»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend! | feather will shut down permanently on 2015-03-31
Set by jnthn on 28 February 2015.
japhb timotimo: Is JSON::Fast now 100% correct? If so, can you replace the Rakudo built-in JSON parser with it? 00:29
TimToady JVM very unhappy tonight, seems 01:13
TimToady 67 test files failing 01:16
gist.github.com/anonymous/9dc73118674e46248155 is the boringly repetitious error message 01:18
--ll-exception is no help 01:20
ugexe its the range operator i believe 01:44
my @encoding-chars = 'A'..'Z','a'..'z','0'..'9','+','/'; <-- exception on jvm... reason for MIME::Base64 failing on jvm
ugexe replaced with an array of explicitly declared characters it works 01:47
ugexe unless you are looking for why/where... cant help you there :) 01:52
although it will happen if you try to run that code, or just precompile it 01:55
TimToady probably my fault somehow then :) 01:58
since ranges turn into sequences, and I was just mucking with sequences
ugexe the only reason i caught that was because i saw you all talking about them earlier 01:59
TimToady well, despite the terrible lack of info, I can probably hunt it down through divide-and-conquer
yeah, say 1...5 blows it up 02:00
well, I shoulda figgered it out myself, but thanks 02:01
dalek kudo/nom: 0354bb6 | TimToady++ | src/core/operators.pm:
avoid last in loop in gather that blows up jvm
 03:32
TimToady .tell jnthn any idea why the lasts I removed in 0354bb6 cause the JVM to puke? 03:33
yoleaux TimToady: I'll pass your message to jnthn.
awwaiid Can I do something like: sub MAIN(IO $filename ~~ :f); to say the param must be a file? 03:34
TimToady use where intead of ~~ maybe
but you might also need IO() to coerce it 03:35
awwaiid ah yes, where! I was looking for that but didn't search right
ub MAIN(IO() $filename where { $filename ~~ :f }); # works perfectly, thanks! 03:36
TimToady where :f oughta work 03:37
it's just a smartmatch
skids m: { .say; succeed 43.say; }(1).perl.say # I can see no use for succeed hitting bottom without finding a when or topicalizer (?), so it should probably warn, at least.
camelia rakudo-moar 5a3f38: OUTPUT«1␤43␤»
TimToady a bare block topicalizes $_, in fact, you used the top in .say 03:38
*topic
awwaiid it does! neat 03:38
skids Yes, and because there was no when { }, succeed did not stop there. It went a the way to the outermost scope. 03:39
*all
m: (when * { succeed 43.say; }).perl.say
camelia rakudo-moar 5a3f38: OUTPUT«43␤»
skids That found the when but no topicalizer. 03:40
TimToady yeah, seems a bit rinky-tink yet 03:41
skids Unless there is a use case, I think if a succeed control exception hits the bottom, it shold just throw.
(something other than a succeed control exception :-) 03:42
TimToady most of the lexotics in rakudo are still a bit rinky-tink 03:43
skids Well, to be fair I don't think the behavior is addressed in spec. 03:44
TimToady S04:952 is pretty detailed, but doesn't say how it should fail to be sure 03:46
m: { when * -> $_ { succeed .say }; say "Oops" }(42).say 03:48
camelia rakudo-moar 0354bb: OUTPUT«Too few positionals passed; expected 1 argument but got 0␤ in block at /tmp/Iz6taeDCrn:1␤ in block <unit> at /tmp/Iz6taeDCrn:1␤␤»
TimToady I guess when -> isn't implemented yet, but would be interesting to see if it is merely scanning for a topicalizer and ignoring the when 03:49
TimToady wonders if there are tests written... 03:50
skids If a bare block topicalizes, is there an example of a non-topicalizing block? 03:51
TimToady m: say "Here { $_ } is a non-topicalizing block" given 42 03:57
camelia rakudo-moar 0354bb: OUTPUT«Here is a non-topicalizing block␤»
TimToady m: say "Here { $^a } is a non-topicalizing block" given 42 03:58
camelia rakudo-moar 0354bb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/DvN6528gib␤Placeholder variable $^a may not be used here because the surrounding block takes no signature␤at /tmp/DvN6528gib:1␤------> 3say "Here { $^a }7⏏5 is a non-topicalizing block" given 42␤ …»
skids m: (given 1 { -> $a { when * { succeed 43 }; }(1).say; 42; }).say; 41.say # Is a pointy considered topicalizing? 03:59
camelia rakudo-moar 0354bb: OUTPUT«43␤42␤41␤»
TimToady only if it declares $_ in the siggie
dalek c: 49755ef | skids++ | lib/Language/control.pod:
Split up some long lines and go OCD on the code sample spacing
 04:35
c: ade043f | skids++ | lib/Language/control.pod:
Subdivide the given/when/default/proceed/succeed section

Show a few examples of when/default without given For now, avoided giving examples with non-topicalized blocks (rakudo NYI)
skids Now I wait to see how badly I screwed up markup.
sjn question. is it meaningful/correct to talk about "sigilless variables" as constants? 04:44
skids m: my \d = 1; d = 2; d.say; 04:45
camelia rakudo-moar 0354bb: OUTPUT«Cannot modify an immutable Int␤ in block <unit> at /tmp/VlcXiG8pVX:1␤␤»
skids hrm. 04:46
m: sub foo (\d is rw) { d++ }; my $a = 1; foo($a); $a.say; 04:47
camelia rakudo-moar 0354bb: OUTPUT«2␤»
skids m: my (\d); d++; d++; d.say; 04:49
camelia rakudo-moar 0354bb: OUTPUT«2␤»
dalek c/sjn-twigil-1: 0949009 | (Salve J. Nilsen)++ | lib/Language/variables.pod:
Update variables.pod

Example should use $.b to demonstrate method calls to $.a		 04:55
c: 0949009 | (Salve J. Nilsen)++ | lib/Language/variables.pod:
Update variables.pod

Example should use $.b to demonstrate method calls to $.a		 05:02
c: 796181f | paultcochrane++ | lib/Language/variables.pod:
Merge pull request #73 from perl6/sjn-twigil-1

Update variables.pod
[ptc] skids++ # doc corrections 05:07
TimToady we try to reserve the term "constant" for things that can never be rebound. at best \d would be a readonly since it can be rebound next time through, and it looks like sometimes it's bound to a container 05:32
m: my (\d); say VAR(d).WHAT; 05:33
camelia rakudo-moar 0354bb: OUTPUT«(Scalar)␤»
TimToady m: my (\d) ::= 42; say VAR(d).WHAT;
camelia rakudo-moar 0354bb: OUTPUT«Too few positionals passed; expected 1 argument but got 0␤ in block <unit> at /tmp/e7a60IHhI6:1␤␤»
TimToady m: my (\d) = 42; say VAR(d).WHAT;
camelia rakudo-moar 0354bb: OUTPUT«(Scalar)␤»
TimToady well, is arguably buggy anyway
m: my \d = 42; say VAR(d).WHAT; 05:34
camelia rakudo-moar 0354bb: OUTPUT«(Int)␤»
TimToady that one is definitely readonly
m: my \d = 42.item; say VAR(d).WHAT;
camelia rakudo-moar 0354bb: OUTPUT«(Int)␤»
TimToady m: my \d = $(42); say VAR(d).WHAT;
camelia rakudo-moar 0354bb: OUTPUT«(Int)␤»
TimToady m: my \d = Scalar.new(42); say VAR(d).WHAT; 05:35
camelia rakudo-moar 0354bb: OUTPUT«Default constructor for 'Scalar' only takes named arguments␤ in method new at src/gen/m-CORE.setting:974␤ in block <unit> at /tmp/PWXfh1bu46:1␤␤»
TimToady m: my \d = Scalar.new; say VAR(d).WHAT;
camelia rakudo-moar 0354bb: OUTPUT«Cannot call method 'BUILDALL' on a null object␤ in method bless at src/gen/m-CORE.setting:985␤ in method new at src/gen/m-CORE.setting:971␤ in block <unit> at /tmp/x7Ep31OViP:1␤␤»
labster m: =begin pod␤ =for pod :number(42) :zebras :!sheep␤=end pod␤ say $=pod[0]; 05:40
camelia rakudo-moar 0354bb: OUTPUT«Pod::Block::Named{:name("pod")}<>␤ Pod::Block::Named{:config("\{:number(\"42\"), :sheep(\"0\"), :zebras(\"1\")}"), :name("pod")}<>␤␤»
labster shouldn't those be :sheep(False) and :zebras(True)? 05:41
skids [ptc]++ docs in the first place 05:43
lizmat NLPW& 05:52
lizmat good morning from the NLPW in the lovely town of Utrecht 07:06
m: m: =begin pod␤ =for pod :number(42) :zebras :!sheep␤=end pod␤ say $=pod[0]; # labster: feels to me that should be False/True rather than 0/1 07:09
camelia rakudo-moar 0354bb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/z1sBA2C2PV␤Preceding context expects a term, but found infix = instead␤at /tmp/z1sBA2C2PV:1␤------> 3m: =7⏏5begin pod␤»
lizmat m: =begin pod␤ =for pod :number(42) :zebras :!sheep␤=end pod␤ say $=pod[0]; # labster: feels to me that should be False/True rather than 0/1
camelia rakudo-moar 0354bb: OUTPUT«Pod::Block::Named{:name("pod")}<>␤ Pod::Block::Named{:config("\{:number(\"42\"), :sheep(\"0\"), :zebras(\"1\")}"), :name("pod")}<>␤␤»
lizmat perhaps a missing nqp::p6bool() wrapper ?
Ven github.com/Microsoft/openssl/blob/...project.pl microsoft totally uses perl as well 07:14
RabidGravy Ven, if my memory serves me they actually paid for the first port of Perl 5 to windows 07:29
FROGGS RabidGravy: they supported ActivePerl IIRC 07:34
RabidGravy yeah but there was a truly horrible port of 5.000m that was on the Windows NT resource kit CD that pre-dated ActivePerl 07:39
FROGGS aye 07:40
lizmat .tell timotimo no, I didn't see your fast JSON parser yet, something to put in core ? 07:47
yoleaux lizmat: I'll pass your message to timotimo.
brrt lizmat: github.com/timo/json_fast 07:56
lizmat cool! 07:58
brrt yeah
lizmat .tell timotimo feels to me it could use a few native arrays and native str in some places still 07:59
yoleaux lizmat: I'll pass your message to timotimo.
dalek Heuristic branch merge: pushed 26 commits to rakudo/newio by lizmat 08:01
brrt .tell timotimo Fast.pm L66:L82 can be a hash? 08:02
yoleaux brrt: I'll pass your message to timotimo.
dalek Heuristic branch merge: pushed 34 commits to roast/newio by lizmat
timotimo yeah, i think a hash would be good, just not sure how to handle the \u case well 08:05
yoleaux 07:47Z <lizmat> timotimo: no, I didn't see your fast JSON parser yet, something to put in core ?
