»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.
00:01 pmichaud left 00:03 pmichaud joined
timotimo i'm slightly surprised the mu repo refers to the article swtch.com/~rsc/regexp/regexp1.html "Regular Expression Matching Can Be Simple And Fast", yet in rakudo the pathological regex is still pathological 00:17
timotimo disappears 00:18
[Coke] ... gist finally updated after a family tech support incident. 00:19
00:26 btyler joined 01:10 FROGGS_ joined 01:13 FROGGS left 01:18 Patterner left 01:42 btyler left 01:50 btyler joined 02:14 SamuraiJack_ joined 02:43 BenGoldberg left
colomon is wondering how long he should give his latest awesome line of p6 code before he gives up on it... 02:49
oh, that wasn't so bad... 02:52
diakopter . 02:56
colomon my %by-name = lines.map(*.path).categorize(*.bas...ep(*.value > 1); 03:01
pass it a list of filenames, and it will find all the base names that are duplicates and build a hash from the base name to an array of duplicate paths. :) 03:03
03:04 kingbeast left
colomon takes 3.5 minutes to process 15,000 filenames under parrakudo 03:08
03:11 atroxaper joined 03:16 benabik joined 03:36 fgomez left
lee_ doe this look like a bug? gist.github.com/leedo/5906625 03:38
er, nevermind 03:39
benabik r: &infix:<+> !=== &infix:<+> 03:42
camelia rakudo d67185: ( no output )
benabik r: say &infix:<+> !=== &infix:<+>
camelia rakudo d67185: OUTPUT«False␤»
lee_ is === the proper way to compare operators? or should I be using =:=? 03:44
TimToady === should work 03:50
03:50 preflex_ joined, ChanServ sets mode: +v preflex_, preflex left, preflex_ is now known as preflex 03:58 FROGGS_ left 04:00 FROGGS_ joined, chayin_ joined 04:01 chayin left 04:02 Chillance left 04:03 btyler left 04:22 FROGGS_ left
sorear good * #perl6 04:33
lee_ r: all(1,2) ~~ Int 04:34
camelia rakudo d67185: ( no output )
lee_ r: say all(1,2) ~~ Int
camelia rakudo d67185: OUTPUT«False␤»
sorear whee, more backlog than fits in my buffer 04:35
how's the jvm port?
04:49 xinming left
JimmyZ sorear: 90% passed 04:49
04:51 xinming joined 04:55 toddr joined
sorear JimmyZ: sweeeet 04:56
05:01 birdwindupbird joined
JimmyZ sorear: 90% passed 05:03
05:03 xilo left
JimmyZ irssi-- 05:03
lee_ does this look curious to anyone else? gist.github.com/leedo/5906863 05:07
JimmyZ lee_: nope 05:08
lee_: why need two (( ? 05:09
lee_ why does the *@flat signature work on n and not m?
sorear because (Any) is more specific than a slurpy 05:10
JimmyZ lee_: try m(1,2,3);
sorear subsignatures are a kind of postconstraint; they apply after the multidispatcher is done
lee_ JimmyZ: the section of code i am working on gets passed a parcel
sorear: ok thanks, i think that makes sense 05:11
sorear although.. 05:13
r: gist.github.com/leedo/5906863
camelia rakudo d67185: OUTPUT«flattened 3␤got Range␤flattened 3␤»
sorear n: gist.github.com/leedo/5906863
camelia niecza v24-86-g39ab531: OUTPUT«Potential difficulties:␤ $range is declared but not used at /tmp/ik76Y12fzw line 9:␤------> multi sub m ((Range ⏏$range)) {␤␤flattened 3␤got Range␤flattened 3␤»
lee_ hmm, i am running on jvm-rakudo
sorear D: 05:14
regresssssion
do we have a test for that?
lee_ i can poke around for one tomorrow
sorear so jvm-rakudo is now being used in the wild? :)
lee_ sorear: if me being bored is in the wild, yes :P
05:19 toddr left, konundra left 05:28 ecocode joined
lee_ sorear: it looks like some related tests are failing in S06-multi/unpackability.t 05:28
(on jvm) 05:29
05:39 Psyche^ joined, Psyche^ is now known as Patterner 05:43 Patterner left 05:47 Psyche^ joined, Psyche^ is now known as Patterner 05:48 zby_home joined 05:51 raiph left 05:52 Patterner left 05:54 Psyche^ joined, Psyche^ is now known as Patterner 06:01 Patterner left 06:09 Psyche^ joined, Psyche^ is now known as Patterner 06:12 Patterner left 06:18 dmol joined 06:24 FROGGS[mobile] joined
FROGGS[mobile] o/ 06:24
sorear o/ 06:25
06:25 berekuk left
JimmyZ lee_: try m(1,2,3); 06:26
irssi-- again
:(
sorear why is your irssi randomly repeating lines 06:27
06:28 berekuk joined 06:35 FROGGS[mobile] left 06:37 FROGGS joined
sorear hmm, if I go ahead with this 9-to-5 thing it'll really hurt my overlap with jnthn :| 06:40
06:40 raiph joined
JimmyZ because I press up arrow to look for '/window next', when I see it, I press enter, but It suddenly switch to another msg 06:41
JimmyZ cant't find another way, i.e: ctrl+r or ctrl+1
sorear alt-right arrow 06:43
JimmyZ It doesn't work in Xshell4
sorear alt-1..alt-o 06:44
try esc, meta
JimmyZ I can only use ctrl+x to switch irc server
sorear++, esc + 1/2/3 works 06:46
06:50 kaleem joined, pmurias joined
pmurias when pushing multiple commits to nqp are some of them allowed to break tests if the last one passes? 06:51
FROGGS if you push all at once, I'd say yes 06:53
because sometimes it is nicer to split things to have smaller commits
sorear I think not, because I think the point is bisectability 06:54
FROGGS hmmm, good point
pmurias sorear: so I should squash them into a single commit? 06:57
moritz either that, or make it a branch and merge
otoh we aren't very strict about such rules in nqp land 06:58
FROGGS bisecting is hard these days anyway, at least for rakudo when you have to rebuild nqp
06:59 domidumont joined
dalek ast: 9c64dd0 | (Tobias Leich)++ | S32-num/rshift_pos_amount.t:
added test for right shifting by a positive amount
 07:12
07:15 domidumont left 07:16 domidumont joined 07:20 bbkr joined 07:36 daxim joined
lizmat good *, #perl6! 07:43
labster good * lizmat!
woosley hello world 07:44
lizmat woosley! labster!
nwc10 FROGGS: script it! 07:45
given that in 5 years time, Rakudo will still need NQP (re)building to bisect problems, the sooner that someone automates the bisection dance, the sooner it stops being hard 07:46
this isn't work that is going to be thrown away
(yes, it *is* work)
woosley hello #perl6, how can I do "get start with nqp and rakudo and Perl6 and even MoraVM" ?
FROGGS nwc10: nice idea
07:47 ztt joined
lizmat .oO{ A virtual machine for snacks } 07:47
sorear o/ labster, lizmat, woosley (ETOOCLOSE2WOOLFY), nwc10
FROGGS virtual snacks?
lizmat Mora is a world renowned brand of snacks in the Netherlands 07:48
sorear world renowned, but only in the Netherlands?
labster most world-renowned things are only renowned by a subset of the world. 07:49
lizmat :-) 07:50
DrEeevil lizmat: ... and belgium!
"if we sell it you can fry it", right?
lizmat yup
07:53 labster left 07:54 labster joined
lizmat .tell jnthn I think the problem is that the default default is not set in the ContainerDescription 07:54
yoleaux lizmat: I'll pass your message to jnthn.
labster woosley: is that a "how do I learn Perl 6" or "how to I help out" question?
lizmat .tell jnthn in Actions.nqp, between lines 1943 and 1951 we need to add a default_value key to %cont_info, but I have no idea where to get that from :-( 07:55
yoleaux lizmat: I'll pass your message to jnthn.
woosley labster: Maybe how do I help out?
FROGGS tools/build/bisect.pl --good-is-when="ok 12 - Some crazy test" - or - tools/build/bisect.pl --bad-is-when="nok 12 - Some crazy test" 07:56
nwc10: would that be what you think? 07:57
labster woosley: well, that's easy: You should just try to have fun with P6. There are simple things, like writing tests, or just making a module and adding it to the ecosystem. There are more difficult things like diving into the guts of the compilers, and writing C/Java/NQP. 07:58
FROGGS I guess it should ingore (?) checkouts somehow that cant be build
sorear ignore
lizmat .tell jnthn scratch that last one: container_type_info is doing that already 07:59
yoleaux lizmat: I'll pass your message to jnthn.
07:59 bbkr left
labster woosley: Do whatever you're comfortable with for now -- even using the language helps us out. And ask here if you have any specific questions/problems. 07:59
woosley labster: Aha, all right, I suppose I can start write some little programs using Perl6 08:03
08:03 xinming left
sorear I guess I just need to get up realllly early and schedule overlap time before $dayjob time 08:05
08:05 xinming joined
sorear on that thought, sleep& 08:06
08:06 zby_home left
labster I get the sense that sorear's sense of "really early" is "about the time normal people wake up" 08:06
sorear labster: thinking more like 5 08:07
labster oh well, that is really early :)
pmurias a parrot code ref is a PMC*? 08:11
lizmat Parrot Magic Cookie, could be anything, afaik 08:13
tadzik good morrning 08:15
pmurias morning 08:19
08:19 Stygia joined
lizmat tadzik! pmurias! 08:22
.g thinko
yoleaux www.urbandictionary.com/define.php?term=thinko 08:23
lizmat .g PMC 08:29
yoleaux www.ncbi.nlm.nih.gov/pmc/
lizmat that's completely out of context for #perl6 08:30
.j
.j thinko
maybe .j should take the lemma's of S99 ?
08:41 dakkar joined
lizmat fitness& 08:43
raiph .tell jnthn fperrad has confirmed that "My future goal is to add a TvmJIT backend to github.com/perl6/nqp/" 08:48
yoleaux raiph: I'll pass your message to jnthn.
08:51 SamuraiJack_ left 08:52 SamuraiJack_ joined
raiph .tell jnthn TvmJIT being more or less LuaJIT (but with s-expression syntax) 08:54
yoleaux raiph: I'll pass your message to jnthn.
08:56 fhelmberger joined
JimmyZ \o/ another backend 09:01
dalek p: afa84b8 | (Pawel Murias)++ | / (9 files):
Wrap subs in a NQPRoutine code object when NQPRoutine is in scope.

We need capturelex to get that to work properly. I's a port of p6capturelex that takes the code objects type as an argument rather than assuming it's a Rakudo Code. We also add a test that ensures subs are wrapped in codeobjects.		 09:03
nwc10 And Unladen Swallow will make Python 5 times faster and remove the GIL
ie, it's only "another backend" when it's done 09:04
but it *is* good news that people are working on things like this
masak good antenoon, #perl6 09:05
man, it's hard to focus on anything with new HPMoR episodes out... 09:06
pmurias hi 09:07
pmurias goes off to read the newest HPMoR chapter while his latest nqp commit is tested on both backends ;) 09:08
FROGGS $another_backend means too that NQP has a great acceptance... and I like that :o)
moritz masak: not so hard to focus on reading HPMoR :-)
09:08 Stygia left 09:11 xinming left 09:12 xinming joined
nwc10 masak: how long will it take you to read them all? 09:12
jnthn yawns
yoleaux 07:54Z <lizmat> jnthn: I think the problem is that the default default is not set in the ContainerDescription
07:55Z <lizmat> jnthn: in Actions.nqp, between lines 1943 and 1951 we need to add a default_value key to %cont_info, but I have no idea where to get that from :-(
07:58Z <lizmat> jnthn: scratch that last one: container_type_info is doing that already
08:48Z <raiph> jnthn: fperrad has confirmed that "My future goal is to add a TvmJIT backend to github.com/perl6/nqp/"
08:54Z <raiph> jnthn: TvmJIT being more or less LuaJIT (but with s-expression syntax)
FROGGS .oO( rm -rf INBOX ) 09:21
09:22 xinming left 09:23 xinming joined
dalek p: 5942262 | (Pawel Murias)++ | / (2 files):
Fix BEGIN blocks, they get outer lexicals and can return a value. Add a test.
 09:23
pmurias jnthn: BEGIN blocks are fixed and subs wrapped in code objects on both backends 09:25
FROGGS dont we have four backends atm?
(and a fifth in the queue?) 09:26
pmurias FROGGS: jvm,parrot,moarvm,js
nwc10 no, we have one working backend
FROGGS true 09:27
nwc10 one more nearly working
FROGGS nwc10: I said nothing about their state :o)
nwc10 and I don't think that any of the others can even self-host NQP yet
09:27 raiph left
nwc10 if you *don't* qualify it, people will assume that it's useful and working 09:28
FROGGS I think pmurias knows what is going on
and others can ask, it is a chat 09:29
09:29 xinming left
nwc10 it's also logged for posterity, and can be taken out of context 09:29
09:29 Stygia joined
nwc10 I wasn't aware of it in context 09:29
FROGGS so, I am thinking it would be nice to smoke *all* in-dev-and-future-backends, if possible
nwc10 but I likely will still be upset by inexact use of terms
a reputation for "vapourware" is a big problem 09:30
FROGGS then I will try to be more exact in future
nwc10 thanks
09:30 xinming joined
nwc10 and I'll try not to be so jumpy 09:30
FROGGS *g*
09:33 pecastro left
jnthn On backends, I think it's worth remembering that JVM has taken ~8 months to get to 90% spectest coverage. 09:39
09:39 pecastro joined
jnthn Now yes, a lot of that was designing portability abstractions, etc. 09:39
But still, it's a non-small piece of work.
FROGGS well, I had expected a few months more to get to this point fwiw 09:42
jnthn :)
FROGGS jnthn: so I can read a nice blog post on my couch tonight? :P 09:43
jnthn FROGGS: Once the numbers in roast-data look good :) 09:48
09:48 bbkr joined
FROGGS I'll buy beer to be on the safe side... 09:49
09:50 bbkr left
FROGGS .oO( because we all now, "blog" means "beer + lying on gouch" ) 09:50
09:51 rindolf joined
nwc10 mmm, roast. To eat with the beer 09:51
jnthn Hmm...$dayjob of Perl 6... :) 09:54
*or
Guess I can try and knock out a few more passes :) 09:55
10:01 rindolf left 10:09 dmol1 joined, dmol left 10:13 dmol1 left 10:14 dmol joined
FROGGS jnthn++ 10:27
:o)
dalek kudo/nom: 54b7dd5 | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/ (2 files):
Enforce scalar container type constraints.

