»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by moritz on 25 December 2014.
TimToady well, I already implemented tests, but only for exchange on Pair, which, er, was the only one specced... 00:31
and I asked for use cases for anything else...
er...speculated... 00:33
TimToady it is not, in particular, clear whether .exchange is the opposite of .pairs, or the opposite of .invert 00:35
if people expect the latter on a list of pairs, they'll be very surprised to get extra keys 00:36
in particular, I was thinking about it as the opposite of invert, not the opposite of pairs, or I'd've named it something different 00:37
dalek ast: c349a13 | TimToady++ | S02-types/pair.t:
unfudge .exchange tests
 00:50
TimToady maybe .antipairs is what lizmat++ implemented 00:52
then .exchange would only work on a Pair or list of Pair 00:55
dalek ecs: db834e7 | TimToady++ | S32-setting-library/Containers.pod:
distinguish antipairs from antipair/invert
 02:21
dalek ecs: 1968457 | skids++ | S03-operators.pod:
Negative binary shifts are indeed tested in roast, so whirl them

useful on infix:«<+», so mention that. (The :signed and :unsigned adverbs are currently only in spec not in roast/implementation)		 03:04
TimToady alternately, we should make that decision based on the left-hand type 03:05
skids Yes it makes no sense for Int. 03:08
But maybe better as a runtime error, rather than requiring tight typing? 03:09
TimToady actually, shifts are fine for Int 03:10
skids Yes, but :unsigned isn't 03:11
erm :!signed
dalek ok: 5a4285b | (Christopher Bottoms)++ | src/operators.pod:
Minor grammar fix
 03:14
TimToady with Int we just assume an infinite number of 0's or 1's to the left 03:15
m: say -1 +< 30
camelia rakudo-moar 3eb446: OUTPUT«-1073741824␤»
TimToady m: say 1 +< 30 03:16
camelia rakudo-moar 3eb446: OUTPUT«1073741824␤»
skids Right so :!signed would either be mute, or should be an error.
*moot
TimToady would like to avoid adverbs entirely if we can make it type based
likewise if we ever get rotate instructions 03:17
skids The point of the adverb is kind of to override signedness of the type, I thought.
TimToady see rosettacode.org/wiki/Bitwise_operations#Perl_6 03:18
TimToady notice how we cheat on the unsigned rightshfit 03:18
and the rotates
that could easily be determined by the native size of the left though
well, called to dinner & 03:19
skids I think that's the intent of the language in specs -- the left hand type determines which adverb would be needed to break the rules. 03:20
skids FWIW if anyone ever wants to rewrite asm in perl6 they would probably also want a "with carry" option on shifts and rotates. 03:22
skids
.oO(Texas ⥀ ... ^< Texas ⥁ ... ^>)
 03:51
TimToady well, probably rol and ror are fine Texas versions 03:57
b2gills m: subset Test of Str where 0 < *.chars < 256; say '0' ~~ Test; say '1' ~~ Test; say 'a' ~~ Test 03:58
camelia rakudo-moar 3eb446: OUTPUT«False␤True␤Cannot call 'Real'; none of these signatures match:␤:(Mu:U \v: *%_)␤ in method Real at src/gen/m-CORE.setting:4685␤ in sub infix:<<> at src/gen/m-CORE.setting:5193␤ in block <unit> at /tmp/EdXKXiyxUm:1␤␤»
skids modernizes gist.github.com/skids/4998260 03:59
b2gills m: subset Test of Str where {0 < $_.chars < 256}; say '0' ~~ Test; say '1' ~~ Test; say 'a' ~~ Test
camelia rakudo-moar 3eb446: OUTPUT«True␤True␤True␤»
b2gills star-m: subset Test of Str where 0 < *.chars < 256; say '0' ~~ Test; say '1' ~~ Test; say 'a' ~~ Test 04:00
camelia star-m 2015.01: OUTPUT«False␤True␤Cannot call 'Real'; none of these signatures match:␤:(Mu:U \v: *%_)␤ in method Real at src/gen/m-CORE.setting:4236␤ in sub infix:<<> at src/gen/m-CORE.setting:4746␤ in block <unit> at /tmp/mxH8DNt_Ga:1␤␤»
TimToady looks kinda like a bug
m: say 0 < 'a'.chars < 256 04:01
camelia rakudo-moar 3eb446: OUTPUT«True␤»
TimToady actually, looks like it's ignoring the .chars
TimToady since they should all be True 04:02
oh, the real closure works
so it's just not handling * with chained ops
b2gills I think that it is seeing that the *.chars is in Real context and is trying to convert it too soon 04:03
TimToady m: say (0 < *.chars < 256).WHAT
camelia rakudo-moar 3eb446: OUTPUT«(WhateverCode)␤»
TimToady m: say (0 < *.chars < 256)('a') 04:04
camelia rakudo-moar 3eb446: OUTPUT«Cannot call 'Real'; none of these signatures match:␤:(Mu:U \v: *%_)␤ in method Real at src/gen/m-CORE.setting:4685␤ in sub infix:<<> at src/gen/m-CORE.setting:5193␤ in block <unit> at /tmp/aYgm5iSKst:1␤␤»
b2gills ignore what I just wrote, it doesn't explain the `0` case 04:06
dalek ast: 9c69eb1 | TimToady++ | S (2 files):
antipairs and inversions
 04:08
kudo/nom: 06f9b89 | TimToady++ | src/core/ (12 files):
untangle antipairs vs invert
 04:09
TimToady m: say (0 < (*.chars) < 256)('a') 04:10
camelia rakudo-moar 3eb446: OUTPUT«Cannot call 'Real'; none of these signatures match:␤:(Mu:U \v: *%_)␤ in method Real at src/gen/m-CORE.setting:4685␤ in sub infix:<<> at src/gen/m-CORE.setting:5193␤ in block <unit> at /tmp/0GxAW99xbc:1␤␤»
TimToady m: say (0 < *.chars)('a')
camelia rakudo-moar 3eb446: OUTPUT«True␤»
TimToady m: say (*.chars < 256)('a')
camelia rakudo-moar 3eb446: OUTPUT«True␤»
TimToady yeah, it's just the chaining fouling it up 04:11
TimToady something tells me it's probably even reported already 04:17
skids m: (0 < (*,).chars < 256)("a").say 04:18
camelia rakudo-moar 3eb446: OUTPUT«Cannot find method 'postcircumfix:<( )>'␤ in block <unit> at /tmp/_WL6_uLKcS:1␤␤»
TimToady that'll never work 04:19
TimToady a list with a Whatever in it is just a list with a Whatever in it 04:19
skids Yeah. Can't blame a guy for trying :-) 04:20
TimToady the autopriming is really only defined on unary and binary ops
b2gills rakudobrew really doesn't work right when called as a symbolic ref 04:24
skids RT#123144 maybe, hit internally? 04:25
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=123144
colomon \o/ 04:46
colomon was o/ the 15 minutes on Raspberry Pi 2 note. backlogging is confusing. 04:50
TimToady let's do math 04:50
raydiak m: sub foo (|args) { args[*] }; foo :bar # known? 05:40
camelia rakudo-moar 06f9b8: OUTPUT«Cannot convert string to number: base-10 number must begin with valid digits or '.' in '⏏bar' (indicated by ⏏)␤ in method Int at src/gen/m-CORE.setting:15021␤ in method Int at src/gen/m-CORE.setting:6832␤ in sub postcircumfix:<[ ]> at src/gen/…»
raydiak wonders if replacing all the SELF.keys in src/core/array_slice.pm with ^SELF.elems or so might fix it 05:45
moritz \o 06:25
raydiak o/ 06:26
TimToady o—
TimToady is testing a fix for local overriding of quote keywords so that my m = 42; say m / 100; prints 0.42 06:27
m: say { a => [1,2] }.invert.perl 06:32
camelia rakudo-moar 06f9b8: OUTPUT«(1 => "a", 2 => "a").list␤»
TimToady m: say { a => [1,2] }.antipairs.perl 06:33
camelia rakudo-moar 06f9b8: OUTPUT«([1, 2] => "a",).list␤»
TimToady 'antipairs' being somewhat dehuffmanized on purpose
m: say <a b c>.invert 06:37
camelia rakudo-moar 06f9b8: OUTPUT«Memory allocation failed; could not allocate 57696 bytes␤»
TimToady yowser
m: say <a b c>.pairs.invert
camelia rakudo-moar 06f9b8: OUTPUT«Memory allocation failed; could not allocate 640 bytes␤» 06:38
TimToady hmm
m: say <a b c>.pairs
camelia rakudo-moar 06f9b8: OUTPUT«0 => a 1 => b 2 => c␤»
TimToady m: say <a b c>.pairs>>.antipair
camelia rakudo-moar 06f9b8: OUTPUT«a => 0 b => 1 c => 2␤»
TimToady well, something screwy on List.invert still 06:39
japhb Does Rakudo have any way to take a path containing .. and properly resolve it? For example, given ('../foo/bar' relative to '/a/b/c/d/e'), which absolutifies to '/a/b/c/d/e/../foo/bar', where any segments of that path could be links (symbolic or otherwise), allows you to do .parent.parent.parent and get a sane result?
