»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 masak on 12 May 2015.
dalek osystem: f5886bd | tony-o++ | META.list:
adding json::faster

provides a significantly faster `to-json` method		 00:00
ShimmerFairy cbk1090: by the way, the /<!URL> ./ part can be written as /<-URL>/ . I just didn't write it that way since I wasn't quite sure of <-rule>'s behavior when I wrote that example :) 00:04
cbk1090 ShimmerFairy, ok I don't think <!URL> works because the program just hangs...? I'll try the other way.. 00:05
00:05 raiph joined
ugexe use Grammar::Tracer; (needs to be installed) and you will get a nice output of what the grammar is doing 00:06
cbk1090 ShimmerFairy, That was it! Thanks mucho! 00:07
ugexe, Ok I'll try that soon.
ShimmerFairy cbk1090: considering the <-URL> version worked, did you forget the dot after <!URL> ?
tony-o grammar::tracer is amazing for debugging grammars 00:08
ShimmerFairy cbk1090: important to note that you get Grammar::Tracer via panda install Grammar::Debugger , assuming you use panda of course :)
cbk1090 yes I'll use panda. But why is it important to do it like that not not just install by hand? 00:09
ugexe he means the name of the distribution is actually Grammar::Debugger (which contains Grammar::Tracer) 00:10
cbk1090 Oh
ShimmerFairy yeah, panda doesn't know how to find packages based on a module you want, which I'd be surprised if it never did. 00:11
ugexe soon im hoping to see like $cpan-or-ecosystem.candidates(whatever) 00:12
ShimmerFairy ugexe: by the way, whoever "he" is, he apparently meant the same thing with his suggestion that I did with my suggestion :P 00:13
tony-o ShimmerFairy: there isn't a gender associated with 'he' in some parts of the us, we just refer to people that way if we don't know 00:15
a group of people is 'you guys' even if there are women or whatever 00:16
ShimmerFairy tony-o: I'm well aware of English being male-default, and don't mind it, I just wanted to offer a correction in any case :)
tony-o it'd probably be more helpful if you wanted ugexe to refer to as some other pronoun 00:17
ShimmerFairy personally uses singular 'they', over-corrective grammarians be damned, so she doesn't confer a default gender on someone 00:18
tony-o: true, I just wanted to be a bit nicer/more subtle about it than "hey, I'm a 'she'", though it of course is more direct :) 00:19
TimToady ShimmerFairy: but singular 'they' is so newfangled, why, it wasn't introduced till like the 16th century! 00:20
tony-o personally, i'm a 12/13th century kind of guy 00:21
TimToady My Hobby: following prescriptive grammarians around and catching them using singular 'they' without realizing it
tony-o mongols win
ShimmerFairy TimToady: precisely why I think the grammarians are over-corrective :P . I also think it's perfectly to end sentences in prepositions, since that's a valid position for any word to be in. And if those infinitives would rather not be split, they need to surely be a one word construct.
*perfectly fine 00:22
tony-o i follow TimToady around and just call everything 'they' 00:23
ShimmerFairy TimToady: it's interesting that we have precedent for taking a plural into the singular (which is why I used "themself" like "yourself", despite wikipedia claiming "themself" is relatively less standard than the other parts of singular they)
skids feels validated about using 'they' most of the time.
TimToady if anyone doesn't, they should feel ashamed 00:24
ShimmerFairy TimToady: I honestly wouldn't be surprised if someday in the future the royal 'we' (or I guess "corporate 'we'" in modern context) got the same treatment, and it became the singular 1st person pronoun as well :)
tony-o TimToady: nice
TimToady then there's the editorial 'we' that means 'you' :) 00:25
tony-o i call that one the managerial 'we'
ugexe you are missing out on the interesting world of ebonics
tony-o 'we' need to do this work, meaning 'i need you to get this work done'
ShimmerFairy TimToady: My other real prediction for English's future is that someday we'll forget the / in "and/or", and then we'll finally have a word for inclusive or: "andor". 00:26
TimToady well, but the manager is planning to take credit :)
tony-o if you have a bad manager, yea
00:26 laouji joined
cbk1090 So I keep getting "Type check failed in binding $filename; expected 'Cool' but got 'IO::Handle' " 00:27
when I try and do this: my $match = url::Grammar.parsefile($fh);
ShimmerFairy
.oO("we need to do the work" means "you" to the employees and "I" to the superiors :P)
ugexe a filename, not a file handle
ShimmerFairy Although I don't think it'd be a bad idea for .parsefile to accept a file handle, now that it's come up. 00:28
cbk1090 here is how I open the file: $fh = open( 'inputFile.dat', :r);
ugexe it would be nice if the error mentioned Str(Cool) or somehow mentioned Str instead of cool
tony-o cbk1090: pass url::Grammar.parsefil('inputFile.dat') 00:29
but spell everything correctly
cbk1090 tony-o, ok....
ShimmerFairy ugexe: maybe "expected 'Cool' (that will be converted to 'Str')" ? (maybe adding "via .Str" after 'Str')
ugexe that would make some things more clear to the fresh meat i think 00:32
skids "Expected to convert Cool to Str, but got a... " maybe?
TimToady .tell laben >>op<< is intended primarily for finite, equal-sized arrays/matrices/vectors; there's little point in duplicating Z's behavior here 00:34
yoleaux TimToady: I'll pass your message to laben.
ShimmerFairy ugexe: like mentioned above, I think we could stand to have messages that weren't so insistent on being terse (once I make my Pod parser public, which'll hopefully be soon, I think I have a couple of nice examples) 00:35
skids I just tend to avoid parens when possible in those, unless it's just around a very short visual pill. 00:37
ShimmerFairy m: sub foo(Int ::INTEGER $a, INTEGER(Cool) $b) { say "OK!" }; foo(1, "2"); # something I noticed a while back. A bug, I would hope
GLRelia rakudo-moar d7ab32: OUTPUT«Method 'INTEGER' not found for invocant of class 'Str'␤ in sub foo at /tmp/ekIbB4ek0A:1␤ in block <unit> at /tmp/ekIbB4ek0A:1␤␤»
camelia rakudo-moar f7cfe9: OUTPUT«Method 'INTEGER' not found for invocant of class 'Str'␤ in sub foo at /tmp/Z1H6Vn1mhO:1␤ in block <unit> at /tmp/Z1H6Vn1mhO:1␤␤»
skids I think when I noticed that jnthn said something like "damn I was hoping nobody would try that" and it would be a while before that worked. 00:39
00:39 laouji left 00:40 laouji joined
ShimmerFairy skids: I can see why that'd be a pain, considering the implementation is Cool.INTEGER atm :) 00:40
m: sub foo(Int ::INTEGER $a) { say "5".INTEGER }; foo(1);
GLRelia rakudo-moar d7ab32: OUTPUT«Method 'INTEGER' not found for invocant of class 'Str'␤ in sub foo at /tmp/DBMmLHklY9:1␤ in block <unit> at /tmp/DBMmLHklY9:1␤␤»
camelia rakudo-moar f7cfe9: OUTPUT«Method 'INTEGER' not found for invocant of class 'Str'␤ in sub foo at /tmp/3yIOY999bj:1␤ in block <unit> at /tmp/3yIOY999bj:1␤␤»
ShimmerFairy m: sub foo(Int ::INTEGER $a) { say INTEGER.new("5") }; foo(1); 00:41
GLRelia rakudo-moar d7ab32: OUTPUT«This type cannot unbox to a native integer␤ in sub foo at /tmp/vufdctKWaR:1␤ in block <unit> at /tmp/vufdctKWaR:1␤␤»
camelia rakudo-moar f7cfe9: OUTPUT«This type cannot unbox to a native integer␤ in sub foo at /tmp/qTgCm1vGa7:1␤ in block <unit> at /tmp/qTgCm1vGa7:1␤␤»
00:42 laouji left, laouji joined 00:43 rurban joined
ShimmerFairy m: sub foo(Int ::INTEGER $a) { say "2"."{INTEGER.^name}"() }; foo(1) 00:43
GLRelia rakudo-moar d7ab32: OUTPUT«2␤»
camelia rakudo-moar f7cfe9: OUTPUT«2␤»
ShimmerFairy ^ skids: well, there's one (terrible) way to do it :P
m: sub foo(Int ::INTEGER $a) { say "2".$(INTEGER.WHAT) }; foo(1) # slightly less bad 00:49
GLRelia rakudo-moar d7ab32: OUTPUT«2␤»
camelia rakudo-moar f7cfe9: OUTPUT«2␤»
00:51 adu joined 00:53 llfourn joined 00:57 adu left 00:58 llfourn left 01:00 aborazmeh joined, aborazmeh left, aborazmeh joined 01:10 rangerprice joined
rangerprice Hi 01:10
Zoffix \o
01:14 amurf left 01:22 Peter_R left 01:37 dayangkun joined 01:43 rurban left 01:56 aborazmeh left 01:59 adu joined
dalek osystem: d5a2d88 | retupmoca++ | META.list:
Steal Text::Markdown name from masak.

As per IRC Logs 2015-05-28. This is somewhat delayed, so please revert if this is now unneeded.
While not yet a 100% implementation, this module implements more Markdown than the current one.		 01:59
01:59 llfourn joined
adu raydiak: hey 02:00
02:04 raiph left 02:09 raiph joined 02:16 vendethiel joined 02:21 noganex_ joined 02:23 noganex left 02:24 rangerprice left
raydiak \o adu 02:32
adu o/
how goes?
raydiak not bad...working more than anything else lately...you? 02:33
02:34 khw joined
adu yeah, same 02:36
labster that sounds suspiciously similar to my life 02:38
raydiak seems to be a common situation :) 02:41
02:45 dwarring joined 02:52 yqt left
dalek ast/glr: 4626a7d | (David Warring [email@hidden.address] | integration/advent2012-day (2 files):
GLR flattening in 2010 advent days 10 & 21
 02:52
dwarring ^^ s/2010/2012/ 02:54
02:54 mohij_ joined 02:58 mohij left 03:02 kaare_ joined 03:16 raiph left 03:19 raiph joined 03:21 rangerprice joined 03:23 adu left 03:24 yeahnoob joined 03:36 vendethiel left 03:46 laouji left
dalek ast/glr: 57cb607 | (David Warring [email@hidden.address] | integration/99problems-01-to-10.t:
GLRify 99problems-01-to-10.t
 04:01
04:05 aborazmeh joined, aborazmeh left, aborazmeh joined, telex left 04:06 telex joined 04:10 davercc joined 04:14 rangerprice left 04:17 AlexDaniel left
dalek ast/glr: 698541c | (David Warring [email@hidden.address] | integration/99problems-11-to-20.t:
GLRify 99problems-11-to-20.t
 04:41
04:44 diana_olhovik joined 04:46 khw left 04:52 laouji joined 04:54 skids left 05:06 espadrine joined 05:14 araujo_ left 05:18 diana_olhovik left 05:19 laouji left, laouji joined 05:20 araujo joined, araujo left 05:29 cbk1090 left 05:35 araujo joined, araujo left
moritz \o 05:56
m: sink map { .say }, 1..4
GLRelia ( no output )
camelia rakudo-moar f7cfe9: OUTPUT«1␤2␤3␤4␤»
05:59 rurban joined 06:07 diana_olhovik_ joined 06:11 brrt joined 06:12 xzenfar joined 06:13 ggherdov left 06:14 colomon left 06:18 CQ2 joined 06:22 ggherdov joined
moritz m: say (map { .say }, 1..4).^name 06:27
GLRelia rakudo-moar d7ab32: OUTPUT«Seq␤»
camelia rakudo-moar f7cfe9: OUTPUT«List␤»
moritz m: say (map { .say }, 1..4).sink
GLRelia rakudo-moar d7ab32: OUTPUT«1␤2␤3␤4␤Nil␤»
camelia rakudo-moar f7cfe9: OUTPUT«1␤2␤3␤4␤Nil␤»
moritz huh, sink statement prefix doesn't call .sink anymore?
06:28 xfix joined
moritz huh, sink calls .eager 06:28
and eager isn't eager
06:29 salva joined
moritz ... and changing it to calling .sink makes session compilation loop/nhang 06:34
masak morning, #perl6 06:36
moritz good morning, masak 06:37
06:41 zakharyas joined
nine dwarring++ # plowing through those advent spec tests 06:44
moritz: I think we have enough evidence now that we need to make .eager really eager 06:45
dalek kudo/glr: d747af1 | (Stefan Seifert)++ | src/core/native_array.pm:
Make array.new behave like Array.new WRT flattening of args