07:59Z <lizmat> timotimo: feels to me it could use a few native arrays and native str in some places still
08:02Z <brrt> timotimo: Fast.pm L66:L82 can be a hash?
timotimo i thought a few more native things would be better, too. i have to do precise timings to see what is worth it and what isn't 08:07
it seems like there's a sort of trade-off between "my int $ord = nqp::ordat($text, $pos)" vs "my $ord := nqp::ordat($text, $pos)"
and i don't have to use parenthesis notation everywhere for nqp "sub" calls, either 08:08
timotimo at least with regards to allocations, it seems waterbeddy at the moment 08:08
as in: either you get Scalar allocations or Str/Int allocations 08:09
and there's a LOT of those allocations
and while $pos++ would work now, it still turns into a full sub call, if i saw correctly
timotimo that's why i put $pos = $pos + 1 everywhere 08:09
brrt kill all the allocations! 08:10
timotimo i'd like to
brrt but yeah, looks pretty good timo :-)
timotimo no, it looks ugly :)
brrt you are not a true C programmer
then :-P
timotimo :D
brrt why are most subs prefixed with my? don't want to expose them? 08:11
timotimo correct
their interface isn't exactly user-friendly
brrt also... i don't agree that a json string ought to start with a { or [
timotimo blame the JSON::Tiny test suite
i was looking for that in the ecma spec for json, too 08:12
brrt i think you can have totally valid numbers and strings
timotimo and couldn't find it
brrt as in "foo" is perfectly valid json imho
but timotimo++
timotimo # stolen from JSON::XS, 18_json_checker.t 08:12
Q<<"A JSON payload should be an object or array, not a string.">>,
possibly stolen from there before that was changed in the spec or something like that? 08:13
JSON::Tiny is pretty old already
RabidGravy fwiw, Node.js won't parse a plain atom either 08:18
timotimo the json checker test file from JSON::XS still seems to want the outermost value having to be an object or array
RabidGravy > var foo = JSON.parse('hshsh');
SyntaxError: Unexpected token h
RabidGravy no I'm talking shit 08:20
omitted the ""
timotimo ah
RabidGravy > var foo = JSON.parse('"hshsh"'); 08:21
undefined
timotimo ... undefined?
RabidGravy the Node.js REPL does that for any object 08:22
timotimo that's weird. is it just so that it doesn't print a thousand lines of output for everything?
RabidGravy console.log(JSON.parse('{}')); 08:23
{}
undefined
> console.log(JSON.parse('"foooo"'));
foooo
undefined
timotimo OK
RabidGravy never been sufficiently curious to find out *why* it prints "undefined" 08:24
bobkare I seem to recall something about non-array/object json documents being a new thing in the new json standard last year 08:27
timotimo i read the 2013 standard and it didn't have any restriction on that 08:29
labster lizmat: Actually I'm pretty sure that it's in Perl6::Pod that's causing that pod error earlier, right under this comment: # and this is the worst hack of them all. # Hide your kids, hide your wife! 08:32
timotimo oh tadzik %) 08:33
FROGGS :D
labster much more entertaining that TODO FIXME
*than
FROGGS m: try use Flubber; 08:37
camelia rakudo-moar 0354bb: OUTPUT«===SORRY!===␤Could not find Flubber in any of: /home/camelia/.perl6/2015.03-223-g0354bb6/lib, /home/camelia/.perl6/2015.03-223-g0354bb6, /home/camelia/rakudo-inst-2/share/perl6/lib, /home/camelia/rakudo-inst-2/share/perl6/vendor/lib, /home/camelia/rakudo…»
RabidGravy I'm trying to remember what software it was where it documented that it *extends* JSON to allow the atoms at the top level 08:41
nwc10 "YAML?" :-) 08:44
Ven json5 probably 08:46
RabidGravy no, no some actual software that uses JSON, I thought it was CouchDB but I can't find what I thought I read in the docs 08:48
RabidGravy though it does have a useful round up of real world json parser behaviour in docs.couchdb.org/en/1.6.1/api/basic...son-basics 08:51
jnthn timotimo: "my" is the defualt for subs 08:52
yoleaux 9 Apr 2015 22:12Z <lizmat> jnthn: that's what b6bc4ecd08b4792d73ab0ee9da01edac60a909c0 did (well, checking, not throwing, not sure what would be best in this case)
9 Apr 2015 22:13Z <lizmat> jnthn: perhaps an X::OutOfRange exception would be best, to stay in line with .substr
03:33Z <TimToady> jnthn: any idea why the lasts I removed in 0354bb6 cause the JVM to puke?
eli-se .botsnack 08:54
yoleaux :D
jnthn .tell TimToady Not immediately, no... :( 08:55
yoleaux jnthn: I'll pass your message to TimToady.
eli-se I think yoleaux likes bones. "bot" is Dutch for "bone"
nine I think we should end calling modules ::Tiny or ::Fast. Tiny things never stay tiny. And fast things get superseeded by things that are even faster. Why not just name it after what it does? It's a JSON::Parser, nothing more, nothing less. And in Perl 6, both JSON::Tiny and JSON::Fast could be called JSON::Parser, just by two different authorities. The one even is a reimplementation of the other. 08:59
yoleaux 9 Apr 2015 22:11Z <lizmat> nine: the solution is simple: Inline::Perl6 -> Inline::Perl5 -> other version :-)
nine .tell lizmat shame on me for not thinking about using Inline::Perl6 :) 09:00
yoleaux nine: I'll pass your message to lizmat.
dalek ast: e35d2cf | labster++ | S (3 files):
test that "0" is now True, fix tests for non-string pod options
 09:06
kudo/nom: 82a1811 | labster++ | src/core/Str.pm:
?"0" is now True per spec change

In particular perl6/specs@befb20c3b		 09:07
moritz labster: do you have some assessment of the ecosystem fallout caused by that change? 09:11
labster not really, though it only broke 4 spectests (and some wrong tests). If you want to revert, go ahead. 09:11
FROGGS no, we can measure the fallout
moritz labster: I don't want to revert right now 09:12
only if it turns out to be a disaster :-)
labster FROGGS: how do we measure now?
FROGGS run a panda smoke, copy the share/panda/reports.<compiler-version>, checkout to the commit before the change, smoke again, and compare both report files 09:13
moritz then we can prepare branches for modules, and do the change right after a release
FROGGS ahh, and you've got to set PANDA_SUBMIT_TESTREPORTS=1
labster :)
RabidGravy right off to the seaside to escape the pollution and pollen 09:25
dalek pan style="color: #395be5">perl6-examples: 533535f | paultcochrane++ | categories/wsg/advanced-2008/event (2 files):
Use consistent filename for event010-dwhipp.p6
 09:42
pan style="color: #395be5">perl6-examples: a294ef5 | paultcochrane++ | categories/wsg/ (8 files):
Document examples for wsg category
[ptc]_ lizmat: does slurp still append a newline? 09:53
lizmat not sure why it would, does it? 09:54
yoleaux 09:00Z <nine> lizmat: shame on me for not thinking about using Inline::Perl6 :)
[ptc]_ lizmat: just found it mentioned in tests in the perl6-examples project, and after Tux' comment, wondered if it had been changed 09:54
lizmat is you look at the result of a slurp with say, it will have a \n after it :-) 09:55
[ptc]_ lizmat: aha, it's still there. Thanks :-)
lizmat std: [42]<> 10:09
camelia std 28329a7: OUTPUT«ok 00:00 137m␤»
lizmat m: say [42]<>.perl # did I miss a memo on that syntax?
camelia rakudo-moar 82a181: OUTPUT«[42]<>␤»
lizmat m: say [42].perl # did I miss a memo on that syntax? 10:10
camelia rakudo-moar 82a181: OUTPUT«[42]␤»
psch m: say [42].flat.perl
camelia rakudo-moar 82a181: OUTPUT«[42]<>␤»
psch lizmat: i understood it as a container-independant zen slice 10:11
masak good noon, #perl6
psch lizmat: i think TimToady++ had some rational about a week back, let me see if i can find it
lizmat "switch to .<> as more visually distinct decont"
masak lizmat: there's a bit of discussion about it in the backlog. 10:11
lizmat ok
masak lizmat: I can't say I'm overly pleased with it.
to me, .<> still rings of "hash access" 10:12
psch right, decont it was
irclog.perlgeek.de/perl6/2015-04-02#i_10382903
eli-se I want to add the <> operator as a slang exercise.
masak why a slang? it's just a normal non-magical operator, sugar for `.decont()` or whatever 10:13
eli-se I mean the Perl 5 operator, like <$fh> 10:14
masak oh! 10:15
moritz that's a postcircumfix operator
masak no, it's a circumfix operator
eli-se normal circumfix operator
moritz erm
sorry
yes
masak we don't often talk about those :)
eli-se m: sub circumfix:<< < > >>($fh) { say $fh }; <'foo'> 10:16
camelia rakudo-moar 82a181: OUTPUT«5===SORRY!5=== Error while compiling /tmp/XmiRRzxwHJ␤Missing required term after infix␤at /tmp/XmiRRzxwHJ:1␤------> 3mfix:<< < > >>($fh) { say $fh }; <'foo'>7⏏5<EOL>␤ expecting any of:␤ prefix␤ term␤»
eli-se lolwat
psch i don't think the rakudo parser is there yet
quote ops are parsed special
psch m: &circumfix:«< >».perl.say 10:16
camelia rakudo-moar 82a181: OUTPUT«5===SORRY!5=== Error while compiling /tmp/QQME8pOr4Z␤Undeclared routine:␤ &circumfix:<< >> used at line 1. Did you mean '&circumfix:<:{ }>', '&circumfix:<[ ]>', '&circumfix:<{ }>'?␤␤»
psch m: &circumfix:«[ ]».perl.say 10:18
camelia rakudo-moar 82a181: OUTPUT«sub circumfix:<[ ]> (*@elems) { #`(Sub|36386456) ... }␤»
eli-se What does #` mean?
psch eli-se: delimited comment
eli-se ok
psch m: say #`[ not this ] "this"
camelia rakudo-moar 82a181: OUTPUT«this␤»
psch wonders how in the world overriding infix:<;> is going to word 10:20
psch *work 10:20
masak psch: doing the overriding will work, but the operator will be of limited use.
psch: as it will be picked up by other rules, like statement-ending rules. 10:21
psch masak: oh, so it'd only get picked up as overridden where it would work as LoL delimiter otherwise?
that's less weird i suppose 10:22
(or maybe "that'd be")
lizmat lunch& 10:32
timotimo yo flussence 11:33
6.49user 0.05system 0:06.58elapsed 99%CPU (0avgtext+0avgdata 150996maxresident)k 11:34
3.23user 0.04system 0:03.29elapsed 99%CPU (0avgtext+0avgdata 114532maxresident)k
this is JSON::Tiny and then JSON::Fast
moritz timotimo: is JSON::Fast a reimplementation from the ground up? 11:35
timotimo: or incremental updates to JSON::Tiny?
