bartolin lizmat: ahh, 'once' was what i wanted :-)
dalek ast: 729f46f | lizmat++ | S32-list/squish.t:
Fix GLR-busted initialization
lizmat hmmm.. have my suspicions about the scoping of once though 21:28
21:28 telex left 21:30 telex joined
dalek ast: f2199d3 | lizmat++ | S32-list/squish.t:
Unfudge #RT 124205 related tests
21:34
lizmat jnthn: a bare "done" and "emit 42" are still silent for me ? 21:36
jnthn You need a newer NQP
detrain and then sleep at last! :) 21:37
back tomorrow &
lizmat good night, jnthn
dalek kudo-star-daily: ae04bda | coke++ | log/ (2 files):
today (automated commit)
21:37 muraiki left
[Coke] finds a network connection but probably can't get a lot done here. :) 21:38
bartolin g'night, #perl6 21:39
21:42 diana_olhovik_ left 21:48 spider-mario left, espadrine left 21:50 telex left 21:51 araujo left 21:52 telex joined
dalek p: 8fd2e96 | coke++ | docs/release_guide.pod:
avoid weird git error that bit me on last release
21:56
21:56 hernanGOA left
lizmat goes to bed while it is still today... 21:58
21:59 telex left, telex joined, travis-ci joined
travis-ci NQP build failed. Will "Coke" Coleda 'avoid weird git error that bit me on last release' 21:59
travis-ci.org/perl6/nqp/builds/80918921 github.com/perl6/nqp/compare/ac856...d2e96edc66
21:59 travis-ci left
[Coke] ~~ 21:59
that was a doc only bug. 22:00
22:00 telex left
[Coke] er, commit. travis-- # noise 22:00
22:02 telex joined 22:06 Sqirrel_ left 22:19 cognominal left
lichtkind good night lizmat 22:21
i clearly not huged you enough in spain 22:23
22:23 pmurias left 22:31 telex left 22:32 telex joined, kid51 joined 22:41 skids joined 22:44 Skarsnik left 22:45 Gruber joined 22:51 Sqirrel joined
raiph m: class A { has $.a; }; say A.new(a => 5) eqv A.new(a => 5); #=> False # <-- from doc.perl6.org/language/operators#infix_eqv 22:54
camelia rakudo-moar e95a7f: OUTPUT«True␤»
23:01 leont left 23:04 telex left 23:05 tokuhiro_ joined 23:06 telex joined
raiph star: class A { has $.a; }; say A.new(a => 5) eqv A.new(a => 5); 23:08
camelia star-m 2015.03: OUTPUT«False␤»
23:10 tokuhiro_ left 23:12 telex left 23:13 labster joined 23:14 telex joined, larion joined
ShimmerFairy Something that's been on my mind: is there any way to make the output more useful for test failures (when using 'make spectest', at least)? I've personally never found the summary helpful when mentioning "Tests 15, 20-25, 31", since tests aren't numbered in the actual file. Could the summary output possibly reference line numbers instead? 23:20
23:22 telex left, telex joined
pink_mist +1 to that idea 23:22
ShimmerFairy Taking a quick look at `man prove`, it doesn't seem to be something it does offhand. I honestly wonder how the summary from `prove` has ever been considered useful -- maybe P5 tests have test numbers in source, or maybe rakudo uses the most tests with it out of anyone :P 23:26
23:30 telex left 23:32 telex joined
pink_mist I've been annoyed with p5 tests before that I had to manually count the tests :/ 23:33
prove -v helps if the tests are given names though
which isn't always the case
ilmari without -v the summary is less necessary, since that only outputs the failures 23:34
(with line numbers)
the summary is more useful with -v, if there's lots of passes between/after the fails
ShimmerFairy pink_mist: with P6's Test at least, failed tests give a line number in the error reporting, IIRC. It's just the summary that isn't so helpful :) 23:38
pink_mist: When I get failures from "spectest", I always end up running the files individually so I can see the errors printed amongst the passes.
pink_mist if the test failed because it die()d I'd get line numbers in perl5 :P 23:39
... but it's getting late, and I must sleep
good night #perl6 23:40
vendethiel \o pink_mist
23:41 telex left
ShimmerFairy pink_mist o/ 23:42
23:42 telex joined 23:44 skids left 23:53 lichtkind left 23:55 TEttinger joined