»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 diakopter on 14 April 2013.
BenGoldberg But somehow, I don't think that that form can be made into an infinite lazy list ;) 00:00
BenGoldberg could be wrong, though :) 00:01
00:11 thou left 00:13 avuserow left
raiph en.wikipedia.org/wiki/Thue-Morse_sequence # odious & evil 00:20
kurahaupo BenGoldberg: isn't that sequence simply the second-to-top bit from a greys coding? 00:21
(or inversion thereof)?
BenGoldberg Huh? 00:22
A gray coding involves an order reversal, I think.
00:22 tgt left
kurahaupo grays encoding: an encoding of binary where counting up or down by one only ever flips one bit 00:22
1, 11, 10, 110, 111, 101, 100, 1100 ... 00:23
read the second-to-left bit from each element, and I think it will read out your putative non-lazy sequence
BenGoldberg I don't see it 00:24
kurahaupo will have to investigate more thoroughly, but they certainly look related
BenGoldberg Of course they're related, they're both made from zeros and ones! 00:27
;)
kurahaupo r: @a = 1; @a := @a, @a, map{1-$_} @a for 1..6; say @a; 00:32
camelia rakudo 37c995: OUTPUT«===SORRY!===␤Variable '@a' is not declared␤at /tmp/_gAtpHMLcb:1␤------> @a⏏ = 1; @a := @a, @a, map{1-$_} @a for 1..␤ expecting any of:␤ postfix␤»
kurahaupo r: my @a = 1; @a := @a, @a, map{1-$_} @a for 1..6; say @a;
camelia rakudo 37c995: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/5YjFPCpg87:1␤------> my @a = 1; @a := @a, @a, map{1-$_} ⏏@a for 1..6; say @a;␤ expecting any of:␤ postfix␤ statement end␤ statement modifier␤ statement modif…
kurahaupo r: my @a = 1; @a := @a, @a, map{1-$_}, @a for 1..6; say @a;
camelia rakudo 37c995: OUTPUT«===SORRY!===␤CHECK FAILED:␤Calling 'map' will never work with no arguments (line 1)␤ Expected any of:␤ :(&code, *@values)␤»
kurahaupo (drat, came in half way through that conversation and didn't scroll back far enough to see the previous mention of Grays coding) 00:37
r: my @a=0; @a := @a, map {1-$_}, @a for 1..6; say @a; my @b=0; @b := map {$_,1-$_}, @b for 1..6; say @b; 00:54
camelia rakudo 37c995: OUTPUT«0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0 1 0 0 1 0 1 1 0 0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0 0 1 1 0 1 0 0 1 0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0␤0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0 1 0 0 1 0 1 1 0 0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0 0 1 1 0 1 0 0 1 0 1 1 0 1 0 0 1 1 0 0 1 0 1 1 0␤»
kurahaupo recognizes the sequence: parity on binary counting
In P5 roughly: @a = map { unpack( "%32b32", pack "V", $_ ) % 2 } 0 .. MAXINT 01:01
01:12 kbenson left 01:13 adu joined 01:42 FROGGS_ joined 01:46 FROGGS left 02:10 rhinux_ left 02:16 japhb_ left 02:32 daniel-s_ joined
diakopter agentzh++ (openresty++) www.techempower.com/benchmarks/#section=data-r3 02:32
so, I look at these benchmark numbers 02:33
and I'm forced to imagine rakudo's http server doing this 02:34
02:35 daniel-s left, census left 02:40 lustlife joined
diakopter my first guess is that rakudo/parrot could maybe get 1,000x slower than the fastest ones in every test 02:40
02:45 orafu left
labster that's only 3 orders of magnitude -- and 1 order of magnitude from PHP 02:49
look at the bright side
02:50 orafu joined
diakopter o_O and in base-1000, 1 order of magnitude 02:51
bonsaikitten geeeh, why is that website so retarded ... I can't see the benchmark results 02:53
people really need to remember how http works I guess
diakopter I suppose you could suspect that if it worked for less than nearly everyone.. 02:54
bonsaikitten oooh. oooooooh. 02:55
you need to enable javascript so that the links work
I have no idea what that's supposed to mean, but at least it's futuristic
labster \o/ finally figured out how to use precompiled regexes in the setting! 03:06
well, that's a new one: Error while constructing error object:Could not locate compile-time value for symbol Syntax::Confused Error while compiling, type X::Syntax::Confused at line 8526, near "my $str_no" Still manages to be informative, though. 03:36
It's an error I made in the setting, so don't worry about fixing it. 03:37
03:43 Nom- left, Nom- joined 04:10 kaleem joined 04:18 lustlife left 04:20 kurahaupo left, kurahaupo1 joined 04:23 SmokeMachine left 04:26 Patterner left 04:30 SmokeMachine joined 04:46 Psyche^ joined, Psyche^ is now known as Patterner 04:52 kaare_ joined 05:05 BenGoldberg left, adu left 05:09 SamuraiJack joined 05:13 kaare_ left 05:14 ilbot6 left, _ilbot left 05:15 moritz left 05:26 frdmn left 05:46 unknown_ joined 05:47 unknown_ left 05:48 fgomez left 06:01 FROGGS_ left 06:14 FROGGS_ joined 06:15 frdmn joined, FROGGS_ is now known as FROGGS, fgomez joined 06:54 moritz joined
moritz \o 06:54
japhb o/ moritz 06:55
BTW ... clogs are clogged. :-/
moritz I know
japhb I figured you might.
moritz the server is unreachable :(
japhb Not even you can reach it? Ouch. 06:56
moritz it's a vm, and the hypervisor is ping-able
japhb At least it's not net-silent
moritz but I don't have access to the hypervisor (the VM is shared among several folks) 06:57
06:57 kurahaupo1 left 07:00 domidumont joined
labster o/ 07:14
FROGGS \o 07:15
hoelzro ahoy #perl6!
FROGGS harrrr! hoelzro
tadzik is it talk like a pirate day again, mateys? 07:17
07:17 sqirrel joined
FROGGS dunno if that is today 07:17
hmmm, looks like today is a non-special day... according to the google page 07:18
07:19 hoelzro joined
tadzik heh 07:19
www.wolframalpha.com/input/?i=today
Chernobyl anniversary
Gestapo establishment anniversary
Confederates surrender anniversary
nothing worth the google page I guess :) 07:20
FROGGS ya
07:20 domidumont left
FROGGS for me it is a "I'm sick"-day -.- 07:20
tadzik aw
07:20 domidumont joined
tadzik hugs FROGGS 07:21
FROGGS thanks ó.ò
tadzik for me it's "I have a Networking exam and then I slack off" day
labster hugme: hug FROGGS
hugme hugs FROGGS
tadzik where "slack off" may end up being "hacking on panda"
labster doesn't want to get the germs on myself
FROGGS *g* 07:22
labster Finally! Just got File::Spec Unix, Win32, and Cygwin working in core. 07:27
Naturally every file I added failed for a different reason.
diakopter labster: cool. :D 07:28
labster++
07:40 kresike joined
FROGGS labster++ 07:40
kresike hello all you happy perl6 people
FROGGS hi kresike
labster hello kresike
kresike FROGGS, labster o/ 07:42
FROGGS does someone gets this error msg too? Missing or wrong version of dependency 'src/Perl6/Pod.nqp' 07:45
lizmat didn't jnthn commit something in that area yesterday ? 07:48
FROGGS ohh, good hint, will look at the commits 07:49
hmmm, just for the jvm build though :/ 07:50
07:52 rindolf joined
FROGGS ohh, might be my fault... 07:56
lizmat rn: my $a; my $x = [postcircumfix:<{ }>] $a,<foo bar>; say $a 07:59
camelia rakudo 37c995: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/BX8792UNA1:1␤------> my $a; my $x = [postcircumfix:<{ }>] ⏏$a,<foo bar>; say $a␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement en…
..niecza v24-37-gf9c8fc2: OUTPUT«===SORRY!===␤␤Two terms in a row (preceding is not a valid reduce operator) at /tmp/yA_wcxH2AA line 1:␤------> my $a; my $x = [postcircumfix:<{ }>] ⏏$a,<foo bar>; say $a␤␤Parse failed␤␤»
08:00 kaleem left
lizmat trying to grok S03:4562 08:01
rn &infix:<dehash> ::= postcircumfix:<{ }>; 08:02
rn: &infix:<dehash> ::= postcircumfix:<{ }>;
camelia niecza v24-37-gf9c8fc2: OUTPUT«===SORRY!===␤␤Cannot find definition for binding??? at /tmp/qwZn7jcUCs line 1:␤------> &infix:<dehash> ::= postcircumfix:<{ }>⏏;␤␤Undeclared routine:␤ 'infix:<dehash>' used at line 1␤␤Unhandled exception: Check failed␤␤…
..rakudo 37c995: OUTPUT«===SORRY!===␤Cannot use bind operator with this left-hand side␤at /tmp/lS9Jbd0V9J:1␤------> &infix:<dehash> ::= postcircumfix:<{ }>⏏;␤ expecting any of:␤ argument list␤ postfix␤»
lizmat I guess NYI
FROGGS :/ 08:03
moritz &infix:<dehash> ::= postcircumfix:<{ }>;
I'm pretty sure that must be &postcircumfix:<{ }>; on the right, no?