FROGGS and without a grammar
timotimo reimplementation 11:36
FROGGS moritz: github.com/timo/json_fast/blob/mas...ON/Fast.pm
timotimo the code is "butt-ugly"
and if i uncomment the to-json proto and such, i get Redeclaration of routine to-json
:(
FROGGS timotimo: nobody forces you to cuddle else :o)
moritz wow, it seems you *can* write Pascal in any language :-)
timotimo thanks
FROGGS m: sub to-json ($s) { } 11:37
camelia ( no output )
FROGGS m: proto to-json ($s) { }
camelia ( no output )
FROGGS hmmm
timotimo moritz: i can make it prettier still, but the recent MoarVM and rakudo "regressions" make a few things a bit worse, performance-wise
FROGGS m: proto to-json is export {*}
camelia ( no output )
timotimo it ought to be faster in general, but lots of blocks that could be inlined don't get inlined at the moment
and we end up allocating a gigantic crapton of BOOTCode objects, for example 11:38
FROGGS m: proto to-json($) is export {*}; multi to-json(Real:D $d) { ~$d }
camelia ( no output )
timotimo and $foo++ doesn't get optimized properly yet, either
FROGGS why don't I get that?
timotimo m: module JSON::Fast; proto to-json($) is export {*}
camelia ( no output )
timotimo i have no clue
FROGGS timotimo: would be awesome to optimize the heck out of it... 11:39
timotimo yes
FROGGS so that it can be inlined etc
timotimo at the moment, a few things are more or less waterbeddy
FROGGS though I don't understand nothing about that :o(
timotimo i tried to make nom-ws not allocate any Scalar or Int objects, but at the moment it has to do either one or the other 11:39
FROGGS it would be also nice of course to get JSON::Tiny faster by smartifying spesh and the jit 11:45
FROGGS but I see that this might not be trivial 11:46
timotimo not only spesh and the jit, but also our regex compiler and optimizer need some love 11:47
FROGGS yeah 11:52
though, one needs a good profile to see what exactly needs love
brrt true enough 11:54
FROGGS I dunno if JSON::Tiny is too big or not for that task.... minimizing it might just make it unreal 11:55
I hope we will have some sort of feature freeze in Q4 this year, and only care about fixing bugs and speedup things 11:56
nwc10 I don't think that you can have a proper Christmas without snow
FROGGS umm, WAT? 11:57
nwc10 (or, less obtusely, I don't think it will be possible to make a release without some sort of observance of a freeze)
FROGGS ahh
yeah
:o=)
err
:o)
grondilu hum weird. Now that ?"0" is True I thought I could remove the C<'' ne> in rosettacode.org/wiki/Temperature_co...on#Perl_6, but it does not work as I'd expect. 12:10
moritz m: say so '0'
camelia rakudo-moar 82a181: OUTPUT«True␤»
moritz shouldn't that be False now? 12:11
psch star-m: say so '0'
camelia star-m 2013.03: OUTPUT«False␤»
moritz erm
yes
I'm confused
FROGGS ===CONFUSED!===
grondilu no it was concluded that it should be true
moritz right 12:12
grondilu github.com/perl6/specs/issues/87
moritz I just mis-remembered the direction of the change
high time for the weekend
grondilu but strangly when I enter 0 the loop stops
FROGGS grondilu: what type is it for the 0 case? 12:13
dalek kudo/nom: 5a38b21 | lizmat++ | src/core/ (2 files):
Make "say/note" look at .nl setting
 12:13
FROGGS m: my $i; while '0' && $i++ < 10 { }; say $i 12:14
camelia rakudo-moar 82a181: OUTPUT«(Any)␤»
grondilu it comes from prompt so it should be Str
grondilu try this: while my $ = prompt "value? " { say "ok" } 12:14
FROGGS m: my $i; while '0e0' && $i++ < 10 { }; say $i
camelia rakudo-moar 82a181: OUTPUT«11␤»
FROGGS :/
FROGGS m: say '0'.Bool 12:14
camelia rakudo-moar 82a181: OUTPUT«True␤»
FROGGS m: my $i; while '0'.Bool && $i++ < 10 { }; say $i
camelia rakudo-moar 82a181: OUTPUT«11␤»
FROGGS so, what does while there?
psch m: say '0'.Stringy.Bool
camelia rakudo-moar 82a181: OUTPUT«True␤»
FROGGS the while op does not seem to call .Bool it seems 12:15
m: say nqp::p6bool('0')
camelia rakudo-moar 82a181: OUTPUT«False␤»
FROGGS ahh
grondilu well spotted
awwaiid Is there a way to do ruby-style "DSL" blocks called like "myfun $p1, $p2 { ... }" (with no comma between the params and the block). Doing it with the comma is easy with just a last &thing param for myfun. Just wondering. 12:16
FROGGS m: say nqp::istrue_s('0')
camelia rakudo-moar 82a181: OUTPUT«===SORRY!===␤No registered operation handler for 'istrue_s'␤»
FROGGS hmmm, that's a MAST::Op
psch awwaiid: i think that's slang or macro territory 12:17
awwaiid I can do a nested sub and then call it as myfun($p1, $p2){ ... } . The lack of space between the ")" and the "{" seems really important
FROGGS MVM_string_get_grapheme_at_nocheck(tc, str, 0) == 48)
m: say '0'.ord
camelia rakudo-moar 82a181: OUTPUT«48␤»
psch j: my $i; while '0'.Bool && $i++ < 10 { }; say $i
camelia rakudo-jvm 82a181: OUTPUT«cannot connect to eval server: Connection refused␤»
psch grml
awwaiid I mean, the lack of space is important to p6, and makes me unhappy :) 12:18
psch .oO( nobody wants the unspace )
awwaiid ah, ok :)
awwaiid alright well I should go to $work anyway :) 12:19
psch anyway, nqp-j also special cases "0" in istrue_s and isfalse_s 12:20
masak awwaiid: didn't quite get your nested sub syntax point above.
awwaiid: `myfun($p1, $p2){ ... }` means "call the sub myfun with those two arguments, then hash-index the result with the expression ..." 12:21
awwaiid: and yes, in that case, the lack of whitespace is significant.
moritz masak: but why are we even trying to parse a generic expression in a declaration? 12:22
masak moritz: awwaiid said "and then call it"
moritz masak: oh 12:23
masak moritz: so I got the impression that he was talking about a callsite, not a declaration.
moritz m: sub f(){ say 42 }
camelia ( no output )
moritz right
moritz not an error in the declaration 12:23
moritz should go to bed right now
masak hugs moritz :)
moritz hugs back 12:24
masak wishes he had an image readily available from the Scandinavia and the World webcomic, of Sweden and Norway hugging :) 12:25
FROGGS labster: are you going to smoke the '0'.Bool change? if not, I'll do it 12:28
labster: because of the nqp::p6bool glitch we found just now
I already have a patch, but still need to spectest and ecosystemtest it... 12:29
[Coke] wonders why it's VAR(d) and not d.VAR 12:31
masak [Coke]: isn't it both?
FROGGS [Coke]: where?
I've only seen d.VAR so far fwiw
[Coke] TT was using VAR() in backscroll 12:43
VAR(d).WHAT looked weird to me. :) 12:44
moritz VAT
[Coke] ugh, I work in tax, that's a dirty word. :P 12:46
timotimo is AFK until the evening 12:50
dalek ecs/newio: 0cd0cde | (Carl Masak)++ | S16-io.pod:
rephrase slightly for clarity
 12:59
masak - eof file pointer reached end of file 13:01
+ eof file pointer reached end of information
masak "end of information" is not a concept I'm familiar with. 13:01
masak the change also breaks the nice implicit expansion/explanation of the acronym "eof" 13:01
yes, I know that not all of the things we reach the end of are files 13:02
but for better or for worse, "eof" means "end of file" even when it's not a file
moritz also, not all data carries information 13:03
masak it's like with `get`, we still say `get next line`. even though the record separator might have been altered to something other than "\n" and it's not actually a line.
FROGGS well, if it still states "file pointer" we can also say "end of file", no? 13:04
masak the total sum of all the possibilities is just too general to sensibly talk about.
all these notions, "file", "line", "end of file" end up being metasyntactic variables standing in for their general equivalents.
FROGGS aye 13:05
masak but we use the simple terms with a shared understanding with the reader that it's more general than that.
I'd much prefer if S16 did the same, perhaps with an early note that such is the case.
moritz
.oO( let's call it .eos, which means "end of stream", and when that's not general enough, it could mean "end of something" )
 13:06
masak :P
- opened is the file open?
+ opened is it not closed?
here, too, I see that the intent was a good one. but the result (IMO) is worse.
moritz masak: I think I agree 13:07
masak we go from clarifying a concept that's fairly clear already, to defining it in terms of its opposite, without clearly defining the opposite.
FROGGS yeah, add a disclaimer and then we stick to file/line 13:08
masak also, I'm reading this as "hasn't it been closed yet?" -- and I'm not 100% sure that's all it takes for something to be opened.
for example, if I just do "foo".IO, what I have is not something opened. 13:09
(I think)
so besides being less clear, the new explanation is actually wrong.
masak (by the way, I know writing these one-line blurbs about methods is quite hard. and easy to critique afterwards. I'll see if I can improve some of them myself.) 13:14
nbdsp Greetings! Could someone advise please the proper way to declare multiline strings with interpolation? Here-docs AFAIK don't interpolate? 13:16
psch nbdsp: heredocs can interpolate 13:16
nbdsp: if you start them with an interpolating quote construct
e.g. qq:to/END/ instead of q:to/END/
nbdsp ohh - qq! thanks! 13:17
masak nbdsp: also, ordinary strings can be multiline (but it might not look so nice)
nbdsp masak: Should ordinary strings be ended with quote on each line, or backslash should be used? 13:19
masak m: say "foo bar"
camelia rakudo-moar 5a38b2: OUTPUT«foo bar␤»
masak m: say "foo␤␤␤bar"
camelia rakudo-moar 5a38b2: OUTPUT«foo␤␤␤bar␤»
masak nbdsp: just newlines works fine.
nbdsp masak: thanks! 13:20
masak nbdsp: but (as opposed to heredocs), you don't get any indentation benefits.
FROGGS nbdsp: as masak said you can indent heredocs, and the common indentation level will be removed from the string, so to say 13:23
which makes the actually nice :o)
masak heredocs are awesome
arnsholt Indeed. I've used them for SQL code (in Perl 5, admittedly) recently 13:24
Very nice way to separate code and (verbose) data
PerlJam good morning #perl6!
FROGGS ours are more awesome then P5's though :o)
hi PerlJam
arnsholt Yes, they are =)
masak FROGGS: I used to think that Perl 6's improvements were just bloat/bells and whistles. but I'm completely on-board with them now, after having used heredocs for a while. 13:25
PerlJam The heredoc discussion makes me wonder if heredocs have supplanted formats for most of the simpler uses of formats.
arnsholt But writing my webapp in Perl 6 was a bit more of an adventure than I felt comfortable with ATM
FROGGS masak: yeah... it is very nice to see here that we taken given things, and put that one little extra on top that is for one very sane make makes the thing twice as powerful 13:26
like I love P6's enums
masak Perl 6 enums turned out really nice too.
it took ~3 tries, but we finally did it. 13:27
well, TimToady++ did it, mostly.
masak - unlink remove as file 13:27
+ unlink remove as local entity
PerlJam masak: Energy and persistence conquer all things. --Ben Franklin
masak what non-file case are we considering here that merits this change?