22:30 < nine> jnthn: array.new takes *@values while Array.new takes **@values.
   Which one is right?
22:30 < jnthn> **		 06:46
06:47 jkva joined 06:49 [Tux] left, [Tux] joined 06:50 araujo joined 06:51 araujo left 06:52 araujo joined, araujo left, araujo joined, araujo left 06:57 domidumont joined 07:01 domidumont left, domidumont joined 07:04 [Sno] left 07:06 [Sno] joined 07:07 abraxxa joined, Ven joined 07:10 laben joined
laben hello p6'ers 07:10
yoleaux 00:34Z <TimToady> laben: >>op<< is intended primarily for finite, equal-sized arrays/matrices/vectors; there's little point in duplicating Z's behavior here
moritz nine: hack.p6c.org/~moritz/0001-Make-eage...be-e.patch is my current attempt 07:15
nine: the test that regresses is my @res = (one X @two)[^20], which dies with "this Seq has already been iterated" 07:17
dalek ok: 7998e63 | (Steve Mynott)++ | src/classes-and-objects.pod:
fix minor typos
 07:19
ok: a5f3c17 | moritz++ | src/classes-and-objects.pod:
Merge pull request #84 from stmuk/master

fix minor typos
07:23 darutoko joined 07:28 mr-foobar left
masak I think www.webkit.org/blog/3846/ will be of interest to some here; especially the bit about compilers at the end. 07:28
07:29 araujo joined
nine moritz: I cannot reproduce an error there? Tried my @one = 1, 2; my @two = 1, 2; my @res = (@one X @two)[^20]; say @res.perl; and I get [$(1, 1), $(1, 2), $(2, 1), $(2, 2)] with your patch applied 07:30
07:30 araujo left
moritz nine: one is a Seq in that test, not an array 07:31
nine: just run t/spec/S02-types/lazy-lists.t with my patch applied
nine moritz: I wonder if it's really your patch's fault or METAOP_CROSS' 07:35
err... infix:<X> of course 07:37
moritz huh 07:38
infix:<X>(|lol) { begins with if lol.hash { ... } 07:39
it tries to coerce its argument list to hash?
nine lol is a Capture, isn't it?
It's just looking for adverbs, though I didn't know you could adverb infix operators 07:40
moritz oh, right
you can, they just go after the right arg, or something like
m: say 1 + 4:weird 07:41
GLRelia rakudo-moar d747af: OUTPUT«Unexpected named parameter 'weird' passed␤ in block <unit> at /tmp/0BgAUq7RmU:1␤␤»
camelia rakudo-moar f7cfe9: OUTPUT«Unexpected named parameter 'weird' passed␤ in block <unit> at /tmp/73GMn2FhHb:1␤␤»
nine infix:<X> is not changed much from the implementation in nom. It even uses .elems which I see as a bad sign. 07:46
moritz and it uses eager, fwiw
nine And it was one of the first things I worked at. I really didn't know what I was doing back then :) 07:48
masak I've heard lots of people say such a thing over the years :) 07:52
yes, infix:<X> should be at least partly lazy. 07:53
moritz infix:<X> should ba lazy in the first argument
masak right
m: constant ll := gather { say "lazy"; take 1; say "so lazy"; take 2 }; say "OH HAI"; say ll[0] 07:54
GLRelia rakudo-moar d747af: OUTPUT«OH HAI␤lazy␤1␤»
camelia rakudo-moar f7cfe9: OUTPUT«OH HAI␤lazy␤1␤»
nine And it should collect the other argument's values in IterationBuffers unless it _knows_ that it can access them multiple times
masak m: constant ll := gather { say "lazy"; take 1; say "so lazy"; take 2 }; say "OH HAI"; say ll X <a b c>
GLRelia rakudo-moar d747af: OUTPUT«OH HAI␤lazy␤so lazy␤1 a 1 b 1 c 2 a 2 b 2 c␤»
camelia rakudo-moar f7cfe9: OUTPUT«OH HAI␤lazy␤so lazy␤1 a 1 b 1 c 2 a 2 b 2 c␤»
moritz m: my \seq = <a b c>.sort; say seq X 1, 2 07:55
GLRelia rakudo-moar d747af: OUTPUT«a 1 a 2 b 1 b 2 c 1 c 2␤»
camelia rakudo-moar f7cfe9: OUTPUT«a 1 a 2 b 1 b 2 c 1 c 2␤»
07:56 lizmat joined 07:57 cgfbee left 08:01 araujo joined, araujo left, araujo joined
Ven waves at #perl6 – o/ 08:01
08:02 RabidGravy joined
masak \o 08:05
lizmat good *, #perl6 from Home 08:08
yoleaux 24 Aug 2015 09:30Z <jnthn> lizmat: I'd be tempted ot say that emit NEVER tries to iterate anything and just emits what it's given verbatim :)
lizmat m: my @a; sub a(\a) { @a.push(a) }; a 1..10; say @a.perl # how do we fix this ? 08:09
camelia rakudo-moar f7cfe9: OUTPUT«[1, 2, 3, 4, 5, 6, 7, 8, 9, 10]<>␤»
GLRelia rakudo-moar d747af: OUTPUT«[1, 2, 3, 4, 5, 6, 7, 8, 9, 10]␤»
lizmat fwiw, I expected to see [1..10] there
08:12 cgfbee joined
moritz lizmat: maybe github.com/rakudo/rakudo/pull/512 fixe it? 08:16
lizmat perhaps, but I'd like to see jnthn's opinion on that PR first 08:17
I mean, this is one of the WAT's of the one argument rule
m: my @a; sub a(\a) { @a.push(a) }; a (1..10,); say @a.perl # does this fix?
camelia rakudo-moar f7cfe9: OUTPUT«[1, 2, 3, 4, 5, 6, 7, 8, 9, 10]<>␤»
GLRelia rakudo-moar d747af: OUTPUT«[1..10]␤»
lizmat so yes, that does fix it, but it transfers responsibility to the user, not to the lib writer 08:18
I mean, a sig like (\a) can't get much untouchable than that, still it is touched by the rule 08:19
moritz it is? 08:20
lizmat I don't see how I can write sub a in such a way that "NEVER tries to iterate anything and just emits what it's given verbatim"
moritz m: sub f(\a) { say a.perl }; f 1..3
GLRelia rakudo-moar d747af: OUTPUT«1..3␤»
camelia rakudo-moar f7cfe9: OUTPUT«1..3␤»
moritz doesn't seem touched.
08:20 rurban left
moritz sub identity(\thing) { thing } 08:21
never tries to iterate anything
well, might need 'is rw' to be a try identity 08:22
lizmat \a is rw ? 08:23
moritz sub identity(\thing) is rw { thing }
lizmat isn't that like double ?
ah
08:23 jkva left
lizmat hmmm... 08:24
08:25 jkva joined, sno joined 08:26 [Sno] left
moritz ok, creeping; returning a Seq in a map block causes it to be iterated 08:27
m: my @seqs = map { <a b c>.sort }, 1, 2; .say for @seqs[0]
GLRelia rakudo-moar d747af: OUTPUT«a␤b␤c␤»
camelia rakudo-moar f7cfe9: OUTPUT«a␤»
moritz or not?
now I'm confused
m: say <a b c>.sort.^name 08:28
camelia rakudo-moar f7cfe9: OUTPUT«Parcel␤»
GLRelia rakudo-moar d747af: OUTPUT«Seq␤»
08:28 rurban joined 08:29 dakkar joined, KCL_ joined
moritz oh, not if it's itmized 08:31
m: my \s = map { $_ }, 1, 2, 3; say s.^name; .say for (map { s }, 1)[0]
GLRelia rakudo-moar d747af: OUTPUT«Seq␤This Seq has already been iterated, and its values consumed␤ in block <unit> at /tmp/IimJ7dpyxr:1␤␤»
camelia rakudo-moar f7cfe9: OUTPUT«List␤1␤»
moritz m: my \s = map { $_ }, 1, 2, 3; say s.^name; .say for (map { s.item }, 1)[0]
GLRelia rakudo-moar d747af: OUTPUT«Seq␤1␤2␤3␤»
camelia rakudo-moar f7cfe9: OUTPUT«List␤1 2 3␤»
moritz is that intentional? 08:32
08:32 KCL left
moritz I thought only a Slip would do that 08:32
lizmat moritz: consider me confused...
but that can very well be attributed to jet lag and lack of sleep in general
if not age :-) 08:33
08:33 |Tux| joined
moritz lizmat: confused by the current behavior? Or confused by what I said? 08:33
lizmat: or by the push thingy?
masak lizmat! \o/
welcome back to Europe :) 08:34
lizmat moritz: by the push thing
masak o/
yeah, much better weather here :-)
.oO( until we hid Grenada )
moritz lizmat: I think you were confused about where the Range is flattened out into a list; it's not by the parameter binding, but by @array.push
lizmat *hit
yeah, I get that now
but in the case of emit not being able to emit a Range, there I still don't see where the iteration happens 08:35
moritz which emit?
lizmat I mean, emit is basically:
moritz lacks context
masak "hit Granada" sounds exceptionally painful. 08:36
lizmat method emit(Supply:D: \msg) { if self.tappers -> \tappers { .emit().(msg) for tappers }
nothing iterating there, afaics
aaaaahhhhh the plot thickens... (I think) 08:37
lemme check some things
laben lizmat: if the tappers got *@a signature, it's flattened/iterated, no?
lizmat yeah, but that's not about the values being emitted, that's just about which taps to serve 08:38
laben m: sub foo(*@m) { for ^@m {.say}}; foo(1..10) 08:39
GLRelia rakudo-moar d747af: OUTPUT«0␤1␤2␤3␤4␤5␤6␤7␤8␤9␤»
camelia rakudo-moar f7cfe9: OUTPUT«0␤1␤2␤3␤4␤5␤6␤7␤8␤9␤»
laben m: sub foo(**@m) { for ^@m {.say}}; foo(1..10)
GLRelia rakudo-moar d747af: OUTPUT«0␤»
camelia rakudo-moar f7cfe9: OUTPUT«0␤»
moritz m: sub foo(**@m) { for @m {.say}}; foo(1..10) 08:40
GLRelia rakudo-moar d747af: OUTPUT«1..10␤»
camelia rakudo-moar f7cfe9: OUTPUT«1..10␤»
moritz m: sub foo(**@m) { say @m.elems }; foo(1..10) 08:43
GLRelia rakudo-moar d747af: OUTPUT«1␤»
camelia rakudo-moar f7cfe9: OUTPUT«1␤»
dalek kudo/glr: 48791be | moritz++ | src/core/ (3 files):
Make eager actually eager, not just a promise to be eager later on

also both List.eager and Seq.eager return a List now. This causes a regression in S02-types/lazy-lists.t, but makes
  "try { eager map { die }, 1 }" work.		 08:44
kudo/glr: 90d45ed | moritz++ | src/core/List.pm:
More idiomatic use of eager
kudo/glr: 508d223 | moritz++ | src/core/List.pm:
Fix infix:<X> with Seq as first argument
kudo/glr: ab408b1 | moritz++ | src/core/Seq.pm:
add Seq.is-ready for debugging purposes