TimToady probably
but I dunno what it is
TimToady m: say <a b c>.pairs>>.invert 06:40
camelia rakudo-moar 06f9b8: OUTPUT«a => 0 b => 1 c => 2␤»
moritz japhb: IO::Path.absolute or something like that 06:41
lizmat only implemented .exchange last night 06:42
off for some swimming& 06:43
japhb moritz: .absolute oddly gets you a Str of the relative base and relative path concatenated by the directory separator. So it gets you that thing I said "absolutifies to" above.
raydiak m: ".".IO.absolute.perl.say 06:52
camelia rakudo-moar 06f9b8: OUTPUT«"/home/camelia"␤»
raydiak m: ".".IO.abspath.perl.say
camelia rakudo-moar 06f9b8: OUTPUT«"/home/camelia"␤»
raydiak m: ".".IO.canonpath.perl.say 06:56
camelia rakudo-moar 06f9b8: OUTPUT«No such method 'canonpath' for invocant of type 'IO::Path'␤ in block <unit> at /tmp/ibumoZm_J8:1␤␤»
raydiak m: $*SPEC.canonpath(".").perl.say 06:57
camelia rakudo-moar 06f9b8: OUTPUT«"."␤»
raydiak m: ".".IO.resolve.perl.say
camelia rakudo-moar 06f9b8: OUTPUT«Failure.new(exception => X::NYI.new(feature => "IO::Path.resolve"))␤»
japhb Sigh, NYI 06:59
dalek kudo/nom: f786d89 | TimToady++ | src/Perl6/Grammar.nqp:
quoters now lexically overridable by terms/subs
 07:00
ast: 6d85d5a | TimToady++ | S0 (2 files):
quoters now lexically overridable by terms/subs
 07:01
dalek c: d12c47b | moritz++ | t/pod-convenience.t:
Fix a test
 07:05
moritz [ptc]: github.com/perl6/doc/issues/56 07:09
dalek kudo/nom: 6f9016e | TimToady++ | src/core/ (2 files):
fix infinite regress on List.invert
 07:13
TimToady m: my \s = 'sigma'; say s;
camelia rakudo-moar f786d8: OUTPUT«sigma␤»
moritz TimToady++
TimToady can override by using any option to get back to the qouter 07:14
*uo
m: my \m = 'millipede'; say m ~~ m:s/\w*/ 07:15
camelia rakudo-moar f786d8: OUTPUT«｢millipede｣␤»
TimToady m: my \m = 'millipede'; say m ~~ ms/\w*/
camelia rakudo-moar f786d8: OUTPUT«｢millipede｣␤»
TimToady m: my \m = 'millipede'; say m ~~ m/\w*/ 07:16
camelia rakudo-moar f786d8: OUTPUT«5===SORRY!5===␤Regex not terminated.␤at /tmp/y9RJoxQNZw:1␤------> 3my \m = 'millipede'; say m ~~ m/\w*/7⏏5<EOL>␤Regex not terminated.␤at /tmp/y9RJoxQNZw:1␤------> 3my \m = 'millipede'; say m ~~ m/\w*/7⏏5<EOL>␤Unable to parse r…»
TimToady m: my \m = 'millipede'; say m ~~ m /\w*/ 07:16
camelia rakudo-moar f786d8: OUTPUT«5===SORRY!5===␤Regex not terminated.␤at /tmp/JK9sCZV9qB:1␤------> 3my \m = 'millipede'; say m ~~ m /\w*/7⏏5<EOL>␤Regex not terminated.␤at /tmp/JK9sCZV9qB:1␤------> 3my \m = 'millipede'; say m ~~ m /\w*/7⏏5<EOL>␤Unable to parse…»
moritz colomon: something's off with the smoker. smoke.perl6.org/project/Questhub says Questhub failed its test, but it doesn't even have a t/ directory
colomon: and in the second-to-last report, it succeeded, so it can't be general failure of counting absense of t/ as test failure 07:17
raydiak bugfix PR github.com/rakudo/rakudo/pull/375 07:18
moritz I'd like to leave that for lizmat++ to evaluate, since she wrote the code and is probably aware of performance tradeoffs 07:20
raydiak sounds good, I'd not be surprised if that is a slower-than-necessary way to write it 07:21
moritz maybe we should have positional-keys and associative-keys methods 07:34
(and still have .keys, but [] could use .positional-keys) 07:35
raydiak kv, pairs, elems, etc too?
moritz hmmm 07:37
there's a bite more to it than I initially thought :/
raydiak yeah the overlap is actually pretty sizable 07:38
my patch relies on the fact that things which do both seem to delegate .elems to the array part 07:39
FROGGS[mobile] TimToady++ # overridable quoters 07:41
dalek ast: 08174d6 | TimToady++ | S02-types/pair.t:
test invert on list of pairs
 07:46
TimToady m: say <a b c>.pairs.invert 08:01
camelia rakudo-moar 6f9016: OUTPUT«a => 0 b => 1 c => 2␤»
TimToady m: say <a b c>.invert
camelia rakudo-moar 6f9016: OUTPUT«Cannot call 'invert'; none of these signatures match:␤ in block <unit> at /tmp/emLQR3RLBd:1␤␤»
TimToady m: say <a b c>.list.invert
camelia rakudo-moar 6f9016: OUTPUT«No such method 'value' for invocant of type 'Str'␤ in block at src/gen/m-CORE.setting:10239␤ in method reify at src/gen/m-CORE.setting:9438␤ in block at src/gen/m-CORE.setting:9270␤ in method reify at src/gen/m-CORE.setting:9244␤ in block …»
TimToady maybe that failure mode is a bit LTA 08:02
it *should* fail though
since invert is now only defined on lists of pairs and on hashes
dalek kudo-star-daily: e3ff6cd | coke++ | log/ (9 files):
today (automated commit)
 08:10
timotimo moritz: would you like to walk me through how to update star-m: on camelia? or just do it yourself (or check what went wrong with the automatic update - if it exists) 08:15
moritz timotimo: there is no automatic upgrade 08:21
timotimo: and camelia run on nine_++'s box, so you need to ask him for an account first 08:22
raydiak I think we should have a few range constructor methods in Int or Cool, or even just a single coercion, such that 5.Range eqv ^5 08:26
raydiak m: say 5.&prefix:<^> # this works, though it feels clumsy 08:29
camelia rakudo-moar 6f9016: OUTPUT«0..^5␤»
raydiak 5.Range.map looks and feels better in some ways than (^5).map...or maybe something more like .from-zero for ^5, .from-zero-to for 0..5, and s/zero/one/ variants for 0^.. 08:36
timotimo oh, right, it's nine's box 08:43
i knew that, but i forgot about it briefly 08:44
moritz I guess I should just document it 08:46
dalek kudo/nom: 4f854b3 | raydiak++ | src/core/array_slice.pm:
Attempt to fix array slicing of Capture
 08:47
kudo/nom: e35414b | lizmat++ | src/core/array_slice.pm:
Merge pull request #375 from raydiak/arrayhashslicefix

Fix array slicing of Capture, Match, etc. Brings about a ~5% performance gain on [*] slices
timotimo has a look at the whatever currying problem 08:50
raydiak thank you lizmat :)
lizmat raydiak: thank you!
timotimo thank you, raydiak and lizmat :)
raydiak timotimo: what I had before for the currying is on the assumingwhatevs branch of my rakudo fork if it's any use 08:51
timotimo oh 08:52
no, not that
i mean this:
m: say (0 < *.chars < 255)("a")
camelia rakudo-moar 6f9016: OUTPUT«Cannot call 'Real'; none of these signatures match:␤:(Mu:U \v: *%_)␤ in method Real at src/gen/m-CORE.setting:4683␤ in sub infix:<<> at src/gen/m-CORE.setting:5191␤ in block <unit> at /tmp/6dZ75MJkVg:1␤␤»
timotimo the .chars gets dropped somewhere in the inlining process
raydiak ah I see 08:53
timotimo a couple of months ago i built some code that would generate a single code object from this rather than three nested ones that were sort-of recursively called 08:53
timotimo i'm a tiny bit annoyed that there was no test that caught this problem back then 08:54
but also apparently no code in the ecosystem stumbled upon this
masak good morning, #perl6
raydiak good morning masak 08:55
timotimo i think i have a clue what's happened, but not 100% sure
nine_ sjn: a little example of what actions can do: github.com/niner/Grammar-Highlighter 08:56
masak m: say prefix:<^> 5
masak come back, camelia! :/ 08:56
anyway, locally I get `0..^5`
I like it.
it seems we already have a listop for this :) 08:57
raydiak I was looking for a way to get it as an easier-to-type postfix, b/c e.g. (^5).map feels awkward 08:58
if you have method(s), you don't have to type any circumfix at all
dalek kudo/nom: 35de750 | lizmat++ | src/core/Any.pm:
Fix copy-pasto
 08:59
raydiak but 5.&prefix:<^> is no better
masak m: sub up($n) { ^$n }; say 5.&up
raydiak: that works, too.
timotimo raydiak: do you know about map ^5:; ?
m: say map ^5: *.say;
er
map ^5: *.say works, anyway. 09:00
i'm beginning to like indirect method call syntax
raydiak masak: yes that's not a terrible way to do it, but I probably won't bother if it's not in core
timotimo: neat, I hadn't thought of that
masak raydiak: that's my point. you can define something locally to make it easier for you. 09:02
timotimo nine_: some time ago i wrote this fun piece of code: github.com/timo/text_to_cell/blob/...r.h#L21-23 09:03
masak raydiak: why do people get so hung up on "if it's [not] in core"? I don't understand.
timotimo nine_: it's kind of sort of similar to the grammar highlighter
masak you have this wonderfully extensible language where, at the snap of your fingers, you can turn it into what you want.
and then people want to stuff their favorite DSL into core, or they refuse to play.