nwc10 we're not being logged? So I can say impolite things about Perl 3? :-) 08:04
lizmat not sure… I'm cutting/pasting from S03:4562
moritz nwc10: countless monkeys are working on restoring the server
lizmat nwc10: we are being logged and you can say impolite things about Perl 3
moritz fwiw ilogger2 is still active 08:06
(I guess)
though I don't know where it logs to :-) (not my bot)
08:12 berekuk joined
lizmat away until later today& 08:17
08:36 berekuk left
dalek ecs: 2f51c10 | moritz++ | S03-operators.pod:
[S03] fix a small thinko
 08:40
08:49 spider-mario joined 08:54 berekuk joined 09:00 MrMeek joined 09:02 MrMeek-afk left 09:07 daxim joined 09:14 kaleem joined 09:28 bbkr joined
jnthn yawns 09:29
FROGGS g'yawning jnthn 09:31
09:39 raiph left 09:41 sqirrel left
masak good antenoon, #perl6 09:58
moritz \o masak
masak: I'm a wee bit disappointed that you didn't mention the trick of how my t2 solution recognizes nouns, verbs etc. :-) 09:59
lunsj& 10:02
masak moritz: oh!
yes, sounds like there was something I missed pointing out there.
masak looks again 10:03
nwc10 wonders what keyboard layout puts lunsj near to lunch 10:05
jnthn nwc10: I think that's how you actually spell it. 10:06
nwc10 aha.
jnthn nwc10: In Norwegian.
masak .oO( any sufficiently advanced spelling system is indistinguishable from a typo ) 10:09
jnthn Jes. 10:10
tadzik Tak jest.
10:12 lustlife joined
masak :P 10:22
bbkr when registration for PLPW 2013 ends? 10:32
masak bbkr: there doesn't seem to be a deadline for registration. 10:35
bbkr: act.yapc.eu/plpw2013/
I don't see anything on that page.
jnthn I love how on act.yapc.eu/plpw2013/talks masak is the only person undecided whether to give his talk in Polish or English :P
bbkr :) 10:36
masak I've decided to late-bind that decision. 10:37
who knows what will happen between now and then.
(fixed) 10:38
tadzik so there are no Polish talks :) 10:41
how international :)
bbkr: we don't plan a deadline
maybe we'll close it when we order the t-shirts :)
bbkr nice 10:42
tadzik I just sent a new invitation that I expect to be a Great Success 10:43
I was supposed to give a Perl talk on my university, but the class was cancelled that day
so I took a liberty to invite all the students on the mailing list, saying "come, it's fun, you learn a lot and free beer"
there's at least one thing there that'll easily trigger their attention :) 10:44
jnthn There's free beer?!
tadzik there'll be at least some free beer
10:46 tgt joined
sorear moritz: 03:51 <+g0ph3r_> anyone else getting 504 on irclog.perlgeek.de/rosettacode/today ? 11:06
moritz: i get a connection timeout; just referred them here as well
moritz the host is down 11:08
11:50 snarkyboojum joined
masak snarkyboojum! \o/ 11:50
timotimo i'd like to hear your opinion on if (True, False).pick { ... } else { ... } vs ({ ... }, { ... }).pick.() 11:58
masak timotimo: first off, 'if Book.pick' 11:59
you're welcome. :) (I made sure that worked by spec.)
timotimo my preference is the latter, especially since it more naturally expands to more than two branches
even though for that a given ^3).pick { when 0 { .. }; when 1 { ... } ...} may work as well
masak timotimo: second, not knowing the context, the former one feels much easier to read.
timotimo: and a great example of not having too much end-weight.
timotimo not having to manually write numbers for each branch is a plus, too 12:00
masak put all the choices in an array. do @choices.pick().()
timotimo that^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^Hthat's not considered end-heavy, because it's a new line?
masak because it's short :) 12:01
colomon masak: does pick work like Bool.pick on enu,s?
*enums
daxim ololol, literal ^H
masak colomon: yes.
daxim ␡eted! 12:02
masak timotimo: the "end weight" problem is when you first see a big expression and the statement appears to do one thing, but then you see the end and it turns out to do something completely different.
timotimo: '.pick.()' in your second alternative has that problem. 'if Bool.pick' doesn't. 12:03
timotimo i see why that's so. thanks! 12:04
colomon masak, timotimo: then maybe defining an enum with a descriptive name for each branch is the way to go? 12:05
12:05 bluescreen10 joined
cognominal nqp: sub a($a-a) {} 12:05
camelia nqp: OUTPUT«Routine declaration requires a signature at line 2, near "($a-a) {}"␤current instr.: 'panic' pc 14721 (src/stage2/gen/NQPHLL.pir:5232) (src/stage2/gen/NQPHLL.nqp:279)␤»
timotimo would you consider pick(1, {...}, {...}).() to be better?
masak .oO( "end weight" will be one of the topics in the FREE #masakism workshop on Wednesday. sign up! )
cognominal nqp: my $a-a
camelia nqp: ( no output )
timotimo already looking forward to the workshop :)
i really dislike not being ble to tell if the tssh connection is just very laggy/disconnected or if the irc channel's just very slow. can't even tell by the clock in the lower right corner, because it's usually one to four minutes out of sync with my local one >_> 12:10
12:11 arlinius left
masak colomon: yes, descripting enum is a very nice idea. 12:13
descriptive*
12:13 spider-mario left 12:25 tgt left
timotimo agreed. although it offers only a little bit of "added value" over a list of strings (or am i missing something cool?) 12:27
12:28 census joined
jnthn Can use enums as type constraints too 12:28
12:28 xinming left
timotimo masak: also, when you said "I made sure that worked by spec.", do you mean you made sure that it would get put into the spec or that you checked with the specs if that is correct? 12:28
i imagine this use case to be my enum Decision <use-rule ignore-rule>; given decision.pick { when ... { ... } ... } and then forget about the enum; not sure where the type constraint w^Hcomes in 12:29
masak timotimo: the former, IIRC.
12:30 xinming joined, kaleem left
timotimo good thinking :) 12:31
^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^Hmakes me wonder what the neatest way would be to write a decision where each branch has a given "probability" to be chosen near it
[Coke] waves 12:32
timotimo i am aware of the given a number between 0 and 1 { when 0 .. 0.25 { ... }; when 0.25 .. 0.75 { ... }; default { ... } } way already, which seems to be what most people would write 12:33
i dislike that because it expresses the probability as the difference between two numbers and changing probabilities in the middle requires some extra work every time 12:34
anyway, gotta get out of the tram now. will check in later for a few clever answers :3 12:35
pmichaud when * < 0.25 { ... }; when * < 0.75 { ... }; default { ... } 12:36
12:42 domidumont left 12:43 sergot_ is now known as sergot, mattp_ left, skids left 12:45 frdmn|afk joined, frdmn|afk left
[Coke] sergot++ # noted the module list was updated at some point. 12:46
12:46 frdmn left
tadzik I think it updates daily now 12:46
which is very cool
masak timotimo: you might want to check out KeyBag's .pick and .roll, which weight things for you. 12:51
12:55 kaleem joined 13:04 donaldh joined
dalek : 101f503 | (Tobias Leich)++ | rakudo.patch:
this patch contains unapplied patches for rakudo

These are usually WIP, and must be confirmed by Commit Committee and the Bad Ideas Reversal Squad(R) before they can be applied.		 13:05
13:19 shinobicl left
gfldex could the Bad Ideas Reversl Squad have a talk with my government please? 13:25
13:26 fhelmberger joined
timotimo pmichaud: ah, of course given/when is serial. but this just makes the individual probabilities even harder to spot 13:29
especially if the blocks are big-ish
moritz maybe you want something along the lines of 13:30
my @probs = 0.25 => &sub1, 0.3 => &sub2, ...;
my $sum = 0;
for @probs { $sum += .key; if $value <= $sum { .value.(); last } } 13:31
timotimo maybe a macro would be cool for this. 13:33
jnthn r: say Macro ~~ Cool
camelia rakudo 37c995: OUTPUT«False␤»
jnthn Nope. :P
pmichaud use a bag and .pick, then. 13:36
13:36 fhelmberger left
pmichaud or .roll 13:37
moritz or .rick-roll 13:38
timotimo that's nice, i like KeyBag. 13:39
13:39 SmokeMachine left
timotimo that's a thing i should try out more often 13:39
13:40 SmokeMachine joined 13:48 crab2313 joined
donaldh jnthn: hai 13:50
dalek : c4d4cb2 | (Tobias Leich)++ | STATUS.md:
update
donaldh jnthn: I am just catching up with NQP and JVM after much dayjob travel and vacation. 13:51
dalek : ee6aaf8 | (Tobias Leich)++ | Makefile:
adding blib is dangerous when v5 is installed, directory gets changed and Perl 5 scripts get require-d
 13:52
13:52 crab2313 left
donaldh jnthn: I see that JVM has landed in the NQP repo now 13:52
jnthn: what exciting things are on the todo list?