it returns False if this Seq has already been iterated
laben mybe a better name would be already-consumed or is-done or is-produced... 08:48
lizmat
.oO( dog in the pot )
 08:49
masak laben: .spent 08:51
aw, it's only morning, and I already used up my bikeshedding token... :P 08:52
laben yeah or even better is-spent
masak only if there are other 'is-' methods already on that object. I think hyphens have a very high design cost. 08:53
that ain't really motivated just to prepend a redundant 'is'
lizmat
.oO( postfix a ? :-)
masak and then slap a Ruby 8.0 on the box, and ship it!
if someone creates a slang where you can add a `?` to the end of function and method names, please call it Slang::Riddler :P 08:54
laben masak: rejected. hyphens are very cool and much better than doing CamelCase or under_case or strangeCase. or maybe it's Lisp speaking to me...
08:55 krunen joined
masak I only had one bikeshedding token, so... 08:55
laben masak: the sky is the limit .u upwards arrow
.u upwards arrow
yoleaux U+034E COMBINING UPWARDS ARROW BELOW [Mn] (◌͎)
U+2191 UPWARDS ARROW [Sm] (↑) 08:56
08:57 jkva left
masak .u 2250 08:57
yoleaux U+2250 APPROACHES THE LIMIT [Sm] (≐)
08:57 colomon joined
masak .u limitation 08:57
yoleaux U+4DFB HEXAGRAM FOR LIMITATION [So] (䷻)
08:58 jkva joined
dalek ast/glr: 06d424d | lizmat++ | packages/Test/Tap.pm:
Make sure we don't iterate when checking emissions
 09:02
09:02 espadrine left
lizmat this does *not* fix the problem, fwiw 09:03
09:03 FROGGS_ joined
moritz fwiw Seq already as an is-lazy method 09:03
lizmat
.oO( didn't know moritz was French )
TimToady e
09:04 rindolf joined, FROGGS left 09:08 colomon left
Ven masak: someone (FROGGS++) already created such a slang. I think you elected to give someone a book for that then.. 09:08
masak: github.com/FROGGS/p6-Slang-Piersing 09:09
masak oh right, that's what Piersing does 09:12
slangs as they stand today -- it's both fantastic that they exist and have the reach they do, but there's also the feeling that they could be much nicer/more consistent, somehow 09:14
I like the look of <[?!]>?
:)
JimmyZ selang ....
masak it occurs twice in github.com/FROGGS/p6-Slang-Piersin...iersing.pm
clearly an opportunity was missed in factoring it into <hunh>? 09:15
JimmyZ masak: you may know what selang means in chinese :P
masak :) 09:16
DrForr Dangerously close to interrobang :)
09:16 yeahnoob left
masak JimmyZ: seems to mean en.wikipedia.org/wiki/Satyr 09:16
JimmyZ masak: yeah 09:17
09:19 Ven left 09:21 FROGGS_ is now known as FROGGS
FROGGS masak: clearly the way we implement slangs today is just a bad cheat 09:22
09:24 colomon joined 09:26 espadrine joined
masak FROGGS: will we meet in the next two weeks? maybe we can talk about it in person. 09:29
09:29 colomon left
_itz the .ch act site /. ed? 09:30
or my browser crashed
masak _itz: act.perl-workshop.ch/spw2015/ works here. 09:32
09:34 colomon joined
_itz act never likes me :/ 09:35
09:35 colomon left 09:37 virtualsue joined 09:48 jkva left
masak is it fair to say that the new Slip() thing represents Perl 5's (auto-)flattening, but that it's now opt-in in many situations? 09:50
09:50 brrt left
masak trying to understand GLR a bit here. 09:50
09:50 jkva joined
sergot hi #perl6 \o 09:51
09:53 colomon joined
nine masak: yes, it is. 09:54
masak: There's flat and | to flatten from the outside and Slip() to flatten from the inside. 09:55
TimToady well, I'd say it's more like P5's use of @{} to deref a scalar 09:56
09:58 trnh joined
nine moritz: FWIW I think we should get rid of the my @l = do for loop in infix:<X> (and others that are similiarly structured), as the flattening problem doesn't exist anymore anyway. A plain map would n 09:59
moritz: FWIW I think we should get rid of the my @l = do for loop in infix:<X> (and others that are similiarly structured), as the flattening problem doesn't exist anymore anyway. A plain map would do nicely.
virtualsue "your online payment has been approved" but i didn't make one o-O 10:01
jdv79 virtualsue: ditto. hope they take cash at the door. 10:02
nine And since I'm at spewing wisdom: I think we can simply replace LoL by an actual List of Lists as in 1,2;2,3 becomes ((1, 2), (2, 3)) 10:03
virtualsue yeah. i don't think i'll be able to effect a bank transfer particularly easily
10:03 anaeem1 joined
_itz I'd feel sorry for the Nigerian Perl Workshop organisers should there ever be one 10:04
moritz nine: yes, I suspected that much; will actually try now
10:04 aborazmeh left 10:05 colomon left
ShimmerFairy nine: naw, that's crazy talk! .oO(It should be an Array of Lists, obviously, to proudly show off that we have a new list design that has just two main list types) 10:05
moritz nine: ah, but we need the index inside anyway, because the left-most arg isn't part of the Infinity consideration
nine moritz: I still think a plain map would be easier to reason about. But that could just be my limited experience with Perl 6 speaking. 10:07
jdv79 would it be a bad idea to allow multiple doc langs? 10:08
if people like markdown why not allow that - that's the sorta vague idea i had.
moritz jdv79: and here I thought you meant documentation in different natural languages 10:09
jdv79 i was just thinking about ShimmerFairy's latest blog post about fixing up pod6. but isn't that just another re-invention that may not be necessary? 10:10
ShimmerFairy jdv79: not really, it's just fixing up what's already there
nine moritz: also don't we have loop phasers now for skipping the first iteration? 10:11
jdv79 but i can mix p5 and p6 and python. why not pod6 and markdown and whatever other doc format? just a random thought. 10:13
ShimmerFairy jdv79: you can subvert the Pod parser with a DOC INIT (there's a theoretical example in S26) if you really insisted on your own doc format :) 10:14
As well any documentation format (like Doxygen IIRC) that can integrate with existing languages' comment syntax and such, of course. 10:15
jdv79 nice 10:16
ShimmerFairy (Also, I personally think Pod should get its own spot on the language braid, which would mean that it can be accessed through something like $~POD and modified like any other component of the language)
jdv79 my motivation was just laking people learn less. learn p6 - fine. but don't want to learn pod6 - use whatever you already know. 10:17
*making
ShimmerFairy Well, I don't think Pod6 is asking anything much more from people wanting to learn Perl 6, but in any case that's what slangs are for, modifying the language where you just can't agree with it :) 10:18
lizmat wow, bare startup in GLR now at 93 milliseconds for me :-)
jdv79 its def a weak argument of a use case but at least its possible 10:19
10:20 Ven joined
jdv79 lizmat: what is bare startup to you? 10:20
lizmat perl6 -e '' 10:21
nine It's 74ms here :)
lizmat nine: then you have a faster machine then I do (or a superior OS :-)
ShimmerFairy The biggest change I'm imagining for Pod6 at the moment is to have #= be the only declarator syntax, since I find having both leading and trailing declarator comments to be confusing and not very useful. (I'd have #= be the leading style, since I think doc comments generally work better there. Use of '=' despite conflicting with existing syntax just because = looks like a Pod thing) 10:22
jdv79 i see nom at ~115ms and glr at ~ 20ms here
nine Intel(R) Core(TM) i7-4700MQ CPU @ 2.40GHz
20ms!?
jdv79 pretty sweet but wonder exactly what caused that diff
oh, hold on. errored. 10:23
TimToady leading comments don't work well on things like parameters
Ven ShimmerFairy: if you do that, please notice me so I can change my talk from #| to #=
jdv79 oh that's less interesting. 102ms on glr 10:24
ShimmerFairy Ven: I will, but I really don't like conflicting with existing syntax anyway :)
lizmat nine: i7 2.8 Ghz here, so it must be a superior OS :-) 10:25
or perhaps a faster disk (SSD in my case)
ShimmerFairy TimToady: my thinking was that documenting at the parameter level is a level of detail that's already best served by having detailed documentation elsewhere, as opposed to inline.
nine lizmat: or just a difference in measurement. I just use time perl6 -e '' and look at the "real" line
lizmat: the importent part is: we're becoming fast :) 10:26
lizmat nine: that's what I do
yes...
ShimmerFairy TimToady: I think my biggest gripe with the two styles is that it's a bit too unrestricted for my tastes (e.g. can have multiple leading or trailing on something, can have both on one thing), and that the existing spellings are not at all indicative of what kind of declarator comment they are.
TimToady I am -1 to that change because the point of | is to make a false left margin, visually, while the purpose of = is to tie the left thing to the right thing 10:28
nine Ah, as I feared. If we replace LoL by a real List of Lists, it's kinda hard to write a multi candidate for sub postcircumfix:<[ ]> 10:29
But maybe we can make LoL a role instead?
ShimmerFairy I see, though that isn't clear to me without that explanation :) . And there's an argument the #= could be the leading comment because it looks like a pod directive, so it makes sense to start of something with that
TimToady nine: yes, that's what we said some time ago, but maybe subscripts can bet special
*be
ShimmerFairy: then it would be visually confusing if it's on the left, I suspect 10:30
ShimmerFairy To me, #^ and #$ would more clearly illustrate the leading/trailing distinction, but of course those look like regex-y comments, rather than Pod comments :)
nine If we use a List.new does LoL, we can use that for multi dispatch and otherwise don't have to worry about LoL anywhere. 10:31
TimToady the parser could notice your subscript has a ; in it, and dispatch to, say postcircumfix:<[ ;]> instad 10:32
I think that would be simplest 10:33
nine Then we couldn't do: my $position = (1;1;2); say @multidim[|$position];
_itz .tell [ptc] examples.perl6.org appears to have stopped updating
yoleaux _itz: I'll pass your message to [ptc].
10:34 brrt joined
TimToady @multidim[|$position;] maybe 10:35
or [|$position] is always assumed to be multi 10:36
nine That would very probably hurt the overwhelming majority of single dim array access performance wise. 10:37
TimToady only if you use |, otherwise it goes directly to the [] insted of the [ ;] function 10:43
YuviPanda Wrote my first (and very small!) perl6 module! 10:44
needs some more cleanup and docs and maybe tests...
TimToady or we actually implement the originally specced || vs |
YuviPanda++
YuviPanda (github.com/yuvipanda/perl6-Ident-Client) 10:46
TimToady but certainly we have proof that the LoL approach works, and I suppose the extra type doesn't hurt performance all that much
10:49 dayangkun left
YuviPanda so the DESTROY in github.com/yuvipanda/perl6-Ident-C...Client.pm6 10:51
can I expect that to be called consistently enough to leave the closing of the socket to it?
nine YuviPanda: better not
YuviPanda so it's mostly 'best effort'? 10:52
what's the reccomended way to deal with these things?
have an explicit .close() method?
nine yes 10:54
YuviPanda ok 10:55
so what kind of things can you rely on DESTROY for?
moritz you can rely on the fact that Perl 6 will call it if it's under memory pressure 10:59
10:59 vendethiel joined 11:00 TEttinger left
YuviPanda hmm right 11:00
so how do I do something like python's 'with' 11:01
moritz there are LEAVE phasers (though the caller's responsiblity) 11:02
given open($filename, :w) { LEAVE .close; # do stuff here } 11:03
YuviPanda aaah, nice 11:04
cool
jdv79 there's no way to get attr defaults before BUILD is completed, right?
11:10 vytas joined
TimToady my $handle will leave { .close } = open($filename, :w); # another way 11:16
11:16 Ven left
laben jdv79: what do you mean? you can pass args to new which will pass them to BUILD 11:16
11:17 ab6tract joined, colomon joined
ab6tract TimToady: that's quite interesting indeed! are all phasers exposed via 'will' ? 11:18
yoleaux 21 Aug 2015 21:34Z <RabidGravy> ab6tract: the touchstone of any language is whether you can write great software in it not whether the language is perfect. It's going to be great
21 Aug 2015 21:45Z <raiph> ab6tract: I share your passion to get things right. If I upset you, that was the opposite of my intent and I apologize.
22 Aug 2015 02:34Z <lizmat> ab6tract: re irclog.perlgeek.de/perl6/2015-08-21#i_11096207 , .uniq looks like the Unix uniq command, but that actually works like .squish
11:18 Ven joined
lizmat afk for a few hours& 11:19
11:20 trnh left
ab6tract p6doc -f will: No documentation found for a routine named 'will' 11:20
;(
oha ab5tract, it will 11:21
ab6tract .tell raiph Not to worry, mate. I just had to run, and am always bothered by a "then fix it" response to a (relatively) major language design discussion 11:22
yoleaux ab6tract: I'll pass your message to raiph.
ab6tract lizmat: re: .uniq ... it's a shame that only a naming conflict with a crypti-nomic unix command can lead to de-crypto-nomification of a p6 routine :( 11:23
11:25 vendethiel left
ab6tract laben: if I understood jdv79 correctly, it was about the default values that were specified by the class. `has $!var = 'foo' ` 11:25
TimToady m: say "Before"; { my $x will enter { say "Entered!" } = 42; }; say "After" 11:27
GLRelia rakudo-moar ab408b: OUTPUT«Before␤Entered!␤After␤»
camelia rakudo-moar f7cfe9: OUTPUT«Before␤Entered!␤After␤»
TimToady ab6tract: pretty much, except for maybe CATCH and CONTROL 11:28
moritz m: my $x will leave { "say $x" }; $x = 5 11:29
GLRelia rakudo-moar ab408b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/sxu5Y0Z4Ys␤Variable '$x' is not declared␤at /tmp/sxu5Y0Z4Ys:1␤------> 3my $x will leave { "say 7⏏5$x" }; $x = 5␤»
camelia rakudo-moar f7cfe9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/HBm06yJTOm␤Variable '$x' is not declared␤at /tmp/HBm06yJTOm:1␤------> 3my $x will leave { "say 7⏏5$x" }; $x = 5␤»
moritz huh?
that looks wrong 11:30
masak looks like a bug to me.
$x is declared at that point.
ab6tract TimToady: that's pretty neat :) .. though I wonder if this is one of those things that could have maybe been implemented adverbially but couldn't be due to the era of P6 when it was introduced?
masak moritz: want me to rakudobug it?
moritz masak: yes please 11:31
masak does 11:32
laben ab6tract: i dont understand what he was trying to do. do you mean trying to access the attr inside BUILD expecting it to have the default value? 11:33
masak m: my $x will enter { say $x } = 2 11:34
GLRelia rakudo-moar ab408b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/WAJGS0aaQC␤Variable '$x' is not declared␤at /tmp/WAJGS0aaQC:1␤------> 3my $x will enter { say 7⏏5$x } = 2␤»
camelia rakudo-moar f7cfe9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/qLbm4QEvKz␤Variable '$x' is not declared␤at /tmp/qLbm4QEvKz:1␤------> 3my $x will enter { say 7⏏5$x } = 2␤»
ab6tract laben: I can't speak for what they wanted it for, but I took it as "can I introspect the default value within BUILD" 11:35
laben then why would you think the attr has been initialised before BUILD? 11:36
ab6tract laben: what do you mean? first, this was not my question. second, I don't think that. but if attr's arrived in BUILD with the default value already installed, you could introspect that default value 11:38
laben sorry, the "you" was not referring to you specifically but generically like 'why would a person <etc>'. about the attr, if the default is not installed within BUILD, there's no other place, or is there? 11:41
ab6tract laben: you can set a default via the attribute definition without bothering with a build 11:42
11:43 RabidGravy left
ab6tract m: class T { has $.s = "silly string" }; my $t = T.new; say $t.s 11:43
camelia rakudo-moar f7cfe9: OUTPUT«silly string␤»
GLRelia rakudo-moar ab408b: OUTPUT«silly string␤»
laben yes, but if you do write the BUILD, there's no auto default. or am i wrong?
_itz is print buffered? I ask since I get different results with print "A"; sleep 2; print "B";
on perl5 and perl6?
ab6tract m: class T { has $.s = "silly string"; submethod BUILD { } }; my $t = T.new; say $t.s 11:44
GLRelia rakudo-moar ab408b: OUTPUT«silly string␤»
camelia rakudo-moar f7cfe9: OUTPUT«silly string␤»
ab6tract m: class T { has $.s = "silly string"; submethod BUILD { $!s = "not so silly now are you"; } }; my $t = T.new; say $t.s
camelia rakudo-moar f7cfe9: OUTPUT«not so silly now are you␤»
GLRelia rakudo-moar ab408b: OUTPUT«not so silly now are you␤»
ab6tract If I dig for a use case for introspecting defaults, it would relate to roles 11:45
laben m: class T { has $.s = "silly string"; submethod BUILD { say $!s; } }; my $t = T.new; say $t.s
camelia rakudo-moar f7cfe9: OUTPUT«(Any)␤(Any)␤»
GLRelia rakudo-moar ab408b: OUTPUT«(Any)␤(Any)␤»
laben m: class T { has $.s = "silly string"; submethod BUILD { say $.s; } }; my $t = T.new; say $t.s
GLRelia rakudo-moar ab408b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/u_4XqAbDye␤Virtual call $.s may not be used on partially constructed objects␤at /tmp/u_4XqAbDye:1␤------> 3silly string"; submethod BUILD { say $.s7⏏5; } }; my $t = T.new; say $t.s␤ expecting …»
camelia rakudo-moar f7cfe9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/MxiIxuau9H␤Virtual call $.s may not be used on partially constructed objects␤at /tmp/MxiIxuau9H:1␤------> 3silly string"; submethod BUILD { say $.s7⏏5; } }; my $t = T.new; say $t.s␤ expecting …»
laben m: class T { has $.s = "silly string"; submethod BUILD { say $!s; $!s = "after"; } }; my $t = T.new; say $t.s 11:46
GLRelia rakudo-moar ab408b: OUTPUT«(Any)␤after␤»
camelia rakudo-moar f7cfe9: OUTPUT«(Any)␤after␤»
laben ok, so this was what jdv79 was asking...
ab6tract it looks like if you poke at the attribute, it won't take the default anymore. but just having a BUILD does not by itself remove the default
m: class T { has $.s = "silly string"; has $.v; submethod BUILD { $!v = "goo" } }; my $t = T.new; say $t.s 11:47
GLRelia rakudo-moar ab408b: OUTPUT«silly string␤»
camelia rakudo-moar f7cfe9: OUTPUT«silly string␤»
laben still, why would it not take the default if i don't reassign to it?
ab6tract laben: I have no idea why poking at the attribute would do that. only others could say whether it's a bug though 11:49
laben will try to check it out, hope it's not the jit 11:50
ab6tract well, I do have an idea, and it relates to the poking causing some level of defined-ness/containerization that results in the skipping of the default assingment 11:51
11:52 rurban left
nine _itz: STDOUT is buffered by default on Perl 5 11:53
_itz but not on perl6?
nine _itz: try perl -e 'STDOUT->autoflush(1); print "A"; sleep 2; print "B";'
_itz: seems like there's no buffering in Perl 6, yes.
_itz ah thanks 11:54
nine _itz: though I wouldn't depend on it. Probably just an implementation artifact :)
_itz open has :autoflush in perl6
I guess buffering is just NYI 11:55
moritz IO is still very immature 11:57
12:01 larion joined, cognominal left
nine I wonder why we haven't seen any mention of GLR benchmarks yet since jnthn's initial one. 12:02
12:03 pmurias joined 12:04 AlexDaniel joined 12:16 Ven left
_itz lies, damm lies and benchmarks :) 12:16
12:18 cognominal joined
masak I'd be interested in some benchmarks. 12:19
I think the reason we haven't seen any might be something mundane like, it takes effort to produce them.
nine Or like panda still not working...
masak it isn't? :/ 12:20
12:22 virtualsue left 12:23 Ven joined
dalek kudo/glr: b1fa1c2 | (Stefan Seifert)++ | src/core/Mu.pm:
Fix "Method 'is-lazy' not found" regression in META_CROSS

