»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
tzictli I apologize if I've missed something obvious, but is there some way to verify the integrity and authenticity of Rakudo source downloads? I'm not seeing any OpenPGP signatures or hashes, and the 2014.07 tag from the github source isn't signed, either. 01:40
dwarring tzictli: try this: gzip -t rakudo-2014.03.tar.gz 03:19
"gzip -t --test: Test. Check the compressed file integrity" 03:20
xiaomiao dwarring: that's not what tzictli asked for 03:22
dwarring ok, that just checks that a gzipped file is intact 03:24
xiaomiao yes. doesn't tell you if the content is what you ordered
TimToady
.oO(immutable official modules)
 03:42
ingy
.oO(ingymutable modules!)
 03:43
ingy IngyInTheMiddle attacks! 03:44
japhb m: multi postfix:<d6>(Int $n) { [+] (1..6).roll($n) }; say 3d6 + 100 xx 20; 04:07
camelia rakudo-moar 1cf7ca: OUTPUT«110 113 112 109 111 113 107 113 111 117 109 110 107 109 113 116 116 112 109 112␤»
japhb Or somewhat more verbosely: 04:08
m: multi infix:<d>(Int $n, Int $sides) { [+] (1..$sides).roll($n) }; say 3 d 6 + 100 xx 20;
camelia rakudo-moar 1cf7ca: OUTPUT«113 111 113 111 108 111 106 110 110 106 105 110 113 114 109 110 116 110 110 112␤»
japhb I just like how those turn out. :-)
tzictli dwarring: thanks, I've been using Unix-like systems for several years, but I've never used the -t option until now, so that was informative; integrity down, still have authenticity to go 04:49
xiaomiao: quite right
awwaiid I'm going to a talk on the Julia programming language tomorrow at Polyglot-DC. Looks like there's a lot of feature overlap with perl6 :) 04:56
at least at my first glance
(as far as operator / multi-dispatch goes) 04:57
xiaomiao awwaiid: the big difference is that julia upstream is willing to do quick hacks to get results ;) 05:20
masak good morning, #perl6 06:59
moritz \o masak, #perl6 07:12
Ven o/, * 07:16
sergot hi o/ 07:21
brrt \o 07:23
timotimo o/ 08:01
masak re-discovers and is re-inspired by worrydream.com/LearnableProgramming/ 08:29
Ven masak: Perl 6 lighttable ! 08:30
desugars your code as you type it ( ͡° ͜ʖ ͡°)
masak :)
masak I haven't been paying attention to lighttable. should I? 08:31
Ven masak: I don't know, actually :D. Last time I looked at it, it was early 2013 (I think) and they had just removed everything cool 08:32
I think they put it back in now.
masak taking a brief look around on their main page tells me I should pay at least some attention. 08:35
Ven julia's "multi-dispatch" reminds me of magpie 08:50
masak what's magpie? 08:51
Ven masak: magpie-lang.org 08:52
masak: magpie-lang.org/multimethods.html 08:53
lizmat sightseeing&
dalek kudo/nom: dc9083f | (Rob Hoelz)++ | src/Perl6/Actions.nqp:
Fix broken DOC INIT blocks
 09:44
kudo/nom: 35d714b | moritz++ | src/Perl6/Actions.nqp:
Merge pull request #298 from hoelzro/nom

Fix broken DOC INIT blocks
moritz hoelzro++
timotimo .o( can someone please fix ($)? ) 11:38
hoelzro morning #perl6 11:40
timotimo hey hoelzro 11:41
Unsupported use of $ variable
okay, now you're just fucking with me.
m: sub test($) { } 11:42
camelia ( no output )
hoelzro moritz: thanks for the merge!
hoelzro o/ timotimo 11:42
psch hi #perl6 11:45
timotimo: a bare ($) throws the error for me
m: ($)
camelia rakudo-moar 35d714: OUTPUT«===SORRY!=== Error while compiling /tmp/HXW7_4aA8D␤Unsupported use of $) variable; in Perl 6 please use $*EGID␤at /tmp/HXW7_4aA8D:1␤------> (⏏$)␤ expecting any of:␤ semicolon list␤ prefix …»
psch m: my $capt = \($)
camelia rakudo-moar 35d714: OUTPUT«===SORRY!=== Error while compiling /tmp/0Vh6i17Cln␤Unsupported use of $) variable; in Perl 6 please use $*EGID␤at /tmp/0Vh6i17Cln:1␤------> my $capt = \(⏏$)␤ expecting any of:␤ prefix or term␤ …»
psch ^- that probably wants to work though 11:46
timotimo m: class Foo; method bar { Proxy.new( FETCH => sub($) { 1 } ) };
camelia rakudo-moar 35d714: OUTPUT«===SORRY!=== Error while compiling /tmp/1z7y2vd5zN␤Unsupported use of $) variable; in Perl 6 please use $*EGID␤at /tmp/1z7y2vd5zN:1␤------> method bar { Proxy.new( FETCH => sub($)⏏ { 1 } ) };␤»
timotimo m: sub ($) { }
camelia ( no output )
timotimo m: my &foo = sub ($) { }
camelia ( no output )
timotimo m: my &foo = YOINK => sub ($) { }
camelia rakudo-moar 35d714: OUTPUT«Type check failed in assignment to '&foo'; expected 'Callable' but got 'Pair'␤ in block at /tmp/thDFapoNLs:1␤␤»
timotimo m: my $foo = YOINK => sub ($) { }
camelia ( no output )
timotimo m: my $foo = YOINK => sub ($) { 1 }
camelia ( no output )
timotimo the fuck?