Fixes various tests that relied on them being enforced.		 10:31
jnthn Down to 166 test files with failures by now. 10:35
FROGGS jnthn: if you have time, can you run t/spec/S32-num/rshift_pos_amount.t and see if it passes/fails on jakudo? 10:37
I added it but it is not in spectest.data yet 10:38
jnthn pmurias: Your changes broke the Rakudo JVM build :( 10:39
lizmat r: my $a; say $a; $a=Nil; say $a # incorrect, untested bahviour
camelia rakudo d67185: OUTPUT«(Any)␤Nil␤»
jnthn Hm, probably Rakudo Parrot on latest NQP too. 10:40
FROGGS will test
lizmat fwiw, I broke the Nil behaviour!
jnthn FROGGS: Can't try that until I unbust the build... :(
FROGGS jnthn: of course 10:41
lizmat .tell TimToady if we're going to put Nil outside of Any, things like this will break, as they now break with Mu: my $a; say $a === Mu 10:46
yoleaux lizmat: I'll pass your message to TimToady.
lizmat r: my $a; say $a === Mu
camelia rakudo d67185: OUTPUT«Nominal type check failed for parameter ''; expected Any but got Mu instead␤ in sub infix:<===> at src/gen/CORE.setting:2041␤ in block at /tmp/viS4a7jFuG:1␤␤» 10:47
lizmat r: say Mu === Mu
camelia rakudo d67185: OUTPUT«Nominal type check failed for parameter ''; expected Any but got Mu instead␤ in sub infix:<===> at src/gen/CORE.setting:2041␤ in block at /tmp/qywc7ybuFt:1␤␤»
lizmat r: say Mu =:= Mu # works! :-) 10:48
camelia rakudo d67185: OUTPUT«True␤»
jnthn Bother. The thing I thought would fix it didn't. :(
FROGGS jnthn: yes, pakudo doent like it: Method 'get_lexinfo' not found for invocant of class 'NQPRoutine'
jnthn :( 10:49
Yes, making substantial NQP changes and not testing them against Rakudo is a very good way to get me to revert stuff....
Especially when you change the semantics of BEGIN, which was only put in NQP for the sake of Rakduo's BOOTSTRAP. 10:51
10:51 cognominal left
tadzik hmm. Why do Perl 6 modules like Actions.nqp, Pod.nqp etc end up being jars, and not just classes? 10:52
jnthn tadzik: Because we can more easily package the serialization stuff into the JAR
tadzik: Rather than having to shove it in the class file constant pool, which has quite strict limits and required encoding. 10:53
tadzik I see
also, shouldn't make clean remove blib? :) 10:54
but then I have to manually create blib/Perl6 10:55
jnthn yes, we have a .gitignore in there to make sure it exists. :) 10:56
nwc10 I'd never thought of the idea of using .gitignore as a hack to avoid needing an '.exists' file or other directory-dropping
dalek p: 8ca10b2 | jnthn++ | / (2 files):
Revert BEGIN work, as it breaks Rakudo.

It now lives in a branch begin-lexicals.		 10:59
p: f40a989 | jnthn++ | / (9 files):
Revert routine wrapping; breaks Rakudo build.

This commit can also be found in begin-lexicals branch now.
11:00 atroxaper left
jnthn pmurias: The patches caused serious breakage to Rakudo's BOOTSTRAP. It may even *rely* on BEGIN not being fully implemented in NQP. The only reason NQP got BEGIN was for the bootstrap in Rakudo... The work is now in a branch that I've pushed. 11:01
pmurias: I don't actually much care that BEGIN is only half-implemented in NQP; NQP is, after all, just a subset. I guess you want it for something, though? 11:02
pmurias yes
I wanted that to test deserializing closures 11:04
I planned to create them a compile time with BEGIN
jnthn: routine wrapping breaks Rakudo or just BEGIN? 11:05
FROGGS pmurias: I'll be around if you want me to test commits of a given branch (other than master/nom) of parrot/nqp/rakudo
jnthn pmurias: Wrapping. I did this, gist.github.com/jnthn/5908412 , but it was then broken with another error. 11:11
So maybe the BEGIN stuff is to blame for the secondary breakage. 11:12
pmurias: Aren't there already some closure serialization tests? 11:15
11:16 mstefanko left
jnthn t/serialization/03-closures.t 11:16
?
jnthn uses those when porting the JVM stuff...
*to 11:17
pmurias jnthn: I don't have serialization yet, only deserialization so I wanted the compiler to serialize stuff for me 11:19
jnthn: if I want to create a pull request to rakudo, how should I check that it still passes all the tests? 11:20
jnthn pmurias: On Parrot, make spectest should come out pretty clean. On JVM, make sure it at least passes sanity tests and runs spectest... 11:22
11:25 mstefanko joined
tadzik duh. act.yapc.eu/ye2013/talk/4943 11:26
"Perl6 is dead, long live Perl5"
FROGGS hehe 11:28
tadzik: I gues he isnt that serious when looking at the tags
+s
tadzik heh 11:29
I'd have trouble treating that serious anyway :P
jnthn It's funny, I'm fairly sure Perl 5 actually is much better resourced than Perl 6 :P 11:30
nwc10 I believe that it is
and many people doing the ranting aren't contributing resources
tadzik there are plenty of "let's jump on the perl 7 bandwagon" talks submitted
I wonder how many will get through 11:31
11:33 Timbus left 11:35 grondilu joined, Timbus joined
grondilu Hello. Not really perl6 related, but here's an article about a new language for general purpose GPU programming: www.wired.com/wiredenterprise/2013/...-language/ 11:35
tadzik TIL: I've been always using "duh" wrong. 11:36
nwc10 does anyone who wants to *talk* about Perl "7" have any ability to code?
er, to *deliver* code 11:37
lizmat actually, I don't think so
jnthn Or, difficulter, to deliver language design? 11:39
nwc10 or difficultest, to deliver language redesign
everyone forgets that there is already a design.
jnthn *nod*
lizmat language design? What? you need to design? a language? that just develops itself, no? like php, right? 11:40
nwc10 and the more of that design you think you can throw away, the more you throw away of the ecosystem you were actually relying on
lizmat thought of the day: is there a difference between Mu and Nil ? 11:41
nwc10 I will be surprised if there is anyone *talking* who has actually run `git clone git://perl5.git.perl.org/perl.git`
let alone tweaked anything
lizmat nwc10: you're going to be at YAPC::EU ?
jnthn FROGGS: Um, well rshift_pos_amount.t seems to give a stack overflow error 11:42
nwc10 ACT thinks that I am
jnthn (on JVM)
lizmat cool!
do you?
FROGGS jnthn: uhh
I'll build jakudo later and do some testing 11:43
masak lizmat: there's definitely a difference between Mu and Nil. even considering that Nil has meant a number of things lately. 11:46
dalek ecs: a120d5a | (Elizabeth Mattijsen)++ | S99-glossary.pod:
Add parakudo|jakudo
 11:47
hoelzro hehe
lizmat masak: I came to the thought of asking it, as TimToady seemed to play with the idea to place Nil outside of Any
hoelzro can Rakudo on MoarVM be called Mokudo?
mo' kudo, mo' problems.
tadzik Marudo
.tr pl en marudo 11:48
yoleaux pl in marudo (eo → en)
tadzik que?
lizmat hoelzro/tadzik: we'll see which vernacular will stick
nwc10 lizmat: I've paid for it and bought a plane ticket. I think it's likely
hoelzro Mokudo!
lizmat and if Nil is outside of any, what makes it different from Mu then?
if we understand the difference, we know that we need two terms 11:49
if we don't, why don't we stick with Mu then?
playing advocate of the devil here, yes
masak "Mu" is just a really wonky name for "Object". it's a blank slate, representing something entirely without features.
"Nil" has more information than that. 11:50
it means "you looked for something, but it is not here".
11:50 ecocode` joined
masak it's a built-in sentinel. 11:50
11:50 ecocode left
lizmat not seeing much of a difference yet, I mean Nil is also a blank slate, something without features 11:51
even so, that if you assign it to something, it reverts back to its original state
JimmyZ what? not pakudo?
so rakudo is akudo on Ruby? 11:52
lizmat lakudo, rakudo on Lua
back in 15min&
FROGGS damnit! I wanna use Perl6 now for $work, I dont wanna do if( anymore, and I dont wanna pass args like date => $date anymore >:o[ 11:53
JimmyZ parakudo is not like jakudo?
masak FROGGS: that's the spirit! 11:56
FROGGS: use that frustration to further Perl 6 in various ways :) 11:57
FROGGS masak: I do
like 6 hours a day
masak lizmat: 'Mu' is the primordial stateless goo out of which the universe is built, the seed of being and nothingness, but not yet either of them. 'Nil' is that goo specialized into nothingness. 11:58
'Any' is that goo specialized into being.
'Junction' is that goo specialized into parallel being.
or superposed being. 11:59
12:13 grondilu left 12:15 konundra joined
lizmat masak++ 12:23
masak well, 老子++, really. :) 12:25
JimmyZ :D 12:26
ztt 哈哈哈哈 12:28
JimmyZ Claiming 老子 means I'm arrogant 12:29
timotimo "Development of Perl6 is sucking in huge amounts of community creativity, attention and effort." wait what. 12:30
hoelzro timotimo: source?
12:31 konundra left
timotimo act.yapc.eu/ye2013/talk/4943 12:31
masak JimmyZ: in what way? he said lots of good things. 12:32
timotimo: I'm just so tired of that zero-sum bullshit. 12:33
12:33 atroxaper joined, atroxaper left
timotimo yeah, i can understand it 12:33
but that's even *worse*
JimmyZ masak: I can't follow you ...
timotimo that's blowing the work that's going into perl6 so ridiculously out of perspective
JimmyZ masak: oh I know 12:34
timotimo "damnit, if those two hundred perl6 hackers would just come back to perl5!", wat?
JimmyZ I know what you mean
lizmat otoh, I do think the Perl community as a whole, has lost a lot of good people in the past 10 years 12:35
JimmyZ masak: we say the different things
lizmat getting *them* back is one of *my* long term goals :-)
BinGOs "We know where you live" 12:36
lizmat actually, I wish we did, only the NSA knows :-( 12:37
ztt that act smells like a joke 12:38
lizmat on the way to the NLPM meeting& 12:39
aka Amsterdam.PM meeting 12:40
jnthn Grr. I fixed two things, won a bunch of tests...but lost another load
o/ lizmat
12:40 lizmat left 12:41 ajr joined, ajr is now known as Guest93975 12:42 Guest93975 is now known as ajr_
timotimo jnthn: hopefully these other tests have been injustly passing before you fixed a bug that would bring the injustice to light .... or something 12:43
12:43 konundra joined
jnthn timotimo: Yeah, am hunting down what's going on now :) 12:44
I did manage to win us the transliteration tests. :)
[Coke] I can sympathize with the p5 people that think p6 is pulling cycles away from p5. (see moarvm vs. parrot, as a smaller example of a very similar thing) 12:49
s/p5:first/perl/, maybe.
masak I can sympathize, too. 12:51
I just don't think it's true in, you know, a factual sense. 12:52
colomon [Coke]: I've never understood that logic. I mean, it's not like the p6 developers (TimToady excepted) are established p5 hackers who were lured away by something shiny.
moritz colomon: I think that was different in the beginning
colomon moritz: sure, a decade ago 12:53
12:58 rindolf joined, jaldhar joined
[Coke] colomon: I'm not saying they are right, of course. 12:59
12:59 mtk left
[Coke] jnthn++ # more jvm tests 12:59
jnthn, sorear: soon we'll have to fix actual failing tests instead of exploding ones. 13:00
13:00 mtk joined
jnthn [Coke]: Well, doing a bit of that now too 13:00
nwc10 [Coke]: "p5 people" - I'm not aware of much overlap between the people blowing off about Perl 6 and the people contributing code to Perl 5 (either the core, or (show me to be wrong) important things on CPAN) 13:01
[Coke] not porters. 13:03
13:04 woosley1 joined
nwc10 so people who think it's more useful to contribute by talking. But don't actually contribute in any way to the thing that they say matters to them. 13:04
13:09 daniel-s_ joined 13:10 PacoAir joined
Ulti I think there is a mismatch in perl-land between people who are employed to program in perl, and those that just like it as a go to language for fun things as well as bits of work 13:10
the people who only program in perl5 every day at work and not in the evening are the ones most bitter as far as I can see... its where all the production ready cries come from
because until its the mythical production ready release they aren't even going to start learning the language or writing a single line 13:11
nwc10 not thought of it that way (work vs play). Interesting
Ulti so basically for them its been a decade of nothing
nwc10 but those people *also* don't contribute to Perl 5
Ulti rather than a decade of play and hopeful steady improvements 13:12
nwc10: well probably not core, but they likely have that once CPAN module only they use :)
*one
nwc10 use? or wrote? 13:13
Ulti both
jnthn ah, think I found why I busted stuff :) 13:14
Ulti I groan when I read the standard perlmonks perl6 troll about production readyness and waiting a decade *blah* *blah*... I'm in academia you just use whatever is a good tool for the job, I've used perl6 in "production" just to parse an especially nasty custom text format into something usable by my perl5 scripts
jnthn *and* another bug :)
13:15 jaldhar left
[Coke] jnthn: as long as the net # of passing tests increases! 13:17
13:17 jaldhar joined
jnthn [Coke]: yeah, I managed to lose some, but think I've saved them again :) 13:18
13:19 mtk left
masak Ulti: the "work vs play" notion is brilliant. thank you. Ulti++ 13:19
nwc10 I also see a bunch of other, um, factual inaccuracies such as grumbling that for PHP there's now Hip-Hop, which is a drop-in faster replacement. Which, well, means that the author hasn't read github.com/facebook/hiphop-php/blo...sistencies
jnthn Indeed 13:22
"HipHop VM currently does not support PHP namespaces."
That feels like a fairly sizable missing thing...though maybe namespaces ain't used much in PHP. :) 13:23
13:23 mtk joined
nwc10 well, they are new, and they use the most awesome character ever as the namespace separator 13:23
jnthn Yeah!
Nice for Windows users. ;) 13:24
"oh, so it's like a directory" :)
tadzik :D
nwc10 oh, so it's part of a meta-character in double-quoted strings 13:25
dalek p: fa3d25f | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Invoke accepts_type on correct meta-object.