Commit 7bc1a99ad91deb8888ab5a51adc909ea1d26ce1a replaced calls to
  .infinite by the new .is-lazy. META_CROSS may call .is-lazy now on
simple scalars like Int. Mu provided a method infinite for such cases. Replace that by a method is-lazy that defaults to False.
Thus the migration from infinite to is-lazy is complete.		 12:23
nine masak: I think laben has done much work on that but it's not yet merged.
12:24 virtualsue joined, virtualsue left 12:25 virtualsue joined
masak ok 12:25
12:25 rurban joined
laben masak: i did publish the PR (github.com/tadzik/panda/pull/202), panda works fine with those and yes i do use bootstrap 12:27
i would like some feedback on other machines
masak oh, panda is broken *on GLR*. I see. 12:28
nine laben: I left a couple of comments on your commits in the panda pull request 12:34
12:35 virtualsue left 12:37 cognominal left
nine m: sub foo() { my $a = 1; my $b = 2; :$b, :$a}; say foo.perl; my (:$a, :$b) = foo; say $a, $b; 12:37
GLRelia rakudo-moar b1fa1c: OUTPUT«(:b(2), :a(1))␤b => 2a => 1␤»
camelia rakudo-moar f7cfe9: OUTPUT«(:b(2), :a(1))␤b => 2a => 1␤»
nine m: sub foo() { my $a = 1; my $b = 2; :$b, :$a}; say foo.perl; my (:$a, :$b) := foo; say $a, $b;
camelia rakudo-moar f7cfe9: OUTPUT«(:b(2), :a(1))␤12␤»
GLRelia rakudo-moar b1fa1c: OUTPUT«(:b(2), :a(1))␤Too many positionals passed; expected 0 arguments but got 2␤ in block <unit> at /tmp/sDRvZ1axk3:1␤␤»
nine Sucks that we lost this magic. I wonder if we can get it back? 12:38
masak why do I have a sense that there's probably just a `flat` missing somewhere? 12:39
nine Hmm...the error comes from bind_cap_to_sig which indicates that perl6 is still trying to do the right thing. It's just failing. 12:41
laben nine: about github.com/laben/panda/commit/d97f...c1b83ca9d1 if it's not slip, it would leave the empty list inside. i understood this was an intended change 12:43
nine laben: but you flatten that empty list both in the @deps.push: line and the other place where you call get-deps 12:44
laben: so, yes, a Slip is a safe choice to return but not strictly necessary. Since you need a workaround for nom there, I'd just leave it out. 12:45
12:46 virtualsue joined 12:47 FROGGS left 12:56 anaeem1 left 13:02 FROGGS joined
nine laben: oh and I just remembered: Empty is an empty Slip 13:03
laben m: dd Empty
GLRelia rakudo-moar b1fa1c: OUTPUT«()␤»
camelia rakudo-moar f7cfe9: OUTPUT«Empty␤»
nine m: sub a { return Empty; }; my @a = a; say @a.perl;
GLRelia rakudo-moar b1fa1c: OUTPUT«[]␤»
camelia rakudo-moar f7cfe9: OUTPUT«[]<>␤»
nine works on nom and glr :)
laben pushed commit fixing run-and-gather, let's see if i can change the commit with Slip, else i add another to change again and return Empty 13:05
nine m: sub foo() { my $a = 1; my $b = 2; \(:$b, :$a)}; my (:$a, :$b) := foo; say $a, $b; 13:08
GLRelia rakudo-moar b1fa1c: OUTPUT«12␤»
camelia rakudo-moar f7cfe9: OUTPUT«12␤»
nine laben: maybe we fix the capture binding problem this way^^^?
m: sub foo() { my $a = 1; my $b = 2; \(:$b, :$a)}; my ($a, $b) = foo; say $a, $b; 13:09
GLRelia rakudo-moar b1fa1c: OUTPUT«\(:a(1), :b(2))(Any)␤»
camelia rakudo-moar f7cfe9: OUTPUT«\(:a(1), :b(2))(Any)␤»
nine m: sub foo() { my $a = 1; my $b = 2; :$b, :$a}; my ($a, $b) = foo; say $a, $b;
GLRelia rakudo-moar b1fa1c: OUTPUT«b => 2a => 1␤»
camelia rakudo-moar f7cfe9: OUTPUT«b => 2a => 1␤»
laben nine: you prefer the capture to the list return?
13:11 skids joined
nine With the capture we only have to change run-and-gather-output instead of both its callers. 13:11
nebuchadnezzar The Slip type is really fun in french, because it's the french word for en.wiktionary.org/wiki/briefs
héhé
nine nebuchadnezzar: same in German 13:12
Well we even had a type called LoL, so what do you expect? ;)
laben a type called Dota
nine: mh, yeah, seems good, now i only need to understand how to clean up the PR commits 13:13
nine git rebase -i origin/glr
13:14 rmgk is now known as Guest7040, rmgk_ joined, Guest7040 left, rmgk_ is now known as rmgk
nebuchadnezzar nine: sure, but when I read “Empty is an empty Slip”, I read “Empty is an empty brief”, which sounds strange ;-) 13:14
nine re-order the commits so the fixups are right after the commits they change and change the "pick" to "fixup" to merge those commits
colomon nebuchadnezzar: in English (American English?) a slip is also a kind of woman’s underwear. 13:15
ab6tract nebuchadnezzar: in English, it refers more to an 'under dress' type of underwear. the dress under the dress 13:16
colomon: jinx ;)
and you are probably right, I bet it is an American English thinq 13:17
timotimo tony-o: i'd be interested to see if json-faster could be made even faster still by pushing bits of string into an array and then joining it at the end 13:19
colomon ab6tract: I don’t know that, I just know I’m American and my wife occasionally wears a slip. :)
13:19 virtualsue left
ab6tract colomon: :) 13:21
nine timotimo: use an IterationBuffer instead of an array
timotimo right, i forgot we have a nqp::-less nqp::list now :) 13:28
skids lizmat: Without any protection by itemization, the way to iter-proof .push($_) is .push(($_,)). Dispatch cannot tell the difference between $_ and $_, at the top level.
(apparently)
PR#512 is semi-rhetorical BTW: it is indeed the right way to do first-arg-rule-does-not-check-itemization, but also show that the fallout from that is a bit harsh. 13:30
(when we deflatten sub map that is another bit of craziness)
13:33 diana_olhovik_ left 13:34 telex left, jkva left 13:36 telex joined 13:37 Ven left
laben nine: i think i made a mess, i cant push anymore to my branch which means i will have to make a new PR with the squashed commits. is it alright? 13:38
skids Actually. Now that I think of it, how does [[],] even work. 13:40
13:42 FROGGS left 13:44 brrt left
skids Hah. Array.STORE checks containerization. So if we fix that, there will be *no* way to create [[]]. I think the idea that $ does not protect against first-arg-rule is going to crash and burn. 13:45
Unless dispatch learns the difference between foo and foo, 13:46
So we have two ways forward and someone has to choose which. 13:47
13:50 jkva joined
ShimmerFairy $ can't protect against single-arg unless you want stuff like ($@a).map to break :) 13:51
m: my @a = 1,2,3; say ($@a).map: * + 1
camelia rakudo-moar f7cfe9: OUTPUT«2 3 4␤»
GLRelia rakudo-moar b1fa1c: OUTPUT«2 3 4␤»
skids Why would that be "broken", you'd write it as @a.map or (@a).map if you wanted to descend into @a. 13:53
(A third alternative is to never use multi foo (\item), always just only foo (**values) and then check values[0] for iterator and for !values.EXIST-POS(1) -- **everywhere**) 13:55
pflanze Does Perl6 attempt to use common interfaces for things like hash maps, sequences, collections? Where should I look to get to learn about them? 13:56
ShimmerFairy m: sub foo { 1,2,3,5 }; my $result = foo; say $result.map: * * 2 # more realistic example 13:57
GLRelia rakudo-moar b1fa1c: OUTPUT«2 4 6 10␤»
camelia rakudo-moar f7cfe9: OUTPUT«2 4 6 10␤»
timotimo we have at least AT-POS and AT-KEY as well as BIND-POS and BIND-KEY
skids pflanze: doc.perl6.org would probably be a good start, but it is incomplete in places, so the final recourse is the Perl6 design docs.
timotimo i'd say the final recourse would be the contents of src/core ;) 13:58
pflanze ok.
Are there any attempts at porting such interfaces to Perl5?
skids ShimmerFairy: what would you expect that to do if $ protected against single-arg-rule? 13:59
pflanze: Moose and Moo do some of that, IIRC.
timotimo hm, porting in what way?
pflanze Since I'm currently coding in Perl5 (also, functional perl5), and am interested in using some.
timotimo, tie doesn't cut it for arrays, hashes in a purely functional context. Also, there's no API for (lazy) sequences. 14:00
timotimo ah
pflanze So, I'm looking for how I should go about it.
skids (And nine++ has done some work on running P6 from inside P5 IIRC)
timotimo has actually no clue about perl5
right, we have Inline::P6 for p5 and Inline::P5 for p6
ShimmerFairy skids: I'd expect it to return just 8 ($list.elems * 2), since $ protection would either have to consistently apply to Scalar containers, or else cause some ungodly mess of special rules 14:01
m: sub foo { 1,2,3,5 }; my $result = foo; for $result { say $_ * 2 }
camelia rakudo-moar f7cfe9: OUTPUT«8␤»
GLRelia rakudo-moar b1fa1c: OUTPUT«2␤4␤6␤10␤»
ShimmerFairy skids: ^^^ camelia's current behavior in for loop form is what we're trying to _avoid_, not perpetuate :)
skids ShimmerFairy: No, because you invoked method map on $foo, $ does not protect against method invocation, of course. 14:02
If you wanted 8 you have to say $$foo.map:
14:02 yqt joined
skids The method invocation eats the $. 14:03
(because you don't even have a list to apply a $ to until you have used it) 14:04
nine laben: try push -f to push to your branch
14:05 diana_olhovik_ joined
nine laben: with git rebase -i you changed commits and git prevents you from pushing changed commits by default as it will hurt others who work on the same branch. A -f is the right thing in your case however. 14:05
skids (or actually precedence may require $($foo.map: ...)) 14:07
14:07 xzenfar left 14:08 TimToady joined
nine m: sub foo { 1,2,3,5 }; my $result = foo; for $result, { say $_ * 2 } 14:08
GLRelia rakudo-moar b1fa1c: OUTPUT«5===SORRY!5===␤Expression needs parens to avoid gobbling block␤at /tmp/lyF7Vf5QYJ:1␤------> 3esult = foo; for $result, { say $_ * 2 }7⏏5<EOL>␤Missing block (apparently taken by expression)␤at /tmp/lyF7Vf5QYJ:1␤------> 3esult = foo; f…»