timotimo m: my &foo = ( sub($) { 1 } ); 11:47
camelia rakudo-moar 35d714: OUTPUT«===SORRY!=== Error while compiling /tmp/NGYqWDNKeB␤Unsupported use of $) variable; in Perl 6 please use $*EGID␤at /tmp/NGYqWDNKeB:1␤------> my &foo = ( sub($)⏏ { 1 } );␤»
timotimo i think that's weird.
timotimo oh, ufo may need a fix for the recent Foobar.pm.moarvm and friends change 11:50
masak oh, ok.
jnthn timotimo: You're forgetting the foo(...) is always a function call rule :)
masak contributions on that front heartily accepted. I will merge them as soon as I see them. 11:51
timotimo masak: now i wonder if i should put a version guard in there so that it only removes the .pm6? if the rakudo version is new enough? 11:52
or if there's a clever way to detect that feature/behavior 11:53
masak don't know. need more background info on that. 11:58
timotimo mhm 12:00
how should i model zmq_recv, which now takes a void* to a buffer that is supposed to be filled with data and a length for that buffer? 12:01
(a NativeCall question)
jnthn CArray 12:05
psch m: class A { method list() { <a b c> }; method Str() { "foo" } }; my $a = A.new; say $a.join("|"); say @$a.join("|"); # this is the same as i'm seeing with my patch for nested match objects for :g. 12:06
camelia rakudo-moar 35d714: OUTPUT«foo␤a|b|c␤»
psch i suppose that means that the tests assuming m:g// gets a list are wrong.
s/gets/gives/ # still not over those typos, grrr
timotimo it feels like i've not written anything but the simplest code in Perl6 for a loooong time 12:07
hoelzro I've just discovered something intersting with World.add_constant with regards to its caching 12:14
one this is that the named parameters are made part of the key, but the names aren't sorted in advance 12:15
can one depend on hash traversal order being consistent from call to call?
moritz only if the hash keys remain unchanged inbetween calls 12:16
hoelzro moritz: so if I did $*W.add_constant(..., :foo, :bar) twice, there'd be no guaranteed that the generated key would be the same? 12:17
moritz hoelzro: seems like 12:18
hoelzro =/ 12:19
well, that would at worst result in cache misses when there are none
the other is more insidious: you could make a call like this: $*W.add_constant('Array', 'type_new', ~$object) and later like this: $*W.add_constant('Array', 'type_new', |@array_containing_object_from_before) 12:20
and the cache key for both would be identical, so the generated constant would result in ['ObjectStringified'] rather than [$object]
this just hit me with the S26 work
I think that the positional parameters' types should be made part of the key; does anyone object? 12:21
moritz wait wait wait
why wouldn't the cache key identical? 12:22
you're dong the same thing twice, after all
hoelzro but I'm not
$object isn't a Str
the first time, I'm stringifying it
so the cached constant is an Array with a single element, namely a St
*Str
hoelzro the second time, I'm not stringifying it, but the array's members are stringified to make the key 12:23
Vendethiel- eh, vim and ruby don't work well together. braaaaceees, it screams, even.
hoelzro so I put in an Array with a single object member, and I get back a constant that contains a single string 12:23
moritz huh? perl uses braces too, no?
hoelzro: ok, I understand the problem know 12:24
hoelzro: but I don't know enough about add_constant usage to decide anything
hoelzro alright 12:26
I'll just leave it here for others' consideration, then =)
liztormato timotimo: You should be able to create a CompUnit object from a path. The .has-precomp method will then tell you whether there is a precompiled version available. 12:38
Please note that I haven't written any tests for CompUnit usage like that 12:39
timotimo i ... don't really know how that will help me %)
i'm trying to fix ufo up, which creates a makefile for your "project"
Vendethiel- .u pen 12:40
yoleaux U+20B0 GERMAN PENNY SIGN [Sc] (₰)
U+23D9 METRICAL PENTASEME [So] (⏙)
U+26E4 PENTAGRAM [So] (⛤)
Vendethiel- erm...thanks?
masak :D
Vendethiel-: ✐ 12:41
liztormato Sightseeing continues&
masak U+2710 UPPER RIGHT PENCIL
Vendethiel- masak: ✎ I got this one. FIGHT 12:41
masak heh.