 13:28
kudo/nom: dceb13c | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Binder.java:
Implement checking post constraints.
 13:30
kudo/nom: faa0f9e | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Ops.java:
Implement nqp::p6isbindable.

Fixes various of the remaining multi-dispatch issues, and other things that relied on multi-dispatch, such as transliteration.
kudo/nom: 7dfcd66 | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/RakudoContainerSpec.java:
Fix error reporting.
colomon what's the file op for creation date?
moritz nqp::stat maybe? 13:32
colomon in p6, I mean
moritz r: say 'README'.path.created 13:33
camelia rakudo 54b7dd: OUTPUT«No such method 'created' for invocant of type 'IO::Path'␤ in block at /tmp/LLLUUTZdkF:1␤␤»
colomon hmm.... changed or modified? (probably either is close enough, but what's the difference?)
timotimo i think one is for data and the other is for metadata
moritz one is about content, one is for meta data
oh, and POSIX doesn't track the 'created' date of a file 13:34
colomon is pretty certain these files have never been changed...
moritz++ timotimo++
colomon needs to blog on this project. :) 13:37
timotimo now i'm interested to know what it is 13:38
colomon ;)
I'm coding up a p6 script to figure out which photo files on my hard drives are duplicates that can be safely deleted 13:39
timotimo ah
while that's not useful to me in particular, that does sound nice :)
colomon it's got lots of fun bits, like sort(*.changed)
to get the oldest file first 13:40
tadzik colomon: install fdupes :)
or fdups
I did the same thing a while ao
colomon and categorize(*.basename) to find which files have the same filename but different paths 13:42
masak why 'categorize' and not (the simpler) 'classify'? 13:43
I'm asking because I don't think I see the point of 'categorize' yet. I've never used it, only 'classify'.
colomon .... good point, I think
think-o 13:44
masak yes. S32/Containers seems to suggest 'classify' is what to use here.
colomon funny, in that I did use classify a couple of lines later...
classify({ MD5($_) }
masak :)
colomon I had a great use for categorize a few days ago, let me see if I can dig it up 13:45
timotimo colomon: wouldn't classify: *.&MD5 also work? 13:47
colomon timotimo: calling a sub as a method?
masak classify: &MD5
timotimo oh, even better still
masak yes, it's the only one that removes a level of invocation. 13:48
colomon masak++
timotimo r: sub do-something($a) { say "someone think of the $a!" }; "children".&do-something
camelia rakudo 54b7dd: OUTPUT«someone think of the children!␤»
colomon I used categorize to invert a hash
I'm now trying to figure out if invert would have worked as well. It seems like it's not quite the same thing? 13:49
timotimo well, invert gives you a list of pairs, if you push that into an empty hash, you'll get a hash with once-values-now-keys and list of once-keys-now-values
colomon actually, it was a bit more than inverting the hash. but it's an interesting fine line. 13:53
most ponder
*must 13:56
13:57 daniel-s_ left
masak sounds interesting. please share if the pondering yields something. :) 13:58
13:59 daniel-s- joined 14:05 FROGGS left 14:06 lizmat joined, kaleem left 14:09 daniel-s- left 14:10 notjack joined
lizmat is in the car, on the way to Amsterdam 14:12
and not driving
masak :)
14:20 ztt left 14:21 xilo joined
jnthn grr...working out this given.t fail is a bit tricky 14:24
14:26 FROGGS[mobile] joined 14:33 skids left 14:41 ecocode` left
dalek ecs: c4f1fe8 | (Elizabeth Mattijsen)++ | S11-modules.pod:
Some preliminary work on CompUnitRepo's interface
 14:42
14:43 lizmat left
jnthn aha.... 14:44
yay. given.t passes in full now. 14:46
masak yay
timotimo hooray!
jnthn Gets us another 45. 14:47
notjack perl6: my $t = 'pass'; given $t {when 'pass' {say 'yay'} when 'fail' {say 'boo'}} 14:48
camelia rakudo 7dfcd6: OUTPUT«===SORRY!===␤Confused␤at /tmp/s2B0BGUsuv:1␤------> ass'; given $t {when 'pass' {say 'yay'} ⏏when 'fail' {say 'boo'}}␤ expecting any of:␤ postfix␤ statement end␤ statement modifier␤ statement modifi…
..niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Strange text after block (missing comma, semicolon, comment marker?) at /tmp/Sb0zHQGy15 line 1:␤------> pass'; given $t {when 'pass' {say 'yay'}⏏ when 'fail' {say 'boo'}}␤␤Parse failed␤␤»
timotimo semicolon
.o(rakudo could totally steal that error message, no?
)
maybe i'll look into that later today
jnthn I thought I tried that once and ran into something, but I totally forget what. Maybe it was nothing big. 14:49
notjack confused at 'ass' 14:50
I've been there, camelia.
jnthn ;)
14:52 raiph joined
timotimo runs away scared instead 14:52
jnthn timotimo: nah, give it a try! 14:53
dalek p: 800df5e | jnthn++ | src/vm/jvm/QAST/Compiler.nqp:
Be sure to keep tc.curFrame in sync.
 14:55
TimToady always wonders whether LuaJIT can possibly have the type finesse to handle Perl 6 when the type system of Lua is so impoverished^Wsimple
yoleaux 10:46Z <lizmat> TimToady: if we're going to put Nil outside of Any, things like this will break, as they now break with Mu: my $a; say $a === Mu
hoelzro hmm
I've thought about a Lua backend for NQP; I wonder how much work that would take? 14:56
bonus: really easy-to-write C extensions =)
jnthn hoelzro: The answer, whether or not you s/Lua//, is "lots" :)
TimToady tell lizmat If we put Nil outside of Any, we probably need a .is-nill primitive, or some way of training people to use =:= when they try to use ===
.tell lizmat If we put Nil outside of Any, we probably need a .is-nill primitive, or some way of training people to use =:= when they try to use ===
yoleaux TimToady: I'll pass your message to lizmat.
14:57 Chillance joined
hoelzro heh 14:57
TimToady much like we currently do with smartmatching
std: 42 when Nil
camelia std d4cc5ab: OUTPUT«Potential difficulties:␤ Smartmatch against Nil always fails; perhaps you should use * === Nil or :!defined instead at /tmp/ZaNr1otX5x line 1:␤------> 42 when ⏏Nil␤ok 00:00 42m␤»
jnthn oh...the loop code-gen doesn't know how to handle arity 1 loop blocks...
TimToady of course, now that message is wrong, unless we fix === Nil :) 14:58
as I mentioned yesterday, we can do something similar to junctions to force === to work on Nil, though perhaps aliasing to none() is overkill 14:59
TimToady is actually a bit surprised that =:= Nil works, which probably means he is thinking of =:= wrongishly... 15:00
15:00 snoopy left
jnthn Or implementations are thinking of =:= wrongishly... ;) 15:00
I think it means at the momnet something like "are the things on each side at the same memory address" 15:01
TimToady doesn't think of those things as containers, but it's really just a more primitive container notion
and thus perhaps the right thing here
nr: constant Mu2 = Mu; say Mu2 =:= Mu 15:03
camelia rakudo 7dfcd6, niecza v24-86-g39ab531: OUTPUT«True␤»
TimToady \o/
TimToady notest that the only instance of '=== Nil' in either niecza or rakudo are actually down in roast 15:07
*instances
all in the pick and roll tests
raiph does anyone know if any gather/take implementation does batching yet? 15:12
is sorear's work on continuations relevant to laziness being useful/performant? 15:13
dalek p: 98a7d2e | jnthn++ | src/vm/jvm/QAST/Compiler.nqp:
Handle arity-1 loop constructs.
raiph (for gather/take)
jnthn The continuation work on JVM was done for gather/take. 15:14
(Without that, we wouldn't have been able to support it at all on JVM.)
raiph jnthn: thanks 15:16
jnthn: does the rakudo gather/take do batching?
15:17 rindolf left, woosley1 left
raiph jnthn: is/was batching a tricky thing to implement? 15:18
TimToady raiph: it's unlikely to do that before we figure out how to optimize some gather/takes to use interthread queues
but once there's a queue, there's a natural buffer for batching
it might be worth batching some in the continuation model as well, if there's a place to put the results, and if taking a continuation is expensive 15:19
jnthn raiph: I believe it doesn't do batching.
I think it'll be a worthwhile optimization.
TimToady and if we can detect where strict laziness isn't required
jnthn Yeah. 15:20
I think pmichaud++, when he gets tuits for it, is planning on some work on letting us know such things.
Or at least, know more... :)
TimToady recursive constant definitions will tend to want strict laziness, for instance, but maybe this falls out of recognizing we're asking for something that hasn't been computed yet
15:21 domidumont left
dalek p: 65a2aef | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
nqp::decont shouldn't explode on nulls.
 15:21
TimToady would still like to be able to define Hamming numbers recursively: 15:23
constant @hamming = 1, dedup (@hamming X* 2) M (@hamming X* 3) M (@hamming X* 5);
I guess that'd be s/dedup/squish/ these days, now that it's a builtin 15:24
15:24 raiph left
TimToady (and assuming M is defined somehow to be a lazy n-way merge) 15:24
nr: say squish 1,2,2,3 15:31
15:31 raiph joined
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Undeclared routine:␤ 'squish' used at line 1␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 1502 (die @ 5) ␤ at /home/p6eval/niecza/src/STD.pm6 line 1147 (P6.comp_unit @ 36) ␤ at /h… 15:31
..rakudo 7dfcd6: OUTPUT«===SORRY!===␤Undeclared routine:␤ squish used at line 1␤␤»
TimToady NYI I guess
is LHF though 15:32
timotimo the sun is so lovely out, but even with a light terminal and full backlight i can hardly use my editor or irc client :| 15:33
also the keyboard gets pretty darn hot
TimToady you need to find a cave 15:34
timotimo just 5 meters to my right is a hackspace ;)
but it's warm in there. it's direltcy under a roof and the roof isn't insulated properly
TimToady that's what IR clogs are for 15:35
maybe it would help to turn the hackspace over so that the floor is radiating instead of the ceiling... 15:36
timotimo hehe 15:37
lee_ does this look like a useful optimization for [+] 1..10000? gist.github.com/leedo/735a5f09350bd2d50d33
it seems to cut the time down quite a bit 15:38
timotimo screed heating? :)
imgur.com/a/BLsoE <- my spot right now 15:40
lee_: watch out, that may blow up if someone defines their own infix:<+>
at least i always had to make sure that didn't happen when i worked with the optimizer 15:41
oh, i'm talking crap, since it's lexical, you'll always compare with the "correct" infix:<+>
15:42 crab2313 joined
lee_ i wasn't sure what the best way to test that min and max are both Ints 15:43
colomon I'd say $values.min ~~ Int && $values.max ~~ Int 15:44
15:50 crab2313 left 15:57 FROGGS joined
colomon just erased 24 gigs worth of duplicated photos 15:59
15:59 domidumont joined
FROGGS ohh, I read "duplicated p*rnos" for a moment 16:00
16:00 PacoAir left
TimToady lee_: there's something a little wrong with it in regards to flattening, since a range in list context flattens, but you're kinda bypassing the flattening, such that [+] $range would "flatten" $range when it shouldn't, seems to me 16:00
jnthn Pro tip: an epoch in milliseconds is quite different from an epoc in seconds. 16:01
TimToady epochs shouldn't be integers at all, in P6-think...
TimToady realizes that this is a minority view in the world today :) 16:02
jnthn Well, we gotta get what we want in terms of what platforms think ;)
TimToady that's part of the world I was referrin' to
pmurias jnthn: the way I'm solving the lack of BEGIN to generate a convenient example of a serialized closure is it insert vim folds into generated code so that I can view more complex stuff
jnthn bah, all that and the test that wants $*TZ explodes 'cus we don't have sprintf that works yet... 16:04
dalek kudo/nom: 284a76f | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Ops.java:
Implement nqp::p6decodelocaltime.
kudo/nom: 5b1775f | jnthn++ | src/core/Temporal.pm:
Enable $*TZ on JVM.
TimToady lee_: also, I'm not sure the form you have it would be amenable to constant folding--you'd like [+] 1..10000 to fold at compile time 16:05
16:05 konundra left
TimToady nr: my $range = 1..10000; say [+] $range 16:06
camelia niecza v24-86-g39ab531: OUTPUT«1..10000␤»
..rakudo 7dfcd6: OUTPUT«10000␤»
TimToady rakudo++
nr: my $range = 1..10000; say [+] @$range
camelia rakudo 7dfcd6, niecza v24-86-g39ab531: OUTPUT«50005000␤»
TimToady lee_: I don't think your optimization distinguishes those as it ought 16:07
lee_ hmm yeah, i see your point
TimToady probably need something slightly more ASTy 16:08
16:10 kaleem joined
TimToady nr: sub foo (*@args) { say +@args }; sub bar (**@args) { say +@args }; foo 1..100, 1..100; bar 1..100, 1..100 16:10
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Slice parameters NYI at /tmp/5rPlwdIOLf line 1:␤------> *@args) { say +@args }; sub bar (**@args⏏) { say +@args }; foo 1..100, 1..100; ba␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/l…
..rakudo 7dfcd6: OUTPUT«200␤2␤»
TimToady r: sub bar (**@args) { say @args[0].WHAT }; bar 1..100 16:12
camelia rakudo 7dfcd6: OUTPUT«(Range)␤»
TimToady r: sub bar (**@args) { say @args[0].WHAT }; my $range = 1..100; bar $range
camelia rakudo 7dfcd6: OUTPUT«(Range)␤»
TimToady rakudobug
(I suspect) 16:13
r: sub bar (|args) { say args[0].WHAT }; my $range = 1..100; bar $range
camelia rakudo 7dfcd6: OUTPUT«(Range)␤»
lee_ looks up *@args vs **args, i'm new!