Ven masak: what were the 2 previous ones? 13:28
masak Ven: just iterations towards what we have now. they were decidedly less good.
moritz I don't remember much about them, but I do remember being frustrated by them 13:29
masak Ven: I remember at one point throwing a bunch of use cases towards TimToady, which (I think) made him go off and ponder, and come back with a new better model for enums.
Ven: it's all there in the spec commit log if you're really curious.
moritz however, I'm also frustrated when I try to hack at the current enum implementation, because I simply can't figure out how it all fits together
masak moritz: anything in particular?
moritz m: enum A <b c>; say A ~~ Enumeration 13:30
camelia rakudo-moar 5a38b2: OUTPUT«False␤»
moritz m: enum A <b c>; say b ~~ Enumeration
camelia rakudo-moar 5a38b2: OUTPUT«False␤»
moritz masak: what's that role for?
masak individual key/value pairs, it seems. 13:31
moritz so it's an EnumItem or so 13:32
masak right. 13:32
moritz but, isn't b in the example that?
PerlJam moritz: I'd go with "nothing" right now. It's not used anywhere that I can see
moritz or if not, how do I get one?
masak moritz: no, b is just one of the halves. the value, I'd say.
masak m: enum A <b c>; say A.enums 13:32
camelia rakudo-moar 5a38b2: OUTPUT«b => 0, c => 1␤»
masak m: enum A <b c>; say A.enums.WHAT
moritz m: enum A <b c>; say b.name
camelia rakudo-moar 5a38b2: OUTPUT«(Hash)␤»
rakudo-moar 5a38b2: OUTPUT«No such method 'name' for invocant of type 'A'␤ in block <unit> at /tmp/s13kdhDz0U:1␤␤»
masak m: enum A <b c>; say A.enums.list[0].WHAT 13:33
camelia rakudo-moar 5a38b2: OUTPUT«(Pair)␤»
masak hm.
masak I see that Actions.nqp goes about creating Enumeration objects and (I think) stuffing them into the enum thing it's creating. 13:33
but that doesn't tell us how to get them out. 13:34
moritz m: enum A <b c>; A ~~ Enum 13:34
camelia ( no output )
moritz m: enum A <b c>; say A ~~ Enum
camelia rakudo-moar 5a38b2: OUTPUT«False␤»
moritz masak: second mystery: what's Enum? 13:35
masak hehe
moritz cause class Enum { has $.key; has $.value }
masak waitaminute
moritz that *also* looks like an EnumItem to me
masak it does
something is rotten in the state of enums 13:36
PerlJam moritz++ maybe enums aren't quite as nice as masak thinks ;)
masak to be fair, none of the above has hindered me in *using* enums in nice ways ;)
but it's still an indication that something is... off
PerlJam yeah, same here
masak in the implementation
moritz so, we have two types that I can't figure out, but seem to represent a single enum item (k/v pair)
moritz oh 13:36
m: say Pair.^mro
camelia rakudo-moar 5a38b2: OUTPUT«(Pair) (Enum) (Any) (Mu)␤»
moritz right, Enum is simply an immutable Pair 13:37
the comment about role Enumeration says: # Method that we have on enumeration types
and it has methods pick and roll 13:38
m: enum A <b c>; say A.^roles
camelia rakudo-moar 5a38b2: OUTPUT«No such method 'roles' for invocant of type 'Perl6::Metamodel::EnumHOW'␤ in block <unit> at /tmp/4ovsaGqH8a:1␤␤»
masak when I become dictator, the act of using inheritance to create a new mutable type from an immutable one is going to be a punishable crime.
kinslayer I have a problem figuring out why this code fails lpaste.net/130484 13:39
moritz masak: what's your approach?
kinslayer it only fails at the last line.
masak m: role R { method foo { say "OH HAI" } }; enum A does R <b c>; b.foo
camelia rakudo-moar 5a38b2: OUTPUT«OH HAI␤»
masak moritz: at the very least, wrap and delegate. 13:40
moritz: that way, the new class is a "mutable wrapper" for the old one.
moritz kinslayer: what do you mean by "fail"?
m: role R { method foo { say "OH HAI" } }; enum A does R <b c>; say A ~~ R
camelia rakudo-moar 5a38b2: OUTPUT«False␤»
moritz oh
maybe it's just type-checking enums against roles that's broken 13:41
kinslayer moritz: that it does do what I want that is giving me a hash I can do stuff with.
masak moritz: without fail, immutability will have been part of the contract/invariants of the old class. inheriting and adding mutability means breaking Liskov.
moritz masak: yes, I know
kinslayer It just says that postcircumfix {} not defined for array
masak m: role R { method foo { say "OH HAI" } }; enum A does R <b c>; A.foo
camelia rakudo-moar 5a38b2: OUTPUT«OH HAI␤»
moritz kinslayer: no line number? 13:42
masak yeah, something is off there too.
how can A.foo work if A ~~ R is false?
kinslayer moritz 15 31
moritz lpaste.net/130484#line11 seems to be the problem 13:43
no dereferencing going on
either my @rooms := $json<rooms> or my @rooms = @( $json<rooms> )
kinslayer: also, %room-graph<"$i.north"> probably wants to be %room-graph{"$i.north"} 13:44
kinslayer yeah the last one I would fix by myself, just found the other syntax but the other things seems more natural 13:45
PerlJam kinslayer: you could also have used $rooms rather than @rooms 13:46
moritz PerlJam: @rooms is more idiomatic
kinslayer moritz so what does := do spefically just so I can remember.
moritz and the iterate with for @rooms.kv -> $idx, $oom { ... } 13:47
kinslayer: binding
doc.perl6.org/language/containers
kinslayer okay I might remember now that I now. :D
kinslayer now to test whether this works. 13:47
PerlJam I dunno. Part of the benefit of Perl6 was supposed to be that you didn't have to be concerned about arrays versus array references; either would Just Work.
kinslayer now to test whether this works. 13:49
sorry wrong window.
so now it works at least on a small scale. 13:51
lucasb While reading the source, I thought about decreasing the indentation level of reify methods in GatherIter, HashIter, ListIter, MapIter and native_array. Unfortunately, the whitespace changes are messing with the diff output. :( 14:15
Maybe it is not worth it, but if you want to take a look, it is here:
gist.github.com/lucasbuchala/28795...ed951663fa
moritz lucasb: those methods are *carefully* optimized, and explicit returns are kinda slow 14:18
(because return is just a code object to be called)
lucasb: so don't change anything unless you have also carefully benchmarked them 14:19
dalek pan style="color: #395be5">perl6-examples: d326a14 | paultcochrane++ | lib/ (26 files):
Purge old library code and tests

After discussing this code with mberends++ at NLPW2015 it was decided that it was no longer current and could be removed.		 14:22
pan style="color: #395be5">perl6-examples: 8fe55f9 | paultcochrane++ | t/categories/99-problems.t:
[99-problems] Correct test filename after rename
pan style="color: #395be5">perl6-examples: 5171ca4 | paultcochrane++ | categories/shootout/fasta.p6.pl:
Document the FASTA shootout example
pan style="color: #395be5">perl6-examples: 120c653 | paultcochrane++ | categories/shootout/revcomp.p6-v2.pl:
Add author to revcomp-v2 example
pan style="color: #395be5">perl6-examples: e8dfb31 | paultcochrane++ | categories/shootout/n-body.p6-v2.pl:
Add author info for nbody-v2
pan style="color: #395be5">perl6-examples: d0ad0ef | paultcochrane++ | categories/shootout/ (14 files):
[shootout] simplify shootout example filenames
pan style="color: #395be5">perl6-examples: aa6eb51 | paultcochrane++ | lib/Pod/Htmlify.pm6:
Match .pl/.p6 at the end of the filename
lizmat hmmm... [ptc]++ 14:22
lizmat hmmm... seems that my spec change of line-input-separator to nl has struck back 14:23
lizmat .nl doesn't say whether it is input or output 14:24
my last patch makes uses it for output
but what if you want different separators for input and output ?
lucasb I thought that maybe an explicit return could be less costly than having an if block, but I didn't think about optimization really, just code layout. Thanks for the tip, moritz++ 14:25
[ptc]_ moritz: I'd like to reserve "examples.perl6.org" if I may (just so you know, you don't have to do anything at this stage) 14:26
moritz: btw: should I set up the examples.perl6.org site for you? Is there any way I can help reduce the admin load?
moritz [ptc]_: yes, do it :-) 14:27
[ptc]_: do you have access to www.p6c.org?
[ptc]_ no, was just about to ask 14:28
I'll make the examples.perl6.org user and then set up an rsync in much the same way doc.perl6.org works
only have access to hack
moritz [ptc]_: +1 14:29
lizmat m: sub a { 42 }; for ^1000000 { a }; say now - INIT now # lucasb
camelia rakudo-moar 5a38b2: OUTPUT«0.2423389␤»
lizmat m: sub a { return 42 }; for ^1000000 { a }; say now - INIT now # lucasb
camelia rakudo-moar 5a38b2: OUTPUT«1.3662083␤»
moritz [ptc]_: and skim github.com/perl6/infrastructure-do...6c.org.pod if you haven't already
(and update it to add yourself to the sudo list) 14:30
lucasb lizmat: Thanks for the return tip. I hope this thing can get optimized in the future. 14:39
lizmat it's on the radar, afaik 14:40
[ptc]_ moritz: ok, thanks for the link 14:52
masak am I missing something, or do the synopses not mention heredoc semantics at all? 15:01
masak (I mean the thing where the string from the heredoc gets de-indented to the level of the terminator) 15:02
lizmat S02:4468 15:04
masak thank you.
lizmat "Leading whitespace equivalent to the indentation of the delimiter will be removed from all preceding lines."
masak that's exactly what I was looking for. some very specific type of blindness made me miss it, apparently. 15:05
lizmat we've all been there :-) 15:06
lizmat m: say +"" # didn't this used to warn / should it not warn ? 15:15
camelia rakudo-moar 5a38b2: OUTPUT«0␤»
masak it should most definitely, without any hesitation or doubt, warn. 15:17
"" is not a number
[Coke] m: say ?+"0" == ?"0"
camelia rakudo-moar 5a38b2: OUTPUT«False␤»
[Coke] whee.