moritz make rakudo run on top of NQP+JVM :-) 13:53
13:53 PacoAir joined 13:54 crab2313 joined, kbenson joined
jnthn donaldh: I got your asm stuff merged and we're using invokedynamic in a couple of places by now. 13:54
donaldh jnthn++ 13:55
jnthn donaldh: Still more to come; I need to get method calls working with invokedynamic. May do that this evening...
donaldh: Mostly though it's working through getting Rakudo to run.
There's still some nqp:: ops missing; a handful of big int related ones, and various IO related ones come to mind. 13:56
Also, I hacked together a simple readline thing but it doesn't quite work yet, so the NQP REPL doesn't function. That should be quite approachable to work on.
donaldh jnthn: okay, I could explore some of that.
jnthn :) 13:57
donaldh I get test failures on nqp/jvm on MacOS
Are some to be expected just now ?
nwc10 jnthn: 5% speedup for that Levenstein benchmark from using invokedynamic 13:58
jnthn Yeah; they're ones that never were in the nqp-jvm-prep.
donaldh So some of them will target the missing ops, etc. ?
jnthn donaldh: I think only one of them (nqpop) fails due to those kinds of reasons. 13:59
FROGGS btw, there are issues with &is-prime on mac+rakudo+parrot too
jnthn donaldh: I think the missing ops are largely things that Rakudo uses and are thus indirectly covered by spectest, but don't have direct NQP tests. 14:00
[Coke] FROGGS: yes, there's finally a ticket for that.
FROGGS ya, just wanted to point on that
[Coke] notes that [WEIRD] is not really a useful diagnostic bug label, he thinks.
r: say 643 - 70 14:01
camelia rakudo 37c995: OUTPUT«573␤»
moritz [Coke]: it helps me ignore some bug reports when I want to work on stuff that I want to be helpful to users immediately 14:02
dalek : f8bd973 | (Tobias Leich)++ | t/v5/ (37 files):
strip 'use perl5;', because there is no such thing
[Coke] moritz: hokay 14:05
dalek kudo/nom: 71ea142 | (Timothy Totten)++ | src/core/Temporal.pm:
DateTime uses Int timezone offsets.

Signed-off-by: Moritz Lenz [email@hidden.address]		 14:06
moritz [Coke]: there, that should make your rakudo spectests clean(er) again
14:08 skids joined
[Coke] masak: rt.perl.org/rt3/Ticket/Display.html?id=73230 - please followup? 14:08
14:09 tgt joined
[Coke] FROGGS: didn't you fix a bug like this recently: rt.perl.org/rt3/Ticket/Display.html?id=73742 14:11
FROGGS r: class JSON::Tiny::Actions 14:14
camelia rakudo 71ea14: OUTPUT«===SORRY!===␤Unable to parse class definition␤at /tmp/MSBdkl1WdH:1␤------> class JSON::Tiny::Actions⏏<EOL>␤ expecting any of:␤ statement list␤ prefix or term␤ prefix or meta-prefix␤ generic role␤»…
FROGGS [Coke]: I'm not aware of fixing it 14:15
r: { use Test }
camelia rakudo 71ea14: ( no output )
FROGGS (that one fails locally using v5 )
14:25 arlinius joined
skids r: for (rand xx 50) -> $p is copy { my sub P ($chance) { $p -= $chance; $p < 0 }; when P(0.25) { "a".print }; when P(0.5) { "b".print }; default { "c".print } } 14:28
camelia rakudo 71ea14: OUTPUT«cabcbabbbbbbabbcaccbcbbabacbcbcbbbbbbbcaaabbbcacbb»
[Coke] github.com/blog/1227-commit-status-api - this could be nifty.
it was "PRE ; " vs. "PRE {}" I think.
14:30 bbkr left, frdmn joined
FROGGS ya, there was a PHASER ; issue once 14:30
because <blorst> was faulty 14:31
skids
.oO(yes I know that's very procedural but I'm not a fan of functionality over functionality... oh wait... crap.)
 14:32
14:34 SamuraiJack left 14:39 domidumont joined 14:42 mattp_ joined
gtodd o/ all 14:43
in perl5 land there are things like PAR and I think other approaches to python eggish deployment (not sure how popular or widely used these are) is there a synopsis or set of discussions on current/future tools for perl6 for cross platform packaging/deployment ? 14:47
tangentstorm we should bring back penguin :) 14:48
14:48 kaleem left
tangentstorm www.hpcf.upr.edu/~humberto/document...pting.html 14:49
actually nothing like python eggs but that reminded me of this for some reason. :)
gtodd heh ok wasn't sure if panda/penguin etc were prototypes for future even better stuff or ... 14:50
tangentstorm panda is the cpan-alike for perl6 as far as i can tell.
penguin is a very old thing for distributed computing and mobile code. 14:51
(it actually does have to package up (running!) code so it can run in different places)
gtodd ah ... perl6 features/approaches seem to cross pollinate with perl5 (e.g. Moose) so was wondering if packaging was doing that too :-)
14:52 mobi_d joined
skids Probably the banging out the module loading filestructure will come before serious efforts. 14:52
gtodd tangentstorm: someone in mojolicious channel mentioned wanting to run perl/mojo on a cell phone so the demand is there :-) 14:53
skids (not to call panda "unserious")
gtodd skids: ok I thought maybe it was a chariot/horses thing ... like a much better cart will be possible when our horses are ready 14:54
tangentstorm gtodd: cool, but not that kind of mobile... mobile like it's running on one machine, then you freeze it and send it over to another machine and it thaws out and continues running there. 14:56
like distributed computing
gtodd right
tangentstorm but i guess there could be a lot of overlap between that and the modern sense of "mobile" :)
gtodd that is cool though 14:57
tangentstorm though reading this again, i'm not sure penguin actually does this :/
gtodd more cool than cloud computing ... sort of more like a swarm/flock 14:58
tangentstorm this paper says it's new (in 1996)... i think the perl journal article was from a little later in the development cycle 15:02
15:03 spider-mario joined
gtodd tangentstorm: is that the sort of thing erlang can do currently? 15:03
tangentstorm not sure. it's been probably 10 years since i read this article. looking for it now. 15:04
15:04 donaldh left
tangentstorm www.foo.be/docs/tpj/issues/vol4_4/t...-0011.html <- weren't people talking about getting perl on the jvm in here recently? :D 15:05
15:07 raiph joined
gtodd erlang's vm errm I mean abstract machine only runs erlang code ... not perl :) 15:08
oh wait you said jvm
yes
tangentstorm found it.. looks like there were two penguin articles in TPJ... both from 1996... i guess it died out after that.
www.foo.be/docs/tpj/issues/vol1_2/t...-0002.html
www.foo.be/docs/tpj/issues/vol1_3/t...-0004.html 15:09
gtodd interesting
15:11 crab2313 left 15:12 arlinius left
kresike bye folks 15:12
15:12 kresike left, raiph left 15:13 mobi_d left
nwc10 tangentstorm: there's a heck of a lot of difference between getting some simple test code working, and having a full-on language 15:13
let alone one that is actually as fast as the "current" implementation 15:14
tangentstorm nwc10: yeah, i know... i just thought it was funny to see that from so long ago. 15:15
jnthn
.oO( And don't I know it... )
tangentstorm blast from the past... in 1998-1999 perl was such a dominant languange... it was every bit as big as java. how times change. :) 15:16
nwc10 perl has fewer CVEs.