16:13 kaleem_ joined
TimToady r: sub bar (|args) { say args[0].WHAT }; my $range = 1..100; bar |$range 16:13
camelia rakudo 5b1775: OUTPUT«(Int)␤» 16:14
TimToady r: sub bar (|args) { say args[0].WHAT }; my $range = 1..100; bar @$range
camelia rakudo 5b1775: OUTPUT«(List)␤»
jnthn timotimo: Just did the Makefile re-ordering we discussed last night. :)
timotimo: Gets rid of some errors :)
[Coke] jnthn: almost time for the daily run. should I wait 10m? 16:15
jnthn [Coke]: Yes ;)
Will push in a moment
(within 10)
TimToady r: sub bar (|args) { say args[0].WHAT }; my @range := 1..100; bar @range
camelia rakudo 5b1775: OUTPUT«(Range)␤»
TimToady r: sub bar (|args) { say args[0].WHAT }; my $range := 1..100; bar $range
camelia rakudo 5b1775: OUTPUT«(Range)␤»
16:15 Stygia left
TimToady so where's the $ vs @ difference? 16:16
16:16 kaleem left
TimToady such that one could emulate normal slurpy by examining |args? 16:17
jnthn
.oO( in the mind of whoever wants binding not to be a low level operation... :S )
16:17 jlaire left
TimToady n: sub bar (|args) { say args[0].WHAT }; my @range := 1..100; bar @range 16:17
camelia niecza v24-86-g39ab531: OUTPUT«Unhandled exception: System.NullReferenceException: Object reference not set to an instance of an object␤ at Builtins.InvokeSub (Niecza.P6any obj, Niecza.Variable[] pos) [0x00000] in <filename unknown>:0 ␤ at Niecza.Kernel.ToComposable (Niecza.STable arg…
jnthn r: say Range ~~ Positional 16:18
camelia rakudo 5b1775: OUTPUT«True␤»
TimToady |args should be very raw
jnthn That's why it's allowed to be bound directly to :=
It is. And @a and $a are just names of storage locations.
16:19 SamuraiJack_ left
TimToady and how does normal binding distinguish $ from @ then? 16:19
jnthn As in, why is it an error to do:
r: my @a := 'foo'
camelia rakudo 5b1775: OUTPUT«Type check failed in binding; expected 'Positional' but got 'Str'␤ in block at /tmp/O4nHjGxa_F:1␤␤»
jnthn ?
TimToady as in, how does *@args know to flatten @ but not $ 16:20
jnthn *thought* it was a combination of type and "not in a scalar container"
jnthn has no idea how much Pm expects the current details to change in this regard... 16:21
TimToady n: sub bar (|args) { say args[0].WHAT }; my $range = 1..100; bar $range 16:22
camelia niecza v24-86-g39ab531: OUTPUT«Unhandled exception: System.NullReferenceException: Object reference not set to an instance of an object␤ at Builtins.InvokeSub (Niecza.P6any obj, Niecza.Variable[] pos) [0x00000] in <filename unknown>:0 ␤ at Niecza.Kernel.ToComposable (Niecza.STable arg…
TimToady r: sub bar (|args) { say args[0].WHAT }; my $range = 1..100; bar $range
camelia rakudo 5b1775: OUTPUT«(Range)␤»
TimToady no scalar container there...
jnthn Um, they're invisible, though, no?
r: my $x = 1..100; say $x.WHAT
camelia rakudo 5b1775: OUTPUT«(Range)␤»
16:22 jlaire joined
jnthn r: my $x := 1..100; say $x.WHAT 16:22
camelia rakudo 5b1775: OUTPUT«(Range)␤»
FROGGS ohh, Stage parse : 59,851 for jakudo... I should have tried it earlier :o)
dalek kudo/nom: f2b0f89 | jnthn++ | tools/build/Makefile-JVM.in:
Sync CORE.setting ordering with Parrot Makefile.

No reason for them to be different. timotimo++ for noticing.		 16:23
jnthn FROGGS: How much is it for you on Parrot? :)
[Coke]: Fire!
TimToady r: my $x = 1..100; my @a = $x; say +@a
camelia rakudo 5b1775: OUTPUT«1␤»
TimToady r: my $x = 1..100; my $a = $x; say +$a
camelia rakudo 5b1775: OUTPUT«100␤»
FROGGS jnthn: 130s 16:24
jnthn FROGGS: Hm, difference is a bit more pronounced than that for me.
FROGGS damn it, I forgot to buy beer :/
jnthn :( 16:25
TimToady r: sub bar (|args) { say VAR(args[0]).WHAT }; my $range = 1..100; bar $range
camelia rakudo 5b1775: OUTPUT«===SORRY!===␤Undeclared name:␤ VAR used at line 1␤␤»
jnthn .VAR works
TimToady r: sub bar (|args) { say args[0].VAR.WHAT }; my $range = 1..100; bar $range
camelia rakudo 5b1775: OUTPUT«(Scalar)␤»
TimToady there we go
r: sub bar (|args) { say args[0].VAR.WHAT }; my @range := 1..100; bar @range
jnthn ah, that's what you were after...
camelia rakudo 5b1775: OUTPUT«(Range)␤» 16:26
16:26 arnsholt joined
TimToady what lee_++ would need to emulate flattening to do the optimization correctly by that method 16:26
arnsholt o/
jnthn o/ arnsholt 16:27
TimToady r: sub bar (|args) { say args[0].VAR }; my @range := 1..100; bar @range
camelia rakudo 5b1775: OUTPUT«1..100␤»
arnsholt jnthn: From a cursory look at JNA, it looks like it passes pointers around in Java longs, so I think I'll try to mirror the Parrot approach to NativeCall on JVM as a first whack 16:28
TimToady r: sub bar (|args) { say args[0].VAR }; my $range = 1..100; bar $range
camelia rakudo 5b1775: OUTPUT«1..100␤»
TimToady I guess .gist loses the distinction
16:28 fgomez joined, pmurias left
TimToady r: sub bar (|args) { say args[0].VAR.Str }; my $range = 1..100; bar $range 16:29
camelia rakudo 5b1775: OUTPUT«Range<-109147205>␤»
TimToady heh
jnthn arnsholt: ok
arnsholt: Good luck! :) 16:30
arnsholt Thanks =)
TimToady r: sub bar (|args) { my $which = say args[0].VAR; say $which.WHAT }; my $range = 1..100; bar $range 16:31
arnsholt Passing pointers around in integers is kind of horrible, but I think it'll work
camelia rakudo 5b1775: OUTPUT«1..100␤(Bool)␤»
FROGGS I think I'm trying j5 today...
jnthn arnsholt: It's only cheating if you get caught ;)
TimToady r: sub bar (|args) { my \which = say args[0].VAR; say which.WHAT }; my $range = 1..100; bar $range
camelia rakudo 5b1775: OUTPUT«1..100␤(Bool)␤»
TimToady jnthn: why Bool?
duh 16:32
moritz say
TimToady r: sub bar (|args) { my \which = args[0].VAR; say which.WHAT }; my $range = 1..100; bar $range
camelia rakudo 5b1775: OUTPUT«(Scalar)␤»
arnsholt jnthn: I'll hide it as well as I can, in that case =D
djanatyn is there something equivalent to template::toolkit or jinja2 or mustache for perl6
moritz maybe we should spec say() return "you are looking at the return value from &say"
[Coke] jnthn: fired. 16:33
TimToady well, that would work as a True value, but not as a False value :)
16:33 kaleem joined 16:34 kaleem_ left
TimToady "you are looking at the return value from &say" but False, obviously... :) 16:34
jnthn [Coke]: yay. Hopefully we make the 90% :)
Especially since I seem to have spent the day patching Rakudo rather than doing $dayjob :)
[Coke] jnthn: oh, that's a given. unless you screwed something up! 16:35
jnthn Yes, there is that potential ;)
TimToady .tell lizmat A STD.pm6 commit message of "Same change as in rakudo" is not very informative when I do 'git log' :P 16:39
yoleaux TimToady: I'll pass your message to lizmat.
dalek d: 102c740 | larry++ | STD.pm6:
Recommend =:= Nil rather than ===
 16:40
sorear good * #perl6 16:42
jnthn o/ sorear
pmichaud yes, when I get tuits I want to review the whole list model, especially in light of what I learned from preparing the parallelism talk 16:44
so can clean up Nil and stuff then as needed.
sorear o/ pmichaud
pmichaud sorear: btw, your CLA arrived. 16:45
colomon \o
\o/
sorear pmichaud: \o/ \o/
colomon he's official!
TimToady now he just needs to learn to be officious like me :)
sorear
.oO( any idea how many cities it went through 'cus USPS can't read my atrocious handwriting? )
 16:46
TimToady: :p
16:48 PacoAir joined
pmichaud sorear: no, don't know that, sorry. :) 16:49
sorear did it at least get recorded under the correct name? 16:50
pmichaud looking
16:51 mstefanko left
TimToady and that would be Murgatroyd McGillacudy? 16:51
16:53 kaleem left
pmichaud sorear: I think it's recorded under the correct name, since I just now recorded it :) 16:53
16:54 Lorn_ is now known as Lorn, Lorn left, Lorn joined
TimToady been thinking about the list context thing, and if we're not to pass context down to listops saying how eager we are, then the listop has to return an uncommitted list back up, and have lazy/eager/hyper/race select the algorithm lazily after the return somehow 16:55
16:55 dakkar left
TimToady lazy and eager can just be signaled by how you pull args off the list, but hyper and race have to have some other means of telling the algorithm to behave differently 16:55
16:56 mstefanko joined
TimToady and it would be nice if sink could do that too, rather than just being eager and throwing stuff away 16:56
it would also be nice if the mechanism we choose is sufficiently amenable to static analysis that we can pass context down the calls at compile time where the decision doesn't have to be deferred to run time 16:58
pmichaud I think the iterators can signal hyperness downward, though.
at least, that's the model I've been playing with a bit
(mentally)
TimToady what I'm saying is that the compile-time aspects of context could be considered an optimization, and the general solution does not have to involve having passing the context to each thing that might return a list, but merely to the interator object that is returned, as you say 16:59
pmichaud right 17:00
moritz so that means we're getting rid of inward-propagated sink context?
TimToady by that view, thinking of special calls that pass context downward can be viewed as a premature optimization
pmichaud I'm thinking inward-propagated sink is just a form of hyper/race 17:01
TimToady moritz: it means we could
I'm saying it only has to propagate into the iterator, not into the map/grep/whatever
pmichaud well, map/grep/whatever *is* an iterator at the underlying level.
it's just an iterator wrapped inside of a List
anyway, I've been thinking that I have the List/ListIter boundary slightly off in Rakudo 17:02
TimToady otoh, it might mean that map/grep/whatever have to provide different algorithms that this "inversion of control" could select
pmichaud and that fixing that will clean up a lot of the hyper/sink/eager/other stuff.
anyway, I don't know when tuits will arrive again. 17:04
all of our predictions about my tuit availability have been way off thus far.
TimToady or we need to find the primitives that we can implement map in a context-ignorant way and have it Just Work as a hyper map or race map 17:05
pmichaud TimToady: that's what I've been thinking will work.
well, not exactly.
MapIter becomes hyper and race aware, at any rate.
indeed, all of the *Iters do
TimToady and where the language accidentally enforces ordering, we need to address that
pmichaud (there aren't that many... ListIter, MapIter, GatherIter, Range)
timotimo jnthn: i tried to do the reordering of the makefile myself and couldn't get it to compile properly :( 17:06
jnthn timotimo: oh...weird.
timotimo: Well, it's done now
timotimo but if you were successful, i'm glad :) 17:07
jnthn And helped :)
yeah, IO spec tests are in better shape now
TimToady (which, btw, probably means I'm gonna stick to my guns on the non-eagerness of for loops)
but we'll see
17:07 ajr_ left
jnthn uh, IO::Spec spectests, that is :) 17:07
TimToady does anyone foresee a difficulty with changing all the === Nil spectests to =:= Nil? 17:08
moritz
=:= is meant to test containers, not identity 17:09
r: my $x = Nil; say $x =:= Nil;
camelia rakudo f2b0f8: OUTPUT«False␤»
TimToady that's supposed to put Any into $x
r: my $x := Nil; say $x =:= Nil 17:10
camelia rakudo f2b0f8: OUTPUT«True␤»
TimToady so it turns out that =:= is using a more primitive notion of "container" than somoe of us have been thinking, but that seems fine 17:11
17:12 konundra joined
TimToady r: my $x := Mu; say $x =:= Mu 17:12
pmichaud I'm afk again
camelia rakudo f2b0f8: OUTPUT«True␤»
TimToady r: my $x = Mu; say $x =:= Mu
camelia rakudo f2b0f8: OUTPUT«False␤»
TimToady hmm
r: my $x = Mu; say |$x =:= Mu 17:13
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤CHECK FAILED:␤Undefined routine '&prefix:<|>' called (lines 1, 1)␤»
17:13 birdwindupbird left
TimToady how to deref a $x containing Mu (or Nil)... 17:13
masak r: my $x = Mu; say $x.WHICH; say Mu.WHICH
camelia rakudo f2b0f8: OUTPUT«Mu␤Mu␤»
masak hm.
TimToady r: my \x = Mu; say x =:= Mu
camelia rakudo f2b0f8: OUTPUT«True␤»
TimToady I guess don't put it into $x in the first place 17:14
r: my $x = Mu; my \x = $x; say x =:= Mu
camelia rakudo f2b0f8: OUTPUT«False␤»
TimToady hard to strip off that (Scalar) object... 17:15
r: my $x = Mu; my \x = |$x; say x =:= Mu
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤CHECK FAILED:␤Undefined routine '&prefix:<|>' called (line 1)␤»
TimToady we seem to be missing a primitive
moritz r: my $x = Mu; say nqp::decont($x) =:= Mu 17:16
camelia rakudo f2b0f8: OUTPUT«True␤»
moritz r: say Mu =:= Any
camelia rakudo f2b0f8: OUTPUT«False␤»
TimToady maybe we should expose decont? 17:17
17:17 raiph left
moritz what's the use case? 17:17
(besides using =:= for stuf it wasn't intended to do) 17:18
17:18 fhelmberger left
TimToady looking at a user-supplied value in $foo to see if it's Nil? 17:18
well, =:= was intended to compare two pointers, really, and that's still what it's doing
we just don't have a language-level way to name the pointer in $x yet 17:19
17:19 raiph joined 17:22 xilo left
TimToady otoh, there might be a way of getting === to work even if Nil is outside of Any, if Nil is just a pseudo-junction of some sort 17:23
17:23 spider-mario joined
TimToady or more generally, if the failover to Junction can be generalized to other conceptual types outside of Any 17:24
diakopter "Ranges from S02 and S03 which are nearly frozen" 17:25
labster: ...
TimToady diakopter!! \o/
colomon \o/ 17:26
diakopter .seen diakopter 17:27
yoleaux You're right there.
17:29 xilo_ left
masak .seen yoleaux 17:30
yoleaux I'm right here.
masak wherever you go, there you are.