Vendethiel- I'd actually like it to be the other way, but meh.
timotimo .u pencil 12:42
yoleaux U+270E LOWER RIGHT PENCIL [So] (✎)
U+270F PENCIL [So] (✏)
tadzik we need moar pencils 12:43
Vendethiel- .u lower left pencil
yoleaux No characters found
Vendethiel- y u do dis unicode. 12:43
jnthn hoelzro: add_constant's caching was designed to cope with a fairly small set of use cases 12:44
masak Vendethiel-: emojipedia.org/lower-left-pencil/ 12:45
jnthn hoelzro: The reliance on hash ordering is probably suboptimal in that, as you note, could easily cause misses. But it's only suboptimal rather than wrong...
Vendethiel- Soon(TM).
hoelzro ok, the hash keys thing doesn't bother me that much 12:46
jnthn: what about the positional arguments problem?
dalek ast/S26-WHY: abfd5a2 | (Rob Hoelz)++ | S26-TODO.md:
Add more declarations to TODO
 12:48
ast/S26-WHY: 92742b8 | (Rob Hoelz)++ | S26-TODO.md:
Get rid of known bugs

Because I fixed them!
ast/S26-WHY: 73d4888 | (Rob Hoelz)++ | S26-TODO.md:
Make a fancy completion table for WHYs matching
jnthn hoelzro: I'm rather surprised add_constant is being used for a reference type...
ast/S26-WHY: 9a6489e | (Rob Hoelz)++ | S26-TODO.md:
Check off what's been done so far
ast/S26-WHY: f9c0392 | (Rob Hoelz)++ | S26-TODO.md:
Remove bad decls from checklist

Either they weren't in the spec, they don't make sense, or they were folded into another entry
jnthn hoelzro: I didn't design it for that.
The Pod.nqp code using it dates, I think, to a time before we had proper serialization.
hoelzro yeah, I'm not too clear why it's needed...I'm just adapting the code that was there before I started working on this 12:49
jnthn These days walking an object and add_constant'ing parts of it is kinda odd (though at the time there was likely no choice). Might as well just stick the entire data structure in the SC. so far as I can see 12:50
hoelzro alright
jnthn That avoids the issue. Not to mention saving building up a cache on something that's not really cacheable... 12:51
Or not going to be getting re-used
hoelzro should I just not trying caching it, then? 12:52
timotimo i tried to bring the pod code up to date with the new serialization stuff, but failed at some point :(
jnthn hoelzro: I'd say so 12:53
hoelzro: I mean, the only way to get a cache hit would be identical Pod appearing in multiple places in a file
#= This does something
gah such doc fail :)
hoelzro well, the way it's being used right now is that the $=pod object is being constructed at the end of the comp_unit 12:54
jnthn oh
So...it'll never get a cache hit 'cus the cache is scoped to a comp unit
hoelzro but for each declarative comment, a constant is created for that one comment
so this problem I'm seeing is only triggering for files with a single declarative comment 12:55
jnthn Yeah, sounds like you don't want to be using the cache at all 12:55
hoelzro indeed 12:56
jnthn For the strings maybe but certainly not for anything larger...
hoelzro I'll just use :nocache for the $=pod creation call, then
hoelzro commute & 12:58
timotimo i wonder how much slower start-up becomes if this file does exist: /home/timo/perl6/moarvm/../install/languages/perl6/share/libraries.json 13:03
Ven m: say dir.grep(*.d).map({ $_.Str => dir($_)>>.Str }) 14:08
camelia rakudo-moar 35d714: OUTPUT«"star-2014.04" => $("star-2014.04/camelia.sqlite3", "star-2014.04/version", "star-2014.04/src", "star-2014.04/bin", "star-2014.04/man", "star-2014.04/languages", "star-2014.04/include", "star-2014.04/share", "star-2014.04/lib") "star-2014.03" => $("star-20…»
Ven m: say dir.grep(*.d).map({ .Str => dir($_)>>.Str }) # why doesn't that work ??
camelia rakudo-moar 35d714: OUTPUT«"star-2014.04" => $("star-2014.04/camelia.sqlite3", "star-2014.04/version", "star-2014.04/src", "star-2014.04/bin", "star-2014.04/man", "star-2014.04/languages", "star-2014.04/include", "star-2014.04/share", "star-2014.04/lib") "star-2014.03" => $("star-20…»
Ven uh-oh, that doesn't work here.
actually, it does now. Well, it was changing my error message earlier. uh. 14:09
Ven I wonder if I can use X=>, but probably not, since the RHS doesn't autothunk with $_. 14:14
moritz m: say <a b c> X=> 1, 2 14:21
camelia rakudo-moar 35d714: OUTPUT«"a" => 1 "a" => 2 "b" => 1 "b" => 2 "c" => 1 "c" => 2␤»
Ven moritz: yes, but I need $_ in the RHS 14:23
moritz then X isn't the right tool 14:24
as you have noted already
Ven which is why I use .map 14:24
moritz aye 14:26
hoelzro timotimo: regarding bringing the POD code up to speed on the SC stuff, what would that involve? 14:28
Ven Hash[Dir['*'].map {|x| [x, Dir["#{x}/*"]]}] <- ruby version. I despise Hash#[].