Be nice if it were zero 15:17
tangentstorm well there were plenty of buggy and insecure cgi scripts written *in it* though :) 15:18
tpj was really awesome: www.foo.be/docs/tpj/
15:22 raiph joined
gtodd some people say java took over the world ... but it was probably just a pre-beta version of the world ... perl will be there when World 2.0 is ready :) 15:26
sorear don't forget that the market has not remained a constant size 15:27
dalek p: 746aa35 | jnthn++ | src/vm/jvm/ (2 files):
Eliminate arg array assembly in sub call code.

Instead, we use invokedynamic to do it, which reduces the amount of code we generate for sub argument passing. Also should hopefully give better performance, since we're delegating this assembly work to the combinators provided by the JVM, which presumably it knows how to work with efficiently.		 15:28
p: 1d725ae | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/IndyBootstrap.java:
Fix invokedynamic handling of lexotic.
p: 2696e10 | jnthn++ | src/NQP/Actions.nqp:
Make return code-gen something more optimizable.
15:29 FROGGS left 15:34 dmol joined 15:36 daxim left
spider-mario trying to compile nqp for the jvm outputs: 15:40
Unknown compilation target 'classfile'
am I doing it wrong?
(I ran perl ConfigureJVM.pl --prefix=(…); then make) 15:41
jnthn spider-mario: Sounds like something I fixed yesterday, if it comes towards the end of the build
spider-mario oh, I think I know 15:42
I built a parrot nqp in the same directory
I’ll try to clean and do it again
15:43 TimToady_ is now known as TimToady, TimToady left, TimToady joined 15:49 domidumont left, tgt left
spider-mario it worked 15:56
:)
sjn moritz: heya, what's up with the irclog? (Can't seem to reach it :-\) 15:59
masak sjn: <moritz> the host is down 16:00
[Coke]: yes, rt.perl.org/rt3/Ticket/Display.html?id=73230 is closeable now. 16:01
sjn ook
16:05 kshannon left 16:12 kshannon joined 16:20 orevdiabl left, revdiablo joined 16:25 kaleem joined 16:30 FROGGS joined 16:33 sqirrel joined 16:34 kaleem left 16:56 sjohnson joined, sjohnson left, sjohnson joined 16:58 Chillance joined 17:19 SmokeMachine left, SmokeMachine joined 17:22 thou joined 17:31 domidumont joined
cognominal r: 'hi' ~~ / hi { say matched 'hi '} /; 17:54
camelia rakudo 71ea14: OUTPUT«===SORRY!===␤Undeclared routine:␤ matched used at line 1␤␤»
cognominal oops 17:55
r: 'hi' ~~ / hi { say 'matched hi '} /;
camelia rakudo 71ea14: OUTPUT«matched hi ␤»
18:05 geekosaur left, ingy left 18:06 ingy joined 18:07 shachaf joined 18:09 geekosaur joined 18:11 shachaf left 18:14 SmokeMac_ joined 18:15 SmokeMachine left, shachaf joined 18:17 geekosaur left 18:19 colomon left 18:20 shachaf left, colomon joined 18:25 grondilu joined
grondilu rn: package Foo { our $msg; our sub talk { say $msg } }; $Foo::msg = "hello"; Foo::talk; 18:25
camelia rakudo 71ea14: OUTPUT«(Any)␤»
..niecza v24-37-gf9c8fc2: OUTPUT«hello␤»
18:26 geekosaur joined 18:27 domidumont left 18:33 geekosaur left 18:34 shachaf joined 18:38 bluescreen100 joined, shachaf left, bluescreen10 left 18:40 frdmn left 18:42 rurban joined 18:49 tgt joined 18:58 lustlife left 18:59 lustlife joined 19:00 raiph left 19:01 SamuraiJack joined 19:02 SmokeMachine joined, sqirrel left
masak n: say val("003") 19:03
camelia niecza v24-37-gf9c8fc2: OUTPUT«003␤»
masak n: say +val("030")
camelia niecza v24-37-gf9c8fc2: OUTPUT«30␤»
19:04 SmokeMac_ left
lizmat r: say val("030") 19:06
camelia rakudo 71ea14: OUTPUT«===SORRY!===␤Undeclared routine:␤ val used at line 1. Did you mean '&eval'?␤␤»
lizmat NYI
check
19:11 rindolf left 19:12 rindolf joined
lizmat n: say ~val(30) 19:14
camelia niecza v24-37-gf9c8fc2: OUTPUT«30␤»
lizmat n: say ~val(30).WHAT
camelia niecza v24-37-gf9c8fc2: OUTPUT«Int()␤»
lizmat shouldn't that be Str ?
19:17 japhb_ joined
masak lizmat: method calls bind tighter than prefixes. 19:22
n: say (~val(30)).WHAT
camelia niecza v24-37-gf9c8fc2: OUTPUT«(Str)␤»
masak heh, "Int()", "(Str)". 19:23
lizmat n: say (+val("030")).WHAT
camelia niecza v24-37-gf9c8fc2: OUTPUT«(Int)␤»
dalek rl6-roast-data: 41d6282 | coke++ | / (4 files):
today (automated commit)
lizmat n: say 30.WHAT 19:24
camelia niecza v24-37-gf9c8fc2: OUTPUT«(Int)␤»
lizmat n: say val(30).WHAT
camelia niecza v24-37-gf9c8fc2: OUTPUT«(Int)␤»
lizmat n: say ~val(30).WHAT
camelia niecza v24-37-gf9c8fc2: OUTPUT«Int()␤»
lizmat weird
[Coke] rakudo has been clean for 1 day. 19:25
timotimo r: my $kb = KeyBag.new(0.5 => "a", 1.5 => "b"); say $kb.gist; # how do i make proper numbers as keys here?
camelia rakudo 71ea14: OUTPUT«keybag("0.5" => "a", "1.5" => "b")␤»
grondilu rn: package Foo { our $msg; our sub talk { say $msg } }; $Foo::msg = "hello"; Foo::talk; # posted again, just to make sure I hve your attention
camelia rakudo 71ea14: OUTPUT«(Any)␤»
..niecza v24-37-gf9c8fc2: OUTPUT«hello␤»
19:26 daniel-s__ joined
grondilu rn: package Foo { our $msg = "hello"; our sub talk { say $msg } }; $Foo::msg = "bye"; Foo::talk; 19:26
camelia rakudo 71ea14, niecza v24-37-gf9c8fc2: OUTPUT«bye␤»
timotimo so it only works in rakudo if Foo::msg already has a value?! that's strange :) 19:27
grondilu strange indeed.
rn: package Foo { our Str $msg; our sub talk { say $msg } }; $Foo::msg = "bye"; Foo::talk; 19:28
camelia rakudo 71ea14: OUTPUT«===SORRY!===␤Cannot put a type constraint on an 'our'-scoped variable␤at /tmp/wJimU9qRlP:1␤------> package Foo { our Str $msg⏏; our sub talk { say $msg } }; $Foo::msg␤ expecting any of:␤ scoped declarator␤ const…
..niecza v24-37-gf9c8fc2: OUTPUT«===SORRY!===␤␤Common variables are not unique definitions and may not have types at /tmp/JNJzkBNUGj line 1:␤------> package Foo { our Str $msg⏏; our sub talk { say $msg } }; $Foo::msg␤␤Unhandled exception: Check failed␤…
grondilu oops, didn't know that.
timotimo oh? maybe that's the key.