[Coke]: by the way, thank you for the tip about Buckaroo Banzai ;)
17:32 rindolf joined 17:35 ajr joined, ajr is now known as Guest50674 17:36 Guest50674 is now known as ajr_ 17:37 daxim left 17:41 xilo joined
notjack perl6: sub f(*%args) { say %args;} f(1,2,'three', four=>'five'); 17:43
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/PuSqco7PSV:1␤------> sub f(*%args) { say %args;} ⏏f(1,2,'three', four=>'five');␤ expecting any of:␤ postfix␤ statement end␤ statement modifier␤ statement mod…
..niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Strange text after block (missing comma, semicolon, comment marker?) at /tmp/fzhjxMEy4R line 1:␤------> sub f(*%args) { say %args;}⏏ f(1,2,'three', four=>'five');␤␤Parse failed␤␤»
TimToady masak: we used to live a couple miles away from that dam they use in the end credits :)
masak ooh 17:44
masak .oO( we used to live dam close )
TimToady one of those dams that gets water behind it every twenty years or so, but in so doing prevents downtown LA from washing away, which could be construed as either good or bad
masak :) 17:45
TimToady r: sub decont($var) { $var }; my $x = Mu; say decont($x) =:= Mu 17:48
camelia rakudo f2b0f8: OUTPUT«Nominal type check failed for parameter '$var'; expected Any but got Mu instead␤ in sub decont at /tmp/kHY49emIC3:1␤ in block at /tmp/kHY49emIC3:1␤␤»
TimToady r: sub decont(Mu $var) { $var }; my $x = Mu; say decont($x) =:= Mu
camelia rakudo f2b0f8: OUTPUT«False␤»
TimToady r: sub decont(\var) { var }; my $x = Mu; say decont($x) =:= Mu 17:49
camelia rakudo f2b0f8: OUTPUT«Nominal type check failed for parameter 'var'; expected Any but got Mu instead␤ in sub decont at /tmp/FCAYkOSda8:1␤ in block at /tmp/FCAYkOSda8:1␤␤»
TimToady r: sub decont(Mu \var) { var }; my $x = Mu; say decont($x) =:= Mu
camelia rakudo f2b0f8: OUTPUT«False␤»
TimToady rats
r: sub decont(Mu \var) { var }; my $x = Mu; say decont($x).WHAT 17:50
camelia rakudo f2b0f8: OUTPUT«(Mu)␤»
TimToady r: sub decont(Mu $var) { $var }; my $x = Mu; say decont($x).WHAT 17:51
camelia rakudo f2b0f8: OUTPUT«(Mu)␤»
TimToady r: sub decont(Mu $var) { $var }; my $x = Mu; say $x.WHICH; say decont($x).WHICH; say Mu.WHICH 17:55
camelia rakudo f2b0f8: OUTPUT«Mu␤Mu␤Mu␤»
17:55 xilo left
TimToady chews his cud 17:55
r: sub decont(Mu $var) { $var }; my $x = Mu; say $x.WHERE; say decont($x).WHERE; say Mu.WHERE
camelia rakudo f2b0f8: OUTPUT«459420040␤459420040␤459420040␤»
jnthn Um. 17:56
.WHERE decontainerizes?
TimToady r: sub decont(Mu \var) { var }; my $x = Mu; say decont($x) =:= Mu; # why doesn't work?
camelia rakudo f2b0f8: OUTPUT«False␤»
TimToady r: sub decont(Mu $var) { $var }; my $x = Mu; say decont($x) =:= Mu; # why doesn't work? 17:57
camelia rakudo f2b0f8: OUTPUT«False␤»
TimToady r: sub decont(Mu $var) { $var }; my $x = Mu; say nqp::decont($x) =:= Mu; # why doesn't work?
camelia rakudo f2b0f8: OUTPUT«True␤»
jnthn r: sub decont(Mu $var) { $var }; say decont(Mu).VAR
camelia rakudo f2b0f8: OUTPUT«(Mu)␤»
jnthn r: sub decont(Mu $var) { $var }; say decont(Mu) =:= Mu 17:58
camelia rakudo f2b0f8: OUTPUT«True␤»
jnthn r: sub decont(Mu $var) { $var }; my $x := Mu; say decont($x) =:= Mu
camelia rakudo f2b0f8: OUTPUT«True␤»
jnthn r: sub decont(Mu $var) { $var }; my $x = Mu; say decont($x) =:= Mu
camelia rakudo f2b0f8: OUTPUT«False␤»
17:58 domidumont left
jnthn Um. That's surprising... 17:58
TimToady me2
jnthn r: sub decont(Mu $var) { $var }; my $x = Mu; say decont($x).VAR
camelia rakudo f2b0f8: OUTPUT«Mu␤» 17:59
jnthn r: sub decont(Mu $var) { $var }; my $x = Mu; say nqp::iscont(decont($x))
camelia rakudo f2b0f8: OUTPUT«1␤»
jnthn I wonder if this is something to do with the "don't flatten" thing...
17:59 skids joined
TimToady then why doesn't \var work? 17:59
r: sub decont(Mu \var) { var }; my $x = Mu; say nqp::iscont(decont($x)) 18:00
camelia rakudo f2b0f8: OUTPUT«1␤»
jnthn r: sub decont(Mu \var) { say nqp::iscont($var); $var }; my $x = Mu; say nqp::iscont(decont($x))
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤Variable '$var' is not declared␤at /tmp/9X3e9QuPxR:1␤------> b decont(Mu \var) { say nqp::iscont($var⏏); $var }; my $x = Mu; say nqp::iscont(d␤ expecting any of:␤ postfix␤»
jnthn r: sub decont(Mu \var) { say nqp::iscont(var); var }; my $x = Mu; say nqp::iscont(decont($x))
camelia rakudo f2b0f8: OUTPUT«1␤1␤»
jnthn \var doesn't mean decontainerize, it means "take whatever was passed" 18:01
TimToady but returning var should decont?
r: sub decont(Mu \var) { return var }; my $x = Mu; say nqp::iscont(decont($x))
camelia rakudo f2b0f8: OUTPUT«1␤»
TimToady r: sub decont(Mu \var) { |var }; my $x = Mu; say nqp::iscont(decont($x)) 18:02
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤CHECK FAILED:␤Undefined routine '&prefix:<|>' called (lines 1, 1)␤»
jnthn TimToady: I think it reconts.
TimToady: To not lose the scalar nature of it, but to make sure it's ro, it re-wraps it in a read-only container. 18:03
TimToady r: my &decont := &nqp::decont; my $x = Mu; say nqp::iscont(decont($x)) 18:04
camelia rakudo f2b0f8: OUTPUT«Type check failed in binding; expected 'Callable' but got 'Any'␤ in block at /tmp/In723T7kNf:1␤␤»
TimToady r: my &decont = &nqp::decont; my $x = Mu; say nqp::iscont(decont($x))
camelia rakudo f2b0f8: OUTPUT«Type check failed in assignment to '&decont'; expected 'Callable' but got 'Any'␤ in block at /tmp/NLN214_Q6X:1␤␤»
TimToady r: my \decont = &nqp::decont; my $x = Mu; say nqp::iscont(decont($x)) 18:05
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤CHECK FAILED:␤Undefined routine '&decont' called (lines 1, 1)␤»
FROGGS timotimo: an if is allowed
TimToady r: my \decont = &nqp::decont; my $x = Mu; say nqp::iscont(decont.($x))
camelia rakudo f2b0f8: OUTPUT«No such method 'Any' for invocant of type 'Mu'␤ in block at /tmp/RXIMdwFzt_:1␤␤»
TimToady r: my $decont = &nqp::decont; my $x = Mu; say nqp::iscont($decont.($x))
camelia rakudo f2b0f8: OUTPUT«No such method 'Any' for invocant of type 'Mu'␤ in block at /tmp/TGWKSyNgqq:1␤␤»
TimToady
.oO(...)
 18:06
r: say &nqp::decont ~~ Callable 18:07
camelia rakudo f2b0f8: OUTPUT«False␤»
timotimo oh, of course it is. good catch.
but is the basic approach sensible?
FROGGS timotimo: same for when
jnthn TimToady: nqp:: things a macro-ish, not sub-ish
timotimo hmm
FROGGS timotimo: but yes, I think it is useful, for say, routine_decl after block
timotimo so you think i should parse a full thingie after a block and if it succeeds i should explode? 18:08
TimToady jnthn: but then there's the paragraph at S06:3149 18:09
dalek rl6-roast-data: b3c45da | coke++ | / (5 files):
today (automated commit)
[Coke] # 07/02/2013 9c64dd0 rakudo++ (26088); niecza (81.73%); pugs (36.42%); rakudo.jvm (92.34%) 18:10
FROGGS timotimo: yes because I would think that these are the usual mistakes
yay
\o/
colomon \o/ 18:11
[Coke] masak: (BB) glad you liked it!
r: say 26088*.95-24092 18:12
camelia rakudo f2b0f8: OUTPUT«691.6␤»
masak labster: re github.com/perl6/specs/commit/71e9...382a2bd92d -- the descriptions of floor and ceiling don't make sense. the description of round partially doesn't make sense. 18:13
labster: if you just say "the highest integer", it sounds like you mean "real integer".
labster: and complex numbers aren't ordered.
timotimo big ups, everyone \o/ 18:14
jnthn Wow, that's some way past 90%! 18:19
colomon is thinking he is going to install OS X 10.8.X as soon as his current Time Machine backup is finished.... JVM 1.7 here we come! 18:21
jnthn TimToady: The -ish was quite a big -ish ;)
FROGGS jnthn: about that stack overflow from rshift.t, on my machine 0xx3700 blows up while 0xx3500 does not
jnthn: is that a limitation of my machine?
jnthn FROGGS: Wait, is there actually some recursion expected to be going on here? 18:22
jnthn thought it was a test for bit-shifting...
FROGGS jnthn: dunno, does the X+> a recursion? 18:23
yes, but it is just the size of the list
even the hardcoded list blows up
(6010 elements)
jnthn X is implemented iteratively, iirc
FROGGS that is why I tested 0 xx 3700
to find out what the max list size is 18:24
and it is somewhere arround 3500 .. 3700 here
wonder if that is about memory limits or something VM-ish
jnthn yeah, $i looks iterative to me 18:25
oops
X looks iterative tome
the meta-op form 18:26
jnthn wonders if there's something weird going on with the gather/take
notjack is it possible to implement a function composition operator such that it takes two functions as input (foo, bar) and returns a function which, when called, applies foo to the result of bar, then the inverse-of-bar (if it exists) to the result of foo? 18:27
FROGGS if RANGE X+> RANGE and 0 xx NUMBER show the same limit, does this still point to gather/take?
moritz notjack: yes, if you can find a way to find the inverse of bar
jnthn Well, I'm only wondering about gather/take 'cus X meta-op is implemented in terms of it 18:28
notjack I was under the impression that functions can define their own inverses
18:28 lizmat joined
notjack like they can declare that they're associative, etc, I thought they could also declare themselves invertible 18:28
moritz not out of the box
notjack (and provide the inverse definition)
moritz Perl 6 is not a computer algebra system 18:29
notjack hmm, ok, my misapprehension
moritz (though you could certainly build on top of it, and I've been tempted before :-)
*build one
notjack I didn' tmean P6 should "figure out" the inverse, only allow subs to declare themselves invertible, and provide and inverse
moritz you can certainy do that with traits
notjack like sub exp is invertible {...} inverse is log {...}
[Coke] gist.github.com/coke/5879701 updated.
timotimo std: given "1" { when Str { say "yup" } when Int { say "oh what?" } } 18:30
camelia std 102c740: OUTPUT«===SORRY!===␤Strange text after block (missing comma, semicolon, comment marker?) at /tmp/okQWqFuRj0 line 1:␤------> given "1" { when Str { say "yup" }⏏ when Int { say "oh what?" } }␤ expecting horizontal whitespace␤Parse faile…
TimToady we never did invertable, but we did at one point consider 'is commutative'
timotimo yeah, i can port that possibly 18:31
but not right now
notjack yeah, commutitivity would be helpful for optimizing reductions
if you knew + was commutitive then you could optimize the operation [+] 18:32
moritz r: multi sub trait_mod:<is>(&r, :$inverse!) { &r does role { method inverse() { $inverse } } }; sub f($x) is inverse({$_ - 2}) { $x + 2 }
camelia rakudo f2b0f8: ( no output )
moritz r: multi sub trait_mod:<is>(&r, :$inverse!) { &r does role { method inverse() { $inverse } } }; sub f($x) is inverse({$_ - 2}) { $x + 2 }; say &f.inverse.(f(42))
camelia rakudo f2b0f8: OUTPUT«42␤»
moritz there you go
notjack here's a test case: define an invertible function log, whose inverse is exp, then an composition operator "under" and say mult = sum under log such that mult (3,4) is the same as 3*4 18:33
oh, that's cool 18:34
TimToady precision suffers...
notjack traits are neat
TimToady unless you get to real symbolic algebra
[Coke] ah. someone added a reference to nqp::chdir() in the jvm, but never actually implemented it.
notjack TimToady: the goal is more code clarity in certain contexts
TimToady but this is partly why we are trying to keep the typespace outside of Any so that we can do conceptual types 18:35
notjack where it's easier to think of transforming the problem to a new domain, expressing a calculation that's simple (in the new domain), and projecting back to the original domain
moritz like schwartz' transform
TimToady sure, I think of that as a kind of strength reduction optimization
notjack think of FFT, e.g.
right
then of course you have to start thinking of identity functions, left- vs right-inverses, etc 18:37
but even a coarse implementation is a boon for the user
raiph .ask sorear is niecza's gather/take lazy?
yoleaux raiph: I'll pass your message to sorear.
TimToady raiph: yes, it is
raiph TimToady: thx
moritz n: my @a := gather { for ^5 { take $_; .say } }; my $ = @a[2] 18:38
camelia niecza v24-86-g39ab531: OUTPUT«0␤1␤»
TimToady ooh, In-n-Out magically happened, yum &
[Coke] ugh. chdir on the jvm is not straightforward.
FROGGS jnthn: if the things I posted are connected, it is more about memory limits or so rather that a fixed maximum list size... 0..54 X+ 0..54 works while 0..55 X+ 0..55 blows up 18:39
(according to my 6guts :P)
FROGGS .oO( a straightbackward chbackdir ) 18:40
18:41 ilbot_nextgen joined
notjack another sugar-y question 18:41
moritz or maybe a curly chdir?
notjack is it possible to define a function that, given a seq of function names, produces a function which is a chain of those subs? 18:42
moritz yes.
notjack like chain(qw<sqroot sum square>) ?