camelia rakudo-moar f7cfe9: OUTPUT«5===SORRY!5===␤Expression needs parens to avoid gobbling block␤at /tmp/fm7cDdLwqi:1␤------> 3esult = foo; for $result, { say $_ * 2 }7⏏5<EOL>␤Missing block (apparently taken by expression)␤at /tmp/fm7cDdLwqi:1␤------> 3esult = foo; f…»
nine m: sub foo { 1,2,3,5 }; my $result = foo; for ($result,) { say $_ * 2 }
GLRelia rakudo-moar b1fa1c: OUTPUT«8␤»
camelia rakudo-moar f7cfe9: OUTPUT«8␤»
14:08 FROGGS joined
nine skids: if you really want for to run only one iteration, you have to give it a list of size 1 14:09
laben nine++ incredible stuff, it did work and github updated github.com/tadzik/panda/pull/202. dunno why it erased the old commits and comments
FROGGS masak: I'll be in Olten but not at the YAPC
skids nine: if we fix Array.STORE, it will be impossible to construct such a list.
ShimmerFairy As far as I understand, the single-arg rule exists to fix some big issues in pre-GLR design, so it's almost certainly not going anywhere :)
14:10 laouji left
nine laben: git rebase -i replaced the old commits. It did entirely new commits with your new changes. Those new commits have new hashes since the commit hash identifies the whole content of your whole repo in exactly the state where you committed. Since github identifies commits by hash (as makes sense) the comments are still on the old commits. 14:10
skids: in what way does STORE need fixing? I may have missed some discussion there. 14:11
PerlJam good $localtime all
ShimmerFairy skids: if you really wanted for $@a { } to do what you want, that should really be written as given @a { } , assuming [@a,] and such ever does become impossible.
14:12 TimToady left
skids ShimmerFairy: I'm not saying it should be going anywhere. I'm saying as we fully implement, one of two things has to start working: either $ has to protect against it, or foo(thing,) has to MMD to foo(**@values) over foo(\value). If neither of thise two things happen, it will get ugly. 14:12
nine laben: the old commits with comments are still there btw.: github.com/laben/panda/commit/f578...679bc5e7ff They are just no longer contained in your branch (and thus pull request)
skids nine: STORE checks containerization, so it allows $ to protect against signle arg rule.
14:13 TimToady joined
laben nine++ sensei, while fixing glr, teaches me git and github. thanks! 14:13
pmurias pflanze: why doesn't tie cut if for arrays, hashes? 14:14
CQ2 with all the helping he's doing, you have to watch out that nine++ doesn't lead to an integer overflow...
14:15 Ven joined, khw joined
skids m: my @a; @a.STORE($[1,2]); @a.perl.say 14:15
GLRelia rakudo-moar b1fa1c: OUTPUT«[$[1, 2]]␤»
camelia rakudo-moar f7cfe9: OUTPUT«[Any, Any]<>␤»
nine m: my @a; @a.STORE($[1,2]); my @b; @b.STORE(([1,2],)); say @a.perl; say @b.perl; 14:19
GLRelia rakudo-moar b1fa1c: OUTPUT«[$[1, 2]]␤[$[1, 2]]␤»
camelia rakudo-moar f7cfe9: OUTPUT«[Any, Any]<>␤[[1, 2]]<>␤»
nine skids: ^^^ two ways to achieve the same. So we may take away the itemization one?
skids Only if you want to see ((thing,)) everywhere. 14:20
hoelzro good morning #perl6! 14:21
timotimo heyo hoelzro
skids Like to make [[[]]] you would have to write: [([([],),)]
pflanze pmurias, tie assumes that it's used in an eagerly evaluating world that's mutable; taking the list of elements or keys of a hash claims the only enumeration context a hash has to offer, for example.
skids And that assumes there is not a reason to break (thing,)
ShimmerFairy m: my @a; @a.push(a=>1); @a.push((b=>2)); say @a.perl; # not like there aren't other reasons to need (()) already :P 14:22
GLRelia rakudo-moar b1fa1c: OUTPUT«[:b(2)]␤»
camelia rakudo-moar f7cfe9: OUTPUT«[:b(2)]<>␤»
pmurias hoelzro: hi
hoelzro ahoy timotimo, pmurias!
skids push is actually broken without PR512, in thesame way as STORE. JFWIW. 14:23
*[([([],)],)] above. 14:24
nine skids: just how often do you need to create lists containing a single listy item? 14:25
The occasional trailing comma is the price we're paying for sane and consistent semantics. 14:26
skids nine: I would ecourage you to look at what I had to do in PR512 just to fix push/unshift internally.
And it is more than just an occasional trailing comma it is a set of parens around a trailing comma. 14:27
nine But still occasional I'd guess. 14:28
PerlJam figures out that he's totally confused. 14:29
skids ShimmeryFairy: FWIW, push currently does allow $ to override the single element rule and: 14:30
m: my $a = (2,3,4); my @b = 1; (@b.push: $a.map: { $_ }).perl.say;
camelia rakudo-moar f7cfe9: OUTPUT«[1, 2, 3, 4]<>␤»
GLRelia rakudo-moar b1fa1c: OUTPUT«[1, 2, 3, 4]␤»
tony-o timotimo: i'll give that a shot, interesting to know that it would work more quickly that way for sure
timotimo: re: json::faster
timotimo would you accept merging json::faster's to-json into json::fast?
or do you think that's a bad idea as JSON::Fast isn't fully featured yet? 14:31
ShimmerFairy I don't think there are many cases where you want to treat a list as a singular item _and_ you have just one list you wish to treat as such, instead of a list of them.
skids Again, see my PR.
push $@a onto a list of arrays is a rather common idiom. 14:32
nine Sounds like a good topic for Olden 14:33
Olten
ShimmerFairy If nothing else, I think there was talk of reintroducing a LoL type to solve some other issues, which could probably give a good solution to this issue as well :) 14:34
ugexe agree with skids. i was using something like $glr = @($work,) unless $work.list.[0].isa(Pair) 14:35
skids It isn't immediately obvious to me how an LoL type would help, but might be to more veteran folks. 14:36
pflanze Will anyone interested in functional programming on Perl(6) be in Olten this week? I'm pondering going there (I'm the author of the yet-to-be announced functional-perl.org project). 14:38
14:38 colomon left
masak pflanze: I am interested in functional programming, and I'm going to be there. 14:38
nine The discussions about the new list semantics have been going on for a year. Maybe even longer. I haven't been here before that. I just fear that we will not come up with something better than GLR is already. You can't win every battle.
14:38 RabidGravy joined
masak pflanze: I've been trying to understand category theory for the past 1.5 .. 2 years, if that helps :) 14:38
skids nine: GLR "isn't" yet though because there are so many cheats still in the codebase. I don't think the full ramifications of the current guidance have begun to be discovered. 14:42
nine skids: will you be in Olten? 14:45
skids Also I don't want to give the impression that I have a strong inclination towards $ protecting from single-arg-rule, I'm just leaning that way mildly.
nine: I pretty much remain without the same 50 mile radius always.
*within
skids hates travel.
[Coke] waves from the bus to NYC. 14:46
Ven pflanze: I certainly consider myself interested as well!
pflanze Ven, ok, I'm looking into being there on Friday till ~22:00. 14:47
Ven just wrote a one-liner to find recursive calls in a c codebase
pflanze Unless I'd still get a slot in the program, then maybe another day.
Ven fwiw, I'd certainly appreciate something smaller than $*ARGFILES.filename :P. 14:48
(I guess some module could do that, and then I could alias 6l = perl6 -MSixLiner ...) 14:49
moritz finds it appropriately huffmanized
PerlJam finds it totally weird that Ven and moritz said what he was thinking
pflanze Ven, masak: Hm, to prevent misunderstandings, I'm not experienced in Perl6. I've written the functional-perl.org for Perl5 till now, but looking into lerning what Perl6 people are doing in this area.
PerlJam Though in my imagination, the SixLiner module could be autoloaded for command line execution 14:50
pflanze So what I can do is present what I have in Perl5, and learn what Perl6 people think about it or what they know.
14:50 jkva left
Ven moritz: for regular scripts, surely. for my one-liners, no :) 14:50
14:54 xinming_ left, xinming joined
Ven pflanze: seems good 14:56
ab6tract waves at [Coke] 15:01
masak pflanze: out of curiosity, have you read www.amazon.com/Higher-Order-Perl-Tr...558607013/ ? 15:03
15:03 brrt joined
pmurias any ideas how to test nqp::getstderr in nqp? 15:03
pflanze masak, Yes, a long time ago
moritz project idea: HOP code examples ported to Perl 6 15:04
masak yeah, that'd be nice 15:05
timotimo pmurias: only thing i can think of is either capturing the output or checking the underlying fileno
masak pflanze: I also seem to recall eiro++ did something a bit like this. 15:06
pflanze masak, what my project goes further than the book (IIRC) is that it's meant for real-world use, and hence cares about memory deallocation, and for proper modularity.
masak, what it comes short in comparison is that there's no parsing libraries or such yet (which the book has, IIRC).
15:08 diana_olhovik_ left
pflanze (I mention the book on the project home page, and will have to reread it when time permits.) 15:08
pmurias timotimo: capturing the output seems better, as I'm not sure the fileno is the same on windows etc. 15:12
timotimo oh
masak pflanze: I found what eiro has done: github.com/eiro/p5-perlude/blob/ma...erlude.pod 15:14
pflanze: maybe you two can combine your efforts somehow?
dalek kudo/glr: 1bf053a | (Stefan Seifert)++ | src/core/Mu.pm:
Remove now obsolete XXX GLR comment