19:28 daniel-s_ left
timotimo it may even be to spec in rakudo and niecza is out of sync with the specs? dunno 19:30
19:31 rindolf left
timotimo std: package Foo { our $msg = "hello"; our sub talk { say $msg } }; $Foo::msg = "bye"; Foo::talk; 19:32
camelia std 86b102f: OUTPUT«ok 00:00 44m␤» 19:33
timotimo std: package Foo { our $msg = "hello"; our sub talk { say $msg } }; $Foo::msg = "bye"; Foo::talk;
camelia std 86b102f: OUTPUT«ok 00:00 44m␤»
timotimo oh?
labster good morning, everyone
masak hi, labster 19:35
nwc10 jnthn: WTF did you just do? nqp head is 28% faster than 5 hours ago 19:36
timotimo whoa.
there is obviously much performance to be had.
nwc10: what's the comparison to parrot-nqp by now?
nwc10 fbf65e4041e306e288436a3e7192082edda22677 is 2.0683e+02 +/- 8.2e-01
2696e10b27803eb7b0d1f8317fcdba340d4b8e3c is 1.4725e+02 +/- 7.2e-01 19:37
19:37 woolfy1 joined
nwc10 there will be a delay while I try to answer that 19:37
5 hours might be an exaggeration. Possibly it's 8 19:38
timotimo whoa.
do you take that long to build? or do you have a huge benchmark suite?
nwc10 no, it's one program run under dumbbench that takes a while 19:39
it most definately isn't a benchmark suite, or representative
but it is the *same* program :-)
jnthn nwc10: fbf65e was from last night just before I started doing invokedynamic changes
nwc10 it's the levenstein code for the variable names, extracted, and looping
oh, OK. Nearly 24 hours 19:40
that's much less impressive :-)
jnthn Well, it depends what is to thank for it
timotimo ah, i'm glad my code still gets used for miscellaneous stuff :D
nwc10 yes, about 6 commits. After I try to answer the parrot question, I'll see if I can work if if any particular commit was the "problem" 19:41
jnthn 2a34af7 was the initial introduction for sub calls. 55e954c were some minor improvements. 746aa35 greatly improved arg handling for sub calls.
nwc10 actually, so it's not fbf6
jnthn And from there to HEAD optimized return
nwc10 I think I had one that was 1.9+e02 from today
so, sorry, it's probably only 20%. Not 28%
lizmat I'll take 20% any day 19:42
jnthn Sure, that still means 28% since < 24 hours, though? ;)
nwc10 <aol>me too</aol>
jnthn: yes, I guess so :-)
lizmat I assume this has nothing to do with fixing the slowdown that masak saw with string concatenation I believe? 19:43
nwc10 I don't know for sure, but I think it's far more likely to be with jnthn reducing sub call overhead
something like "I love it when a plan comes together" 19:44
lizmat imagines jnthn with a big cigar and a glass of bourbon 19:47
19:51 mathw_ is now known as mathw 19:54 geekosaur joined
masak 19 signups for github.com/perl6/mu/wiki/perl6-wor...p-may-2013 -- amazing. 19:56
I'm planning it now, and from what I can see, it's gonna be great :>
mathw \o/
labster I'm looking forward to it 19:57
masak in a week, we'll all be looking backward to it :)
lizmat and be much wiser :-)
labster and much more beautiful 19:58
lizmat ah, maybe on the inside, I don't think the outside can be fixed on IRC
census masak++ 20:00
mathw it's going to be grand I'm sure
Although I won't be there I will be keen to hear about it afterwards 20:01
20:01 thou left
masak chances are we'll get _ilbot back by then :) 20:02
mathw :) 20:03
It's on my birthday, as it happens, and I'll be spending the evening at aikido helping teach one of my favourite techniques :)
masak ooh! enjoy! 20:05
mathw I need to talk to the instructor for that night to make sure he's actually going to do that
it's on the schedule, but I will make a special request
Sometimes he decides to change the timetable 20:06
20:06 census left
lizmat rn: constant foo=1; sub foo {say "hi"}; foo; foo 20:07
camelia rakudo 71ea14, niecza v24-37-gf9c8fc2: ( no output )
lizmat should the definition of a sub with the same name as a constant generate a warning?
rn: sub foo { say "hello" }; sub foo {say "hi"}; foo 20:08
camelia rakudo 71ea14: OUTPUT«===SORRY!===␤Redeclaration of routine foo␤at /tmp/j8eu8XAmDw:1␤------> foo { say "hello" }; sub foo {say "hi"}⏏; foo␤ expecting any of:␤ postfix␤ statement end␤ statement modifier␤ statement modif…
..niecza v24-37-gf9c8fc2: OUTPUT«===SORRY!===␤␤Illegal redeclaration of routine 'foo' (see line 1) at /tmp/MoFgdeW1EM line 1:␤------> sub foo { say "hello" }; sub foo ⏏{say "hi"}; foo␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/li…
lizmat or even bomb like this? 20:09
jnthn The sub is installed as &foo, not foo, so there's no immediate conflict.
masak lizmat: same name as a constant: no, because the names aren't the same. 20:10
jnthn The warning I might expect is that a constant is in sink context.
masak what jnthn said.
lizmat constant foo=1; sub foo {say "hi"}; foo; &foo 20:11
r: constant foo=1; sub foo {say "hi"}; foo; &foo
camelia rakudo 71ea14: ( no output )
lizmat r: constant foo=1; sub foo {say "hi"}; foo; foo()
camelia rakudo 71ea14: OUTPUT«hi␤»
lizmat shouldn't &foo and foo() be the same here then?
masak no. 20:12
lizmat or is that my perl 5 background confusing me?
masak that's Perl 5 think :)
the sigil doesn't make the call happen, the parens do.
jnthn &foo is a noun
lizmat has met a false friend 20:13
tadzik hello hello
mathw heh that makes it look like tadzik is lizmat's new false friend 20:14
lizmat r: foo; constant foo=1; sub foo {say "hi"} 20:15
camelia rakudo 71ea14: OUTPUT«hi␤»
lizmat r: foo; constant foo=1; sub foo {say "hi"}; foo 20:16
camelia rakudo 71ea14: OUTPUT«hi␤»
lizmat S02:4154 "You still must declare your subroutines, but a bareword with an unrecognized
name is provisionally compiled as a subroutine call, on the assumption that
such a declaration will occur by the end of the current compilation unit:"
masak tadzik! \o/
lizmat r: foo; constant foo=1; sub foo {say "hi"}; say foo
camelia rakudo 71ea14: OUTPUT«hi␤1␤»
lizmat r: say foo; constant foo=1; sub foo {"hi"}; say foo 20:17
camelia rakudo 71ea14: OUTPUT«hi␤1␤»
lizmat I assume this is logical, but it is counter-intuitive at least to me
tadzik masak! \o/ 20:19
mathw did it call foo() and then output the value of foo??
tadzik mathw: how's your t430? :)
mathw tadzik: lovely!
tadzik mathw: you know what
I ordered mine on tuesday
mathw hahaha
I hope you like it as much as I like mine
masak lizmat: what's your desired semantics? name collision?
tadzik I asked the guy yesterday, "when is it going to arrive?"
lizmat r: say foo; constant foo="constant"; sub foo {"sub"}; say foo
camelia rakudo 71ea14: OUTPUT«sub␤constant␤»
tadzik and he answered "I'm probably going to send it on monday"
what a basterd :| 20:20
lizmat well, either that or disallow a bareword as a provisional subroutine call
mathw faster than mine
tadzik well, I expected it to arrive today at worst
jnthn
.oO( that's a crappy of "bastard" :P )
*crappy spelling
tadzik my sl500 is quite falling apart
which doesn't surprise me, considering what I put it through 20:21
lizmat masak: it shouldn't matter whether a sub definition has been seen 20:23
whether "foo" is a sub call or a constant 20:24
as you just said: the parens make a sub call, so why aren't they needed when the sub is not defined yet? 20:25
I would therefore suggest that S02:4158 changes thus: 20:26
- foo; # provisional call if neither &foo nor ::foo is defined so far
+ foo; # not a provisional call since only () force calling the sub
*forces 20:27
masak lizmat: hold on, hold on.
20:27 frdmn joined
lizmat is holding on 20:27
masak lizmat: as another data point, nothing you've said or shown so far really surprises me as a sixer.
lizmat that "foo" does different things depending on order of compilation?
masak lizmat: besides, this is a very narrow corner case we're talking about, where the programmer *in her own scope* overloads a subroutine name with a constant name. 20:28
20:28 SamuraiJack left
masak lizmat: we agree that Perl 6 is one-pass, yes? by necessity. 20:28
lizmat agree, as per spec
masak good.
in that light, I find the current semantics very reasonable.
it's sort of a "best guess". 20:29
jnthn Well, we could consider it an illegal post-decl.
lizmat r: say foo; sub foo {"hi"}; say foo
camelia rakudo 71ea14: OUTPUT«hi␤hi␤»
jnthn r: A; class A { }
camelia rakudo 71ea14: OUTPUT«===SORRY!===␤Illegally post-declared type:␤ A used at line 1␤␤»
masak jnthn: yes, that's a variant. I could accept that.
that's like "you implicitly agreed to the assumption to treat this as a function call, and then you went and invalidated that same assumption. ERROR" 20:30
lizmat: what do you think?
lizmat that seems logical to me
20:30 sorear left
mathw I'm still finding it weird that constants and subs don't seem to share a namespace 20:30
lizmat r: foo; constant foo=1
camelia rakudo 71ea14: OUTPUT«===SORRY!===␤CHECK FAILED:␤Undefined routine '&foo' called (lines 1, 1)␤»
gtodd /j #parrot
oops 20:31
mathw or rather they kind of maybe do. Maybe I've been doing too much C# lately.
masak mathw: it's more like, subs get to sometimes hang out in the constants' namespace.
lizmat masak,jnthn: so "foo; constant foo=1" would become such an error?
masak lizmat: yes.
lizmat that would make sense, yes 20:32
mathw I wouldn't object to postdeclared subs being illegal too, but that's not very Perlish is it :)
masak I'd still like to hear what TimToady thinks about this. but in the meantime, maybe one of us could write this up as a spec issue?