(sorry, I'm still speaking pidgin perl5) 18:43
FROGGS sub chainy( *@subs ) { ::($_) for @subs }
sub chainy( *@subs ) { ::($_)() for @subs } # invoke them
moritz not quite
arnsholt jnthn: On JVM, should a non-serializable type throw an exception in deserialize_stub, or is something else preferred? 18:44
moritz well, if you do it that way, you have to pass the names with &
18:44 Rotwang joined
moritz so chainy <&sqrt &sin> or so 18:44
jnthn arnsholt: There is fine
FROGGS r: sub test { say 42 }; ::('test')()
camelia rakudo f2b0f8: OUTPUT«No such symbol 'test'␤ in method <anon> at src/gen/CORE.setting:10482␤ in any at src/gen/Metamodel.nqp:2580␤ in any find_method_fallback at src/gen/Metamodel.nqp:2568␤ in any find_method at src/gen/Metamodel.nqp:939␤ in block at /tmp/awrTvwhJoM:1␤␤»…
FROGGS r: sub test { say 42 }; ::('&test')()
camelia rakudo f2b0f8: OUTPUT«42␤»
FROGGS moritz++
arnsholt jnthn: Throw, you mean? 18:45
jnthn arnsholt: It's fine to do it there; maybe it's better to use ExceptionHandling.dieInternal(...) than a plain throw
timotimo bleeeeh, contentEditable="true" is dumb :(
FROGGS r: sub chainy( :$args, *@subs ) { ::("&$_")($args) for @subs }; chainy( qw<say print>, args => 42 ) 18:46
camelia rakudo f2b0f8: OUTPUT«42␤42»
notjack r: sqrt(15);
camelia rakudo f2b0f8: ( no output )
notjack r: say sqrt(15);
camelia rakudo f2b0f8: OUTPUT«3.87298334620742␤»
arnsholt Right. dieInternal sounds promising
notjack r: sub sum {[+] @_;} say sum(1,2,3); 18:47
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/iDF1MkkiMC:1␤------> sub sum {[+] @_;} ⏏say sum(1,2,3);␤ expecting any of:␤ postfix␤ statement end␤ statement modifier␤ statement modifier loop␤ horiz…
moritz notjack: the ; should go after the }
notjack r: sub sum {[+] @_}; say sum(1,2,3); 18:48
camelia rakudo f2b0f8: OUTPUT«6␤»
notjack r: sub sum {[+] @_}; sub prod {[*] @_}; sub chainy( *@subs ) { ::($_)() for @subs }; say chainy(&sqrt &sum &prod)(3,4); 18:49
camelia rakudo f2b0f8: OUTPUT«Cannot call 'Numeric'; none of these signatures match:␤:(Mu:U \v: Mu *%_)␤ in method Numeric at src/gen/CORE.setting:885␤ in sub infix:<+> at src/gen/CORE.setting:3263␤ in sub infix:<+> at src/gen/CORE.setting:3262␤ in sub at src/gen/CORE.setting:14408␤ in su…
notjack r: sub sum {[+] @_}; sub prod {[*] @_}; sub chainy( *@subs ) { ::($_)() for @subs }; say (chainy(&sqrt &sum &prod))(3,4); 18:50
camelia rakudo f2b0f8: OUTPUT«Cannot call 'Numeric'; none of these signatures match:␤:(Mu:U \v: Mu *%_)␤ in method Numeric at src/gen/CORE.setting:885␤ in sub infix:<+> at src/gen/CORE.setting:3263␤ in sub infix:<+> at src/gen/CORE.setting:3262␤ in sub at src/gen/CORE.setting:14408␤ in su…
notjack oh, it's backwards! 18:51
I defined prod wrong
18:52 arnsholt_ joined 18:53 xilo joined 18:55 arnsholt left 18:59 dmol left 19:00 cc joined 19:02 cc left
dalek ast: 1bafd26 | (Elizabeth Mattijsen)++ | S02-types/nil.t:
Fix nil regression
 19:04
19:07 ggoebel2 left
FROGGS wow, fetching $beer is equal to ten lines of irc 19:16
moritz wants 10 commits 19:17
FROGGS I could push something, but that lowers jakudo's percentage... 19:20
btw moritz, can you confirm that a rshift operation by a negative amount is correct when you see one? 19:21
I really dont know what to expect...
n: say -10 +> -1
camelia niecza v24-86-g39ab531: OUTPUT«-1␤»
FROGGS n: say -10 +> -9999999
camelia niecza v24-86-g39ab531: OUTPUT«-5␤» 19:22
moritz isn't that just a left-shift with the neg?
FROGGS ?
moritz p: say -16 +> -1
camelia pugs: OUTPUT«-32␤»
moritz p: say -16 +> 1
camelia pugs: OUTPUT«-8␤»
moritz that's kinda what I'd expect, intuitively
r: say 16 +> 1
camelia rakudo f2b0f8: OUTPUT«8␤»
FROGGS yeah, that was my first guess
pnr: say -10 +> -1
camelia pugs: OUTPUT«-20␤» 19:23
..rakudo f2b0f8: OUTPUT«-10␤»
..niecza v24-86-g39ab531: OUTPUT«-1␤»
FROGGS pnr: say -10 +< 1
camelia rakudo f2b0f8, niecza v24-86-g39ab531, pugs: OUTPUT«-20␤»
notjack what's the general pattern for applying a unary function foo to each value of an @array?
FROGGS I can implement it that way...
colomon has the 4 GB OS X 10.8 installer downloaded...
FROGGS any objections? anybody?
notjack I tried gather take foo($_) for @array but p6 yelled at me
FROGGS good.
:P
notjack and >>foo<< @array gave me a weird error 19:24
moritz notjack: @array.map: &foo;
FROGGS >>foo<< needs a LHS too, no?
notjack foo is unary
FROGGS brb
moritz notjack: or gather { take foo($_) for @array }
notjack moritz: ah, thanks
I was missing the { } around gather 19:25
moritz >>foo<< would only work with an infix foo
19:25 xilo left
skids FROGGS: beware rshift and a few bitops broken at the BigInt/native int border currently. 19:26
moritz that's what FROGGS is trying to fix, no?
notjack oh, I know why I skipped the {} after gather: if I put them in, then the block can't see @_ inside a sub
dalek ecs: adbf8b0 | (Brent Laabs)++ | S32-setting-library/Numeric.pod:
clarify Complex rounders
moritz just give your sub a proper signature :-)
notjack sub baz { gather { take sqrt $_ for @_ } } # :( doesn't work 19:27
masak useless use of gather.
skids Fix is already in rt.
moritz sub baz(*@v) { @v>>.sqrt }
masak r: sub baz { sqrt $_ for @_ }; say baz 1 4 9
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/18eL0jX7rc:1␤------> sub baz { sqrt $_ for @_ }; say baz 1 ⏏4 9␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement end␤ statem…
masak hrm :)
r: sub baz { sqrt $_ for @_ }; say baz <1 4 9> 19:28
camelia rakudo f2b0f8: OUTPUT«1 2 3␤»
huf wait wait wait, for really *is* map?
19:28 colomon left
notjack hmm, I imagined that would only return the last value 19:28
moritz huf: yes :(
anyway, map or >> is fine
notjack hence gather
huf i should read some docs eventually, instead of asking basic questions one by one... :(
moritz if you want to do it with gather, you can day
masak huf: I don't mind it so much for statement_mod 'for'.
huf: because that's how Perl 6 does list comprehensions. 19:29
moritz r: sub baz(*@v) { gather { take sqrt($_) for @v } }; say baz 1, 4, 9, 25
camelia rakudo f2b0f8: OUTPUT«1 2 3 5␤»
moritz but yes, gather is overkill
skids RT #115958 and...
huf masak: i dont think i mind, since in p5 map made "sense" in void context (and yet, no sensible person ever used it like that)
segomos moritz: is it a bug that DBIish fails to release a prepared SELECT in a way that i can insert afterwards (Commands out of sync error)? If it is then i'll debug a little more and send a pull request with a fix
huf the way a thing works is no license to use it like that :D
segomos i can't insert^ (if i don't request rows from the SELECT) 19:30
skids ... #115966
moritz segomos: yes, it's a bug
labster masak: hopefully my clarifications on the rounders spec is better. I suspect that was the result of me trying to spec something too late at night. 19:31
notjack I'm still having trouble with the "chained functions" thing 19:43
r: sub sqar {@_ »*« @_}; sub sum { [+] @_}; sub sqrtt {@_>>.sqrt}; sub chainy( :$args, *@subs ) { ::("&$_")($args) for reverse @subs }; say chainy(qw<sqrtt sum sqar>, args => (3,4) )
camelia rakudo f2b0f8: OUTPUT«4 2 1.4142135623731␤»
notjack what I want is "5" 19:44
(sqrt of sum of squares) (3,4) #pythogorean triple is 3 4 5
dalek p: 88892f7 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/CallCaptureInstance.java:
Make CallCapture REPR clonable.
 19:45
notjack whereas this works:
moritz notjack: $args is a scalar, yet you try to treat it as a list
notjack I stole "chainy" from FROGGS 19:46
it should be passed as an opaque argument list to the first funcion in the chain
moritz ... indendently of who wrote the function :-)
notjack maybe that's |*args?
moritz I don't think you can have an argument that is both named and without context 19:47
jnthn std: sub foo(:\bar) { }
camelia std 102c740: OUTPUT«ok 00:00 43m␤»
jnthn r: sub foo(:\bar) { }
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤Malformed parameter␤at /tmp/IYwqZYTlhh:1␤------> sub foo(⏏:\bar) { }␤ expecting any of:␤ statement list␤ prefix or term␤ prefix or meta-prefix␤ new name to be defined␤»…
jnthn Hm. :)
moritz not in rakudo, at least :-)
notjack this works:
r: sub sqar {@_ »*« @_}; sub sum { [+] @_}; sub sqrtt {@_>>.sqrt}; say sqrtt(sum(sqar(3,4)));
camelia rakudo f2b0f8: OUTPUT«5␤»
19:47 rindolf left
notjack what I want is a chainy such that chainy(qw<sqrtt sum sqar), ...whatever...) is sqrtt(sum(squar(...whatever...))); 19:48
or, ideally,chainy simply constructs and returns the function (sqrtt of sum of sqar) such that I can assign it to a name and use it later 19:49
moritz then do that :-)
notjack ok, how? :)
moritz do you know how currying works? 19:50
notjack not in p6
jnthn walk, bbiab
moritz you can do it just like in perl 5
return an anonymous function
notjack but how do I get chainy to compose the function? 19:51
given that it returns a (novel, anonymous) function derived from the functions named in its arguments?
and the argument list can name any number of functions, from 1 to N 19:52
moritz do you know how to do it with just one function?
notjack In p5 I'd evoke it with &
moritz or let me rephrase 19:53
do you know how to compose/chain two functions?
notjack only by stringing them out
moritz r: sub compose(&a, &b) { return sub (\args) { b(a(args)) } } 19:54
camelia rakudo f2b0f8: ( no output )
moritz r: sub compose(&a, &b) { return sub (\args) { b(a(args)) } }; my &f = compose(&sqrt, &cos); f(0)
camelia rakudo f2b0f8: ( no output )
moritz r: sub compose(&a, &b) { return sub (\args) { b(a(args)) } }; my &f = compose(&sqrt, &cos); say f(0)
camelia rakudo f2b0f8: OUTPUT«1␤»
moritz now you just have to do it in a loop, and with the lookup by name
notjack given that b(a(...)) is syntax, it's not clear to me hwo to make that a loop 19:55
19:55 btyler joined
notjack I can see that's amenable to recursion 19:55
moritz recursion and loops are just the same, really 19:56
notjack I mean, an easy way is to literally compose a string and execute it 19:57
> r: sub compose(&a, &b) { return sub (\args) { b(a(args)) } }; my &f = [compose] (&sqrt, &cos, &cos); say f(0); 19:59
FROGGS skids: yes, I'm working on that... your patch btw breaks compiling. but I think #i can supply a fix till tomorrow for rshift
notjack r: sub compose(&a, &b) { return sub (\args) { b(a(args)) } }; my &f = [compose] (&sqrt, &cos, &cos); say f(0);
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/vMHWDpmbBk:1␤------> gs) { b(a(args)) } }; my &f = [compose] ⏏(&sqrt, &cos, &cos); say f(0);␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ …
dalek Iish: 1d48c17 | segomos++ | lib (2 files):
Bug work-around where a prepared SELECT w/o .fetch causes the result set to not be released and subsequent prepared statements fail
notjack to me, that indicates there's a difference between loops and recursion in this context 20:00
because functions aren't quite values in the same way that (for example) numbers are, in p6 20:01
moritz well, syntactic differences, yes
but what you can do with recursion, you can also do with loops
or vice versa
notjack yes, in general, recursion can be implemented with a loop and a stack
moritz r: sub compose(&a, &b) { return sub (\args) { b(a(args)) } }; my &composed = reduce &compose, &sqrt, &cos, &cos; say composed(0) 20:02
camelia rakudo f2b0f8: OUTPUT«0.54030230586814␤»
notjack but I thought the topic of discussion was exactly those syntactic differences
segomos moritz: thanks dude
TimToady .tell lizmat The biggest practical difference between Mu and Nil is that Mu.ACCEPTS($x) is always true, and Nil.ACCEPTS($x) is always false
yoleaux TimToady: I'll pass your message to lizmat.
moritz segomos: thank you for your fix
notjack ok, i'm going to have to study reduce 20:03
I thought [] should suffice for reduction
s/should/would/
moritz [] is for infixes
notjack ah...
thank you.