method infinite has not survived.
pflanze masak, thanks for the link. I'll contact him. 15:18
ab6tract Ven: care to share that one liner? :) 15:28
YuviPanda is new here
where can I learn about the 'GLR'?
and what it means?
Ven ab6tract: oh, it's the dumbest possible. perl6 -ne 'BEGIN { my $f }; if /:i ^(<[a..z]>+) "("/ { say $*ARGFILES.filename ~ ":" ~ $0; $f = $0; } elsif $_ eq "}" { $f = "" } elsif $f && $_ ~~ /$f/ { say $f ~ ": " ~ $_ }' *.c
masak YuviPanda: welcome! 15:29
YuviPanda hello! 15:30
masak YuviPanda: briefly, it's a big refactor before release later this year.
YuviPanda ah. can I ask what it is refactoring?
masak YuviPanda: some things about lists and how they work that need to change, mostly to make them more performant.
YuviPanda (as background, I shared a car ride with sjn from CAMP and spent a hour and something talking about perl6, and now I've been learning the language / writing small things in it)
masak YuviPanda: this post has some detail: pmthium.com/2014/10/apw2014/
YuviPanda masak: ah, I see. so no non-performance user facing features
YuviPanda clicks
masak YuviPanda: yes, but there are some syntactic changes too. definitely things the user will see sometimes. 15:31
AlexDaniel Hmm, I feel adventurous 15:32
FROGGS m: use Test; is 'a b c d'.subst(/\w/, 'x', :nth(0)), 'a b c d', '.subst and :nth(0)' 15:33
camelia rakudo-moar f7cfe9: OUTPUT«ok 1 - .subst and :nth(0)␤»
GLRelia rakudo-moar 1bf053: OUTPUT«Attempt to retrieve before :1st match -- :nth(0)␤ in block <unit> at /tmp/8Ear49WCOz:1␤␤Actually thrown at:␤ in block <unit> at /tmp/8Ear49WCOz:1␤␤»
FROGGS m: use Test; is 'a b c d'.subst(/\w/, 'x', :nth(5)), 'a b c d', '.subst and :nth(5)' 15:35
camelia rakudo-moar f7cfe9: OUTPUT«ok 1 - .subst and :nth(5)␤»
GLRelia rakudo-moar 1bf053: OUTPUT«ok 1 - .subst and :nth(5)␤»
15:36 Ven left
tony-o timotimo: i wouldn't mind merging the two (Fast - Faster), mine does less type checking iirc. it only doesn't quote Rat | Int 15:36
timotimo: i don't plan on doing a from-json because the built-in is about as fast as any pp6 i'd do, i was considering a nativecall solution for both but, honestly, how fast does json parsing need to be *right now* 15:37
timotimo it doesn't quote rat | int? 15:38
we already have a nativecall json parser 15:39
15:41 mr-foobar joined
skids FROGGS: See S05-modifier/counted-match.t I didn't realize there was a duplicate test. 15:41
FROGGS hmmm 15:42
skids It should fail, per the RT, unless we think somehow the behavior is useful. 15:43
(versus :nth(-1), why would one fail and not the other)
FROGGS std: say 'a b c d'.subst(/\w/, 'x', :nth(5)) 15:44
15:44 GLRelia left
camelia std 28329a7: OUTPUT«ok 00:00 144m␤» 15:44
FROGGS hmpf
std: say 'a b c d'.subst(/\w/, 'x', :nth(0)) 15:45
camelia std 28329a7: OUTPUT«ok 00:00 144m␤»
FROGGS std: say 'a b c d'.subst(/\w/, 'x', :nth(-1))
camelia std 28329a7: OUTPUT«ok 00:00 144m␤»
sjn YuviPanda: hey, nice to see you here! :)
FROGGS gah, perhaps std does not care
timotimo probably
YuviPanda sjn: hey! :)
15:46 GLRelia joined
sjn have you found anything interesting yet? :) 15:46
15:46 virtualsue joined
GLRelia std : OUTPUT«Can't chdir to '/home/steve/std/snap': No such file or directory at lib/EvalbotExecuter.pm line 171.␤ EvalbotExecuter::_auto_execute(HASH(0x253d638), "say 'a b c d'.subst(/\\w/, 'x', :nth(5))", GLOB(0x28fd6d0), "/tmp/mDV37SKL1g", "std") called at lib/EvalbotExecuter.p…» 15:46
tony-o timotimo: correct, if the value is a Rat or Int then it just outputs the number
sjn wonders if he over-sold perl6 on that car ride :)
timotimo is that a problem?
tony-o i doubt it, i haven't run into it with projects.json or various other json things i tried
sjn _did_ manage to bore a quarter of the passengers into sleep <_< 15:47
tony-o i just noticed that ::fast (iirc, it could have been a different json generator) was checking a bunch of different types
FROGGS hmmm, the design docs do not tell... 15:48
ab6tract sjn: Unfortunately I know that feeling 15:52
skids FROGGS: The way I read the design docs when I reimplemented that was :nth(3,0,-1,2,4) would trim to (3,4) but there's nothing about just nth(-1) or nth(0), however the existing code was trying to throw on those so I just fixed it to actually do so.
ab6tract But as long as one person is interested, it's hard for me to stop talking about it ;) 15:53
YuviPanda sjn: yeah, I wrote a small module 15:54
FROGGS skids: and are the values 0,-1,2 ignored as per design docs?
YuviPanda sjn: heh, she was just super tired and had had about 2h of sleep that night :D
sjn: I've found grammars *super* fun
FROGGS m: say 'a b c d'.subst(/\w/, 'x', :nth(3,0,-1,2,4))
camelia rakudo-moar f7cfe9: OUTPUT«a b c d␤»
GLRelia rakudo-moar 1bf053: OUTPUT«a b x x␤»
skids It says sequences must be monotonic, and if they aren't the nonmonotonic values will be ignored.
YuviPanda since I used to hate writing any form of string manipulation code
sjn: github.com/yuvipanda/perl6-Ident-Client
FROGGS skids: are, seems to work fine
15:55 zakharyas left
timotimo JSON::Fast is just a 1:1 copy of JSON::Tiny for the json generator parts 15:55
skids lunch already? OK then.
YuviPanda sjn: so slowly poking around and enjoying the language as a pure fun thing :) 15:56
dalek ast/glr: f76ca69 | FROGGS++ | S05-substitution/subst.t:
apply nth(0) fix from S05-modifier/counted-match.t, skids++
 15:58
[Coke] colomon has a process on hack that's been running for 4.5h and consuming3 cpus. 15:59
also has 4 processes that all seem to be running with the same smoker.json parameter. 16:00
the other 3 are all up to an hour. 16:01
16:02 muraiki joined
[Coke] .seen colomon 16:02
yoleaux I saw colomon 13:19Z in #perl6: <colomon> ab6tract: I don’t know that, I just know I’m American and my wife occasionally wears a slip. :)
dalek kudo/glr: 2ffd6ab | FROGGS++ | src/core/Str.pm:
fix Str.samespace
 16:08
16:13 larion left 16:17 Gardner left 16:18 dwarring left 16:22 Gardner joined
[Tux] does panda install on glr yet? 16:24
timotimo there's a pullrequest that makes it work
16:29 ab6tract left
sjn YuviPanda: yeah, grammars rock :D 16:30
YuviPanda sjn: yeah, totally. multimethods and what appears to me to be forms of contract programming also sound great :) 16:31
sjn YuviPanda: have you found MAIN yet? :)
YuviPanda sjn: oh yah :)
that was really nice surprise too
sjn loves that stuff 16:32
16:32 Ven joined
sjn I still point people to a talk jnthn had in Oslo quite a few years back, about how you can start with a simple one-liner, and graduate it to a command line script using MAIN and grammars and later turn it into a serious all enterprisey OO-buzzwordy program 16:34
sjn tries to find it again
it may be getting old now though, but I still think it should be all valid 16:35
www.nuug.no/aktiviteter/20100914-li...arge-apps/ # in Norwegian, but the important bits are in English 16:36
16:36 Ven left
sjn (please ignore the guy at the start of the video ) 16:37
:)
16:41 yqt left
laben nine: thanks for merging my panda PR! i just noticed. 16:42
16:43 brrt left
AlexDaniel nine is so awesome lately 16:44
16:47 anaeem1 joined
masak nine has always been awesome. 16:51
"hey, I wrote my first Perl 6 program! it implements Inline::Perl5!" :D
PerlJam nine++ 16:52
16:56 abraxxa left 17:03 CQ2 left, anaeem1__ joined 17:06 anaeem1 left 17:08 Ven joined 17:19 dakkar left, espadrine left
jdv79 how many ++'s does it take until he's ten? 17:29
nein++ # as an american in germany at the moment that's amusing for some reason. sorry. 17:30
17:30 Ven left
lizmat is reminded of a German guy who had only seen "Monty Python and the Holy Grail" dubbed in German 17:31
then, when he finally saw the movie in English, was surprised to hear it was *not* "The Knights Who Say Never" but "...Say Ni!" 17:32
timotimo tee hee 17:33
17:41 Peter_R joined 17:48 brrt joined, ab6tract joined 17:56 nowan left 17:59 nowan joined 18:02 llfourn left 18:09 nowan left 18:10 nowan joined 18:11 domidumont left 18:14 darutoko- joined
muraiki if I want to kill the parent process from the callback provided to .act (sorry if I'm using the wrong terminology), what should I do? I guess exit() just exits the thread the .act is running in? sorry, I'm pretty new to this 18:15
basically I only want to act once :)
18:16 anaeem1_ joined 18:17 anaeem1__ left, brrt left, darutoko left
AlexDaniel m: sub foo('test') { say 'ok' }; foo('test'); 18:21
GLRelia rakudo-moar 2ffd6a: OUTPUT«ok␤»
camelia rakudo-moar f7cfe9: OUTPUT«ok␤»
18:21 ribasushi joined
AlexDaniel makes me want to try 18:21
m: my 'hello' $test
camelia rakudo-moar f7cfe9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/8fFgIGLCA6␤Malformed my␤at /tmp/8fFgIGLCA6:1␤------> 3my7⏏5 'hello' $test␤»
GLRelia rakudo-moar 2ffd6a: OUTPUT«5===SORRY!5=== Error while compiling /tmp/6pJglZ0rl8␤Malformed my␤at /tmp/6pJglZ0rl8:1␤------> 3my7⏏5 'hello' $test␤»
AlexDaniel m: my Str where 'hello' $test; 18:23
camelia rakudo-moar f7cfe9: OUTPUT«5===SORRY!5===␤Type 'Str' is not declared. Did you mean 'str'?␤at /tmp/1hszJKLnIJ:1␤------> 3my Str 7⏏5where 'hello' $test;␤Malformed my␤at /tmp/1hszJKLnIJ:1␤------> 3my Str 7⏏5where 'hello' $test;␤␤»
GLRelia rakudo-moar 2ffd6a: OUTPUT«5===SORRY!5===␤Type 'Str' is not declared. Did you mean 'str'?␤at /tmp/hOAUR4_3gt:1␤------> 3my Str 7⏏5where 'hello' $test;␤Malformed my␤at /tmp/hOAUR4_3gt:1␤------> 3my Str 7⏏5where 'hello' $test;␤␤»
AlexDaniel type Str is not declared? Hmm?
FROGGS O.o 18:24
AlexDaniel m: my str where 'hello' $test;
camelia rakudo-moar f7cfe9: OUTPUT«5===SORRY!5===␤Type 'str' is not declared. Did you mean 'Str'?␤at /tmp/eflb_uvwk1:1␤------> 3my str 7⏏5where 'hello' $test;␤Malformed my␤at /tmp/eflb_uvwk1:1␤------> 3my str 7⏏5where 'hello' $test;␤␤»
GLRelia rakudo-moar 2ffd6a: OUTPUT«5===SORRY!5===␤Type 'str' is not declared. Did you mean 'Str'?␤at /tmp/tnplEesp_X:1␤------> 3my str 7⏏5where 'hello' $test;␤Malformed my␤at /tmp/tnplEesp_X:1␤------> 3my str 7⏏5where 'hello' $test;␤␤»
AlexDaniel what is going on here… I don't really understand
I thought that this code was valid?
FROGGS m: my Str $test where 'hello'; 18:26
camelia ( no output )
GLRelia ( no output )
AlexDaniel ooooooohhh
right
FROGGS still... can you rakudobug that as a misleading error message?
AlexDaniel FROGGS: of course!
FROGGS AlexDaniel++ 18:27
dalek ast/glr: 857c7c2 | (Stefan Seifert)++ | S03-metaops/zip.t:
Replace usage of internal "munch" method in zip.t
 18:32
ast/glr: 5e181ff | (Stefan Seifert)++ | S03-metaops/zip.t:
Adapt zip.t tests to new (non-)flattening semantics
MilkmanDan devopsreactions.tumblr.com/post/620...-manifests
dalek kudo/glr: 42c2d81 | (Stefan Seifert)++ | src/core/List.pm:
Fix "Method 'iterator' not found" in infix:<Z>
kudo/glr: 27cb277 | (Stefan Seifert)++ | src/core/metaops.pm:
Fix "Too many positionals passed" in METAOP_ZIP