hoelzro would someone mind marking rt.perl.org/Ticket/Display.html?id=122056 as resolved? it got fixed in my PR from yesterday 14:57
nwc10 done. 14:59
seems that I have privs to do that queue
um, did you actually want me to add a comment as to why? 15:00
PerlJam A comment that links to the commit that fixes it might be nice :)
hoelzro I guess you could point out the commit that fixed it?
nwc10 yes, but I don't know that 15:01
I'm just the minion here
hoelzro looks for the commit 15:05
PerlJam hoelzro: why does perl6 --doc -e'DOC INIT {}' output a newline? I would have expected it to output nothing.
hoelzro dc9083f
PerlJam: that's a...good question 15:06
I don't know; I only made sure it didn't crash =)
PerlJam Hmm. probably because the default DOC INIT phase calls &say with $pod2text (which is empty) 15:09
hoelzro that makes sense
PerlJam tests a change 15:10
PerlJam rakudo compiles faster now, but it still seems to take forever when you're waiting on it :) 15:11
jnthn On my box we're not too far off Rakudo builds taking around a minute... 15:14
(For the whole thing, not just CORE.setting, which is now under 45s here)
But yes, faster would be nice :) 15:16
PerlJam for me the CORE.setting is taking about 80s 15:22
(circa 2009 hardware)
[Coke] push 15:30
dalek kudo/nom: 0de574f | duff++ | src/Perl6/Actions.nqp:
make DOC INIT phaser only call &say when there's something to output
[Coke] ww!
dalek kudo-star-daily: 437f3ed | coke++ | log/ (14 files):
today (automated commit)
rl6-roast-data: 724c809 | coke++ | / (5 files):
today (automated commit)
PerlJam Coke: If you have to type "push" in another window, is it *really* and automated commit? ;) 15:31
PerlJam s/and/an/ 15:31
moritz PerlJam: commit != push 15:34
PerlJam sure, but don't ruin my inference from a highly coincidental juxtaposition :) 15:35
moritz PerlJam: you're in #git too, I expect some amount of precision from you! :-) 15:35
anyway, afk 15:36
timotimo hoelzro: that would entail among other things not having any add_constant calls any more 15:43
hoelzro timotimo: I see; so what would the code taking its place look like? 15:45
I've been ignorant of this SC stuff
the S26 stuff has been a good tour of the compiler, but not that =) 15:46
timotimo ideally, it would just be creating List objects and putting stuff in them and boxing Str objects and such
psch github.com/rakudo/rakudo/pull/295 fairly confident this is now to spec :) 15:51
dalek blets: 5b0497a | (Herbert Breunung)++ | docs/appendix-a-index.txt:
stub for make, remove 2 typos
psch should i push the roast branch to perl6/roast and link it from the comments? i'm a bit scared it'd be a messy thing to do... 15:52
but then i should probably at least push it to my fork of roast, and i was told i got a commit bit to roast so that i don't have to PR for it all the time
dalek ast/matching_quote_ops: e9f732d | (Pepe Schwarz)++ | S05-substitution/subst.t:
Correct some tests for s///.
 15:53
ast/matching_quote_ops: 52a08c3 | (Pepe Schwarz)++ | S05-modifier/overlapping.t:
Remove P5ism, call .from instead of .to for Match position.
ast/matching_quote_ops: b019f3b | (Pepe Schwarz)++ | S05-modifier/ (2 files):
More of tests to fall in line with my rt82108 branch.
ast/matching_quote_ops: 4253e72 | (Pepe Schwarz)++ | S05-modifier/overlapping.t:
Correct overlapping tests.
ast/matching_quote_ops: 500a45e | (Pepe Schwarz)++ | integration/advent2012-day22.t:
Add a working version of RT#121789.

Arguably, fixing the test doesn't fix the ticket, but my understanding of what the snippet is supposed to do is getting the matched integers, which a raw .list call would not. Calling .list on a $/ obtained from :g should list the submatches for each repetition of the match, and not the captures, similarly to how calling .list on a 'normal' $/ gives the matches and not their captures.
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=121789
hoelzro psch: you can always push to your fork, and then push to perl6/roast later
psch th-thanks, dalek
psch hoelzro: too late! :) 15:53
hoelzro oh well =)
FROGGS o/
hoelzro branches are ok too 15:54
o/ FROGGS
psch FROGGS \o
FROGGS I like branches... because I usually do not see commits of forks...
timotimo psch++ :) 15:55
hoelzro yeah, that's a good point
PerlJam just realized his recent commit will generate no output for embedded pod that consists of a single 0
FROGGS which means that collab is much easier that way
psch FROGGS: so you're in favor of me applying for a commit bit for rakudo? ;)
FROGGS psch: ohh yes
but send your CLA first :o)
hoelzro who ends up approving those, btw? pmichaud, right?