masak I'm fine with that one :) 15:18
[Coke] m: :m 15:19
camelia ( no output )
[Coke] m: say.say :m
camelia rakudo-moar 5a38b2: OUTPUT«5===SORRY!5===␤Argument to "say" seems to be malformed␤at /tmp/MzvwgzT6Wi:1␤------> 3say7⏏5.say :m␤Other potential difficulties:␤ Unsupported use of bare "say"; in Perl 6 please use .say if you meant $_, or use an explicit invocant o…»
awwaiid m: sub wrap($p1){ sub with(&f) { say "$p1 wrap"; &f() }}; wrap("hello"){ say "yep" } 15:29
camelia rakudo-moar 5a38b2: OUTPUT«yep␤Unhandled exception: postcircumfix:<{ }> not defined for type Sub␤ at <unknown>:1 (/home/camelia/rakudo-inst-2/share/perl6/runtime/CORE.setting.moarvm:throw:4294967295)␤ from src/gen/m-CORE.setting:15991 (/home/camelia/rakudo-inst-2/share/per…»
awwaiid ohhh.... I see. "works" in the REPL but not in -e. Interesting. 15:35
masak psch: even with the current '*' continuation on the REPL, heredocs still don't work :( 15:41
which means it's annoying *and* not useful enough
masak (I'm not saying I think heredocs should work on the REPL. but with the '*' continuation in place by default, I kind of expected them to) 15:42
raydiak morning o/ 15:48
n0vacane good morning sir and/or madam 15:51
masak good morning radiant raydiak
may your bright light shine on the antipode 15:52
raydiak wow thanks masak :) hello n0vacane
masak: a few days ago, I decided to see what interesting thing I could do in the span of time you were asleep...a lot of mental circles and 80 lines later I had the cheesiest cheatingest self-compiling compiler ever, which targets P6 16:02
btyler raydiak: github link? :D 16:03
raydiak btyler: I have a habit of not putting things out there that will embarass me more than I do already :) but gimme a sec I'll gist it 16:04
raydiak gist.github.com/raydiak/b6497a9641a6e53ce831 16:08
so many cheatings I can't even tell you...but I just wanted to see what the result would look like and what ideas the exercise gave me
Ven raydiak: lol. that's pretty amazing 16:09
raydiak heh thanks :) it'll be interesting to see how a real version turns out some day 16:12
masak raydiak: wow 16:16
raydiak if it had a little more structure and a little less brain-damage, it'd be simple to get it to target numerous languages...of course that's because it doesn't do much of anything else, but still... 16:17
raydiak one of the things I found interesting to think about is the way you have to develop in "generations" once you throw away the bootstrap implementation...e.g. I can add any feature I want just by writing it in itself, but you have to add the feature using existing features, then recompile before you can actually make use of that feature in the *following* iteration 16:34
which totally makes sense, but nevertheless feels like walking down a hall of mirrors... 16:35
masak raydiak: at this point you would do yourself a great favor by googling "reflections on trusting trust ken thompson" 16:36
it's a not-too-long read, but very effective.
raydiak: also, I guess, "I am a strange loop" by Hofstadter comes warmly recommended :) 16:37
raydiak masak: thanks! reading the first one now...the other one sounds familiar, I think I have it on a list somewhere...maybe even one of the list items I got from you :) 16:39
psch masak: actually, heredoc is to me the strongest argument yet against a multiline REPL in core 16:41
maybe inline documentation is another 16:42
'cause it feels a bit too special-cased to me to react to a q:to with a continuation prompt even if the line ended on ; 16:43
ugexe is precompiling without reinvoking $*EXECUTABLE_NAME --target for each module something that could done? or is there a reason why it should be that way? i'm wondering if its worthwhile to add such functionality to reduce the time required to install a bunch of modules under JVM (which I assume is getting the startup penalty each time .precomp is called) 16:47
masak psch: note how that problem, too, completely goes away if Enter is restored to meaning "execute" and something like Shift+Enter means "one more line" 17:25
fewer heuristics, fewer special cases, more predictability.
raydiak: heh, I don't think I've recommended you (or anyone) "trusting trust" before. though it is a classic. 17:26
TimToady timotimo: 'my sub' is redundant, since subs are already 'my' by default
yoleaux 08:55Z <jnthn> TimToady: Not immediately, no... :(
TimToady oh, jnthn++ pointed it out already 17:31
psch masak: «my $s = q:to/END/;» and then enter can never compile succesfully then, and always needs shift+enter 17:34
which still strikes me as inconsistent 17:35
masak not me. the Enter keystroke under my model signifies "what I have here is a program unto itself" 17:37
psch ah 17:37
masak not including the heredoc terminator will fail under that model, becuase it *should* fail
psch i had a misconception then, i realize 17:38
eli-se I want to make a video game.
masak again, this is just basically copying Chrome Devtools
TimToady imo, shift+enter just puts the blame onto the user when things go wrong; a good parser should always know when it has all its ducks in a row 17:39
and a heredoc is just another duck that can get out of line 17:40
masak I can buy that argument, too. it's consistent.
there's always a well-defined, detectable "end-of-statement" point
but both TimToady's semantics and mine are better than the current semicolon/double-Enter semantics that's currently in place 17:41
either one would be an improvement
mine has prior art :)
TimToady so does mine, if you don't count parser complexity :)
psch fwiw, i've been trying to improve it during the last week, it's just hard :P 17:42
i've definitely merged too early, sorry about that 17:43
masak no, it's good
gives us something concrete to discuss
things are allowed to get worse for a bit before they get better
on the Internet, if you want to arrive at the correct answer, you should start by claiming something outrageous and disprovable :) 17:44
TimToady would have fun working on that, if he didn't have eⁿ things on his plate already
psch well, the current implementation cheats (and breaks) for <blockoid> 17:48
the next implementation i could arrive at never meets the closing bracket
which confuses me to no end, and points at some oversight
TimToady how are you counting "ducks in a row" currently? 17:49
TimToady would probably try a dynvar first, since those automatically clear at the appropriate moment
psch TimToady: i wanted to try checking !highexpect next, because i evidently need something to know if the parser can stop
TimToady thinks relying on highexpect would be fraught 17:50
psch TimToady: currently there's a $*MOREINPUT_BLOCK_DEPTH, but that's not granular enough
(i'm not sure granular is the right word there...)
in any case, blocks aren't the only thing left to expect, but many other things seem to work out fine on their own 17:51
e.g. «1 +\n1» works with just calling readlineintfh in _ws
(basically)
the different ways to signify end-of-statement aren't quite transparent to me yet either 17:52
e.g. the first <?before> in eat_terminator doesn't actually *eat* the terminator 17:53
and it can't because the last statement in a block doesn't need a ;
TimToady well, inside a block you'll never want to eval anyway 17:54
dalek osystem: 72f9a4a | tony-o++ | META.list:
Extendable JS style 'Events'

Can use regex, callable, or whatever to pick up events. Can use supplies or channels out of the box. Neat.
TimToady the final decision probably needs to be in the reduction of <statement> 17:55
though I don't know offhand whether that includes the heredoc 17:56
ah well, now here I am working on it; told you it'd be fun for me :) 17:57
back to backlogging...
raydiak masak: trusting trust definitely brings up one or two things I hadn't considered...apparently your recommendation rattling around in my memory was "books by Hofstadter" a couple months ago
psch is bound to learn loads about CURSOR it seems 17:59
dinner &
[Coke] m: 3.123456789 ~~ 3.123456788 18:02
camelia ( no output )
[Coke] m: say 3.123456789 ~~ 3.123456788 18:02
camelia rakudo-moar 5a38b2: OUTPUT«False␤»
masak raydiak: well, now that you've been delighted by Thompson, maybe Hofstadter is next ;) 18:04
TimToady so, is there a backdoor hidden in our bootstrap yet?
masak ssssssh!
I mean, um, no. of course not.
masak glances around furtively 18:05
cschwenz is there a perl6 equivalent of starman? 18:06
TimToady no, we only do et and independenceday 18:07
cschwenz :-P 18:08
cschwenz okay, so what is needed to make something in perl6 approximating the functionality of starman? 18:09
colomon “Starman? Why the cute name instead of more descriptive namespace? Are you on drugs?” Ah, CPAN. 18:10
tadzik :)
NO FUN ALLOWED
raydiak masak: sounds like it...though I'd s/delighted/more appropriately terrified and interested/ :)
colomon hopes tadzik can answer cschwenz’s question
TimToady the first N pages of Google have nothing about a starman 18:11
except mostly the movie
cschwenz metacpan.org/pod/Starman 18:11
raydiak it's a psgi thing iirc
TimToady ah
colomon TimToady: I got search.cpan.org/~miyagawa/Starman-0...Starman.pm on the first page of my Google search
TimToady hmm, personalization strikes again... 18:12
colomon Very last thing on the first page.
cschwenz or, first, is there a perl6 webserver?
colomon TimToady: you must really love the movie, eh? ;)
TimToady ah, there it is, I just dintseeyit
TimToady I thought someone did something P6ish with PSGI... 18:13
or PSGIish with P6...
tony-o timotimo: that IO::Socket::INET gist i posted seems to work now, do you know of any changes 18:14
TimToady modules.perl6.org lists 3 modules that mention PSGI 18:14
cschwenz Hmm, okay. I'll have a look there. :-) 18:15
Is modules.perl6.org the extent of CPAN for Perl6? 18:16
raydiak yep there are some http servers and psgi things listed there, though no idea what the state of any of them is precisely atm
yes 18:17
cschwenz okay thanks! :-)
raydiak holy ssh lag batman
you're welcome :)
TimToady blames the microwave oven
our new one is very leaky at 2.4Ghz 18:18
maybe we should wrap it with tin foil...
raydiak well it does sit about 5 feet away, but that's how I know it's not running
TimToady neighbors?
oyse I have a problem building Rakudo for the JVM. I get jvm::runtime.jars value not available from <my path here>/rakudo/install/bin/nqp-j --show-config when doing perl Configure.pl --gen-nqp --backends=jvm 18:19
raydiak roommates more likely...every time they walk around in the wrong places or play their MMOs it gets sluggish :)
oyse Is this a know problem?
TimToady where we live, sometimes it's the president flying into moffat jamming everything, and sometimes its someone arc-welding in their garage...
oyse I am on Ubuntu 14.04 with openjdk-7 and gcc installed 18:20
raydiak imagines his stimulant-addictied neighboars playing with arc welders
timotimo is back on-line 18:21
tony-o: i don't know of any changes :(
raydiak wb timo o/ 18:22
timotimo o/
tony-o timotimo: really strange, i can curl from that HTTP::Server::Threaded without any issues now
TimToady oyse: on Ubuntu I succeed in building both backends with: /usr/bin/perl Configure.pl --backends=ALL --gen-nqp --gen-moar
timotimo super weird.
tony-o timotimo: nevermind, i'm silly
timotimo so it's still b0rken? :( 18:23
tony-o yea
TimToady oyse: are you using the latest git version?
that came out wrong
are you cloning the nom branch, or using some brewer?
tadzik cschwenz, colomon we don't have any good webservers afaik :( 18:24
tony-o cschwenz: i'm working on a couple but have some bugs
oyse timtoady: I just started with the instructions from INSTALL.txt on Github: github.com/rakudo/rakudo/blob/nom/INSTALL.txt 18:25
timtoady: I am trying the command you suggested now
FROGGS oyse: what's your exact javac version? 18:26
--> javac -version
oyse FROGGS: javac 1.7.0_75
FROGGS that's new enough
oyse timtoady: The --backends=ALL when worse than the other command: Perl API version v5.14.0 of Encode does not match v5.18.0 at /usr/share/perl/5.18/XSLoader.pm line 92. 18:27
TimToady I'm using that one, as well as javac 1.7.0_51
ugexe modules are failing to install due to 'No such method print-nl' on a fresh build
FROGGS oyse: what rakudo version do you try to build? 18:28
oyse: because just recently I applied a build fix for jvm...