The METAOP_REDUCE_* reducers expect only a single iterable while plain ops require arguments passed separately. Have to reify the mapped iterators to a List to avoid "This Seq has already been iterated" errors.
18:33 dha joined 18:35 dha left 18:36 dha joined 18:38 larion joined
dalek ast/glr: 21d17ec | (Stefan Seifert)++ | S32-container/zip.t:
Fix remaining infix:<Z> tests

Need to explicitly flatten those ranges		 18:41
AlexDaniel m: my Str $test where 'test'; $test = 25; 18:43
camelia rakudo-moar f7cfe9: OUTPUT«Type check failed in assignment to '$test'; expected '<anon>' but got 'Int'␤ in block <unit> at /tmp/309lVbuFOs:1␤␤»
GLRelia rakudo-moar 27cb27: OUTPUT«Type check failed in assignment to '$test'; expected '<anon>' but got 'Int'␤ in block <unit> at /tmp/abKlIJY5S7:1␤␤»
AlexDaniel expected '<anon>'?
18:44 larion left 18:45 diana_olhovik joined 18:46 yqt joined 18:50 vendethiel joined
dalek ast/does-ok: e960b57 | (David Warring)++ | / (4 files):
utilize does-ok test assertion
 18:53
AlexDaniel can anybody delete this latest RT? 18:54
Looks like a virus: www.virustotal.com/en/file/b57d12c...440528830/
nine Why is my %h = bag <a b o p a p o o>; not allowed to work? (it does currently, thereby failing a spec test) 18:55
FROGGS nine: I am spectesting a fix for t/spec/S05-substitution/subst.rakudo.moar now btw 18:57
nine: dunno why it is not meant to work... 18:58
18:59 dwarring joined, anaeem1 joined
AlexDaniel someone just did, thanks 18:59
19:00 anaeem1_ left
FROGGS I marked a ticket a spam which a subject Resume or so 19:00
dalek ast/glr: d2a3cfc | (Stefan Seifert)++ | S02-types/range.t:
Range.pick and .roll are expected to return a Seq per GLR

We return a Seq now to avoid unnecessary copying and storing in cases where we just iterate over the results once.
nine That was really LHF ^^^ :) 19:01
FROGGS nice :o)
nine Should be at 104 now :)
AlexDaniel wasn't it 80 already? 19:02
nine That would have been nice...
AlexDaniel huh what, was it my dream? 19:04
muraiki ok, so it turns out what was happening is that inside of the tap callback I was calling exit $return_code, where $return_code was the result of a call to run(). if run succeeds it returns 0, but if it fails it returns some kind of object, which was causing exit to not actually exit
AlexDaniel I remember staring at IRC window and seeing that it is just 80 now… 19:05
19:05 sno left
muraiki I thought I read somewhere that run() should always return the return code, so maybe those docs were out of date 19:05
dalek ast/glr: c94cd06 | (Stefan Seifert)++ | S03-junctions/boolean-context.t:
Fix "Too few positionals passed" in S03-junctions/boolean-context.t