>:-(
raydiak mostly out of a desire to not write code which is difficult for people who are not me to follow. if I came up with an idea with a more tangible benefit, I'd likely use it, but for .&up, which is hard for me to type in the dark anyway which was my original problem? no, the ratio does not meet my own personal threshhold. sorry if that makes anyone frown. 09:07
masak sorry for the rant. it's not just you. 09:08
of course "it doesn't cross my personal threshold for ease/readability" *is* a respectable argument.
raydiak no biggie. I know it's a common thing, and I'm sure more than none of it is still the same stuck-in-the-past thinking which you're rightly opposed to 09:09
masak but I do recommend freely creating ops and functions in small scopes when you need them. it becomes addictive, and it changes the way you think about the code :)
masak like, if you really wanted a postfix:<upto>, there's nothing to stop you from defining one. 09:10
raydiak yeah the syntactic mutability is something I am really attracted to, though I sometimes make a mess quickly when I start down those paths, so I often go to the opposite extreme; still looking for a good balance, as with all things 09:12
raydiak good night #perl6 09:33
timotimo is micro-optimizing whatever_curry, as it seems to be called quite often in stage parse
vendethiel TimToady: smalltalk doesn't have "leave", but return ("^") from a lambda returns from the outerlying function, that's what I was referring to 09:41
timotimo why must my stage parse measurements be so noisy? 09:43
vendethiel psch: yes, that was *basically* the use-case 09:47
timotimo huh. my "optimizations" seem to have made it worse
Kristien I don't like functions that return from their definer. 09:51
timotimo .o( that's me in a nutshell: starts looking at correctness, tries performance tweaks instead ) 09:55
psch vendethiel: the «public String evalP6(…) { }» bit, right? not the bit about nesting p6 with interop i hope :) 09:57
also hi #perl6 \o 09:58
masak hi psch 10:10
smls o/ 10:20
smls backlogs (had very little computer time these last 4 days due to health issue) 10:24
smls "* vendethiel would really, really like to have :D as the default" -- same here, though I agree with TimToady that the :_ syntax is not great. 10:30
Especially since the concept of "whatever / I don't care / all of the above" is usually encoded using an asterisk in Perl 6, not an underscore.
Although :* won't win any beauty contests either.
masak I always thought it was a bit of a nod to p5 that the thing passed in can be either a type object or a concrete object. 10:31
smls Maybe call it :UD or :DU instead?
timotimo :DUI?
smls ;)
or :U:D to make it look like stacking adverbs 10:33
masak doesn't feel quite right, though. 10:34
timotimo i dislike how that collides with the final : for invocant
then you could have Dog:, Dog:D:, Dog:U:, Dog:D:U and Dog:D:U:
masak I think the consequences of making :D the default should be seriously thought through. regardless of what syntax ends up replacing :_ , if any. 10:35
smls if the invocant : requires a space after it, it might not be so bad
masak feels to me that making :D the default "hides" type objects quite a bit further back in the language.
smls: that feels weird, especially for invocant-only methods.
smls one disadvantage of making :D the default in function signatures, is that it doesn't make sense to also make it the default for my declarations 10:36
which, in the general case, are also signatures 10:37
so either it's inconsistent, or one of them is suboptimal
my ($x, $y); # a "type check failed" error here would be rather WAT 10:38
timotimo lacs
er
timotimo lacks permission to restart cameli 10:39
camelia
dalek c: ab78bc1 | paultcochrane++ | t/pod-htmlify.t:
Don't run pod-htmlify if URI::Escape isn't installed
 10:56
smls Being able to say "my-method(:U: $foo)" insead of "my-method(Current::Class::Name:U: $foo)" is nice. moritz++ 10:57
timotimo did moritz already implement that, or only suggest? 10:58
smls he spec'd it
timotimo OK 10:59
smls timotimo: Once good Perl 6 syntax highlighting proliferates, the different kinds of : will hopefully be less confusing. 11:01
e.g. "Dog:U:" --> "<typespec>Dog:U</typespec><interpunctuation>:</interpunctuation>" 11:02
timotimo i disagree with requiring to have syntax highlighting to make code understandable
smls Well there are different levels of understanding 11:03
If newbies can recognize a construct and even use it without understanding the more general rule of how it is contructed from its parts, that's ok imo
vendethiel psch: the problem is -- "evalp6code" will be a real performance problem :) 11:04
psch vendethiel: to the extend that 'perl6-j' is a performance problem, yes 11:08
vendethiel *g* 11:09
no, to the extend that the jvm can't optimize it
(in the long run, that is)
psch why not?
vendethiel because it's re-eval'd everytime
dalek albot/timo-require-config-file-exists: 067570b | timo++ | evalbot.pl:
exit the program if the config file wasn't found

also writes the cwd to the console.
timotimo can has review, please? i don't has perl5 skills 11:10
psch i don't know if that's true. caching CompUnits (e.g. CORE.setting) might be possible?
timotimo: curlies don't interpolate in qq in p5, afair 11:11
timotimo OK. how do i do it instead? 11:12
psch i'd do «"...not found in " . getcwd();»
although those parens might be superstitious 11:13
smls "wasn't found in @{[ getcwd ]}." would also work, though it's a hack. 11:14
jnthn afternoon, #perl6 11:15
timotimo test test
psch o/ jnthn
dalek albot/timo-require-config-file-exists: 56af077 | timo++ | evalbot.pl:
qq doesn't interpolate curlies in perl5 by default

  psch++		 11:16
timotimo looks mergable? 11:18
[ptc] is there a "panda uninstall [package-name]"? 11:26
dalek c: b0c06de | timotimo++ | t/pod-htmlify.t:
simplify URI::Escape requirement in pod-htmlify.t
 11:27
timotimo there is not
[ptc] pity 11:28
is there a simple way to uninstall a package?
timotimo only rm'ing the files in the install directory 11:29
[ptc] ok. been there, done that. However wondered if I was missing something 11:30
timotimo Mouq: are you attentive? 11:31
jnthn TimToady++ # fixing the s and m term thing 11:32
jnthn is gonna have to update the native-ref branch with nom before his next test run, after all these patches :) 11:33
But first, get the JVM porting done... :)
itz is native-ref to be used for NSA? 11:35
timotimo itz: indeed
Mouq: you're wrapping $past with &HYPERWHATEVER near the bottom of whatever_curry after the annotations have been set that are used for inlining curries into each other; do you know if that has negative effects? 11:36
itz [ptc]: I was wondering whether panda should record installed files with SHA256 sums maybe in a JSON structure
jnthn itz: Yeah, it's one of the main building blocks I needed for native arrays, but it also takes care of many language usability issues with native types 11:37
Like the infamous "wtf I can't $native-int++"
timotimo yes! 11:38
psch coercing to string here github.com/rakudo/rakudo/blob/nom/....nqp#L6733 fixes this todo github.com/perl6/roast/blob/master...ders.t#L66
timotimo and "cannot use 'is rw' on a native attribute"
psch i'm unsure if moar is doing the right thing (i.e. not getting hung up on Mu vs Any) or jvm (i.e. getting hung up on *) 11:38
psch 'cause $*IN_DECL, which is what gets annotated into .ann('IN_DECL') on nqp level should be Mu on both backends, right? 11:39
(coercing to string seems a bit like fixing the symptoinstead of the system 11:40
s/oin/om in/
smls jnthn: If @a is a compact array of native int's, what will @a[0] return? Some kind of Proxy container? 11:43
timotimo a native reference to the int in the array 11:44
it is "a kind of container", but not really
smls Is that specunspecified on a language level and users are told to think of @[0] = 4 as magic syntax and not introspect it?
jnthn Well, it is a container in the same sense Scalar is
itz m: my int $i=2; $i++
jnthn Just doing a rather different thing when you store into it
timotimo itz: camelia needs a restart right now, nine isn't at the computer to do it right now, though 11:45
jnthn The key thing is that for common cases, the compiler can elide a lot of the reference takes.
itz ah
jnthn And Moar will be able to eliminate a bunch more, especially in combination with inlining. 11:45
jnthn smls: Well, you can expect some kind of "first class" l-value thing 11:46
smls ok
jnthn smls: But the spec doesn't nail down details beyond that.
smls Currently @a[$i] already returns a proxy for out-of-bounds elements, right?
to make autovivification work
jnthn Yes 11:47
> my int $a = 42; my $b := $a; $b++; say $a;
43
That's one example of the "first class" nature of the refs
smls nice 11:48
timotimo that's supposed to work for compact arrays, too, right?
jnthn Not quite sure how horribly that will fall out of the optimizer, but in combination with lowering to locals and the stuff spesh does, the answer may well be "just fine"
timotimo: Yeah, and already does in native-ref if you count Buf as a kind of native array 11:49
timotimo neato!
jnthn > my $b = Buf.new(1, 2, 3); my $x := $b[1]; $x++; say $b;
Buf:0x<01 03 03>
colomon moritz: I think the problem is emmentaler / panda cannot tell the difference between a module failing tests and one of its dependencies failing tests. Questhub depends on HTTP::UserAgent, which also failed... 11:50
vendethiel nine_++ # highlighter 11:54
timotimo vendethiel: take a screenshot please!
vendethiel doesn't have one :P
timotimo OK 11:55
i'd like a parallel version of that that outputs pretty html 11:57
timotimo jnthn: can we have a way to throw out blocks that have been added into the $*W at some point? 11:59
i'm a bit too distracted to figure the whatever currying stuff out right now 12:02
jnthn timotimo: Umm...at some point maybe 12:03
timotimo: Do we need to do it often? 12:04
timotimo i should measure before i make more "demands"
whatever_curry "inlining" creates many, many throw-away blocks
well, one for each operator that's involved
jnthn Ah, I see 12:05
dalek p/native-ref: ceb047a | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/ (2 files):
HLL entries for native ref types.