TimToady that sounds like perl5 versions fighting each other
oyse FROGGS: I just started with a git clone git://github.com/rakudo/rakudo.git
TimToady try just perl instead of /usr/bin/perl 18:29
oyse git branch says I am on nom
FROGGS oyse: hmmm, that really should do... my patch is six days old
TimToady you probably have a newer perl than the system perl in your path
dalek rl6-roast-data: 25845ff | coke++ | / (9 files):
today (automated commit)
FROGGS oyse: that's the patch btw: github.com/perl6/nqp/commit/52b834...7f73210fc3 18:30
oyse: it touches the jar's you've got problems with
[Coke] raydiak - are you killing hack?
raydiak [Coke]: running a rakudobrew build 18:31
oyse TimToady: quite possible I have installed something a while back. I will check
masak lol, I blogged! strangelyconsistent.org/blog/here-be-heredocs
raydiak [Coke]: nothing important, can kill it if you need the resources for something meaningful :)
[Coke] running an awful lot of java processes. 18:32
raydiak just a single rakudobrew build
just finisheds
s/s$//
FROGGS oyse: if nothing helps, please nopaste your entire build output 18:34
raydiak [Coke]: if you're still seeing sluggishness, I see a stuck module build maxing out a core and taking 8 gigs+ of ram with over 4 cpu hours under its belt so far :) 18:37
dalek c: a2b0e77 | moritz++ | lib/Type/Str.pod:
Str.starts-with
 18:40
ugexe m: class IO::Capture::Simple { sub capture_stderr(Callable $code) is export { my $result; my $*ERR = class { method print(*@args) { $result ~= @args.join; }; }; $code.(); $result; }; }; import IO::Capture::Simple; capture_stderr { note "blah" } 18:41
camelia rakudo-moar 5a38b2: OUTPUT«No such method 'print-nl' for invocant of type '<anon>'␤ in sub note at src/gen/m-CORE.setting:18840␤ in block <unit> at /tmp/guUOFRYET7:1␤␤»
oyse timtoady: --backends=ALL worked better after I had delete some old stuff in the perl path. At least I could start make now :) 18:45
raydiak m: class IO::Capture::Simple { sub capture_stderr(Callable $code) is export { my $result; my $*ERR = class { method print(*@args) { $result ~= @args.join; }; method print-nl(*@args) { $result ~= @args.join ~ "\n" }; }; $code.(); $result; }; }; import IO::Capture::Simple; capture_stderr { note "blah" } # ugexe 18:46
camelia ( no output )
raydiak m: class IO::Capture::Simple { sub capture_stderr(Callable $code) is export { my $result; my $*ERR = class { method print(*@args) { $result ~= @args.join; }; method print-nl(*@args) { $result ~= @args.join ~ "\n" }; }; $code.(); $result; }; }; import IO::Capture::Simple; say capture_stderr { note "blah" } 18:47
camelia rakudo-moar 5a38b2: OUTPUT«blah␤␤»
PerlJam S99:WIP 18:48
synbot6 Link: design.perl6.org/S99.html#WIP
PerlJam S99:Weekly_Changes 18:49
synbot6 Link: design.perl6.org/S99.html#Weekly_Changes
PerlJam S02:2345
synbot6 Link: design.perl6.org/S02.html#line_2345
ugexe raydiak: is that because STDOUT/ERR/etc use print-nl now? 18:50
raydiak ugexe: yeah looking at it now, it changed just hours ago...and that isn't what print-nl should do (my example is wrong I mean) 18:51
ugexe: github.com/rakudo/rakudo/commit/5a...65c12dc6c1
oyse FROGGS: seems like the JVM build worked with --backends=ALL. At least I have a perl6-j now. I assume that is the jvm version
FROGGS oyse: aye 18:52
oyse: also check if you got a perl6-m
that's the MoarVM version which starts up way faster 18:53
oyse FROGGS: it is there
yes, it seems more responsive
In the JVM version is there an API into the VM for doing introspection from Java? Could I for instance ask for all class, their methods etc?
FROGGS you'd have to ask psch++ for JVM interop 18:54
but in Perl 6 land it would be:
m: say Int.^methods 18:55
camelia rakudo-moar 5a38b2: OUTPUT«Int Num Rat FatRat abs Bridge chr sqrt base polymod expmod is-prime floor ceiling round lsb msb narrow Range sign conj rand sin asin cos acos tan atan atan2 sec asec cosec acosec cotan acotan sinh asinh cosh acosh tanh atanh sech asech cosech acosech cotan…»
FROGGS or:
m: say Int.HOW.methods(Int)
camelia rakudo-moar 5a38b2: OUTPUT«Int Num Rat FatRat abs Bridge chr sqrt base polymod expmod is-prime floor ceiling round lsb msb narrow Range sign conj rand sin asin cos acos tan atan atan2 sec asec cosec acosec cotan acotan sinh asinh cosh acosh tanh atanh sech asech cosech acosech cotan…»
ugexe raydiak: i see. yeah s/print/print-nl/ in the actual IO::Capture::Simple module then complains about method print
FROGGS oyse: perhaps you can call the latter from Java somehow
raydiak ugexe: yeah looks like print-nl is supposed to literally just print a newline 18:56
oyse FROGGS: that seems like a good starting point at least. I have to poke around I bit I guess. 18:57
When are the Java people usually here?
psch oyse: check src/vm/jvm/runtime/org/perl6/rakudo/RakudoEvalServer.java
oyse: and ./perl6-eval-server
both of those relative to the git root 18:58
FROGGS oyse: here is the opposite way as an example: github.com/rakudo/rakudo/blob/nom/...-interop.t
cdc Hello #perl6 18:59
raydiak m: class IO::Capture::Simple { sub capture_stderr(Callable $code) is export { my $result; temp $*ERR = $*ERR but role { method print(*@args) { $result ~= @args.join; }; }; $code.(); $result; }; }; import IO::Capture::Simple; say capture_stderr { note "blah" } # ugexe you could do something like this too if you wanted
camelia rakudo-moar 5a38b2: OUTPUT«␤blah␤»
psch o/ cdc
cdc psch: \o
raydiak hello cdc 18:59
psch cdc: did you get somewhere with that typed exception of your? i have to admit i forgot what exactly it was about...
*yours
TimToady ⌽ <-- samurai about to attack 19:00
tony-o looks like the info symbol to me
i with a circle
TimToady fontfeyl 19:01
.u ⌽
yoleaux U+233D APL FUNCTIONAL SYMBOL CIRCLE STILE [So] (⌽)
cdc psch: No, I failed to get a correct patch. That was about getting typed execeptions for yada operators, like X::CodeStub::Executed
masak looks like "this is as high as the waterline gets" on a ship that's been tilted 90 degrees
raydiak looks like phi to me but I suggest dodging instead of examining when a samurai is about to attack :)
labster FROGGS: go ahead, I was very asleep when you asked about smoketesting (and still backlogging) 19:02
TimToady but the "don't be there" defense only works after he's committed to where "there" is 19:03
FROGGS labster: k
oyse psch, FROGGS: thanks!
ugexe raydiak: i went with method print-nl { nextsame };
TimToady or she, if we're in the universe that admits female samurai...
cdc FROGGS (or anyone having commit bit on panda), please could you take a look at this patch (I need it to get Panda working @work): github.com/tadzik/panda/pull/147 19:05
FROGGS cdc: you do we get our hands at MIME::Base64 there? 19:07
psch cdc: do you want guidance wrt X::CodeStub::Executed? (although i'd prefer a shorter name)
raydiak ugexe: nice 19:09
FROGGS cdc: beside the problem with MIME::Base64 I like it
cdc FROGGS: sorry I don't understand. What's the problem with MIME::Base64? 19:11
ugexe how do you install it before panda 19:12
and, at that point, you may as well pull in http::useragent and/or uri
raydiak TimToady: can I hold my action for a dodge until after he's committed, then? or maybe just dodge with every action until another party member handles him/her? :) 19:13
raydiak usually ends up laying in a pool of his own hit points (all three of them) 19:14
cdc ugexe: right, I installed it with a previous version of panda + a trick to bypass proxy authentication. Do I have to borrow code from the module/rosettacode instead? 19:15
TimToady raydiak: that depends totally on whether the sword is an it, or a him/her... 19:15
cdc psch: with pleasure! Maybe next week? 19:16
psch: OK for a shorter name. 19:17
psch cdc: sure, i'll be around at times like now during the next week
cdc: ISTR the first name idea was "X::Stub", but i might be making that up 19:18
cdc FROGGS, ugexe: is it possible to use the bootstrap mechanism to get MIME::Base64? 19:20
FROGGS cdc: we'd have to put it in panda/ext
ugexe and you'd probably want to make it optional
FROGGS hmmm, I guess it is fine to put it in there 19:21
I'd like to know what tadzik thinks about that though
cdc no problem 19:25
FROGGS hmmm, the p6bool change breaks the rakudo build :o(
TimToady we have a file named 0?!?
masak who's "we"? the rakudo repo? no. 19:27
ugexe wouldnt the first module that gets built be installed to site as 0? 19:27
FROGGS TimToady: we have, but that's not my current problem... 19:28
ahh, I can just dump a backtrace when p6bool gets a "0" 19:30
tadzik cdc: looking 19:48
oh, hmm
we seem to desperately need HTTP::Tiny of sorts 19:50
everyone is reimplementing this
FROGGS I'd prefer HTTP::UserAgent though 19:51
tadzik we have one
it's very dependency-heavy tohugh
which is the reason why I didn't pull it into panda
FROGGS ohh yeah :/ 19:51
tadzik anyway, if it helps you cdc I'm all for it :)
RabidGravy only URI and Mime::Base64 19:52
TimToady r: gather loop { take my $foo; last; }
camelia ( no output )
..rakudo-jvm 5a38b2: OUTPUT«cannot connect to eval server: Connection refused␤»
FROGGS let's just pull in MIME::Base64 for now
RabidGravy isn't it?
FROGGS and yeah, perhaps URI
though, that will kill the jvm, right?
ahh no, I fixed it :o) 19:53
TimToady masak: ^^^ there's a minimal test case to blow up perl6-j
tadzik cdc: your patch seems to use MIME::Base64 but it's not 'use'-ing it
oh, it is
but it is now a dependency of panda, can you make it 'require' so it's optional?
FROGGS try require* 19:54
TimToady masak: (that produces the error at gist.github.com/anonymous/9dc73118674e46248155)
FROGGS and then check for ::('MIME::Base64') being true or so
masak submits rakudobug 19:55
FROGGS masak: hold on
might be the same as rt.perl.org/Ticket/Display.html?id=122134
TimToady it might at that 19:56
otoh, if I remove the gather, it passes
so maybe not 19:57
masak TimToady: re "why do people keep thinkingI'm not in favor of this change" -- I like how many people are surprised that Perl 6 still keeps the hash key auto-quoting syntax. it's almost as if they'd expect the higher standard set by Perl 6 to somehow exclude the sinning committed by using "barewords" in hash key position. 19:58
not sure why one sin reminded me of the other. anyway, I'm glad you're on board with the rest of the community on ?"0" ;) 19:59
TimToady say wot? we don't privilege hash keys anymore...pair keys maybe... 20:00
masak meant pair keys.