I applied for one something like 6 months ago =/
PerlJam approving the CLA? 15:55
That's a TPF thing, not a pmichaud thing. 15:56
FROGGS hoelzro: no, more like jnthn and moritz and masak... ppl who know of the to-be-contributor
hoelzro ah ha
hoelzro I got the CLA in 15:56
PerlJam sometimes wishes he could think faster
hoelzro I just needed approval from someone else, iirc
FROGGS that is what I was talking about, aye 15:57
psch a great side effect of my fixing of m:g// is that m:ov// works as well, for free
PerlJam approves of hoelzro ;) 15:57
psch i *think* :ex should work too, but i didn't see any notable tests in roast that deal with it, so maybe it already did
hoelzro well, if any of those fine gentlemen would like to add me to the rakudo org, I would appreciate it =)
thanks!
psch where "notable" in this case means "failing or passing TODOs" 15:58
FROGGS [Coke]: you were in a position to check if a CLA was received? like the CLA of hoelzro? 15:59
PerlJam btw, is Coke our only source of CLA receipt information? (he's the only one I know to ask) Who else would know? 16:02
FROGGS PerlJam: pmichaud most likely 16:03
FROGGS PerlJam: do you mean that Coke++ could gain some money from telling us that a CLA was received? :D 16:03
PerlJam So ... I can give hoelzro access to the rakudo repo right now if no one objects. (I don't know a thing about his CLA though) 16:05
hoelzro it's not critical =) 16:06
PerlJam sure, but I'm a strike-while-the-iron-is-hot kind of guy. 16:06
FROGGS I'd wait at least for jnthn's opinion :o)
timotimo so, zmq_msg_data used to be bound like this: # ZMQ_EXPORT void *zmq_msg_data (zmq_msg_t *msg); 16:07
my sub zmq_msg_data(Net::ZMQ::Message --> Str) is native('libzmq') { * }
timotimo and i get "invalid utf8 at line 1 col 3", i'm expecting to receive "foo" ... could this be the \0 that's used to terminate the string? 16:07
[Coke] Yes, we have Hoelz's CLA. 16:12
hoelzro \o/
PerlJam sends hoelzro an invitation 16:13
[Coke]: who else knows about CLAs?
timotimo m: say "hey".encode('utf8').WHAT
hoelzro hooray
camelia rakudo-moar 35d714: OUTPUT«(utf8)␤»
timotimo m: say "hey".encode('utf8').^mro 16:14
hoelzro thanks [Coke], PerlJam
camelia rakudo-moar 35d714: OUTPUT«(utf8) (Any) (Mu)␤»
timotimo hm.
PerlJam [Coke]: or, more to the point, how can one of us know without necessarily bothering you?
[Coke] PerlJam: here on irc? no one I've seen talk in months and months.
timotimo # got: 'p9L'
# expected: 'foo'
so close! :)
# got: 'qu' 16:15
# expected: 'quux'
(but no cigar)
[Coke] Perljam: you seem trustworthy; what's your preferred gdocs email?
timotimo (unfortunately the exploded character didn't get copypasted)
PerlJam [Coke]: [email@hidden.address]
[Coke] alsowik: timo is on the list. 16:16
alsoalsowik: a niemeyer? 16:17
timotimo huh?
[Coke] there is a recent timo on the CLA list. 16:18
timotimo yeah, i got a cla a year ago or so 16:26
[Coke] timotimo: there are no dates in the file. :) 17:27
timotimo [Coke]: maybe there are dates in the meta-file? :) 17:51
timotimo i'm gone for 1.5 hours and no commits happen? :< 17:54
except for one, that is
FROGGS jnthn: btw, about my NFA problem... perl6-j is also okay 18:05
jnthn FROGGS: Interesting... 18:07
FROGGS: And it's in the construction of the NFA ratehr than the evaluation of it?