 12:06
moritz star-m: say "version?" 12:54
ENOCAMELIA
timotimo moritz: do you have permission to systemctl restart [email@hidden.address] ? 12:59
moritz timotimo yes, but it doesn't help :( 13:00
timotimo oh?
moritz missing module, it seems
timotimo did you see the log?
yeah
dalek p/native-ref: 89d2119 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/NativeRef (5 files):
Fill out native lexical reference REPR bits.
p/native-ref: de6e270 | jnthn++ | src/vm/jvm/QAST/Compiler.nqp:
Native ref lexical ops need to receive the tc.
p/native-ref: 228f135 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Fix native ref type HLL config thinko.
moritz timotimo: installing stuff now.. 13:01
timotimo thanks. and can you review my pull-request in perl6/evalbot?
it adds a early program abort if the specified config file doesn't exist
moritz github.com/perl6/evalbot/pulls is empty 13:02
where can I find it?
timotimo moritz: how did you get permission to systemctl?
moritz timotimo: sudo
timotimo oh
i didn't make an actual pull request
github.com/perl6/evalbot/compare/t...ile-exists
moritz timotimo: +1 13:03
timotimo: and +2 if you explicitly import getcwd from Cwd
timotimo oh 13:04
i can do that
dalek albot/timo-require-config-file-exists: 78eaa68 | timo++ | evalbot.pl:
explicitly pull in getcwd from Cwd
 13:04
timotimo there we go
moritz m: say 42
camelia rakudo-moar 35de75: OUTPUT«42␤» 13:05
moritz star-m: say 'version?'
camelia star-m 2015.02: OUTPUT«version?␤»
dalek albot: a9ae692 | moritz++ | README:
Document how to update the star build
 13:06
timotimo moritz: if the run script for the evalbot would output (at least errors) to stdout/stderr, journalctl and "systemctl status" would display that helpfully when something goes wrong 13:07
though it seems like the evalbot will log everything executed?
moritz timotimo: it usually does output errors to stderr 13:08
(or to IRC, if the errors are with compiler invoked)
timotimo right, i only meant errors like "crashed due to out of memory" 13:09
or stuff like that
timotimo oh, when updating star i forgot to echo YYYY.MM > install/VERSION 13:15
timotimo fosdem videos still very sparsely uploaded; not a single main-track-languages or devroom-perl video appeared :( 13:24
dalek p/native-ref: 52f64dd | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Implement lexical native ref taking ops.
 13:27
timotimo jnthn: whatever_curry gets called crazy often; maybe there'd be a way to prevent calling it very early? 13:55
masak .oO( Breaking News: the Haskell community rages over the insensitive tone of the internal Rakudo routine `whatever_curry` ) 13:56
timotimo m) 13:57
masak .oO( in retaliation, they rename `foldr` to the incredibly immature and not at all well-thought-out `yeahRightWall` ) 13:58
timotimo i'm not sure if that was a good joke or i'm just easily amused :) 14:04
masak :) 14:05
Kristien masak: does the Haskell community every not rage about everything not Haskell? 14:07
masak yes, but only inside a RageIO monad. so you'll never really notice from the outside. 14:08
timotimo i didn't know the haskell community rages a lot
masak raging is a side effect.
vendethiel there are 2 sides of the haskell community... 14:10
timotimo my sides!
vendethiel hence the #haskell and #nothaskell split,btw...
vendethiel would love to toss names around, but won't :)
masak vendethiel: I did not know about that.
timotimo video.fosdem.org/2015/STATUS.txt - cool
masak vendethiel: what do they talk about in #nothaskell, I wonder? 14:11
colomon haskell, I imagine. :)
vendethiel masak: being open ;-)
masak: it's mostly the same, really, they're just more open about people. and don't treat you like a p*** of sh** because you don't know everything about CT 14:12
masak :/
vendethiel has watched smalltalk videos yesterday
timotimo what's CT again?
vendethiel category theory
masak Category Theory.
timotimo ah
masak vendethiel: I will never understand why people choose to be mean/derogatory based on lack of knowledge. 14:13
vendethiel one of the people that's hated by the #nothaskell community is the scalaz guy, that also split the scalaz community
timotimo there was even an XKCD about that topic and that still didn't help!
masak timotimo: *rolls eyes* "what, you didn't even know about the xkcd strip about the topic?"
:P
timotimo :D 14:14
vendethiel is really sad aobut this :(
and the problem is that most people are "afraid" to call them out because they "paved the way". (I've been asked to leave #scalaz once because of that..) 14:15
timotimo but you're an angry frenchman, i can understand why people would ask you to leave places! :P
masak vendethiel: sometimes it seems to me that *transmitting knowledge*, teaching and learning really efficiently, is a still-uncracked challenge. and that whoever cracks it is going to be fairly influential going forards.
vendethiel right. 14:17
timotimo: you know it ;-) 14:18
timotimo *hug*
vendethiel has heard that for real, though...
*hug*
masak *hug* 14:19
timotimo ugh 14:20
i kind of wish the whatever code inlining code could be simplified
i'm still splitting my attention between it and watching a zero-k tournament being cast, but ...
timotimo and looking at the code the whatevercode inliner generates ... holy crap that's ugly 14:21
masak you're not supposed to look directly at it :P 14:22
timotimo probably
FROGGS "This representation can not unbox another" :o( 14:36
b2gills m: say Int.Range; say 8.Range 14:44
camelia rakudo-moar 35de75: OUTPUT«-Inf..Inf␤Invocant requires a type object, but an object instance was passed␤ in method Range at src/gen/m-CORE.setting:5531␤ in block <unit> at /tmp/I0Vfs5Mkv5:1␤␤»
timotimo m: say Int.range
camelia rakudo-moar 35de75: OUTPUT«No such method 'range' for invocant of type 'Int'␤ in block <unit> at /tmp/Mw0J1Xbn8d:1␤␤»
timotimo m: say int8.range
camelia rakudo-moar 35de75: OUTPUT«No such method 'range' for invocant of type 'int8'␤ in block <unit> at /tmp/2soVLIS3SR:1␤␤»
timotimo m: say int8.Range
camelia rakudo-moar 35de75: OUTPUT«-128..127␤»
timotimo m: say uint8.Range
camelia rakudo-moar 35de75: OUTPUT«0..255␤»
masak m: say Str.Range 14:45
camelia rakudo-moar 35de75: OUTPUT«No such method 'Range' for invocant of type 'Str'␤ in block <unit> at /tmp/IC9SzQKqm3:1␤␤»
masak shouldn't that one also be -Inf..Inf ?
m: say Num.Range
camelia rakudo-moar 35de75: OUTPUT«No such method 'Range' for invocant of type 'Num'␤ in block <unit> at /tmp/WS62og84QG:1␤␤»
masak and that one?
m: say Real.Range
camelia rakudo-moar 35de75: OUTPUT«No such method 'Range' for invocant of type 'Real'␤ in block <unit> at /tmp/EzDNPPmyB6:1␤␤»
masak and probably that one, too?
m: say Rat.Range
camelia rakudo-moar 35de75: OUTPUT«No such method 'Range' for invocant of type 'Rat'␤ in block <unit> at /tmp/OhsQCuYc9h:1␤␤»
masak seems there are a couple of them missing :)
b2gills All of them are missing except Int 14:46
I tried `$_.?Range // ^$_` and it didn't work 14:47
timotimo m: say Num.?NoSuchMethod 14:48
camelia rakudo-moar 35de75: OUTPUT«Nil␤»
timotimo m: my $_ = 1; say Str // $_.WHAT.say
camelia rakudo-moar 35de75: OUTPUT«Potential difficulties:␤ Redeclaration of symbol $_␤ at /tmp/qzYZFfyrtk:1␤ ------> 3my $_ 7⏏5= 1; say Str // $_.WHAT.say␤(Int)␤True␤»
timotimo m: my $_ = 1; say (Str // $_.WHAT.say) 14:49
camelia rakudo-moar 35de75: OUTPUT«Potential difficulties:␤ Redeclaration of symbol $_␤ at /tmp/QZUrvHhVCc:1␤ ------> 3my $_ 7⏏5= 1; say (Str // $_.WHAT.say)␤(Int)␤True␤»
timotimo ah
jnthn timotimo: Not sure we need to prevent calling it, more try to make sure it does the most common/cheap check first to quickly return. 14:51
timotimo right, i've looked into that 14:52
i put one of the checks earlier and it catches about 1/10 of cases
geekosaur hmmm. older perl 6 links go into the ether
timotimo but for the rest of the cases we still need to iterate over all arguments to check if they .returns WhateverCode
geekosaur I have dev.perl.org/perl6/syn/ in my history which goes to feather.perl6.nl/syn/ which is 404. likewise the top level links on feather.perl6.nl 14:53
timotimo ah
timotimo perl6.feather.nl seems *very* outdated 14:54
at least partially?
it still refers to pugs, for example 14:55
at least pl6anet.org is being linked to instead of the old planetsix
somehow every WhateverCode we compile still ends up with one lexical per argument 14:56
but there's also locals for lowered params and lowered lexes 14:57
moritz doesn't know why feather still runs at all
jnthn timotimo: We have to keep the lexicals around too
timotimo oh
jnthn timotimo: We don't use them normally
timotimo but they aren't being used :)
OK, fair enough
moritz Juerd: services have been migrated away from feather, so if you want, you can take it down
timotimo so i won't energy into trying to remove them 14:58
jnthn timotimo: But if we get a bind failure, we fall back to the slow path binder, and it gets upset if they ain't there.