moritz nr: sub a($a, $b) { $a + $b }; say [&a] 1, 2, 3 20:04
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Two terms in a row (preceding is not a valid reduce operator) at /tmp/N0OLrjfRwq line 1:␤------> sub a($a, $b) { $a + $b }; say [&a] ⏏1, 2, 3␤␤Parse failed␤␤»
..rakudo f2b0f8: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/5G6u4mz3yM:1␤------> sub a($a, $b) { $a + $b }; say [&a] ⏏1, 2, 3␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement end␤ stat…
TimToady try [[&a]]
moritz nr: sub a($a, $b) { $a + $b }; say [[&a]] 1, 2, 3
camelia niecza v24-86-g39ab531: OUTPUT«6␤»
..rakudo f2b0f8: OUTPUT«===SORRY!===␤Unable to parse expression in bracketed infix; couldn't find final ']'␤at /tmp/MXHCOrJWEd:1␤------> sub a($a, $b) { $a + $b }; say [[&⏏a]] 1, 2, 3␤»
moritz NYI in rakudo
[Coke] wonders why the spec list was backed up
moritz notjack: [] is basically syntactic sugar for &reduce, plus it pays attention to associativity
Ayiko r: sub infix:<compose>(&a, &b) { return sub (\args) { b(a(args)) } }; my &f = [compose] (&sqrt, &cos, &cos); say f(0); 20:06
camelia rakudo f2b0f8: OUTPUT«0.54030230586814␤»
TimToady none of those parens are necessary :) 20:07
except around the signature
r: sub infix:<compose>(&a, &b) { return sub (\args) { b a args } }; my &f = [compose] &sqrt, &cos, &cos; say f 0
camelia rakudo f2b0f8: OUTPUT«0.54030230586814␤»
moritz Ayiko++ 20:08
TimToady lessee, we can do better
r: sub infix:<compose>(&a, &b) { return -> \args { b a args } }; my &f = [compose] &sqrt, &cos, &cos; say f 0 20:09
camelia rakudo f2b0f8: OUTPUT«0.54030230586814␤»
TimToady r: my &infix:<compose> = -> &a, &b { return -> \args { b a args } }; my &f = [compose] &sqrt, &cos, &cos; say f 0
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤Undeclared routine:␤ compose used at line 1␤␤»
20:09 ilbot_nextgen left
TimToady n: my &infix:<compose> = -> &a, &b { return -> \args { b a args } }; my &f = [compose] &sqrt, &cos, &cos; say f 0 20:10
camelia niecza v24-86-g39ab531: OUTPUT«1␤»
TimToady hmm
oh, return...
r: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; my &f = [compose] &sqrt, &cos, &cos; say f 0
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤Undeclared routine:␤ compose used at line 1␤␤»
TimToady still seems like a bug
r: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; my &f = &sqrt compose &cos compose &cos; say f 0 20:11
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/EuhikrJ5oB:1␤------> -> \args { b a args } }; my &f = &sqrt ⏏compose &cos compose &cos; say f 0␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ …
TimToady my &infix<> isn't working right
r: constant infix:<compose> = -> &a, &b { -> \args { b a args } }; my &f = &sqrt compose &cos compose &cos; say f 0 20:12
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤invoke() not implemented in class 'QAST::Want'␤at /tmp/hwpxxON05L:1␤------> ␤»
TimToady n: constant infix:<compose> = -> &a, &b { -> \args { b a args } }; my &f = &sqrt compose &cos compose &cos; say f 0
camelia niecza v24-86-g39ab531: OUTPUT«Unhandled exception: System.NullReferenceException: Object reference not set to an instance of an object␤ at Niecza.StashCursor.Core (System.String key, Boolean final, Niecza.StashCursor& sc, Niecza.Variable& v, Niecza.Variable bind_to) [0x00000] in <file…
TimToady cool!
n: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; my &f = &sqrt compose &cos compose &cos; say f 0 20:13
camelia niecza v24-86-g39ab531: OUTPUT«0.54030230586813977␤»
TimToady n: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; my &f = [compose] &sqrt, &cos, &cos; say f 0 20:14
camelia niecza v24-86-g39ab531: OUTPUT«0.54030230586813977␤»
notjack cool!
TimToady no parens at all :)
notjack can you skip the assignment to &f and call the output of [compose] anonymously, withoug using parens?
TimToady um, not that I know of 20:15
wait, maybe
moritz with reducde
*reduce 20:16
erm, never mind
notjack this is cool
n: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; my &f =
[compose] &sqrt, &cos, &cos; say f 0
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Malformed initializer at /tmp/I4FW_obhiC line 1:␤------> &a, &b { -> \args { b a args } }; my &f ⏏=␤␤Parse failed␤␤»
20:17 colomon_chrome joined
notjack n: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; my &f = [compose] &sqrt, &cos, &cos; say f 0 20:17
camelia niecza v24-86-g39ab531: OUTPUT«0.54030230586813977␤»
notjack r: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; my &f = [compose] &sqrt, &cos, &cos; say f 0
camelia rakudo f2b0f8: OUTPUT«===SORRY!===␤Undeclared routine:␤ compose used at line 1␤␤»
colomon_chrome o/
went looking for LHF on JVM. Discovered indent.t's fatal error 80% through was this line: - warn sprintf('Asked to remove %d spaces, ' ~ - 'but the shortest indent is %d spaces', - $outdent, $common-prefix) if $outdent > $common-prefix; 20:18
notjack n: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; sub chain { [compose] @_; } my &f = chain &cos, &sqrt, &cos; say f(0);
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Strange text after block (missing comma, semicolon, comment marker?) at /tmp/I6XwEu5p6j line 1:␤------> a args } }; sub chain { [compose] @_; }⏏ my &f = chain &cos, &sqrt, &cos; say f(␤␤Parse failed␤␤»…
colomon_chrome is there any real reason to use sprintf instead of normal p6 variable interpolation there?
TimToady n: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; 0.$&({ [compose] &sqrt, &cos, &cos }) # almost 20:19
camelia niecza v24-86-g39ab531: ( no output )
[Coke] colomon_chrome: not based on that example, no.
TimToady n: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; say 0.$&({ [compose] &sqrt, &cos, &cos }) # almost
camelia niecza v24-86-g39ab531: OUTPUT«{ ... }␤»
[Coke] seems reasonable to not drag in stuff not directly related to the test. 20:20
TimToady n: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; say 0.&({ [compose] &sqrt, &cos, &cos }) # almost
camelia niecza v24-86-g39ab531: OUTPUT«{ ... }␤»
colomon_chrome [Coke]: that code is in the indent method
TimToady hmm, oughta work
notjack n: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; my &f = [compose] qw<&sqrt &cos &cos>; say f 0
camelia niecza v24-86-g39ab531: OUTPUT«Unhandled exception: Nominal type check failed in binding '&a' in 'ANON'; got Str, needed Callable␤ at /tmp/xjDVXvkq6W line 0 (ANON @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2773 (reduceop @ 59) ␤ at /tmp/xjDVXvkq6W line 1 (mainline @ 4) ␤ at…
20:21 lizmat left
notjack n: my &infix:<compose> = -> &a, &b { -> \args { b a args } }; my &f = [compose] qw<sqrt cos cos>; say f 0 20:21
[Coke] colomon_chrome: oh. yah, if there's no formatter with the %d, I'd say just use interpolation.
camelia niecza v24-86-g39ab531: OUTPUT«Unhandled exception: Nominal type check failed in binding '&a' in 'ANON'; got Str, needed Callable␤ at /tmp/Ri_hXArZM1 line 0 (ANON @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2773 (reduceop @ 59) ␤ at /tmp/Ri_hXArZM1 line 1 (mainline @ 4) ␤ at…
colomon_chrome [Coke]: so the next thing is, of course I tried it myself before coming here. And here's the crazy thing: it actually breaks the build 20:23
TimToady n: say <sqrt cos cos>».map: { &::($_) }
camelia niecza v24-86-g39ab531: OUTPUT«sub sqrt(Any $x) { ... } sub cos(Any $x) { ... } sub cos(Any $x) { ... }␤»
TimToady notjack: ^^
::() is symbolic lookup syntax
colomon_chrome ooooo, my normal computer is booting up in 10.8.4!
colomon_chrome crosses his fingers.... 20:24
dalek Iish: 0429089 | segomos++ | lib34 (2 files):
Correcting a typo I made in the last bug workaround
20:27 snoopy joined 20:30 colomon joined 20:31 colomon left, colomon joined
colomon o/ 20:32
javap -version
1.7.0_04
20:32 vk joined
jnthn uh-oh :) 20:33
I hope that isn't too early to have the indy fixes...
FROGGS jnthn: do you have any objections that -8 +> -1 acts like -8 +< 1 ?
20:34 colomon_chrome left
colomon jnthn: I'm pretty sure the latest JVM I was able to find online would have been very happy to install on OS X 10.8.4 20:35
jnthn FROGGS: On JVM they come out the same...I dunno if that identity holds up generally? 20:36
skids In C negative rshift is lshift. However newer C compilers will warn if they see this done with literals. 20:38
20:38 konundra left
jnthn k 20:39
colomon fixes the scrolling on his machine
FROGGS k, that is all I wanted to know :o)
thank you all
20:40 fgomez left
colomon BTW, should have mentioned this earlier: JSON::RPC, LWP::Simple, and Sitemap::XML::Parser started failing their tests overnight 20:40
FROGGS colomon: this might be fixed already 20:41
ohh, hmm, nvm
masak labster++ # Complex rounding spec fixes
colomon FROGGS: what were you thinking? I can always trying running the smoke test again...
FROGGS colomon: I was thinking about a bad commit from yesterday, which got fixed by jnthn++ this morning (eu timezone) 20:42
but this bad commit broke building of rakudo
20:42 pmurias joined
FROGGS so, it cant affect just a few modules :o) 20:42
skids Correcting myself, actually it looks like that behavior is compiler-specific. gcc lshifts, but ISO 9899:1999 6.5.7 Bit-wise shift operators §3 says the behavior is undefined. 20:43
FROGGS well, undef is okay too
so we can do something, and spec it
skids I think it could come in handy. 20:44
FROGGS the best thing to do is what the user expects
[Coke] github.com/blog/1547-release-your-software
FROGGS github++ 20:45
dalek kudo/nom: 86975e1 | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Binder.java:
Finish post-constraints trnaslation.
 20:51
kudo/nom: 0ab12f3 | jnthn++ | src/Perl6/Metamodel/BOOTSTRAP.nqp:
Fix over-sharing bug in bind-check multi dispatch
kudo/nom: a026db1 | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Ops.java:
Remove debugging code.
kudo/nom: c96b216 | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Binder.java:
Generic instantiation in signature binding.

Handles constructs like (::T, T).
20:52 lue left
nwc10 jnthn: if there's a prize for figuring out what that bundle of awesome actually enables, I'm not going to win it :-( 20:52
[Coke] jnthn: let me know if I should rerun the aborted files gist tonight. 20:53
timotimo thank you for finishing the trnaslation. that really messed me up!
:þ
[Coke] ah. Generic type parameter binding NYI! that will fix something, I'm sure of it. :)
jnthn nwc10: Most notably, unbusts some parametric roles stuff.
[Coke]: Yes, there was a test failing wiht that error that now passes fully also. 20:54
FROGGS (fully also)++
20:55 vk left
dalek p: ec5448d | (Tobias Leich)++ | src/vm/parrot/ops/nqp_bigint.ops:
special-case negative values in rshift

  ... and treat rshifting by an negative amount like an lshift by
positive amount.		 20:58
colomon I read c96b216 as "Genetic instantiation in signature binding." -- exciting yet worrying
jnthn :P 20:59
20:59 vk joined
FROGGS nr: say -10 +> 1 21:00
camelia rakudo f2b0f8, niecza v24-86-g39ab531: OUTPUT«-5␤»
colomon jnthn: any clue why using string interpolation instead of sprintf in Str.indent would break the Rakudo (jam) build?
FROGGS nr: say -11 +> 1 # this should be equal in a mo'
camelia niecza v24-86-g39ab531: OUTPUT«-6␤»
..rakudo f2b0f8: OUTPUT«-5␤»
jnthn colomon: no 21:01
FROGGS ohh no, I didnt update nqp_head yet...
colomon jnthn: I've temporarily gotten sidetracked by the issue of getting my MBP back to doing what I need now that I've got OS X 10.8.4 on it.
jnthn :) 21:02
FROGGS is there an nqp of for list repetition like there is nqp::x for strings? 21:05
21:05 lue joined 21:06 skids left, arnsholt_ left 21:07 BenGoldberg joined
jnthn No 21:07
FROGGS k
thx
jnthn will have 3 talks at YAPC::EU o.O 21:08
tadzik woosh
masak nice!
FROGGS and I won't be there ó.ò 21:09
jnthn Aww
And...it's in Kiev \o/
FROGGS yeah, it is a bit like family +^ yapc for me
jnthn *nod* 21:10
nwc10 the hackathon is not yet very popular: act.yapc.eu/ye2013/talk/4904
jnthn marks himself as going
21:10 Rotwang left
jnthn I'm guessing that moritz++ and masak++ might attend it too ;) 21:10
nwc10 they haven't starred it yet
jnthn Well, maybe people haven't really got starrted with doing that yet... :) 21:11
FROGGS hehe, then I never met moritz when we are living in the same country >.<
masak starring it now. 21:12
21:12 spider-mario left
jnthn You're a star! 21:14
masak ★ 21:15
nwc10 is starrtled by how many puns can be made with just one word
masak ★.★ 21:16
nwc10 what's the point of trying with anything else? :-)
masak just doing one word might lead to pun starrvation...
TimToady we're only just starring
21:16 lue left
perigrin is a Perl hipster, already registered to attend YAPC::NA 2014. 21:17
nwc10 masak: neighbody is going to starrve on this channel. Plenty of meat :-) 21:18
TimToady is a Perl hipstar
masak nwc10: are you really going to keep beating that d... oh dang. 21:19
jnthn That joke isn't "stable" yet? :P 21:22
21:23 frdmn left
timotimo starbl? 21:23
masak oh here we go again, another lap around the old familiar course.
TimToady That beat has the Ringo truth.
masak just jockeying around.
21:23 frdmn joined
masak frdmn: your quit message makes it look like you're a Python programmer. 21:24
or a parselmouth.
TimToady maybe just a leaky tire
perigrin or a star hooked up to a geiger counter. 21:25
masak or maybe he's just a little horse right now. 21:26
TimToady thinks we should welcome someone from Mongolia...
21:26 raiph left, vk left 21:27 PacoAir left
masak .oO( I don't think we khan do that... ) 21:28
21:28 lue joined
TimToady you are an incorrigible punstar 21:28
masak no, Gengh is. 21:29
TimToady 'cuz Kub is too laconic 21:30
diakopter *groan* 21:31
masak I think we just ran out of famous Khans. 21:32
no-one knows about Ögedei.
huf batu much punning can be bad for you
frdmn masek: whats my quit message again? :P
masak 23:23 -!- frdmn [~frdmn@irc.frd.mn] has quit [Quit: ssssSSSSssss]
frdmn ah. 21:33
i just like minecraft
TimToady are you really in Mongolia? I hear they like horses there...
labster Ulan is betaar
masak haha
huf mondavi and mongolia?
frdmn i aint. i still like horses
masak Huhot is too hot.
diakopter TimToady: yes, especially starlings 21:34
lue wants to get in on the pun action, but doesn't know where to start. Whatever.
perigrin our huhot closed :/
21:35 raiph joined, raiph left 21:39 vk joined
masak perigrin: oh, I was thinking of the capital of Inner Mongolia -- but it seems it's translitted "Hohhot". 21:39
dalek ast: fa26638 | (Brent Laabs)++ | S32-num/rounders.t:
testing fix for RT #118545, round with scale argument
 21:41
ast: eb460a3 | (Brent Laabs)++ | S (4 files):
Merge branch 'master' of github.com:perl6/roast
huf jesus christ that city is huge.