FROGGS jnthn: it is the construction, because the NFA.__dump already is wrong (before running nfarunproto, but perhaps after running nfafromstatelist) 18:09
I guess, I should dump it before running nfafromstatelist also
FROGGS jnthn: also before running nfafromstatelist it is wrong 18:22
m: grammar G { token TOP { <a> }; proto token a {*}; token a:sym<foo> { <b> }; token a:sym<indirect> { \w+ }; proto token b {*}; token b:sym<foo> { <sym> } }; say(G.parse("foo")) 18:26
camelia rakudo-moar 0de574: OUTPUT«｢foo｣␤ a => ｢foo｣␤␤»
FROGGS jnthn: when we hit that <b> here, does that result in a mergesubrule? 18:26
jnthn I don't think we do any of the actual merging until we create the protoregex or alternation NFA 18:27
Until then, subrules remain "virtual"
FROGGS ahh, okay
jnthn So it should be a subrule "b" or so in there 18:28
jnthn (In the NFA attached to the regex itself) 18:28
FROGGS adding a:sym<foo> 18:29
adding b
adding b:sym<foo>
adding a:sym<indirect>
adding b:sym<foo>
nice
that's from nqp-m
FROGGS m: grammar G { token TOP { <a> }; proto token a {*}; token a:sym<foo> { <b> }; token a:sym<indirect> { \w+ }; proto token b {*}; token b:sym<foo> { <sym> } }; say(G.parse("foo").NFA) 18:30
camelia rakudo-moar 0de574: OUTPUT«No such method 'NFA' for invocant of type 'Match'␤ in block at /tmp/fWXxNaZLij:1␤␤»
FROGGS ahh, yes 18:30
timotimo the "fetching panda" phase of rebootstrap does a whole bunch of minor collections that all end with less than 1% of nursery space used afterwards
all except one of those have <0.25% used afterwards 18:31
jnthn
.oO( eats, shoots and gcs )
 18:34
FROGGS adding a:sym<foo> 18:35
adding b
adding a:sym<indirect>
that's perl6-m
timotimo zmq is quite magical ... it's kinda hard to figure out if a connection is going to work or not >_> 18:47
jercos sufficiently advanced technology is difficult to debug without a wizard? 18:50
timotimo oh ... NativeCall doesn't know about Blob yet, eh? 18:55
jnthn 'fraid not 18:57
timotimo oi jnthn, how do i actually create a CArray with starting values? or how do i give it a starting size?
PerlJam
.oO( Blob Hlope? )
jnthn We can likely teach it.
timotimo: CArrays you create are managed
timotimo: So you can just put stuff in them and they'll resize.
timotimo oh
jnthn Ones you get back from C land, it's up to you to boudns check 18:58
timotimo so what's the best way to get from Str via utf8 to CArray[uint8]?
(should i actually make that uint8?)
jnthn For now, copy it into a Blob...
timotimo and then? $mycarray.push $_ for @$my_blob? 18:59
i guess i can push a list
timotimo m: say "höhöhö".encode("utf8").list.perl 19:01
camelia rakudo-moar 0de574: OUTPUT«Array.new(104, 195, 182, 104, 195, 182, 104, 195, 182)␤»
timotimo m: say @("höhöhö".encode("utf8")).perl
camelia rakudo-moar 0de574: OUTPUT«Array.new(104, 195, 182, 104, 195, 182, 104, 195, 182)␤»
jnthn Something like that 19:02
timotimo i can't push into a CArray it seems 19:03
masak m: say "höhöhö".encode("utf8").perl
camelia rakudo-moar 0de574: OUTPUT«utf8.new(104, 195, 182, 104, 195, 182, 104, 195, 182)␤»
timotimo star: use NativeCall; my CArray[int8] $foo .= new; $foo.push: 1; 19:04
camelia star-m 2014.04: OUTPUT«Cannot call 'push'; none of these signatures match:␤:(Any:U \SELF: *@values, *%_)␤ in block at /tmp/tmpfile:1␤␤»
..star-p 2014.04: OUTPUT«Cannot call 'push'; none of these signatures match:␤:(Any:U \SELF: *@values, *%_)␤ in any at gen/parrot/BOOTSTRAP.nqp:1219␤ in method push at gen/parrot/CORE.setting:1616␤ in block at /tmp/tmpfile:1␤␤»
timotimo ... wait what? Any:U?
what was :u for again? 19:05
jnthn That's the auto-vivifying one inherited from Any
timotimo ah
jnthn :U = undefiend = type object
timotimo am i supposed to [$idx] = $val?
jnthn Think so
timotimo i think i want to implement push for CArray. 19:06
timotimo zeeeeeerooooooooomq! >:( 19:10
masak doesn't quite have that ring to it, I'm afraid. 19:11
timotimo jnthn: CArray has the implementation of .elems from Any ... when i asked my CArray for how many elems it has, it says "one." 19:26
that's why only "h" and not "hello there!" arrives at the other end
before that, nothing arrived at the other end because i was passing the $flags where the $buflen was expected and the flags happened to be 0
we need a size_t in perl6 m) 19:27
nwc10 is t/spec/S26-documentation/10-doc-cli.rakudo.moar making unwarranted assumptions about the current directory, and hence failing as spectest? 19:34
it's passing when run stand alone
[Coke] ooh, git fun: 1. make a lot of local changes. 2. realize you have to discard one of them. 3. type git checkout /path/to/ and hit tab for completion. 4. hit enter. 5. cry 19:41
hoelzro =(
that reminds me of a git "extension" I wanted to write called git-safety-net 19:42
everytime you tried to do a git clean/reset/checkout/other-destructive-operation, it would create a ref of your current tree
PerlJam hoelzro: you mean like the reflog? 19:44
timotimo .o( ACID reflog symptom? ) 19:45
PerlJam The only really destructive thing is when you "git reset --hard" but have unstaged changes.
hoelzro PerlJam: kinda, but on non-committed things
exactly
[Coke] right, these were uncommitted changes.