timotimo: We might be able to be smarter in the future on that.
timotimo understood. is thatsomething we'd want to fix at ... yeah
jnthn But it was the most expedient thing :) 14:58
jnthn (Yes, I did learn this the hard way. ;)) 14:58
timotimo FWIW, CORE.setting.moarvm contains 1002 instances of "Execution of block eliminated by optimizer" 15:00
colomon …. huh? 15:01
timotimo colomon: is that huh for my last line? 15:02
colomon timotimo: yes 15:03
timotimo ah
i can explain
vendethiel "i can explain!" 15:04
Juerd moritz: Great, thanks! 15:05
timotimo except my internet connection has the hiccups again
ok, i'm back
so when the optimizer (which works with QAST nodes) inlines a block into another, the original block isn't needed any more 15:06
masak vendethiel: "it's not what it looks like!"
timotimo before removal it'd've been something like call(clone(QAST::BVal(reference_to_block)))
unfortunately, the compiler gets very upset when a block had been added to the World and it's no longer around when we get to compiling 15:07
so what we do is replace the block's contents with just a quick die statement
timotimo it doesn't use up a lot of memory, but since we have 1000 of them in our core setting, whatever memory it uses per block will get moved from bytes to kilobyte or from kilobytes to megabytes :) 15:08
jnthn We inline 1000 of them? Wow, nice :)
timotimo it's better than having calls in their places, for sure 15:09
jnthn Very
timotimo especially since inlined blocks allow more lexical vars to become local vars etc etc
but also call overhead etc
jnthn cranks the music and gets back to native-ref JVM porting 15:10
timotimo don't be so cranky, jnthn
masak don't listen to the cranks, jnthn
jnthn :P
masak their music sucks
colomon timotimo++ 15:12
dalek kudo/nom: d2604ea | timotimo++ | src/Perl6/Actions.nqp:
fewer QAST nodes when eliminating blocks in *-curry
 15:16
kudo/nom: 67a75c1 | timotimo++ | src/Perl6/Optimizer.nqp:
possibly the smallest time save ever achieved.
skids m: role g[::T $t] { multi method foo (T $a) { "OHAI".say }; multi method foo($a) { "ONOES".say }; }; g[Int(Str)].new.foo("100") 15:19
camelia rakudo-moar 35de75: OUTPUT«ONOES␤»
jnthn Bah, I knew somebody would try that :P 15:21
vendethiel *g*
jnthn I only did coercion types so far sufficient that we could migrate away from the deprecated "as" in sigs.
jnthn That one should be possible to make work, anyways. :) 15:21
It's just a bit fiddly. 15:22
skids laughs maniacally
Juerd Who can update the topic? 15:22
timotimo any op 15:23
er
that's not what you're asking, is it? :)
Juerd I want names! :P
timotimo you can /msg ChanServ access #perl6 list 15:24
skids jnthn: I can RT that if you like.
Juerd timotimo: Thanks
Juerd jnthn: You appear to be an op. Could you add to the topic that feather will shut down permanently 2015-03-31? 15:25
jnthn skids: Feel free 15:26
FROGGS "P6opaque representation can not unbox NativeCall" sounds better than "This representation can not unbox another" 15:27
timotimo can someone please /msg ChanServ access add timotimo +Avorit 15:28
jnthn, you could do this
jnthn I'm kinda trying :P
timotimo (gives me a few not very dangerous permissions)
Juerd Is "/msg ChanServ access add timotimo +Avorit" the new "op me plx"? :P 15:29
jnthn »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 15:29
Juerd jnthn++ 15:29
timotimo Juerd: except easier to sneak something malicious into :P
moritz: dalek's whois shows 2001:780:101:ff00::2:9; is there no reverse-dns for that ipv6 yet? 15:32
FROGGS has the feeling that 'is box_target' does not work correctly on jvm 15:37
dalek p/native-ref: cef0fd2 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/NativeRef.java:
Implement NativeRef REPR composition.
 15:37
p/native-ref: 959164f | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/NativeRef.java:
REPR data serialize/deserialize for NativeRef.
jnthn FROGGS: It's very possibly "interesting" 15:38
FROGGS jnthn: I try to fix Inline::C, which applies roles to a routine, and then passes it off to &trait_mod:<is>($r, native => $!dll)
and the result is: "P6opaque representation can not unbox NativeCall" 15:39
(with improved error message)
jnthn FROGGS: Could be a missing decont...
FROGGS I tried nqp::buildnativecall(nqp::decont(self), ... 15:40
so, that explodes right now: call = (NativeCallBody)target.get_boxing_of(tc, ncrepr.ID);
FROGGS tries to pass $!call instead of self 15:41
java.lang.RuntimeException: Cannot access a native attribute as a reference attribute :o( 15:42
dalek p/native-ref: b22a0e7 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/NativeRef.java:
Fix copy-pasta.
 15:48
p/native-ref: 47a347c | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/NativeRefContainerSpec.java:
Native fetch/store for native_ref cont spec.
p/native-ref: 8bebd81 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/NativeRefInstance (3 files):
Correct inheritance for lex ref types.
dalek p/native-ref: 52188d6 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/NativeRefContainerSpec.java:
Non-native fetch/store to native_ref conts.
 15:58
jnthn Whee, get quite a bit further into the CORE.setting compile now.
xtreak p6: my $a = (1, 2, 3, (4, 5,6)); $a.say; 16:17
camelia rakudo-moar 67a75c: OUTPUT«1 2 3 4 5 6␤»
xtreak Doesn't perl6 differentiate a parcel and a normal list element visually? It seems like a 6 element list now 16:18
jnthn Depends how you output it 16:20
p6: my $a = (1, 2, 3, (4, 5,6)); $a.perl.say;
camelia rakudo-moar 67a75c: OUTPUT«$(1, 2, 3, (4, 5, 6))␤»
xtreak What does .perl do? 16:21
skids Prints something that has a chance of being valid perl code that would recreate the same thing.
geekosaur .perl generates something resembling the perl code to produce a value (think Data::Dumper in perl5); .say uses .gist which generates a simplifies "just the facts" representation 16:22
timotimo we have no documentation about exception handling in doc.perl6.org/language yet?
xtreak Thanks. I have installed rakudo star. Is there a way to view documentation for perl 6 like perldoc? 16:23
jnthn I think it's called p6doc, and believe it's included with r* 16:26
timotimo away for most of $evening
BenGoldberg p6: :foo.perl.say 16:28
camelia rakudo-moar 67a75c: OUTPUT«"foo" => Bool::True␤»
BenGoldberg m: sub foo (|bar) { bar[*] }; foo "quux" => Bool::True
camelia ( no output )
BenGoldberg m: sub foo (|bar) { bar[*] }; foo :quux 16:29
camelia ( no output )
dalek p/native-ref: 75d83a9 | jnthn++ | src/vm/jvm/QAST/Compiler.nqp:
Implement attributeref scope compilation.
BenGoldberg m: sub foo (|args) { args[*] }; foo :bar
camelia ( no output )
xtreak Thanks jnthn. It will be better if it has tab completion like perldoc. As I don't know how things are structured. May be am missing something
b2gills xtreak: the only reason you would have tab completion for perldoc is someone added it to your shell 16:31
Is `$*PERL.version` supposed to be the version of the language? 16:32
jnthn It's probably still worth us looking at what was done to make it work nicely for perldoc, so we can replicate it in p6doc :)
b2gills: I think so 16:33
skids xtreak: you might be better of with doc.perl6.org for starters 16:34
b2gills star-m: say $*PERL.version; say $*PERL.compiler.version; # I don't have star installed locally
camelia star-m 2015.02: OUTPUT«vunknown␤v2015.2␤»
dalek p/native-ref: 11b846e | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/NativeRef (2 files):
Extend NativeRef REPR to include attribute refs.
 16:35
p/native-ref: 253b419 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Implement getattrref_[ins] ops.
jnthn b2gills: Given we didn't settle on the exact naming scheme for lang versions yet, unknown is basically true ;) 16:36
One of those things we need to nail down in the next few months. 16:37
b2gills I'm working on a fix for the $] error message
m: $]
camelia rakudo-moar 67a75c: OUTPUT«5===SORRY!5=== Error while compiling /tmp/8heIuDchb5␤Unsupported use of $] variable; in Perl 6 please use $*PERL_VERSION␤at /tmp/8heIuDchb5:1␤------> 3$]7⏏5<EOL>␤»
[ptc] silly question time: how does one slurp a text file into an array, a la in perl5 $fh = open "file"; @lines = <$fh>; ? 16:38
b2gills I was wondering if it should say `... use $*PERL.version or $*PERL.compiler.version`
[ptc] I've tried "file".IO.slurp.split(/\n/), but that gives me one more element than I expect
jnthn [ptc]: $fh.lines
[ptc] I know I've got a thinko in here somewhere...
jnthn [ptc]: Or $filename.IO.lines 16:39
[ptc] jnthn: thanks, I'll document that one
jnthn [ptc]++
[ptc] jnthn: actually, I want to check the (in this case) the last 3 elements of a file for a particular pattern
so, I'd like to split the string on newlines and look at the output (feels sort of like the natural thing to do) 16:40
[ptc] however, split() doesn't do what one as perl5 hacker would expect 16:40
jnthn [ptc]: Last 3 elements = last 3 lines?