TimToady sub foo is really in the & namespace; the use of bare foo is a special case
jnthn std: foo; constant foo=1
camelia std 86b102f: OUTPUT«===SORRY!===␤Illegally post-declared type:␤ 'foo' used at line 1␤Check failed␤FAILED 00:00 41m␤»
masak mathw: postdeclared subs are staying :) that's how you do mutual recursion :)
jnthn bah, to the degree STD is spec, STD already has it that way... 20:33
masak heh.
TimToady constants are in the type namespace as barewords, no &
mathw masak: yeah I know they're useful :)
masak shall I submit a rakudobug?
lizmat masak: fwiw, I was *not* against post-declared subs 20:34
diakopter darn, just missed TimToady
20:34 sorear joined
masak lizmat: yes, I know. 20:34
lizmat just that a bareword would be encoded as a call to a sub before declaration was seen
ok
TimToady diakopter: was someone shooting at me? :) 20:36
diakopter oh, I saw the trailing &
masak *lol* 20:37
20:38 skids left
masak .oO( darn, just missed diakopter ) 20:38
TimToady nr: pi; constant pi = 3; 20:39
camelia niecza v24-37-gf9c8fc2: OUTPUT«===SORRY!===␤␤Lexical symbol 'pi' is already bound to an outer symbol (see ??? line 0);␤ the implicit outer binding at line 1 must be rewritten as pi␤ before you can unambiguously declare a new 'pi' in this scope at /tmp/0L1j5YIuDc line…
..rakudo 71ea14: ( no output )
masak ooh, STDbug.
lizmat rn: pi; constant pi = 3; say pi
camelia niecza v24-37-gf9c8fc2: OUTPUT«===SORRY!===␤␤Lexical symbol 'pi' is already bound to an outer symbol (see ??? line 0);␤ the implicit outer binding at line 1 must be rewritten as pi␤ before you can unambiguously declare a new 'pi' in this scope at /tmp/Zd1bkwxfs0 line…
..rakudo 71ea14: OUTPUT«3␤»
TimToady std: pi; constant pi = 3
camelia std 86b102f: OUTPUT«===SORRY!===␤Lexical symbol 'pi' is already bound to an outer symbol (see CORE.setting line 95);␤ the implicit outer binding at line 1 must be rewritten as OUTER::OUTER::<pi>␤ before you can unambiguously declare a new 'pi' in this scope at /tmp/j…
TimToady huh 20:40
masak oh, it wasn't STD.
the bug's in Niecza.
TimToady it's the OUTER:: limits
lizmat eerie music abounds
TimToady std: OUTER::pi; constant pi = 3;
camelia std 86b102f: OUTPUT«ok 00:00 41m␤»
TimToady wonders why the double OUTER 20:41
diakopter we control the horizontal, but not quite the vertical
lizmat rn: constant foo=1; constant foo=2; say foo 20:43
camelia rakudo 71ea14: OUTPUT«2␤»
..niecza v24-37-gf9c8fc2: OUTPUT«===SORRY!===␤␤Illegal redeclaration of symbol 'foo' (see line 1) at /tmp/d6O2LmsOMT line 1:␤------> constant foo=1; constant foo⏏=2; say foo␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.set…
lizmat that doesn't seem right in rakudo either, or does it?
TimToady std: constant foo=1; constant foo=2; say foo 20:44
camelia std 86b102f: OUTPUT«===SORRY!===␤Illegal redeclaration of symbol 'foo' (see line 1) at /tmp/eiRVcuRNBP line 1:␤------> constant foo=1; constant foo⏏=2; say foo␤Check failed␤FAILED 00:00 42m␤»
diakopter r: eval 'my $OUTER::foo = 45'; say eval '$foo'
camelia rakudo 71ea14: OUTPUT«Cannot modify an immutable value␤ in block at eval_0:1␤ in any at eval_0:1␤ in sub eval at src/gen/CORE.setting:602␤ in sub eval at src/gen/CORE.setting:593␤ in block at /tmp/xrWh4VGeVF:1␤␤»
TimToady the 'immutable value' here being the closed lexpad 20:45
diakopter r: eval 'my \OUTER::foo = 45'; say eval '$foo' 20:46
camelia rakudo 71ea14: OUTPUT«===SORRY!===␤Method 'ast' not found for invocant of class 'NQPMu'␤»
diakopter you're an ast
r: my \OUTER::foo = 45 20:47
camelia rakudo 71ea14: OUTPUT«===SORRY!===␤Method 'ast' not found for invocant of class 'NQPMu'␤»
TimToady std: my \OUTER::foo = 45
camelia std 86b102f: OUTPUT«===SORRY!===␤Term definition requires an initializer at /tmp/5_ATXwfLVG line 1:␤------> my \OUTER⏏::foo = 45␤Confused at /tmp/5_ATXwfLVG line 1:␤------> my \OUTER:⏏:foo = 45␤ expecting any of:␤ coloncircumfix…
TimToady no :: allowed in a term 20:48
TimToady had better sleep off this Novocaine®, not that a geezer needs an excuse to sleep... 20:50
look, ma, no & 20:51
masak TimToady: sleep well. dream of full STD compliance, everywhere. 20:54
lizmat is wondering how something can generate an error in STD, and not in Rakudo? 20:56
mathw because Rakudo doesn't always match STD's rules yet
lizmat ah, so rakudo doesn't use std as its grammar… Hmmmm... 20:57
lizmat hadn't realised that 20:58
mathw Ultimately it would be good if it did, but obviously to start with it couldn't because it wouldn't have been able to cope... LTM being a particular issue 20:59
jnthn Well, it's not like exactly using STD is a goal either. 21:00
lizmat hadn't realised that either
mathw I kind of thought it was
but I'm rather out of touch :)
jnthn Accepting the programs STD does, and rejecting the ones STD does with at least as good error messages is a goal, for sure.
masak Niecza is fairly close to STD. it's basically an implementation of STD, if I understand correctly.
and Rakudo has been nudging closer in *spriit* to STD for a long time now. 21:01
jnthn masak: Even Niecza diverged some amount - because you *have* to if you want to write a Perl 6 implementation, not just a Perl 6 parser.
sorear STD needs some fairly substantial tweaks to be usable as a compiler frontend
masak yeah.
lizmat learned a lot today 21:02
[Coke] "we learned something today." 21:04
mathw no day in which you learn something is a loss 21:07
lizmat wisdom it is you speak 21:10
mathw it's a bit pat really 21:11
but it's true
21:12 BenGoldberg joined 21:15 toddr joined 21:16 mattp__ joined 21:17 mattp_ left 21:29 raiph joined
lizmat rn: my (|$a) := { "hello" } 21:33
camelia niecza v24-37-gf9c8fc2: OUTPUT«===SORRY!===␤␤Cannot use bind operator with this LHS at /tmp/5e4vPb5hmL line 1 (EOF):␤------> my (|$a) := { "hello" }⏏<EOL>␤␤Potential difficulties:␤ Unsupported use of | with sigil; nowadays please use | without sigil…
..rakudo 71ea14: OUTPUT«Obsolete use of | or \ with sigil on param $a␤»
lizmat S02:5175 gives this example
does that imply the spec is outdated as well? 21:34
masak think so. 21:35
it's porbably just |a these days.
but don't take my word for it.
diakopter lizmat: sounds like a candidate for clarification/updating in the list 21:36
lizmat will add
diakopter anyone else familiar with the implementations want to take on another Syn?
lizmat actually, it is S03:5175 21:37
diakopter masak: you're not allowed to volunteer; you owe me a few thousand slides or something
masak yeah, something like that.
probably an even thousand, though. not a few thousand. 21:38
21:41 bluescreen100 left 21:46 bruges_ joined 21:48 bruges left 21:49 xilo left 21:53 xilo joined, Chillance left
lizmat rn: goto "foo" 22:00
camelia niecza v24-37-gf9c8fc2: OUTPUT«Unhandled exception: Illegal control operator: goto(foo, dynamic)␤ at /home/p6eval/niecza/lib/CORE.setting line 1530 (_lexotic @ 8) ␤ at /home/p6eval/niecza/lib/CORE.setting line 1532 (goto @ 4) ␤ at /tmp/JDK814MATF line 1 (mainline @ 3) ␤ at /home/p6e…
..rakudo 71ea14: OUTPUT«===SORRY!===␤Undeclared routine:␤ goto used at line 1␤␤»
diakopter lizmat: :)
dalek p: 18640c0 | jnthn++ | src/vm/jvm/QAST/Compiler.nqp:
Fix some stack management issues.
p: 8bf2ac7 | jnthn++ | src/vm/jvm/QAST/Compiler.nqp:
Scatter some stack spilling operations.