I just keep thinking of them as hash keys.
right, foo => 42, not .{foo}
ooh, maybe it's because one has to do .<foo> nowadays that people are surprised foo => 42 still works 20:01
maybe they connect one with the other, subconsciously
TimToady
.oO( <foo=> 42 )
 20:01
masak aughr
TimToady :D
masak -me .oO( this is the guy we put in charge!? ) :P 20:02
masak .oO( this is the guy we put in charge!? ) :P
masak /msg fail 20:02
er, /me fail 20:03
TimToady
.oO( the light brigade is in charge )
 20:04
masak TimToady: #124279
synbot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=124279
TimToady excellent! 20:05
gah, got so interested in minimizing that test case that I minimized my lunch as well...
masak that happens to me more often than it should 20:06
dalek p: 4409a04 | FROGGS++ | src/HLL/sprintf.nqp:
check string explicitly for not containing zero

  ...since "0" if now true.
p: d7b8b06 | FROGGS++ | src/how/NQPClassHOW.nqp:
pass integer 0 instead of string "0" along to mean falseness
masak wow, so there's nqp fallout for making "0" true? FROGGS++ for fixing it! 20:07
masak reviews those places
FROGGS module Digest suffers from "0" change
masak: aye
and about 15 spectests... 20:08
TimToady I suppose many of the failures are inadvertent Str/Int confusion
masak ok, first one. I don't have much sympathy for a variable $int that contains a string :)
second one looks like it wanted to be (0) all along 20:09
FROGGS TimToady: aye
masak ok, good.
masak in other words, we didn't ruin someone's day. we just taught them to be a bit more careful with types. 20:09
I think that's a win both short-term and long-term
dalek kudo/nom: 0b20921 | FROGGS++ | src/Perl6/Grammar.nqp:
check for "1" to mean truth in method can_meta
 20:10
TimToady we'll make a pythonista of you yet
masak Perl is about stealing the best from other languages. some of those good things just happen to have been in Python first :) 20:11
we're not above stealing them anyway 20:14
eli-se can we make an exception for Ruby? 20:17
PerlJam what sort of exception?
masak I'm not sure Ruby is distinct enough from Perl to be worth stealing from ;) 20:18
PerlJam we already stole monkey patching from ruby ;) 20:19
TimToady we already stole the Ruby sigils and turned them into twigils 20:20
eli-se Sigils are good. 20:21
TimToady other than that, hmm....
TimToady (that is, we stole the idea of including "weird scoping" in our mandatory hungarian notation) 20:22
PerlJam TimToady: we already stole "stealing ideas from Perl" from ruby :)
masak I think we should top rather than steal outright the way Ruby can do cute DSLs.
heh, Ruby was not alone in stealing ideas from Perl :)
eli-se I steal ideas from Perl! 20:23
sjn One might consider renaming the language to "Thief" :-P
avuserow Hi all. I was just chatting with a coworker who was amused about the various ways to spell elsif/elif/elseif, and I was wondering if anyone knew why Perl went with elsif initially 20:24
TimToady we even stole P5 regular expressions, just like everyone else
though they're kinda bolted on the side... 20:25
PerlJam no, we stole them better!
itz nlpw++ # good perl6 presence 20:32
nbdsp Greetings! Could someone advise please the proper way to specify ranges in Str? For example in Python "abcdefghij"[3:-3] returns "defg". How to to it in Perl 6? 20:33
masak m: say "abcdefghij".comb[3..*-4].join 20:34
camelia rakudo-moar 5a38b2: OUTPUT«defg␤»
masak m: say "abcdefghij".comb[3 ..^ *-3].join
camelia rakudo-moar 5a38b2: OUTPUT«defg␤»
masak closenuff
dalek kudo-star-daily: 63912bd | coke++ | log/ (2 files):
today (automated commit)
 20:35
nbdsp masak: thanks! 20:36
flussence m: say 'abcdefghij'.substr(3, *-3);
camelia rakudo-moar 5a38b2: OUTPUT«defg␤»
psch m: say "abcdefghij".substr(3, *-3) 20:37
camelia rakudo-moar 5a38b2: OUTPUT«defg␤»
psch flussence++
masak oh, right 20:38
nbdsp psch: that's more neat! 20:39
psch although my braino surprised me there 20:40
m: say "abcdefghij".substr(*+3, *-3) 20:41
camelia rakudo-moar 5a38b2: OUTPUT«Failure.new(exception => X::OutOfRange.new(what => ("Start", "argument", "to", "substr"), got => 13, range => "0..10", comment => "use *13 if you want to index relative to the end"))␤»
psch but maybe that's actually sensible, cause substr doesn't know it could throw there
and say just .gists
[Coke] m: say 0.deadbeefp2;
camelia rakudo-moar 5a38b2: OUTPUT«No such method 'deadbeefp2' for invocant of type 'Int'␤ in block <unit> at /tmp/2GIiY2QMD0:1␤␤»
masak "use *13"?
psch masak: ... i didn't even see that 20:42
nwc10 why do I see 9 spectest failures? eg
$ ./perl6-m -Ilib t/spec/integration/advent2012-day23.t
1..2
No such method 'print-nl' for invocant of type '<anon>' in sub say at src/gen/m-CORE.setting:18817
psch m: say "abcdefghij".substr(*13, *-3) 20:43
camelia rakudo-moar 5a38b2: OUTPUT«5===SORRY!5=== Error while compiling /tmp/D5DwX5BY3B␤Unable to parse expression in argument list; couldn't find final ')' ␤at /tmp/D5DwX5BY3B:1␤------> 3say "abcdefghij".substr(*7⏏0513, *-3)␤ expecting any of:␤ infix␤ …»
[Coke] 9 failing -files-, causing 31 failures across the board. (jvm, moar, moar-jit). jvm has an additional few failures. 20:45
flussence it looks like say no longer simply calls $*OUT.say? Did whoever made that change document the "new and improved" way to redirect $*OUT? I kinda need that functionality too. 20:51
nwc10 flussence: I think that it's commit 5a38b210988e90f3b5053f3e3c5aaf65c12dc6c1 20:59
(not checked this)
and the fun seems to be because say now assumes that $out.print-nl exists and is usable
and that's not the case in what (appear to be) (now) partial overrides
and it's not clear to me if the more elegant solution is to declare the overrides wrong (and change them) or make say a bit more robust and default to a "\n" if $out.print-nl is not found 21:00
flussence I guess I'll just stop hand-coding it and switch my code to IO::Capture::Simple. Once *that's* fixed... 21:02
masak nwc10: I think it's wrong to assume that a new method should suddenly be there, and die if it isn't. 21:05
[ptc] masak++ # heredoc blog post 21:06
dalek p: 7238985 | FROGGS++ | src/HLL/sprintf.nqp:
track more int/str confusion when checking truthness in sprintf
caymanboy is it true that P6 will make me more well endowed? 21:07
TimToady is your name Stanford or Harvard? 21:09
TimToady the proper way to require a set of methods is to put them in a role 21:10
FROGGS I need to fix istrue in parrot :o( 21:11
vendethiel
.oO( or do you? )
lucasb In rakudo's source, is there any rule for knowing when to use P6 methods/ops and when to use nqp ops directly? Is one preferred over another? 21:12
flussence masak, can you please add a "background: #fff" somewhere on your site so I can continue using unreasonable default fg/bg colours? :)
masak flussence: oh, it doesn't have that? will fix, immediately. 21:13
masak flussence: fixed. might require you to do a hard reload on main.css 21:16
flussence: thanks for reporting
flussence much better, thank you :) 21:17
masak++
masak \o/
dalek pan style="color: #395be5">perl6-examples: c2bc065 | paultcochrane++ | util/ (2 files):
Add scripts to sync examples.perl6.org with repo build
 21:44
p: d2d27cc | FROGGS++ | / (6 files):
tweak istrue_s on jvm, every non-empty string is true now

We also bump MoarVM to get the very same fix in. For Parrot we have to tweak Parrot_str_boolean, that means pull requesting the change and bumping its version.		 21:45
kudo/nom: fd3a7a8 | FROGGS++ | tools/build/NQP_REVISION:
bump nqp version, to get ?"0" fix for moar/jvm
 21:46
pan style="color: #395be5">perl6-examples: 70de7ff | paultcochrane++ | util/update-and-sync:
Use "make html" instead of htmlify
 21:47
pan style="color: #395be5">perl6-examples: 2feb1c8 | paultcochrane++ | util/sync-build-log:
Add the missing sync-build-log script
masak a small report. 21:48
irclog.perlgeek.de/perl6/2014-10-13#i_9500806
dalek href="https://modules.perl6.org:">modules.perl6.org: a57bc71 | (Anthony Parsons)++ | web/index.tmpl:
Ensure text contrast with odd browser colours

  (and add some minor polish on nearby CSS lines)
masak apparently, back in October, it took 12 and a half seconds to run my blogging software.
masak nowadays, it takes around 10 seconds. 21:49
flussence is that for the whole site? 21:50
masak no, only things that are dirty.
I don't have old numbers for whole-site, but let me check what it comes down to.
24.5 seconds. 21:51
i0.kym-cdn.com/photos/images/newsfe...a-face.jpg
now there's a lot of shelling out there to Gruber's Perl 5 Markdown module. 21:52
I'm hoping to eventually replace that, and to be faster than it.
flussence that's... about 15 pages per second. wow.