[ptc] jnthn: yes
jnthn $fh.lines[*-3 .. *] may do it 16:41
[ptc] interestingly enough, split() on an empty file returns a 1 element list, but I've been there before ;-)
jnthn: I'll give it a go
[ptc] Index out of range. Is: -3, should be in 0..Inf 16:42
skids *-3 ..^ *
jnthn b2gills: Are $[ and $^V the same thing in Perl 5, ooc? 16:43
skids hrm.
b2gills No but they are close enough
jnthn OK
nwc10 m say $^V
m: say $^V
camelia rakudo-moar 67a75c: OUTPUT«5===SORRY!5=== Error while compiling /tmp/7dHEmEpx71␤Unsupported use of $^V variable; in Perl 6 please use $*PERL_VERSION␤at /tmp/7dHEmEpx71:1␤------> 3say $^V7⏏5<EOL>␤»
nwc10 m: say ${^PERL_VERSION}
camelia rakudo-moar 67a75c: OUTPUT«5===SORRY!5=== Error while compiling /tmp/2ZOsyAHUNY␤Unsupported use of ${^PERL_VERSION}; in Perl 6 please use $(^PERL_VERSION)␤at /tmp/2ZOsyAHUNY:1␤------> 3say ${^PERL_VERSION}7⏏5<EOL>␤»
dalek kudo/nom: cebde22 | (Brad Gilbert)++ | src/core/Exception.pm:
Update error message for $] to use $*PERL.version
 16:44
kudo/nom: 6959323 | jnthn++ | src/core/Exception.pm:
Merge pull request #376 from b2gills/nom

Update error message for $] to use $*PERL.version
timotimo m: say "".split("\n").perl 16:44
camelia rakudo-moar 67a75c: OUTPUT«("",).list␤»
b2gills $^V is supposed to be an object, while $] is a flot
[ptc] ah, if the file has fewer than 3 lines, it barfs
timotimo we've had a lengthy discussion about this at some point, didn't we?
[ptc] timotimo: do you mean me? 16:45
jnthn We've probably had lengthy discussions about most things by this point... :P
timotimo i mean about the behavior of split on empty strings 16:45
m: say "abacaba".split("b").perl
camelia rakudo-moar 67a75c: OUTPUT«("a", "aca", "a").list␤»
timotimo (just making sure i don't have the arguments to split reversed between invocant and needle) 16:46
m: say ("",).list.join("\n").perl
camelia rakudo-moar 67a75c: OUTPUT«""␤»
timotimo well, at least the behavior is consistent between split and join 16:47
[ptc] m: say "abcd".split("e").elems
camelia rakudo-moar 67a75c: OUTPUT«1␤»
timotimo that's very important.
[ptc] that's something that confuses me. The behaviour is documented though
since it's documented I'm ok with it. Nevertheless, I expect this to trip up more than just a few people (just saying) 16:48
[ptc] I also agree that the consistency with join is important 16:49
b2gills jnthn: $ perl -E'say $^V; say $]'␤v5.20.1␤5.020001
jnthn b2gills: OK, thanks. 16:50
b2gills So perhaps it would be a little more accurate for $] to say `$*PERL.version.Numeric or $*PERL.compiler.version.Numeric`
jnthn I'd probably write it as +$*PERL.version 16:51
I can't tell if changing this is helpful or oddly specific. :)
b2gills The only reason $] still is a float is for backwards compatibility 16:52
dalek p/native-ref: 7008cc2 | jnthn++ | src/vm/jvm/QAST/Compiler.nqp:
Add missing add_lexicalref method.
b2gills It's probably better in the long run to leave off the `+` 16:55
BenGoldberg I would expect perl6 versions, when made into numbers, to be Rats, not floats. 16:59
dalek p/native-ref: 9b7f462 | jnthn++ | src/vm/jvm/QAST/Compiler.nqp:
Fix thinko in lexicalref late-bound comp.
 17:00
jnthn native-ref branch now spits out a CORE.setting.jar on JVM
...which explodes on loading.
TimToady m: my $a = (1,2,3); say $a[].WHAT 17:04
camelia rakudo-moar 67a75c: OUTPUT«(Parcel)␤»
TimToady that is borken, supposed to be List
m: my $a = (1,2,3); say $a.list.WHAT
camelia rakudo-moar 67a75c: OUTPUT«(List)␤»
BenGoldberg m: my $a = (1,2,3); say $a.perl
camelia rakudo-moar 67a75c: OUTPUT«$(1, 2, 3)␤»
japhb Is there a way in the Rakudo setting to detect whether a particular nqp:: op exists? I'd like to do "feature detection" to implement methods like IO::Path.resolve that work if the underlying backend supports the right ops (nqp::readlink, in this case), and NYI's if not. 17:05
BenGoldberg m: my $a = (1,2,3); say $a.list.perl
camelia rakudo-moar 67a75c: OUTPUT«(1, 2, 3).list␤»
TimToady $a[] is supposed to be equivalent to @$a
TimToady m: my $a = (1,2,3); say @$a.WHAT 17:05
camelia rakudo-moar 67a75c: OUTPUT«(List)␤»
japhb jnthn: ^^^ (my question above)
TimToady 'cause we been bragging that we don't have to use ->@* and such, just [] 17:06
(broke one of my quizzing programs too)
timotimo well, if you write a spectest ... :) 17:07
japhb Was pmichaud able to do GLR work, or did the "if I can't, someone else do it" clause get activated?
jnthn japhb: nqp::getcomp('QAST').operations gets you the object that'd be able to answer the question, if only it had a method to do so...
japhb *snort* 17:08
jnthn japhb: However, I don't quite like doing it that way...it's very cross-comp unfriendly if we end up there
japhb How is it cross-comp unfriendly, since it's querying what I presume would be the cross-comp? 17:09
jnthn japhb: nqp::getcomp('perl6').backend.supports-nqp-op('...') would be a nice API for it.
japhb: Trouble is it isn't... :)
japhb Oh, I think I understand your point.
jnthn You want to know the backend of the compiler you're using.
japhb: I actually need to do a similar thing quite soon anyway
japhb Oooh, can haz priority bump? 17:10
jnthn japhb: Yeah, I'll be in exactly the right places to do it.
japhb Thanks muchly.
jnthn Added it to my todo list next to the other thing :) 17:11
japhb heh
jnthn (The other thing is a soft dependency for native-ref merge, so you're likely to get it this weekend.)
japhb Nice
jnthn Depends whether I've got a second 80% to do on native-ref JVM port, or if the 20% remaining really is 20% :) 17:12
japhb :-)
Hmmm, that leaves me with a question, though ... such a feature allows me to detect if an op is supported ... but what happens on a backend that doesn't support it? Will the *compiler* complain that nqp::readlink isn't a known op, even if I make sure it never gets executed on that backend? 17:15
Or have I wandered into macro land? 17:16
japhb waves at amasak
japhb waves at masak
jnthn japhb: Uh...yeah, it'll complain it's missing 17:17
So not immediately sure how to use it...hmm.
japhb Yeah, I'm wondering if this is one of those cases where I really do need to use a macro or something macro-ish
raydiak .tell tadzik minor patch to fix another warning: github.com/tadzik/rakudobrew/pull/30 17:18
yoleaux raydiak: I'll pass your message to tadzik.
raydiak \o #perl6
japhb o/
dalek p/native-ref: b7388c6 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Implement decont_[ins].
 17:20
japhb FROGGS: Is there currently a way to specify a 'use lib' that is script-relative instead of $*CWD relative? 17:26
dalek kudo/native-ref: 631219a | jnthn++ | src/vm/jvm/Perl6/Ops.nqp:
Use decont_* ops to do unboxing.
 17:27
raydiak japhb: I usually use $?FILE.IO.parent or along those lines 17:28
japhb raydiak: This falls into the un-.resolve-d path problem I've been banging my head on. 17:31
raydiak japhb: have you looked through S32 IO? I saw like half a dozen different methods which looked pertinant 17:32
japhb raydiak: I'm thinking I just need to implement the beast myself, but 1) I want to make sure I'm not duplicating something that already exists, possibly in one of the S11 branches, and 2) I run up against stuff like the aforementioned "nqp::readlink only exists in nqp-m" problem. 17:33
raydiak: .resolve is NYI. :-/
tadzik raydiak: thanks :) 17:34
yoleaux 17:18Z <raydiak> tadzik: minor patch to fix another warning: github.com/tadzik/rakudobrew/pull/30
raydiak tadzik: you're welcome :)
japhb: does .canonpath help? I think it already exists (was playing with these things a bit last night, but was too tired to recall now what I saw) 17:35
japhb raydiak: .canonpath explicitly doesn't resolve ../
Because symlinks.