Get us towards enforcing "stack empty on call", which will simplify using invokedynamic for methods but, more importantly, is a prereq for coroutine support which Rakudo will need.
p: 359abcc | jnthn++ | src/vm/jvm/ (2 files):
Basic invokedynamic handling for methods.

Gets the dispatch logic out of the generated code, which in the case of the method-call heavy QAST compiler knocks a third off the size of the output. Not really faster yet; need to stack a PIC-ish thing up in the callsite for that.
timotimo r: my $kb = KeyBag.new(0.5 => "a", 1.5 => "b"); say $kb.gist; say "i didn't get an answer to this before. how do i turn the keys into proper numbers?" 22:02
camelia rakudo 71ea14: OUTPUT«keybag("0.5" => "a", "1.5" => "b")␤i didn't get an answer to this before. how do i turn the keys into proper numbers?␤»
flussence r: my KeyBag[Num] $kb; 22:03
camelia rakudo 71ea14: OUTPUT«===SORRY!===␤Type KeyBag cannot accept type arguments␤»
flussence well there goes that idea...
lizmat rn: my $kb = KeyBag.new(0.5 => "a", 1.5 => "b"); say $kb.keys.gist 22:04
camelia rakudo 71ea14, niecza v24-37-gf9c8fc2: OUTPUT«0.5 1.5␤»
lizmat not sure if I understand the issue
masak flussence: the keys should be proper numbers by default, I *think*...
flussence: unless KeyBag defaults to Str keys, like Hash does.
22:04 lustlife left
masak flussence: I'm actually not sure. TimToady? 22:04
lizmat rn: my $kb = KeyBag.new(0.5 => "a", 1.5 => "b"); say $_.WHAT for $kb.keys
camelia rakudo 71ea14, niecza v24-37-gf9c8fc2: OUTPUT«(Str)␤(Str)␤»
lizmat rn: my $kb = KeyBag.new(0.5 => "a", 1.5 => "b"); say (+$_).WHAT for $kb.keys 22:05
camelia rakudo 71ea14, niecza v24-37-gf9c8fc2: OUTPUT«(Rat)␤(Rat)␤»
lizmat looks up specs on KeyBag 22:06
"A mutable C<Bag> container, represented as C<KeyHash of UInt>." 22:07
so you're saying it should die on using "0.5" as a key?
22:08 arlinius joined
diakopter lizmat: oh, I kinda skimmed over that part on S02 b/c I wasn't familiar with rakudo's status 22:08
lizmat the way I understand KeyBag, is that the *values* are UInt
"A C<KeyHash> represents a mutable set of values, represented as the keys 22:10
of a C<Hash>. When asked to behave as a list it ignores its values
and returns only C<.keys>. C<KeySet> and C<KeyBag> are derived from
this type, but constrain their values to be C<Bool> and C<UInt>,
respectively."
sorear not Int?
lizmat the spec says UInt 22:11
S32/Containers:1212
22:11 jac50 joined 22:12 toddr_ joined
sorear no love for homology eh 22:13
masak please explain. 22:14
what use would KeyBag have from being able to have negative values?
lizmat rn: my $kb = KeyBag.new(a=>"b"); $kb<a>++; say $kb<a>; say $kb<a>.WHAT 22:15
camelia rakudo 71ea14: OUTPUT«1␤(Int)␤»
..niecza v24-37-gf9c8fc2: OUTPUT«Unhandled exception: Excess arguments to KeyBag.new, unused named a␤ at /home/p6eval/niecza/lib/CORE.setting line 0 (KeyBag.new @ 1) ␤ at /tmp/iD6iVRzSeC line 1 (mainline @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4299 (ANON @ 3) ␤ at /home/p6…
22:15 shachaf joined, toddr left
flussence I suspect the spec's using an unreasonably narrow type just for the unsignedness... 22:16
lizmat fwiw, I think not having negative values is right, good and as intended
a bag can contain 0 or more copies of an item
22:17 thou joined
lizmat not -1 copies of an item 22:17
sorear masak: free abelian group on a type of generators
masak: there are natural objects in several fields of math that look like signed keybags
elliptic curve divisors, homology classes 22:18
masak sorear: interesting.
sorear factor representations of rationals / relations in the quadratic sieve and derivatives
masak sorear: fwiw, I'm actively studying group theory and category theory at present. every day I'm more well-equipped to understand what you're saying without using Wikipedia as an adapter :) 22:19
flussence
.oO( it's not too hard to turn those fractions into whole numbers, @values *= ([+] @values).denominator; )
 22:20
masak sorear: but I'm kinda inclined to agree with lizmat here. at least until I understand the benefits of allowing negative integers.
lizmat maybe you want KeyWeight?
sorear the only algorithm I can think of offhand that would use this is the quadratic sieve 22:21
lizmat "A C<KeyHash of FatRat>; like a C<KeyBag> but may have non-integral
weights for use in weighted picking."
sorear lizmat: maybe
lizmat it specifically mentions negative values:
"Keys with fractional weights
are deleted if they go to 0. Negative weights are not deleted, but
the implementation may complain if it notices you attempting to use
such a weight.
"
22:23 grondilu left
masak I don't see why it should complain. depending on the domain, that may be perfectly legit. 22:25
22:26 adu joined 22:28 adu left
lizmat hence the *may* ? 22:29
masak yes, but how will it know? :) 22:30
lue hello world o/ 22:31
22:32 adu joined
lizmat masak: no idea 22:32
more worringly:
rn: say KeyBag.new(a=>"b")<a>
camelia rakudo 71ea14: OUTPUT«0␤»
..niecza v24-37-gf9c8fc2: OUTPUT«Unhandled exception: Excess arguments to KeyBag.new, unused named a␤ at /home/p6eval/niecza/lib/CORE.setting line 0 (KeyBag.new @ 1) ␤ at /tmp/Qqhm9N9Ikq line 1 (mainline @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4299 (ANON @ 3) ␤ at /home/p6…
masak Niecza has it right.
timotimo i was interested in a KeyBag with fractional numbers rather than integers to get the .pick/.roll behavior that uses the values for probability
sorear MAY. or perhaps SHOULD, or SHOULD NOT 22:33
timotimo (and now i realize that using 0.5 => 2 is idiotic, because the key is the value and the value is how often it is in there)
lizmat aha...
masak lizmat: there's a tension there between routine arguments and hash pairs.
lizmat: and in a .() the former wins by default.
lizmat: I believe S02 or S06 details how to get around that. 22:34
rn: say KeyBag.new( (a=>"b") )<a>
camelia rakudo 71ea14, niecza v24-37-gf9c8fc2: OUTPUT«b␤»
timotimo why do the () make a difference?
masak timotimo: because what I just said. 22:35
timotimo oh, in the one case it will be unpacked, in the other it won't?
22:35 dmol left
masak timotimo: there's a tension between routine arguments and hash pairs. 22:35
timotimo ah, i read that as "routing arguments" and was parsefail
masak .oO( "the deneuralizer" )
lizmat but but, shouldn't a keybag only contain uints as values
masak yes. 22:36
lizmat and complain about "b" being passed as a value?
masak yes.