hoelzro it shouldn't be too hard to write; such set up aliases in gitconfig and provide a command to checkout that ref 19:46
PerlJam maybe that's a good use of the stash? :)
hoelzro heh
I have a love-hate relationship with the stash 19:47
hoelzro I think instead of being a stack, it would be a one-ref type deal 19:47
usually, you know if you need that safety net shortly after you've jumped =)
timotimo apparently having zmq_msg_data return Str is wrong, as it seems to have a bogus byte at the end of the string ... 19:54
possibly that bogus 0x4 is followed by a 0x0 and that'd be why it terminates there ...
dalek kudo/nom: c577c87 | TimToady++ | src/core/List.pm:
speed up large pushes by using splice

push is defined to be eager, so there's no point in nibbling the list lazily.		 19:55
masak ooh 19:56
timotimo that's a cute one
masak TimToady++
timotimo jnthn: how should i handle a buffer + length that i get returned from a library function? 20:05
actually ... maybe i want to put a terminating null byte on the sending side? 20:06
dalek p: bc60762 | jnthn++ | tools/build/MOAR_REVISION:
Bump MOAR_REVISION for new extop flags.
jnthn timotimo: Does the library allocate the buffer too? 20:07
dalek kudo/nom: 6117d54 | jnthn++ | / (2 files):
Add new extop flags. Bump NQP_REVISION to get 'em.
 20:08
timotimo jnthn: it does, aye 20:09
timotimo why do i get "cannot unbox type object" from this code that clearly doesn't try to put a type object into the pos NOR the assignee slot for this postcircumfix:<[ ]> 20:20
dalek ast: 616587c | (David Warring [email@hidden.address] | / (2 files):
unfudge for resolved RT #122056 hoelzro++
 20:21
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=122056
hoelzro is there integration set up between GH and RT? 20:22
dwarring would be nice 20:23
PerlJam what sort of "integration"?
dwarring btw S26-documentation/10-doc-cli.t passed for me on a m-spectest 20:24
hoelzro PerlJam: like if a commit sent to GH contained the text "fixed RT #XXXXXX", RT would close that ticket 20:25
PerlJam I dunno if I'd want such a thing. What happens when someone commits a fix, but not the tests? Surely the RT ticket should stay open until both conditions are met. 20:27
hoelzro interestingly enough, there's no existing RT webhook
I think that's up to the project guidelines 20:28
ex. I could just commit nothing and mark a ticket as closed
hoelzro so it's the developers' responsibility to only close the ticket from GH once both tests and fix are present 20:28
another thing that I thought would be nice is if open tickets in RT that have a Perl 6 file attachment would run those files regularly to see if the issue has miraculously been fixed 20:29
granted, that's a huge security hole, but...
hoelzro even if RT added a comment pointing to the resolving commit on GH would integration be nice 20:30
PerlJam hoelzro: sure it's the developers responsibility ... but does everybody know these rules? The way it is now, GH and RT are processed separately so we get independant checks on bug fixes and things.
hoelzro that way RT users would get an e-mail and be able to cureate it
*curate 20:31
timotimo we already want to have a piece of test code in roast for every bug filed :) 20:31
hoelzro I suppose that's true
PerlJam hoelzro: that particular feature I could get behind :)
hoelzro if there's a test in roast or t/integration, it's normally marked skip/todo if it's broken, right? 20:32
hoelzro it might be nice if make test/make spectest detected tests associated with open RT tickets, and let you know if they're passing all of a sudden 20:33
moritz we have that already 20:36
make spectest tells you about passing TODOs
and usually we put the ticket number in the todo message 20:37
hoelzro ah, cool 20:38
PerlJam But a webhook on GH that auto-commented to any tickets mentioned in the log message would be nice. 20:39
that way you can do things like "fixed RT #123123" and also "added tests for RT #123123" and both commits would show up in RT 20:40
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=123123
PerlJam blah! 20:42
stupid RT + stupid human. Somehow I deleted ticket 122056. Anyone happen to know how to undo that?
ah, never mind. It seems I accidentally clicked on the "spam" marker for that ticket. 20:43
moritz PerlJam: there used to be a script that sent mails from roast commits to tickets 20:46
and I think I even wrote it
and Kyle ran it for a while
dunno where the code is; maybe in the mu repo somewhere 20:47
PerlJam test-reporter.pl ? 20:50
moritz might be 20:53
yes, looks right 20:54
and I misremembered, it seems I wasn't involved at all :-)
masak 'night, #perl6
PerlJam maybe intellectually since you remembered it at all :) 20:55
hoelzro o/ masak
PerlJam good night masak! sleep well.