(I do appreciate the ideas, though. You just haven't yet hit on one I haven't tried so far. ;-) 17:37
dalek ast: 51c3b8e | TimToady++ | S02-types/array.t:
make sure .[] decontainerizes
raydiak japhb: so you're trying to get to the actual hardlinked location so that .. might take you somewhere other than where you appeared to start from (in case you were in a symlinked path)? 17:39
japhb raydiak: Well, there's two problems: 1. I need a true absolute path to feed to CompUnitRepo::Locally.new (because otherwise it assumes $*CWD-relative), 2. Someone might call my script using a relative path that contains ../, which means $?FILE and $*PROGRAM_NAME and such contain that, 3. I'm inside a symlinked tree, 4. IO::Path.parent is broken WRT to ../ path components. 17:44
japhb *multiple problems, I guess 17:48
raydiak m: .IO.parent.say for <.. ../foo foo/.. foo/../bar > 17:52
camelia rakudo-moar 695932: OUTPUT«"/home/camelia/../..".IO␤"/home/camelia/../../foo".IO␤"/home/camelia/foo".IO␤"/home/camelia/foo/..".IO␤»
raydiak yeah no brilliant ideas here, guess implementing .resolve is the shortest real solution 17:55
BenGoldberg m: ".".IO.say 18:04
camelia rakudo-moar 695932: OUTPUT«"/home/camelia".IO␤»
BenGoldberg m: ".".IO.parent.say 18:05
camelia rakudo-moar 695932: OUTPUT«"/home/camelia/..".IO␤»
japhb m: my $a = '../foo/bar'.IO.absolute.IO; $a.perl.say; $a.parent.parent.parent.perl.say; 18:08
camelia rakudo-moar 695932: OUTPUT«q|/home/camelia/../foo/bar|.IO(:SPEC(IO::Spec::Unix))␤q|/home/camelia|.IO(:SPEC(IO::Spec::Unix))␤»
japhb BenGoldberg: ^^ that is my pain right now
japhb Also, .IO.absolute.IO is ugly. 18:08
raydiak star-m: use NativeCall; sub realpath (Str:D $, Str $) returns Str is native {*}; realpath("..").IO # japhb: well, there's this... 18:11
camelia ( no output )
raydiak star-m: use NativeCall; sub realpath (Str:D $, Str $) returns Str is native {*}; realpath("..").IO.say
camelia star-m 2015.02: OUTPUT«"/home/camelia/rakudo-star-2015.02".IO␤»
raydiak star-m: use NativeCall; sub realpath (Str:D $, Str $) returns Str is native {*}; realpath(".").say 18:11
camelia star-m 2015.02: OUTPUT«/home/camelia/rakudo-star-2015.02/install␤»
raydiak star-m: say '.'.IO.absolute 18:12
camelia star-m 2015.02: OUTPUT«/home/camelia/rakudo-star-2015.02/install␤»
raydiak star-m: say '..'.IO.absolute
camelia star-m 2015.02: OUTPUT«/home/camelia/rakudo-star-2015.02/install/..␤»
[ptc] timotimo++ # for the pod-htmlify.t improvement 18:22
dalek c: 88af0dc | paultcochrane++ | lib/Type/IO (4 files):
Link to related roles/classes in IO roles/classes
c: bb3a7f5 | paultcochrane++ | lib/Language/5to6.pod:
Mention .lines when reading a text file
c: ae5d229 | paultcochrane++ | t/pod-htmlify.t:
Merge branch 'master' of github.com:perl6/doc
japhb raydiak: Why do you need star-m there?
Also, yeah, I considered the realpath method (though I couldn't remember the C name, thanks) 18:23
raydiak japhb: oh yeah forgot nativecall is in core now duh
timotimo [ptc]: eine meiner leichtesten übungen ;) 18:25
dalek p/native-ref: 40d219c | jnthn++ | src/vm/jvm/QAST/Compiler.nqp:
Elide many lexicalref/attributeref takes.
 18:28
dalek kudo/native-ref: ac026be | jnthn++ | tools/build/create-jvm-runner.pl:
Fix JVM runner generation on Windows.

  %CLASSPATH% may end up with spaces in, and also needs the correct
separator.		 18:29
[ptc] timotimo: ich wünschte mir, ich hätte die Lösung gesehen :-/ 18:31
timotimo: I'm still learning
timotimo that's very all right
i've needed almost this exact pattern in the past
at that point i wanted to react to the presence or absence of Term::ANSIColor 18:32
[ptc] timotimo: I kept having the problem that, even though I'd put the "use" inside a "try" it kept failing at compile time, hence the choice of BEGIN
timotimo ah, i understand that part
[ptc] wonders where this perl of wisdom should be written
timotimo but "EVAL" should have obsoleted the BEGIN 18:33
Kristien vendethiel: is LiveScript still being improved?
[ptc] timotimo: I had: try { use URI::Escape, CATCH { default { say "Need URI::Escape" }}} and that kept failing at compile time 18:34
timotimo yeah
damn you, perl6, for failing so early!
[ptc] timotimo: I also knew it was more complicated than necessary, but couldn't find a way out
timotimo er, hold on 18:35
timotimo running "plan 0;" and "exit" at BEGIN time may not be a good idea if you're in a module, because those get precompiled when modules are installed via panda 18:36
but you could have done something like my $has_escape = BEGIN { ... }
[ptc] yeah, the "plan 0" solution was to stop "prove" from complaining that no tests were planned/run 18:38
timotimo: could you try something out for me? Without URI being installed, run "prove --exec=perl6 -r t" in perl6/doc 18:41
timotimo: for me it fails by not being able to run the subtests (and tries to load URI::Escape in the process) 18:42
vendethiel Kristien: idk
dalek p/native-ref: 6559876 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/NativeRef (2 files):
Extend NativeRef REPR to include positional refs.
 18:43
p/native-ref: 42da59a | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Implement atposref_[ins].
vendethiel Kristien: I pretty much gave up on LS, around a year ago 18:44
Kristien: tho someone is adding sourcemaps
Kristien ok :p
vendethiel Kristien: actually -- that's unfair. I gave up on JS in general.
so I stopped working on both
Kristien Currently working on a lens library in LS 18:45
vendethiel Kristien: I think we have one already:P! 18:46
sorella (notlolita) has done a fair amount of ls stuff
timotimo [ptc]: gimme a minute 18:48
[ptc] timotimo: np
afk for a while # dinner & 18:49
timotimo i get an error from "use-ok" not being defined, because rakudo too old 18:50
nine_ timotimo: github.com/niner/Grammar-Highlight...341587f800 18:52
raydiak wonders why compress::zlib::raw fails to build on the smoker but wfm 18:59
raydiak also wonders what happened to cpandatesters
timotimo nine_: if you have style="..." already, i'd suggest you use the hsl function together with a pseudorandom spread over 0..255 to get nice colors 19:03
also, inside of postcircumfix:<[ ]> you can have a WhateverCode that gets passed the number of elements, so i'd suggest $color % * 19:04
pseudorandom can also mean "actually random but cached with the name of the rule as the key"
nine_ Finding good, distinct colors is a surprisingly difficult problem. 19:05
timotimo oh, agreed 19:07
timotimo [ptc]: you're right, it seems like it's fatal 19:07
oh, derp :)
it "use"s that module a few lines later
timotimo i have to hit the road right now, unfortunately 19:08
you can wrap it in EVAL q{ ... } 19:09
timotimo for now 19:09
[ptc] timotimo: the use-ok isn't as ripe ase it could b
*be
I'll give it a go
vendethiel just learned that groovy copied the " println (4 + 5) * 3" problem from perl5...
timotimo use-ok is all right
the use directly below it is problematic
ttyl :)
Gitu Hello 19:11
vendethiel o/ 19:13
dalek ast: feca340 | TimToady++ | S02-types/ (2 files):
make sure $hash{} decontainerizes too
Gitu My first question here: Can Perl 6 generate Windows executables? Also, can it be used to create COM servers (for Windows)? 19:14
nine_ @colors[$color % *] is really sweet :) 19:18
TimToady even sweeter will be when you can put the modmap into the declaratoin 19:19
which S09 speculates 19:20
TimToady though perhaps with a bogus syntax 19:20
FROGGS[mobile] Gitu: with a little bit of work we can create executables in future, but, the win32 api is not yet there as a module 19:21
Gitu: so com servers would be quite some work...
Gitu Froggs: Thank you. 19:23
dalek ast: 35b8f69 | TimToady++ | S02-types/array.t:
deconting array doesn't necessarily imply List
 19:24
FROGGS[mobile] Gitu: a very good first step would be to port the win32 module from Perl 5 19:25
Gitu Froggs: That module is able to create just COM clients, not servers. Under Perl, only ActiveState's PDK commercial solution can create COM servers. (Or am I wrong?) 19:27
FROGGS[mobile] ohh :o(
avuserow has anyone started a grammar for parsing SQL? in particular I'm wanting to parse .sql files output from mysqldump and similar tools.
FROGGS[mobile] I dont know
FROGGS[mobile] avuserow: tony-o made a slang::sql 19:28
IIRC
look in the irc logs
avuserow FROGGS[mobile]: I saw that one, but it doesn't appear like it actually generates an AST or similar from the SQL 19:30
it has a note in the README saying that it doesn't validate SQL yet :(
raydiak nine_: for the palette, just picking regular intervals around the hue circle can work well...multiplied by a desaturated and/or darkened variant of each hue plus a few greys I wouldn't be surprised if you could generate 20 or 30 adequately distinct colors 19:31
FROGGS[mobile] ohh..
b2gills I think for a COM server you would need at least a C shim 19:45
dalek c: 06ada66 | (Matthias Krull)++ | htmlify.p6:
Fixed generated links to Github files
 19:49
dalek ast: d2134c7 | TimToady++ | S02-types/hash.t:
correcter decontainerized test
 19:54
kudo/nom: 9d41570 | TimToady++ | src/core/ (2 files):
decontainerize bare Zen slices
japhb OK, I've got a locally working IO::Path.resolve, but right now it only works on POSIX filesystems. 20:13
japhb Anyone with both Windows and Linux access interested in fixing it to work on both? I can make guesses, but they'd probably be dead wrong ... 20:17
FROGGS m: use lib $?FILE.IO.dirname ~ '/lib' # japhb 20:19
camelia ( no output )
dalek ast: c5379ec | TimToady++ | S02-lists/tree.t:
yes, the test is correct, since Zen should decont
 20:20
jnthn NQP_VERBOSE_EXCEPTIONS 20:29
oops
japhb FROGGS: I need to handle the case that the script, the lib/, and the user are all in separate directories, possibly multiple levels apart with symlinks between. I'm planning basically 'use lib ($?FILE.IO.dirname ~ "/../../../lib").IO.resolve' 20:35
FROGGS japhb: that's not pretty but it works 20:36
japhb Actually, more like 'use lib ($?FILE.IO.dirname ~ "/../../../foo/bar/lib").IO.resolve'
itz www.topcoder.com/blog/if-programmi...-were-beer
japhb Am I the only one that finds the odd mixture of IO::Path methods that return Str or IO::Path rather odd (and *cough* PHP-like)? 20:38
.oO( odd mixture is odd)
psch japhb: newio exists, and lizmat++ looks like she knows what she's doing in there 20:39
*the branch
japhb nod 20:41
What should .resolve do with paths that don't exist past some point? realpath(3) will error, realpath(1) can work in "resolve symlinks, error on fake paths" or "don't resolve symlinks, but don't error on fake paths" mode. 20:42
japhb I was thinking about a DWIM: Start in "Resolving symlinks, handling real paths" mode, and the first time it hits a path element that doesn't exist, just append the rest of the path elements as-is except for dropping // and /./ 20:44
This has the downside of possibly leaving /../, but iff that could not be successfully resolved.