masak submits rakudobug
adu what does {YOU_ARE_HERE} mean? 22:37
github.com/perl6/nqp/blob/master/s...r.nqp#L197 22:38
22:40 spider-mario left
lizmat adu: S02:2631 and following 22:40
lizmat must admit not completely grokking that section of the spec 22:41
lue my understanding is that it basically means YOUR CODE GOES HERE, where "you" are the user of the language with a P6 script. 22:46
22:47 PacoAir left
lue (in rakudo's setting, it is the very last statement in the generated CORE.setting file) 22:47
lizmat timotimo: maybe you want something like: 22:48
rn: my $kb= KeyBag.new((a => 1, b => 10)); say $kb.pick(*)
camelia niecza v24-37-gf9c8fc2: OUTPUT«b b b b b b a b b b b␤»
..rakudo 71ea14: OUTPUT«b b b b b a b b b b b␤»
lizmat rn: my $kb= KeyBag.new((a => 1, b => 1)); say $kb.pick; say $kb.perl; say $kb.pick 22:50
camelia niecza v24-37-gf9c8fc2: OUTPUT«a␤KeyBag.new({"a" => 1, "b" => 1}.hash)␤b␤»
..rakudo 71ea14: OUTPUT«b␤KeyBag.new(("a" => 1, "b" => 1).hash)␤b␤»
lizmat two "b"?s
is this the Perl 5 bug/feature "each" iterator being reset by "keys" all over again? 22:51
sorear sure
lizmat: no, "pick" selects an element at random 22:52
run it again and you'll get a different answer (with probability 15/16) 22:53
lizmat aha, I read the spec as "pick" attaching some magic that would keep how many times it had already picked values 22:54
sorear no, it only keeps that state within a single call
lizmat but it is only within the .pick method call that the "without" replacement part is actually taking place
indeed
I guess this can be easily subclassed with .pick doing a .pick(*), saving that in an array and then shifting values out as requested 22:56
rn: my $kb= KeyBag.new((a => 1, b => 1)); say $kb.pick(3).elems 22:57
camelia rakudo 71ea14, niecza v24-37-gf9c8fc2: OUTPUT«2␤»
lizmat wonder whether or not that should be a warning/error 22:58
picking N elements from a M picks keybag where N > M 22:59
sorear it's sort of expected that it's not 23:01
rn: say [1..5].pick(10).elems
camelia rakudo 71ea14, niecza v24-37-gf9c8fc2: OUTPUT«5␤»
sorear overpicks always limit
23:02 census joined
lizmat it's not specced what should happen 23:02
afaik
TimToady I believe it's specced somewhere... 23:03
BenGoldberg rn: my @a=^Inf; my @b = @a; shift @a; say @b[^5];
camelia niecza v24-37-gf9c8fc2: OUTPUT«(timeout)»
..rakudo 71ea14: OUTPUT«1 2 3 4 5␤»
TimToady yes, S32/Containers, line 258 23:04
"If C<*> is specified as the number (or if the number of elements in the list is less than the specified number), all the available elements are returned in random order" 23:05
BenGoldberg rn: my @a=^Inf; my @b = @a; shift @b; say @a[^5];
lizmat I've been looking at that paragraph, and did not parse that
masak adu: perhaps it's easiest to understand {YOU_ARE_HERE} in terms of the -n and -p flags. the {YOU_ARE_HERE} terms is simply a mechanism for injecting the program that the user wrote into a larger setting.
camelia niecza v24-37-gf9c8fc2: OUTPUT«(timeout)» 23:06
..rakudo 71ea14: OUTPUT«0 1 2 3 4␤»
masak adu: and that's excatly what's needed to make -n and -p work, for example.
adu masak: thanks, that makes sense
masak adu: but also for the regular setting, called CORE.
23:06 adu left
masak 'night, #perl6 23:06
TimToady YOU_ARE_HERE is just a lazy eval of something that will be specified later :)
lizmat 'night masak 23:07
lue ♞ masak o/
TimToady o/
BenGoldberg Good evening everyone! :)
dalek p: ec08455 | jnthn++ | / (4 files):
Improve method invocation by using guard clauses.

This means we don't have to re-enter the bootstrap method per type of invocant that occurs at the callsite.
diakopter BenGoldberg: 'night
23:07 adu joined
masak {YOU_ARE_HERE} is a templating mechanism with one parameter :) 23:07
really 'night
lizmat as well 23:08
lue Think of it like those maps in public locations. If you're traversing a large setting, it's nice to know where YOU ARE when you run code.
lizmat gnight #perl6
lue ♞ lizmat o/
jnthn OK, that's the majority of the first round of invokedynamic work done. 23:10
TimToady
.oO(the tyranny of the minority...)
BenGoldberg r: my @a := ^Inf; my @b := @a; shift @b; say @a[^5];
camelia rakudo 71ea14: OUTPUT«No such method 'shift' for invocant of type 'Range'␤ in sub shift at src/gen/CORE.setting:6387␤ in block at /tmp/voUECkiWZK:1␤␤»
BenGoldberg is confused. 23:11
TimToady Range is effectively immutable, and iterates by returning a different Range
so a shift in place ain't gonna cut it 23:12
BenGoldberg So why does it work when I use = instead of := ?
TimToady because = is smart enough to switch to lazy semantics in rakudo when it sees infinity 23:13
(but not smart enough in niecza, last I checked)
BenGoldberg Ok.
TimToady and = has copy semantics
r: my @a; @a.plan: ^Inf; shift @a; say @a[^5] 23:14
camelia rakudo 71ea14: OUTPUT«No such method 'plan' for invocant of type 'Array'␤ in block at /tmp/Y4wQXhP0gv:1␤␤»
TimToady hmm
n: my @a; @a.plan: ^Inf; shift @a; say @a[^5] 23:15
camelia niecza v24-37-gf9c8fc2: OUTPUT«1 2 3 4 5␤»
TimToady I guess niecza++ is ahead there
BenGoldberg Speaking of = vs := 23:17
r: my @a = gather { take $_*2 for ^Inf }; say @a[^3]
camelia rakudo 71ea14: OUTPUT«(timeout)»
BenGoldberg r: my @a := gather { take $_*2 for ^Inf }; say @a[^3]
camelia rakudo 71ea14: OUTPUT«0 2 4␤»
TimToady r: my @a := @(^Inf); my @b := @a; shift @b; say @a[^5]; 23:18
camelia rakudo 71ea14: OUTPUT«1 2 3 4 5␤»
TimToady = is "mostly eager"
but it can't see the Inf inside the gather 23:19
23:20 woolfy1 left
pmichaud my Int @a; @a[0] = 'hello'; say @a 23:26
r: my Int @a; @a[0] = 'hello'; say @a
camelia rakudo 71ea14: OUTPUT«Type check failed in assignment to '$v'; expected 'Int' but got 'Str'␤ in block at /tmp/YOgxRxKA1y:1␤␤»
pmichaud r: my Int %s; %s<a> = 'hello'; say %s
camelia rakudo 71ea14: OUTPUT«Type check failed in assignment to '$v'; expected 'Int' but got 'Str'␤ in block at /tmp/ubaGTtEZjh:1␤␤»
jnthn The mystical $v... 23:27
pmichaud I was just looking at RT #117773 real quick.
(KeyBag doesn't enforce int-ness)
r: class XYZ { has Int %!h; method abc() { %!h<a> = 'hello'; } }; XYZ.new.abc(); say 'alive'; 23:29
camelia rakudo 71ea14: OUTPUT«Type check failed in assignment to '$v'; expected 'Int' but got 'Str'␤ in method abc at /tmp/diaJD8BY_d:1␤ in block at /tmp/diaJD8BY_d:1␤␤»
pmichaud r: class A { has Method @.slots; }; A.new.slots.push: [1,2,3]; # RT #112660 23:30
camelia rakudo 71ea14: ( no output )
pmichaud r: class A { has Method @.slots; }; A.new.slots.push: [1,2,3]; say 'alive'; # RT #112660
camelia rakudo 71ea14: OUTPUT«alive␤»
pmichaud r: class A { has Str @.slots; }; A.new.slots.push: [1,2,3]; say 'alive'; # RT #112660
camelia rakudo 71ea14: OUTPUT«alive␤»
pmichaud r: class A { has Str @.slots; }; A.new.slots.push: 1; say 'alive'; # RT #112660 23:34
camelia rakudo 71ea14: OUTPUT«alive␤»
pmichaud I'm afk
adu who was working on the java port of NQP? 23:38
jnthn?
TimToady yes
adu how do I get in on that? 23:39
TimToady ask jnthn, I suppose 23:40
23:42 berekuk left
jnthn adu: It lives in the NQP repository these days; perl ConfigureJVM.pl then make 23:42
adu jnthn: how do I get in on that?
ok
TimToady: did I tell you about my experience with clang? 23:45
jnthn adu: The jvm-support branch in Rakudo is where work is happening to get a Rakudo that builds/runs on NQP on JVM. Nothing too exciting to see yet, alas. 23:46
(I mean, it builds a bunch of stuff...but you can't do anything with it yet. :))
Time for some rest here... 'night 23:47
adu jnthn: I'd like to help if I can
lue ♞ jnthn o/
adu jnthn: good night
jnthn adu: OK; best bet is to get a build and take a look around, then catch me sometime when it's not 2am and I can give some better pointers. :)
o/
adu jnthn: thanks :) 23:48
23:51 thundergnat joined
lue almost mentioned how "2am" without a timezone attached is nearly useless :P 23:58
census hi thundergnat! 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!