(or whatever you really do at night ;)
hmm. I guess I'm not admin on github.com/perl6 20:56
moritz PerlJam: do you need some changes there? or a new repo? 20:58
flussence re: irclog.perlgeek.de/perl6/2014-08-07#i_9146840 - there's a good point... does anyone habitually gpg-sign their git commits or at least tags? I've started doing it on my own code... 20:59
PerlJam moritz: nah, I was just thinking about making a webhook for perl6/roast
moritz PerlJam: well, I can do that for you 21:01
PerlJam no worries. I think I need to let the idea gel in my head a little more first. (consider the downsides, figure out where to actually run the thing, etc.) 21:02
cognome On a fresh rakudo on mac : Stage parse : moar(96372,0x7fff70ccc310) malloc: *** error for object 0x7fd11014c1f0: pointer being freed was not allocated 21:07
... with moar
TimToady report on #moarvm, perhaps? 21:08
cognome yup
hoelzro I always sign my tags 21:15
but not my commits
timotimo maybe i should just get the value i want as a CPointer[int8] and put the ints into a List and create a Buf from that, which i then .decode from utf8 to get the final string 21:56
sounds great!
dalek p/qast_restructure: 26f36ac | jnthn++ | / (5 files):
Add QAST::NodeList; replace direct QAST::Node use.

This is a prerequisite for moving children out of the QAST::Node base class, which will enable us to save an array allocation and pointer on many node types that don't need children.		 22:39
p/qast_restructure: bd41f62 | jnthn++ | / (22 files):
Refactor QAST node structure.

Now only nodes capable of holding children get a slot for them and an array allocated, reducing memory. Also take the time to optimize the new methods so we get away with a lot of the late-bound method calls.
timotimo oooooooh 22:41
that sounds quite worthwhile
dalek kudo/qast_restructure: 1f4cae6 | jnthn++ | src/Perl6/Actions.nqp:
A few changes needed by qast_restructure in NQP.

Only one known regressions remains, which is related to compile-time inlining of multis.		 22:42
jnthn Too tired to track down the regression tonight. It's something to do with inline_info ending up being null. No idea how that can happen.
timotimo OK :( 22:43
jnthn But it looks like it's simply missing, so anybody fancing a bug hunt probably wants to look for inline_info in Actions.nqp :) 22:44
'night 22:46
timotimo jnthn: clone_qast used to bindattr('@!array') on a qast node, but 1) it's not always there any more and 2) it's called @!children now :) 23:02
nqp: role TheRole { }; class TheClass does TheRole { }; my $foo = TheClass.new(); say(nqp::istype($foo, TheRole)); 23:03
camelia nqp-jvm: OUTPUT«(signal )#␤# There is insufficient memory for the Java Runtime Environment to continue.␤# pthread_getattr_np␤»
..nqp-moarvm: OUTPUT«Assignment ("=") not supported in NQP, use ":=" instead at line 2, near " TheClass."␤ at gen/moar/stage2/NQPHLL.nqp:369 (/home/p6eval/rakudo-inst-1/languages/nqp/lib/NQPHLL.moarvm:panic:106)␤ from <unknown>:1 (/home/p6eval/rakudo-inst-1/languages/nqp/lib/n…»
..nqp-parrot: OUTPUT«Assignment ("=") not supported in NQP, use ":=" instead at line 2, near " TheClass."␤current instr.: 'panic' pc 15673 (gen/parrot/stage2/NQPHLL.pir:5731) (gen/parrot/stage2/NQPHLL.nqp:425)␤»
timotimo nqp: role TheRole { }; class TheClass does TheRole { }; my $foo := TheClass.new(); say(nqp::istype($foo, TheRole)); 23:04
camelia nqp-{moarvm,parrot}: OUTPUT«1␤»
..nqp-jvm: OUTPUT«(signal )#␤# There is insufficient memory for the Java Runtime Environment to continue.␤# pthread_getattr_np␤# An error report file with more information is saved as:␤# /home/p6eval/hs_err_pid7478.log␤»
cognome I don't understand that FOREIGN_LANG business :( 23:06
dalek kudo/qast_restructure: 226c44f | (Timo Paulssen)++ | src/Perl6/Actions.nqp:
teach clone_qast (and thus node_walker) about QAST::Children
 23:08
hoelzro incoming... 23:37
dalek Heuristic branch merge: pushed 79 commits to rakudo/S26-WHY by hoelzro 23:38
Mouq So.. Currently when one says "use XYZ;" to switch to the slang XYZ (even though "slang" is NYI..) it expects XYZ to be a complete grammar, because that's how v5 works. But what about the probably more common case where the majority of the language wants to be kept, only modifying a section of the normal grammar, perhaps in a way that is composable with other "use OtherSlang;" statements?
yoleaux 4 Aug 2014 08:54Z <jnthn> Mouq: yes, it is...that error should never happen.
hoelzro oh good; I thought dalek was about to spam us
Mouq Would lexically modifying the Perl 6 grammar be possible on an arbitrary scale like that? 23:39
m: { sub term:<such-term> { 1 }; say such-term + 1 }; say such-term + 1 23:40
camelia rakudo-moar 6117d5: OUTPUT«===SORRY!=== Error while compiling /tmp/qILuv_tTEo␤Undeclared routine:␤ such-term used at line 1␤␤»
Mouq s/possible/feasible without awful performance drawbacks
lue Mouq: aye, trying to figure out how one can modify Perl 6 the language, and how to support it sanely, still kinda hurts to do :P 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!