Which means if the missing intermediate path element were created later, the partially-resolved path would still be correct, and (more-)fully-resolvable. 20:45
japhb So if /usr/bin/X11 -> /usr/bin, then /usr/lib/../bin/X11/foo/../baz would return /usr/bin/foo/../baz, which could later be resolved completely if /usr/bin/foo were created as a directory or as a valid symlink. 20:48
japhb Seem sane, or YTICN? 20:48
Kristien We should call singletons "sigiltons" 20:53
japhb Hmmm, there's also the case of /dir/dir/file/bar/baz to contend with ... :-/ 20:58
Oh, actually, we probably want to let that slide, because 'file' can be a ZIP or TAR file that the filesystem can mount as a dir ...
geekosaur there's also streams in NTFS and HFS+ 20:59
japhb Yeah, the more I think about this, the more I think it's a good idea to be more lax ... after all, if the user wants to be sure the resolved thing is "real", they can always .IO.e on the result. 21:06
Ben_Goldberg On windows "//" right at the beginning of a path name has a special meaning. 21:10
Specifically, all named pipes start with \\ or, equivilantly, // 21:12
geekosaur UNC paths 21:13
/host/share/path 21:14
er
//host/share/path
right, my mental automatic doubling at the start of the line doesn't quite work if it's starting with // for other reasons :p
Ben_Goldberg Err, actually, \\ is the windows equivilant of unix's /dev/ 21:15
Windows named pipes start with "\\<ComputerName>\pipe\" .... and there are other devices in addition to pipes in the \\ namespace. 21:16
geekosaur right, it's a variant of UNC paths 21:17
"pipe" is a virtual share in UNC syntax 21:18
masak m: my $now = DateTime.now; $now.later 21:44
camelia rakudo-moar 9d4157: OUTPUT«use of uninitialized value $unit of type Any in string context in block <unit> at /tmp/RmyvdH1jbH:1␤␤use of uninitialized value $unit of type Any in string context in block <unit> at /tmp/RmyvdH1jbH:1␤␤use of uninitialized value $unit of type Any…»
masak that goes on for a while. ends with "Error while creating error string: No exception handler located for warn" 21:45
masak .oO( who implemented this stuff? ) :P 21:46
dalek p: b91a78d | FROGGS++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/SixModelObject.java:
improve error messages for repr mismatch
masak I see there's a &die in there for "More than one time unit supplied". there should be one for "No time unit supplied" as well, I guess.
masak attempts a quick fix 21:47
by the way, that method has the coolest given block I've ever written. any discussion about the variation in given statements should invlude that given statement as a matter of course. :) 21:49
include*
[ptc] .tell timotimo unfortunately the require version wasn't working at all (errors with and without prove). The check needs to take place at compile time. Don't know what the best solution is though... 21:56
yoleaux [ptc]: I'll pass your message to timotimo.
jnthn yays, native-is-rw.t and native-is-copy.t nowp ass on JVM \o/ 21:59
uh, now pass
colomon \o/
nowp ass!
dalek c: b481d52 | paultcochrane++ | t/pod-htmlify.t:
Wrap check for URI::Escape in BEGIN block

The "unless try" was a good solution, however the check for URI::Escape needs to happen at compile time, since the later "use" will be checked and thus fail since URI::Escape isn't there. What a better solution is, I don't know, however the test works again now.
p/native-ref: e6b895e | jnthn++ | src/vm/jvm/QAST/Compiler.nqp:
Fix a few thinkos in lexicalref compilation.
 22:00
p/native-ref: 134199d | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Don't confuse primitive type and reference kind!
kudo/native-ref: 7b24f43 | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Binder.java:
Update JVM binder impl for native is rw params.
 22:01
jnthn Leaves me with native-lvalue-subroutines.t to fix. 22:02
FROGGS jnthn++ 22:05
jnthn Ah, think I got that one too :) 22:10
colomon :)
colomon is glad someone is being productive today, as he is being spectacularly useless. 22:11
dalek kudo/native-ref: b57292d | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/RakOps.java:
Fix native return value decontainerization.
 22:16
bartolin in case someone is interested: I think I found a reasonable way to write tests for 'multi line input' for the REPL (which works on Windows and Unix platforms): gist.github.com/usev6/f9d0f6ca4e21...nt-1403334 22:25
FROGGS bartolin: ohh nice! 22:26
bartolin .oO ( AWS to the rescue ) 22:28
dalek pan style="color: #395be5">perl6-examples: 1edd74c | paultcochrane++ | rosalind/ (43 files):
Add vim coda to rosalind example files
 22:43
pan style="color: #395be5">perl6-examples: 1f15ddd | paultcochrane++ | rosalind/ (7 files):
Add use v6 pragma to rosalind examples
dalek osystem: 85c5876 | FROGGS++ | META.list:
remove zavolaj (NativeCall) from ecosystem, it lives in rakudo
 22:49
FROGGS jnthn: btw, we have now a way to compare module smoke tests across compiler versions... 22:50
jnthn FROGGS: ooh :) 22:51
FROGGS panda puts not every module test into a file to deduplicate test reports, and this file is created once per compiler version
jnthn JVM spectest is a tad messy, but looks mostly assignable to me missing patches from nom 22:52
FROGGS so, do a smoke run, copy the file to some place, switch rakudo branch, rerun and do a diff
that's how such a file looks like: gist.github.com/FROGGS/2a08ad542c1ebe08e4b2 22:53
(I am doing an ecosystem smoke run right now)
jnthn Yay, the couple of tests I suspected I'd busted something on r-j are also due to missing patches from nom :) 22:54
FROGGS that are the easy ones :o)
jnthn FROGGS: Ah, cool :)
FROGGS gnight 22:55
jnthn 'night, FROGGS++
lizmat gnight FROGGS
jnthn So, I'm down to the r-m perl6-bench performance regressions to look at ahead of native-ref being mergeable, it seems :) 22:56
I'll look at those tomorrow.
masak 'night, #perl6 22:59
dalek pan style="color: #395be5">perl6-examples: bb55df5 | paultcochrane++ | rosalind/afrq-grondilu.pl:
Use default data input for afrq rosalind example
 23:00
lizmat gnight masak
dalek pan style="color: #395be5">perl6-examples: b91954a | paultcochrane++ | rosalind/afrq-grondilu.pl:
Document afrq example origin and expected output
 23:02
lizmat japhb: re .resolve, I'd be very interested in trying that out 23:19
as a status update re the newio branch: I sort of hit a blocker on how to handle symlinks that point to non-existing dir entries 23:20
now that stat() is the default on Moar (as opposed to lstat before)
one of my assumptions was that any dir() entry that I get, by definition exists 23:21
this is no longer true if it is a symlink to a non-existing dir entry
in that case -e is false, and -le is trie
*true
now, one of my speedups was that one would not have to check existence on dir() entries 23:22
and now one needs to :-(
which sort of kicks the legs out from under the whole idea for speedup that I had
still pondering on how to continue...
jnthn lizmat: Where exactly do we need to check existence on them? 23:25
lizmat just before you want to check if something is a regular file or a directory, e.g. 23:26
m: (a => 0).Bag.perl.say # arguably, this should say (a=>0).Bag 23:30
camelia rakudo-moar 9d4157: OUTPUT«().Bag␤»
lizmat m: (a => 0).BagHash.perl.say # this is correct, in think
camelia rakudo-moar 9d4157: OUTPUT«().BagHash␤»
lizmat I'm thinking, that *if* you're creating an immutable structure, the removal of a key is removal of information 23:31
m: (a => -1).Bag.perl.say # but should this be true for any non-positive value? 23:33
camelia rakudo-moar 9d4157: OUTPUT«Type check failed for return value; expected 'Baggy' but got 'Failure'␤ in any return_error at src/vm/moar/Perl6/Ops.nqp:649␤ in method new-from-pairs at src/gen/m-CORE.setting:19715␤ in method Bag at src/gen/m-CORE.setting:2130␤ in block <unit…»
lizmat interesting 23:34
ok, something to look at after some sleep&
jnthn 'night, lizmat
jnthn also heads off to relax &
dalek Iish: 21fdbe8 | (Adrian Kreher)++ | / (2 files):
[mysql] Connect via socket

Add connection option to connect to a server via a (non-default) socket on the filesystem.		 23:50

Powered by Cro and the Raku® Programming Language.

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

Thank you!