We no longer auto-flatten results of meta ops, need to unpack manually		 19:06
nine FROGGS: do you think this ok? ^^^
19:07 [Sno] joined
FROGGS nine: yes, I think so 19:08
nine :) 103
FWIW there still seems to be plenty of LHF in roast. 19:10
FROGGS damn, my patch for subst makes one fail and two hang :(
19:13 telex left 19:14 telex joined 19:16 [Sno] left
nine FROGGS: are those hanger passing before your patch or did you just uncover the hangs? 19:17
19:19 [Sno] joined
FROGGS they were passing 19:20
jdv79 AlexDaniel: maybe the bare startup times are what you are thinging about 19:21
what does it mean to be thinging ^H? 19:22
AlexDaniel I doubt it. It is way more realistic to think that I'm from the future 19:23
jdv79 how is it there?
AlexDaniel not too different 19:24
80 spec test is about… two days from now?
s
19:27 [Sno] left 19:28 [Sno] joined 19:29 yqt left, mvuets joined
hoelzro prepares to head to ORD 19:29
see (many of) you in Olten! 19:30
19:31 larion joined
laben i'm trying to construct the best oneliner to accomplish a cool thing, i'm up to here "([1,1],[2,2],[3,3],[4,4]).map({[+] $_}) Z<=> (2, *+2...*)" but i dont know how to make that map better. anyone got a better idea? 19:34
19:35 ab6tract left 19:37 yqt joined 19:38 anaeem1 left
timotimo yay! 19:39
hoelzro: looking forward to it :)
have a good flight!
skids laben: what needs to be better about the map? 19:41
19:42 diana_olhovik left
timotimo it needs to become the best map that has ever mapped 19:44
skids well, if you knowthe arity is 2 then: 19:45
m: say (flat [1,1],[2,2],[3,3],[4,4]).map(*+*) Z<=> (2, *+2...*) 19:46
camelia rakudo-moar f7cfe9: OUTPUT«More Same␤»
GLRelia rakudo-moar 27cb27: OUTPUT«Same Same Same Same␤»
timotimo how are you? same same.
19:49 diana_olhovik_ joined
dalek rl6-roast-data: a3398ed | coke++ | / (8 files):
today (automated commit)
 19:56
19:57 brrt joined
laben i wanted to do it with Whatever but cannot, i thought there could be a way. arity=2 is only the example here, in reality it differs for probably every array 19:57
19:58 llfourn joined
[Coke] waves from JFK. Slowly making his way to Olten. :) 20:01
20:01 cdc_ joined
cdc_ m: say ([1,1],[2,2],[3,3],[4,4]).map({[+] |$_}) Z<=> (2, *+2...*) 20:02
GLRelia rakudo-moar 27cb27: OUTPUT«Same Same Same Same␤»
camelia rakudo-moar f7cfe9: OUTPUT«Same Same Same Same␤»
20:03 llfourn left 20:04 darutoko- left 20:06 xfix left 20:07 diana_olhovik_ left, diana_olhovik_ joined
masak [Coke]: oh, you'll be in Olten? excellent! 20:11
correct me if I'm wrong, but I don't believe we've met.
dha I, sadly, will not be in Olten. :-( 20:13
20:13 kaare_ left 20:19 diana_olhovik_ left
timotimo masak: i'll be in olten, too, and we haven't met yet, either :) 20:19
20:20 diana_olhovik_ joined
[Coke] masak: no, we have not. I think you missed the one .eu thing I went to. 20:21
20:21 rangerprice joined
pflanze Is there any currying/uncurrying infrastructure in Perl6? 20:23
moritz pflanze: you can automatically curry most operators by passing a * as argument 20:24
m: say map * + 2, 1..5
GLRelia rakudo-moar 27cb27: OUTPUT«3 4 5 6 7␤»
camelia rakudo-moar f7cfe9: OUTPUT«3 4 5 6 7␤»
jnthn Alternatively, you may be looking for .assuming
oooh, I have no messages :)
pflanze Ok, interesting. What about uncurrying? 20:26
20:27 laben left
moritz what's uncurrying? 20:27
pink_mist imagines a .unassuming
pflanze The inverse of currying: take a function that takes 1 argument then returns another function that takes 1 argument, and make that into a fuc
function that takes 2 arguments.
20:27 diana_olhovik_ left
pflanze That's the simplest case, trivially written as a function: 20:28
sub curry { my ($f)=@_; sub { my($a,$b)=@_; $f->($a)->($b) }} # Perl5 syntax
erch.
sub uncurry { my ($f)=@_; sub { my($a,$b)=@_; $f->($a)->($b) }} # Perl5 syntax 20:29
moritz so, just function chaining?
20:29 KCL joined
pflanze But it gets more interesting if you want to combine function layers that take another number of arguments than 1. 20:29
rangerprice Oh
pflanze Hm, function chaining sounds like you mean compose. 20:30
20:30 diana_olhovik_ joined 20:31 rurban left 20:32 KCL_ left
pflanze Example: fun ($x) { fix fun($self, $y) { $y > 0 ? $x * &$self($y-1) } } # Perl5 with Function::Parameters 20:33
_itz does anyone know where the " 20:34
Pre-Hackathon Dinner
" is?
pflanze # and `fix` defined to be the fixpoint combinator; it binds $self to a function representing the second function with just the $y argument.
_itz . o O ( damm line breaks )
pflanze moritz, so, this function now takes first $x then a start value for $y, then recurses till $y reaches 0 and returns a result. 20:35
moritz, so now I want to represent it to the user as a function that takes $x and the start value for $y directly, without needing to chain the calls.
i.e.: fun original ($x) { fix fun($self, $y) { $y > 0 ? $x * &$self($y-1) } } original(10)->(20) 20:36
*pleasant= uncurry *original; pleasant(10,20)
jnthn _itz: I got the impression it may be at the venue, given folks are being pointed in that direction 20:39
20:40 lizmat left
_itz ok thanks 20:41
jnthn Time for some much needed rest 20:42
&
pflanze moritz, here's what I've included in my project for Perl5 currently (definition, and example usage): github.com/pflanze/functional-perl...uncurry.pm github.com/pflanze/functional-perl...es/gen-csv 20:50
timotimo good rest, jnthn 20:51
pflanze Perl5's need to have a subroutine declaration kind of spoils it; perhaps adding syntax is the right thing to do some time. So I was wondering how Perl6 would cope with this.
dalek rl6-roast-data: 4e55de8 | coke++ | / (2 files):
another glr run for today
 20:53
[Coke] anyone else arriving in Zurich early Wednesday? 20:54
I am happy to wait a bit to ride a train with someone. (if not, I'll just head out to olten asap)
20:56 Peter_R left 20:57 llfourn joined
timotimo vendethiel: do you already know when you'll take the train back to your home? if you're leaving the SPW via train at all 20:57
i'd like to take the same train again, like we did after APW 20:58
vendethiel timbunce: yes!
oops. sorry timbunce++. I meant timotimo++ 20:59
[Coke] ~~
21:02 rindolf left, llfourn left
virtualsue we're arriving in zurich around 15:30 21:04
skids pflanze: I think what you are looking for is simply closures. 21:05
pflanze skids, those are all using closures.
Not sure what you would mean.
(Have closures changed in Perl6, new syntax or so?) 21:06
vendethiel pflanze: fwiw, a good way to do it is just by allowing the curried version to take any number of arguments 21:07
pflanze vendethiel, you mean auto-currying, if enough arguments have been passed, call the next level, until reaching the last? 21:08
vendethiel pflanze: yes
pflanze I'm not sold on this, as (1) it's got runtime overhead *and* would be difficult for a compiler to optimize away, (2) it hides what's going on more than the explicit version.
vendethiel pflanze: gist.github.com/jnthn/9772688 21:09
that's an example
muraiki is "say" thread-safe? or if I have multiple says going on in multiple .tap, could they end up mixing up in the console?
skids m: my &f = -> &g { -> &h { -> $x { g(h($x)) } } }; &f(&abs)(&sin)(pi*3/2).say; 21:10
camelia rakudo-moar f7cfe9: OUTPUT«1␤»
GLRelia rakudo-moar 27cb27: OUTPUT«1␤»
skids ^^pflanze
ugexe no, if you have multipl threads doing say/print they certainly dont lock 21:11
you can do it yourself though
pflanze skids, hm, I don't know enough Perl6 to parse this. It looks like you're making a curried function, then calling it in the unchanged curried fashion (no uncurrying going on). 21:12
vendethiel pflanze: -> &a {}; is an anonymous function that takes a hof "&f"
muraiki ugexe: thanks
vendethiel pflanze: and some-fn(&foo); passes the foo function to some-fn
pflanze vendethiel, you mean &a (lexical variable a that's a function)? 21:13
vendethiel pflanze: yes
m: sub p { say "hey hey"; }; sub call(&fn) { fn(); }; call(&p);
camelia rakudo-moar f7cfe9: OUTPUT«hey hey␤»
GLRelia rakudo-moar 27cb27: OUTPUT«hey hey␤»
pflanze I was guessing right, then.
ugexe muraiki: here is a messy as hell example github.com/ugexe/zef/blob/master/l...ar.pm6#L54 21:14
muraiki ugexe: wow, thanks so much! 21:15
pflanze skids, now my &ff= uncurry &f; &ff(&abs, &sin, pi*3/2)
skids, actually, since this is merging 3 levels, my &ff= uncurry_1_1_1 &f; &ff(&abs, &sin, pi*3/2) 21:16
or my &ff= uncurry uncurry &f; &ff(&abs, &sin, pi*3/2)
oh, the latter is actually wrong.
vendethiel m: sub uncurry(&f) { -> $a, $b { f($a, $b); } }; my $add = -> $x { -> $y { $x + $y }; }; my $u-add = uncurry $add; say $u-add(1, 2); 21:17
camelia rakudo-moar f7cfe9: OUTPUT«Too many positionals passed; expected 1 argument but got 2␤ in block <unit> at /tmp/j6iAQkvUoQ:1␤␤»
GLRelia rakudo-moar 27cb27: OUTPUT«Too many positionals passed; expected 1 argument but got 2␤ in block <unit> at /tmp/iIGr9210cB:1␤␤»
vendethiel whoops.
m: sub uncurry(&f) { -> $a, $b { f($a)($b); } }; my $add = -> $x { -> $y { $x + $y }; }; my $u-add = uncurry $add; say $u-add(1, 2);
camelia rakudo-moar f7cfe9: OUTPUT«3␤»
GLRelia rakudo-moar 27cb27: OUTPUT«3␤»
pflanze vendethiel, yep. So that's what I have (see link I provided above); I'm covering the non-standard cases manually by exporting functions with names such as uncurry_1_2 etc. 21:18
Seems Perl6 might be more regular with regards to syntax (Perl5 expects a declaration which it uses syntactically at compile time), which will help. 21:19
ugexe is $a = $b = 0 an atomic assignment?
skids pflanze: OIC. No, when you curry something it compiles. You might be able to get similar results by using wrap, but it is an inefficient way to go about things. 21:20
pflanze skids, what I'm talking about is the uglyness on github.com/pflanze/functional-perl...es/gen-csv
where I need to "sub exps; *exps = "... 21:21
Perl5 needs the "sub exps;" or it won't know that exps exists further down in the code and compile it differently.
(compile exps as a method call instead) 21:22
21:22 Peter_R joined
pflanze I'm hoping Perl6 is saner. 21:22
vendethiel :)
masak 'night, #perl6 21:24
vendethiel \o, masak
masak why does my brain see "\o, masak" and go, "aha, it's the vocative". bad brain, bad!
:P
anywee, night \o #perl6 21:25
skids pflaze: as far as declaring subs you can initialize a &a during declaration. 21:26
m: my &f = &abs; f(-1).say # works fine.
camelia rakudo-moar f7cfe9: OUTPUT«1␤»
GLRelia rakudo-moar 27cb27: OUTPUT«1␤»
skids As far as "uncurrying" you'd probably want to implement that as some sort of slang. I personally could only see it being useful for converting strict FP code into procedural multiargs during a conversion -- otherwise why not define the prototype where people can see it instead of obfuscating it? 21:28
pflanze Good. Looks like there's no difference (syntactically) between lexical and package variables in Perl6.
skids pflanze: depends on what you declare them as -- there's also "our" and "is export" 21:29
pflanze multiargs and FP are orthogonal.
skids Depends on who you ask :-) 21:30
pflanze Even Haskell has them.
It's just that in Haskell the curried form is the default.
21:31 muraiki left
pflanze It makes some sense to have dynamically typed languages default to the uncurried form as it helps debugging. 21:31
vendethiel pflanze: well, haskell doesn't need 0-args function because it's lazy :) 21:32
we're not nearly as lazy...
pflanze ML has solved this with the unit value ("()") when curried.
vendethiel pflanze: and ocaml added back all kind of issues with label and optional parameters :P 21:33
pflanze (strip "when curried"; "when it's the last argument a function needs" is more correct.) 21:34
skids At any rate the "uncurry" construct either has to have functions carry around details about their innards that are normally thrown away, or be satisfied with being a simple wrapper that evals level by level and uses introspection to fugure out what it got at each level.
pflanze You'll always encounter it in cases where an inner function recurses one of the inner levels. 21:35
And takes as start an argument passed from outside.
Then it's natural to leave it curried; and uncurry it outside for convenience.
Doesn't need to be purely functional; just recursive. 21:36
21:37 yqt left
pflanze I'm sure "slang" sounds good, if that means, a module offers it. 21:37
timotimo modules can offer slangs, aye
skids It's a rather deep form of language tinkering, even deeper than macros.
pflanze I'll be happy to provide that module :) (when time comes) 21:38
(The only thing concerning the outside of such a module is that compilers that can infer the types can optimize between curried and uncurried representation; 21:40
skids pflanze: I think the more understandable approach to the recursive thing under Perl6 would be to use multidispatch to seed the start value.
pflanze GHC definitely does it. So, there would be room for a module-provided optimization pass, too, or so.) 21:41
skids, you can also wrap it directly, my &ff= ->&g,&h,&x { my &f= .... } 21:42
skids anyway I should have left a long time ago.
pflanze uncurry is just an abstraction for that pattern (i.e. takes away the need for you typing it out).
skids
.oO(write-once)
 21:43
21:43 skids left 21:46 diana_olhovik_ left 21:57 TimToady left 21:58 mohij_ left 22:06 TimToady joined
jercos m: IO::Socket::Unix 22:09
camelia rakudo-moar f7cfe9: OUTPUT«Could not find symbol '&Unix'␤ in block <unit> at /tmp/9XTnthmePP:1␤␤Actually thrown at:␤ in block <unit> at /tmp/9XTnthmePP:1␤␤»
GLRelia rakudo-moar 27cb27: OUTPUT«Could not find symbol '&Unix'␤ in block <unit> at /tmp/YrQwEPHtRM:1␤␤Actually thrown at:␤ in block <unit> at /tmp/YrQwEPHtRM:1␤␤»
jdv79 the pause fix requires some schema changes. i guess pause can be down for a few minutes. 22:10
if it happens...
22:17 mvuets left
jercos so... are Unix sockets planned in any perl 6 version? 'cause I mean, e.g., nqp::socket() doesn't even take a family. If berkeley sockets were fully exposed I could maybe see writing up IO::Socket::Unix, but I don't think I have the time commitment to put towards changing nqp... 22:23
S32 mentions IO::Socket::Unix, and... that's about it as far as I can see 22:24
brrt jercos: these things are planned, yes, but there are two 'challenges' 22:26
one is that we want this to work on multiple backends, namely jvm and moarvm
pmurias can write the nqp ops for js 22:27
I could even try to write them for the java backend 22:28
jercos yeah, I don't think there's a clean way to access Unix sockets from JVM without JNI... or JS for that matter >.>
brrt moarvm, as far as i can see, would require some hackery iwth libuv
jercos Parrot's socket API seems to be berkeley enough for it to work, but I haven't looked into that.
pmurias jercos: for the io stuff in js I just target node.js 22:29
AlexDaniel m: my $p1 = start { say "1:{foo()}" }; my $p2 = start { say "2:{foo()}" }; say "3:{foo()}"; await $p1, $p2; say foo(); sub foo() is cached { sleep rand * 2; rand } 22:30
GLRelia rakudo-moar 27cb27: OUTPUT«3:0.497086709580827␤1:0.344976470869383␤2:0.446260711123365␤0.446260711123365␤»
camelia rakudo-moar f7cfe9: OUTPUT«3:0.809124497460455␤2:0.381401760822368␤1:0.816585751269165␤0.816585751269165␤»
pmurias obviously the browser can't open unix sockets
AlexDaniel so it is always the last execution
interesting
jercos TIL node.js has Unix sockets. neato. :) 22:31
22:37 RabidGravy left 22:38 virtualsue left 22:41 TimToady left 22:42 TimToady joined 22:44 vendethiel left 22:47 freeze left 22:49 freeze joined 22:54 freeze left 22:55 freeze joined 22:57 TimToady left 22:58 llfourn joined 23:03 llfourn left 23:04 TimToady joined 23:05 brrt left 23:14 skids joined 23:29 pierrot joined 23:31 ribasushi left 23:33 nebuchad` joined 23:35 ribasushi joined 23:50 TimToady left 23:52 pmurias left 23:53 TimToady joined 23:58 TimToady left

Powered by Cro and the Raku® Programming Language.

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

Thank you!