TimToady it's a space startion 21:42
21:42 raiph joined
masak it's a trap! 21:43
diakopter it was a dark and starmy night, the starving dog typed starkly 21:45
labster can someone give me access to close tickets on RT?
21:45 iSlug joined
labster .oO (it was a dark and swarthy night) 21:46
diakopter TimToady: I thought .mn was Minnesota 21:48
21:48 pmurias left
diakopter stares at Van Gogh's Night 21:50
21:50 ajr_ left
jnthn blug: 6guts.wordpress.com/2013/07/02/raku...ss-update/ 21:51
masak jnthn++
timotimo yay
masak who has access to give labster more RT privs?
diakopter afaik, [Coke]
masak pmichaud?
oh, ok.
ooh, two blog posts for the price of one! :D 21:54
jnthn ;) 21:58
labster jnthn++ for both
TimToady nr: class NewNil is Mu { }; multi infix:<===>(Mu \a, NewNil \b) { a.WHAT =:= NewNil }; say 42 === NewNil; say NewNil === NewNil; my $x = NewNil; say $x === NewNil; 22:02
camelia niecza v24-86-g39ab531: OUTPUT«False␤True␤True␤»
..rakudo c96b21: OUTPUT«Nominal type check failed for parameter ''; expected Any but got NewNil instead␤ in sub infix:<===> at src/gen/CORE.setting:2041␤ in block at /tmp/PnvhBvO7YI:1␤␤»
TimToady seems like we might be able to define === for some things outside of Any 22:03
not sure what is bothering rakudo there though...
jnthn Maybe the proto is restrictiver than you want for that. 22:04
TimToady hmm
jnthn proto sub infix:<eqv>(Any $?, Any $?) { * }
22:04 vk left
jnthn So, not a bug in the dispatcher. Phew. :) 22:04
TimToady huh, I don't even see a proto in niecza for ===, so maybe that's how it's getting Mu,Mu 22:06
jnthn I guess the default proto is (|) :) 22:07
oh, but that doesn't work for arity things...
So, no idea... :)
n: say &infix:<===>.arity
camelia niecza v24-86-g39ab531: OUTPUT«2␤»
jnthn n: say &infix:<===>.count
camelia niecza v24-86-g39ab531: OUTPUT«2␤»
jnthn Hm. :)
Must have one somewhere...
TimToady anyway, as long as Mu doesn't have a multi candidate, it should behave the same, triggering junctional logic 22:09
colomon has working gcc again! o/
jnthn Yes, true.
TimToady so Any,Any is probably overspecified
jnthn We can relax it.
TimToady and preventing us from writing === Nil with Nil outside Any
jnthn It just means it'll auto-thread over a failed multi-dispatch rather than a failed proto-dispatch. 22:10
22:10 iSlug left
jnthn (where a proto-dispatch is really just a single dispatch, nothing special) 22:10
22:10 lue left
TimToady well, you have to handle that anyway if you're gonna inline the proto, seems 22:11
jnthn The optimizer considers the proto before considering the candidates, iirc 22:12
r: proto foo($) { * }; multi foo($a) { }; multi foo($a, $b) { }; foo(1, 2)
camelia rakudo c96b21: OUTPUT«===SORRY!===␤CHECK FAILED:␤Calling proto of 'foo' will never work with argument types (Int, Int) (lines 1, 1)␤ Expected: :()␤»
jnthn um, that expected is a bit off :S
But it does say the proto is the problem. :) 22:13
Todo: whine at compile time about multi candidates that could never be reached thanks to the proto. :)
TimToady nr: class NewNil is Mu { }; multi infix:<===>(Mu \a, NewNil \b) { a.WHAT =:= NewNil }; say so 42 === 41 | NewNil | 42; say so NewNil = 41 | NewNil | 42; 22:15
camelia rakudo c96b21: OUTPUT«Nominal type check failed for parameter ''; expected Any but got NewNil instead␤ in sub infix:<===> at src/gen/CORE.setting:2041␤ in block at /tmp/LSxR2m10yb:1␤␤»
..niecza v24-86-g39ab531: OUTPUT«True␤Unhandled exception: Writing to readonly scalar␤ at /tmp/DDyAHFdRax line 1 (mainline @ 10) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4579 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4580 (module-CORE @ 576) ␤ at /home/p6eval/nie…
TimToady hmm
n: class NewNil is Mu { }; multi infix:<===>(Mu \a, NewNil \b) { a.WHAT =:= NewNil }; say so NewNil == 41 | NewNil | 42; 22:16
camelia niecza v24-86-g39ab531: OUTPUT«Potential difficulties:␤ &infix:<===> is declared but not used at /tmp/nGqhQB0JyI line 1:␤------> lass NewNil is Mu { }; multi infix:<===>⏏(Mu \a, NewNil \b) { a.WHAT =:= NewNil }␤␤Unhandled exception: Nominal type check failed for #0 n…
TimToady n: class NewNil is Mu { }; multi infix:<===>(Mu \a, NewNil \b) { a.WHAT =:= NewNil }; say so NewNil === 41 | NewNil | 42;
camelia niecza v24-86-g39ab531: OUTPUT«True␤»
TimToady there we go
22:17 notjack left
TimToady I think we can make === Nil work 22:17
jnthn Same 22:18
TimToady === Mu is still a bit problematic, unless the junction failover can be triggered by Mu,Mu calling nextsame 22:20
well, I suppose the Mu,Mu case could call directly into the autothreader if it wanted to 22:21
raiph .ask sorear is niecza's gather/take batched, or 100% lazy? 22:23
yoleaux raiph: I'll pass your message to sorear.
22:24 lue joined, Timbus left
raiph hi TimToady. Do you know the answer to what I just asked sorear? 22:24
22:26 Timbus joined
TimToady I suspect it's all lazy, but have no proof 22:27
otoh, if it were completely lazy, I'd expect recursive constant definitions to work, and they don't, at least not good enough for hamming numbers 22:29
masak .oO( hamming with faint praise ) 22:30
raiph I see batching as a pragmatic enhancement of P6's laziness capabilities, not a compromise or reflection of some problem in P6's laziness design. Any comments? 22:31
TimToady n: constant @a = 1,2,3, @a.map(*+4); say @a[10]
camelia niecza v24-86-g39ab531: OUTPUT«(Any)␤»
TimToady should be 11, not Any
masak raiph: I sometimes perceive it as a problem.
raiph: mostly when I want the guarantee of laziness.
that is, a batching number of 0.
TimToady well, 1, depending on how you count 22:32
masak yeah.
TimToady n: constant @a = 1,2,3, @a.map(*+4); say @a[^10] 22:33
camelia niecza v24-86-g39ab531: OUTPUT«1 2 3 (Any) (Any) (Any) (Any) (Any) (Any)␤»
TimToady somehow it loses track of @a as a concept, and gets stuck on @a the concrete array we have so far 22:34
masak maybe too little binding, too much assignment?
TimToady the = on constant is binding
or defining, anyway
r: my @a; @a.plan: 1,2,3, @a.map(*+4); say @a[^10] 22:35
camelia rakudo c96b21: OUTPUT«No such method 'plan' for invocant of type 'Array'␤ in block at /tmp/Vt5VeuZknb:1␤␤»
TimToady hmm, I though rakudo had .plan already
masak I thought so too. 22:36
yes, I know the '=' on constant is binding. just thinking maybe Niecza gets it wrong there. 22:37
TimToady anyhoo, not sure that's evidence against strict laziness 22:38
22:38 lizmat joined
TimToady The hard part of knowing when you know comes down to knowing when you don't know. 22:39
lizmat back from Amsterdam.pm meeting 22:40
yoleaux 14:56Z <TimToady> lizmat: If we put Nil outside of Any, we probably need a .is-nill primitive, or some way of training people to use =:= when they try to use ===
16:39Z <TimToady> lizmat: A STD.pm6 commit message of "Same change as in rakudo" is not very informative when I do 'git log' :P
20:02Z <TimToady> lizmat: The biggest practical difference between Mu and Nil is that Mu.ACCEPTS($x) is always true, and Nil.ACCEPTS($x) is always false
lizmat TimToady: good point, seemed like something sensible at the time, clearly lacking in content looking at it now 22:41
TimToady turns out === Nil can probably be made to work anyway 22:42
masak 'night, #perl6 22:44
TimToady I made a prototype NewNil work in niecza, and we probably just need to generalize the proto infix:<===> in rakudo from Any,Any to Mu,Mu to allow definitions of === from outside of Any
lizmat so, Nil in combination with "is default", should this work" sub n { Nil }; my $a is default(fail "No go"); $a=n
I guess === falling out of Junction, would not be a problem ? 22:45
TimToady doesn't need to be in Junction either
we have not defined the ability to pass a thunk or closure as default, so that would fail at compile time currently 22:46
since a trait's arg is evaluated at BEGIN time, basically
not sure what it would mean to supply a closure as a default; when would we call it? 22:47
lizmat but the idea, of putting a Failure object as a default for a scalar, would that make sense? 22:48
TimToady but anyway, yes, a returned Nil is still a Nil, and would still set the default when assigned
lizmat Specifically when receiving a value from a sub, or from a regex
TimToady note though that --> Any would preclude returning Nil if we put it outside of Any 22:49
lizmat true
well
hmmm...
TimToady so --> Int:D becomes a strong guarantee of a real value being returned 22:50
is the latest thinking
lizmat and so BOOM if Nil is returned
22:50 konundra joined
TimToady which means the only way such a routine can signal error is by throwing an exception 22:50
22:50 dmol joined
TimToady and fail would through an exception, presuming Failure is outside of Int 22:50
*throw
lizmat so either it throws an exception itself, or it gets thrown an exception upon return 22:51
TimToady we might end up putting Failure outside of Any, alongside Nil
(this was in the original OKness proposal, btw)
lizmat this is lacking from my memory, can I look that up somewhere?
TimToady as soon as it typechecks against Int, it'd throw 22:52
lizmat indeed
my Int $a is default(42)=Nil
would still set $a to 42 or not? 22:53
raiph masak, TimToady: do you think it likely that rakudo will one day support both strictly lazy (batch=1) and mostly lazy?
TimToady well, it would already be set to 42 by the 'is default', so the = Nil is kinda redundant
lizmat true, I guess the example is contrived 22:54
TimToady my Int $a is default(42); already implies setting to Nil, in a sense
lizmat my Int $a is default(42); $a=23; $a=Nil
TimToady yes, that'd set it to 42
it actually helps to have Nil outside of most types, so you can just check on the type failure of ordinary assignment 22:55
but you and jnthn were discussing that earlier already
22:55 dmol left
lizmat indeed, because currently 22:55
my $a is default(42); $a=23; $a=Nil
would *not* fire up the typecheck 22:56
TimToady because the default constraint is Mu currently
the .of type, I mean
lizmat yup
TimToady there was always some feeling that we should make that Any as well, so that you couldn't accidentally store junctions 22:57
maybe this is an additional argument for that
lizmat r: my Any $a=Nil; say $a # no typecheck, so Nil just gets stored 22:58
camelia rakudo c96b21: OUTPUT«Nil␤»
lizmat r: my Cool $a=Nil; say $a # typecheck, so Nil just gets (wrongly) replaced by default
camelia rakudo c96b21: OUTPUT«Nil␤»
TimToady beause Nil is currently in Any
lizmat r: my Int $a=Nil; say $a # typecheck, so Nil just gets (wrongly) replaced by default
camelia rakudo c96b21: OUTPUT«No such method 'gist' for invocant of type 'NQPMu'␤ in sub say at src/gen/CORE.setting:11365␤ in block at /tmp/QwnYfjAwbt:1␤␤»
TimToady r: class NewNil is Mu {}; my Any $x = NewNil; 22:59
camelia rakudo c96b21: OUTPUT«Nominal type check failed for parameter '$got'; expected Any but got NewNil instead␤ in block at /tmp/WLku4zjaqe:1␤␤»
TimToady it would work more like that
lizmat yup
when I'm awake tomorrow, I'll just do that with Nil and see what happens 23:00
TimToady well, the default default of my Int $a would be Int as well, I presume
lizmat yes, the type Int
TimToady see earlier discussion about how to add an === operator outside of Any
proto infix:<===>(Any,Any) also needs to change to (Mu,Mu) 23:01
lizmat will look at backlog later… pretty tired now… 2 hour drive, not a lot, but enough to stop the brane functions
TimToady understands perfectly :)
lizmat any thoughts on S11 ? 23:02
TimToady not yet
lizmat runs away :-)
23:02 Radvendii joined
lizmat let me know if I should clarify things more 23:02
I guess the basic questions are: 23:03
1. is the compunit the base of use, or is a package?
2. if 1 is compunit, is it a good idea to get meat-info from pos
pod
meta 23:04
I'm tried, I should stop
tired
:-) sleep&
jnthn meat-info! :D 23:05
'night lizmat
labster good work, good night lizmat
flussence github's added binary downloads back, if anyone cares. The only catch is they now have to be associated with a tag 23:08
23:08 skids joined
raiph jnthn: thx for previous answers. my last two: is rakudo's gather/take lazy? and do you think it likely batching will be implemented this year on at least one backend? 23:11
timotimo rakudo's gather/take is totally lazy 23:13
jnthn raiph: It's lazy, don't know about batching. I'm not personally likely to work on it, since gather/take isn't an area I've been much involved with so far, and others know/understand it better. :) 23:14
And I don't like making time estimates on behalf of others :P 23:15
btyler jnthn: I really enjoyed reading the first part of the interview, looking forward to the rest! 23:16
timotimo did y'all see "github releases"? that seems neat 23:17
jnthn btyler: Happy it was interesting :) 23:18
timotimo i agree about the interestingity
raiph jnthn++ # answeringity 23:20
jnthn unconciousity & 23:23
23:31 Radvendii left 23:43 Radvendii joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!