(going by $$('.entry').length in /list-of-posts)
kinslayer masak: what are your blog ?
masak strangelyconsistent.org/ 21:53
flussence: yes, you got that right. plus the list-of-posts page itself and the main page.
kinslayer masak oh it seems I have been by that site before. :D 21:54
masak hehe
flussence I remember being impressed about 5 years ago when I got a php-fcgi site to handle 30 pages/s, and that was with some horribly hacky caching.
kinslayer I think perhaps from a video from a talk or a google search :D
TimToady masak: testing a fix for cheating heredocs 21:58
masak \o/
vendethiel TimToady++ :-)
TimToady works on constant, doesn't work on blorsts yet 21:59
masak constants is quite enough 22:00
masak ponders printing up a community membership badge for himself with the title "The Squeakiest Wheel" :P
dalek ast: e4816ad | FROGGS++ | S0 (2 files):
adjust tests for "0" being true
 22:05
masak well, truthy. :) 22:07
FROGGS btw, fixing Parrot's istrue explodes quite spectacular: gist.github.com/FROGGS/f81a925bc8f5b6d72ddf
masak: I'm tired :o)
masak anyway, FROGGS++ 22:08
vendethiel FROGGS++ ! 22:10
FROGGS the other spectest fails seem to be about that print-nl thingy
gnight 22:12
vendethiel say I wanted to write a "low-level VM" (whatever that means) in C, and have maybe a native array
of u8 (with fixed size). is there everything that needs to be done, say, to read 4 bytes into an int or something? 22:13
night FROGGS!
dalek kudo/nom: d5e2d1e | TimToady++ | src/Perl6/Grammar.nqp:
cheat on constant heredocs that have nothing after
 22:14
FROGGS you can cast the position you care about to an int*, and then deref it
dalek pan style="color: #395be5">perl6-examples: eaed7b7 | paultcochrane++ | lib/Pod/Htmlify.pm6:
Correct the url in pod-links
FROGGS like: int foo = *(int*)(array_ptr + idx); 22:15
vendethiel FROGGS: my VM version has a `char tmp[sizeof(T)];`, I read byte by byte, and then casts the tmp to void* then T* then dereferences it
FROGGS though, the array must of course be idx + 3 bytes long
dalek frastructure-doc: 3b5890d | paultcochrane++ | hosts/hack.p6c.org.pod:
Mention the examples.perl6.org rebuild
 22:16
frastructure-doc: aa1857a | paultcochrane++ | hosts/www.p6c.org.pod:
Mention where cron job runs for examples.perl6.org
FROGGS you have to cast it to a char ptr first...
like: int foo = *(int*)(((char *)array_ptr) + idx);
otherwise the idx would not be about bytes 22:17
vendethiel FROGGS: that works fine: github.com/vendethiel/worldofcore....rs.c#L3-12 I'm just wondering how i'm supposed to do that in perl6 :-)
[ptc] yay! examples.perl6.org just went online. Thanks for your help moritz++! 22:19
FROGGS m: use NativeCall; my $a = CArray[int8].new; $a[$_] = $_ for ^32; say my $ap = nativecast(Pointer, $a); say nativecast(int32, Pointer.new($ap + 4)) 22:20
camelia rakudo-moar 0b2092: OUTPUT«Pointer<0x316ac90>␤117835012␤»
FROGGS vendethiel: ^^
vendethiel FROGGS: amazing, thanks :)
FROGGS :o)
vendethiel but I guess I'll get astray from the school's pdf and have an int @array instead. no reason to follow the stupid stuff 22:21
FROGGS that might also make it endian unaware then...
vendethiel haah 22:22
flussence this is interesting - blog.rust-lang.org/2015/04/10/Fearl...rency.html - it looks like in Rust, function arguments get "used up" if they don't explicitly return them to the caller 22:23
(it explains there's other ways around that, but still, kinda a surprising default) 22:25
timotimo please explain? 22:26
you mean after you pass something to a function you're not allowed to re-use it?
flussence exactly! weird, huh?
timotimo interesting idea 22:27
masak flussence: yes, this is known as "linear types". and it is very, very interesting. 22:31
basically, you get the type system to track ownership. the type system makes sure, statically, that there are never two simultaneous references to some value. 22:32
I also find it interesting that there are parallels here to quantum computing, where this relation holds all the time due to the way qubits work. (if you measure them, you "destroy" them in some sense.) QC calls it the "no-cloning theorem". 22:34
vendethiel it's like they forked c++ value/references/foo before xvalues were added. (I'm just kidding) 22:37
masak citeseerx.ist.psu.edu/viewdoc/downl...p;type=pdf seems relevant, and approachable 22:38
vendethiel philip wadler, can recommend
flussence it does sound like they've put a ton of thought into making concurrency stuff good, rather than just accepting that the state of the art is that 22:39
masak yeah, rust is exciting
vendethiel "shared memory" stuff :-). lifetimes, etc
(that's a problem in general in C++ without having to add concurrency stuff. see move and all that... eh) 22:40
masak gonna go to bed now, but I want to leave you with a parting challenge-oid 22:41
I just took this line gist.github.com/masak/db655cb7ab27...zes-p6-L26
and rewrote it to work on ints instead
gist.github.com/masak/e54c64fc2d51cdd1a0b7
the resulting script runs in 10s if I turn off the `say` instructions 22:42
which is exhiliratingly fast
can you make it faster? :)
I'm sure I've overlooked something -- it is my first attempt.
I have verified that it's a correct replica of the original logic, though.
masak it outputs the exact same set of solutions. 22:43
andreoss m: my @x := do 2 for 1..*; say @x[^10];
m: my @x := gather take 2 for 1..*; say @x[^10];
camelia rakudo-moar 0b2092: OUTPUT«(timeout)»
rakudo-moar 0b2092: OUTPUT«2 2 2 2 2 2 2 2 2 2␤»
masak is not sure he has seen `gather take` together before 22:44
andreoss++
anyway, 'night, #perl6
dalek kudo/nom: 53593e5 | TimToady++ | src/Perl6/Grammar.nqp:
heredoc cheating on blocks and on use as well
masak TimToady++ 22:45
tony-o can arrays not be applied to regex subtraction?
m: @a = qw<a b>; say "abc".subst(/ <[.] - [@a]> /, ''); 22:46
camelia rakudo-moar 0b2092: OUTPUT«5===SORRY!5=== Error while compiling /tmp/ByLm01N8Zl␤Variable '@a' is not declared␤at /tmp/ByLm01N8Zl:1␤------> 3@a7⏏5 = qw<a b>; say "abc".subst(/ <[.] - [@a␤ expecting any of:␤ postfix␤»
tony-o m: my @a = qw<a b>; say "abc".subst(/ <[.] - [@a]> /, '');
camelia rakudo-moar 0b2092: OUTPUT«abc␤»
vendethiel [.]? 22:47
tony-o i'd guess it should be 'ab' after the subst
vendethiel that's a literal one, right?
tony-o it isn't a literal '.'
vendethiel isn't it? i'm fairly sure it's a literal one in this context :o 22:48
tony-o ah maybe it is,
vendethiel m: my @a = qw<a b>; say "abc".subst(/ <[a..z] - [@a]> /, '');
camelia rakudo-moar 0b2092: OUTPUT«ac␤»
tony-o m: my @a = qw<a b>; say "abc".subst(/ <-[@a]> /, '');
camelia rakudo-moar 0b2092: OUTPUT«ac␤»
vendethiel does seem to be a literal one ;-)
tony-o i stand corrected 22:49
why is it getting rid of the b instead of the c though?
m: my @a = qw<a b>; say "abc".subst(/ <-[@a]> /, '');
camelia rakudo-moar 0b2092: OUTPUT«ac␤»
vendethiel no idea, I don't know if @a is correct
I just knew the . was literal because 'foo' warns for duplicated ' :) 22:50
tony-o ah :-)
m: my @a = qw<a b>; say "abc".subst(/ [@a] /, '');
camelia rakudo-moar 0b2092: OUTPUT«bc␤»
tony-o m: my @a = qw<a b>; say "abc".subst(/ @a /, '');
camelia rakudo-moar 0b2092: OUTPUT«bc␤»
dalek ecs: 6fa2d87 | TimToady++ | S02-bits.pod:
doc new heredoc cheating
tony-o maybe it's the sinus infection but none of those seem correct to me 22:51
vendethiel spec say / @a / is the same as / [ $(@a[0]) | $(@a[1]) | ... ] 22:52
TimToady notes that the 'my @herestub_queue;' in the STD role is essentially a global variable, however
vendethiel i'll just go to bed and read backlog tomorrow :). good night, #perl6! 22:53
TimToady o/
those all look correct to me 22:54
tony-o vendethiel: seems like [ 'a' | 'b' ] should match the "ab" section ?
not the "bc" part?..
TimToady there's no quantifier, so no way to match the b after a 22:55
or did you think it was gonna intuit a :g for you?
TimToady knows about them sinus infections...
tony-o sacramento is terrible with it
TimToady just because @a is "plural" doesn't mean it matches more than one thing... 22:56
tony-o m: my @a = qw<a b>; say "abc".subst(:g/ @a /, '');
camelia rakudo-moar 0b2092: OUTPUT«5===SORRY!5=== Error while compiling /tmp/O8wCyXvpPQ␤Preceding context expects a term, but found infix , instead␤at /tmp/O8wCyXvpPQ:1␤------> 3 @a = qw<a b>; say "abc".subst(:g/ @a /,7⏏5 '');␤»
tony-o m: my @a = qw<a b>; say "abc".subst(r:g/ @a /, '');
camelia rakudo-moar 0b2092: OUTPUT«5===SORRY!5=== Error while compiling /tmp/ZqKQLeT6_3␤Preceding context expects a term, but found infix , instead␤at /tmp/ZqKQLeT6_3:1␤------> 3@a = qw<a b>; say "abc".subst(r:g/ @a /,7⏏5 '');␤»
tony-o gah
TimToady: i'll look at it some more if those are right, or just take a break - thanks ^
TimToady m: my @a = qw<a b>; say "abc".subst(:g, / @a /, '')
camelia rakudo-moar 0b2092: OUTPUT«c␤»
TimToady m: my @a = qw<a b>; say "abc".subst(/ @a+ /, '') 22:57
camelia rakudo-moar 0b2092: OUTPUT«c␤»
labster FROGGS++ for fixing all of the "0" things I didn't see 22:59
dalek kudo/nom: be225ce | TimToady++ | src/Perl6/Grammar.nqp:
don't cheat without a newline coming
 23:03
labster design.perl6.org/S26.html#Delimited_blocks mentions that :key<1 2 3> make a list, but right below it looks like it's trying to use :caption<Table of Contents> as a string. 23:04
TimToady well, if you stringify the list, that's what you get 23:05
m: say :caption<Table of Contents>.value.WHAT
camelia rakudo-moar 0b2092: OUTPUT«(Parcel)␤»
TimToady m: say :caption<Table of Contents>.Str
camelia rakudo-moar 0b2092: OUTPUT«caption Table of Contents␤»
TimToady m: say :caption< Table of Contents>.value.Str 23:06
camelia rakudo-moar 0b2092: OUTPUT«Table of Contents␤»
labster true enough. Now I just need to convince rakudo that all colonpairs in pod aren't strings.
skids class B { method foo { "" } }; class A is B { method foo { 0 || nextsame } }; A.new.foo.perl.say; 23:14
m: class B { method foo { "" } }; class A is B { method foo { 0 || nextsame } }; A.new.foo.perl.say;
camelia rakudo-moar 0b2092: OUTPUT«""␤»
skids m: class B { }; class A is B { method foo { 0 || nextsame } }; A.new.foo.perl.say; # Nil is as good as anything I guess 23:15
camelia rakudo-moar 0b2092: OUTPUT«Nil␤»
skids But maybe nextsame should take an arg for use if it finds itself with no recourse. 23:15
eli-se GitHub thinks my Perl 5 code is Perl 6 code. 23:16
My Perl 5 code must be really good!
TimToady the nqp jvm build appears to be busticated 23:23
at least on my downstream server; trying again on my laptop 23:24
eli-se What happens if you next $label; on a different thread than the one the loop is running on?
TimToady yep, kerblooey there too 23:28
hopefully the thread catches the control exception trying to escape, and punishes it for being desirous of freedom
TimToady gist.github.com/anonymous/2d3e3d6de54f621af69e is the jvm build fale 23:31
caymanboy nick caymanboy 23:44

Powered by Cro and the Raku® Programming Language.

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

Thank you!