»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.
masak at least that's the way it feels, deep down in my intuition. 00:00
but yes, I see the deviation from consistency in what an integer literal *is* in the language. 00:05
suddently it's one thing inside of literals, and another thing outside of them.
I wonder if there's a middle way somehow, a way to allow prefix:<-> without corrupting what "literal" means. 00:06
'night, #perl6
vendethiel o/ masak 00:18
ShimmerFairy TimToady: my suspicion is that more people see -1 where the - is a literal part of the int (like '.' in 3.0, or 'e' in 5e2), and don't see it so much as &prefix:<->(1) 01:08
TimToady: but I don't know if actually making the change to reflect is a good idea at this point.
ShimmerFairy notes that S03 shows ** has its own precedence level all by itself, most likely just to make -1 ** 2 mathematically right 01:09
dalek kudo/nom: ea2d21c | TimToady++ | src/Perl6/ (2 files):
allow negative integers as parameters
01:16
ast: 08154ac | TimToady++ | S06-multi/value-based.t:
test negative parameter values
BenGoldberg m: say -1 ** 0.5 01:17
camelia rakudo-moar 6252fc: OUTPUT«-1␤»
BenGoldberg m: sqrt(-1)
camelia rakudo-moar 6252fc: OUTPUT«WARNINGS:␤Useless use of "sqrt(-1)" in expression "sqrt(-1)" in sink context (line 1)␤»
BenGoldberg m: sqrt(-1).say
camelia rakudo-moar 6252fc: OUTPUT«NaN␤»
BenGoldberg m: sqrt(-1 + 0i).say 01:18
camelia rakudo-moar 6252fc: OUTPUT«6.12323399573677e-17+1i␤»
BenGoldberg m: ((-1 + 0i) ** .5 ).say
camelia rakudo-moar 6252fc: OUTPUT«6.12323399573677e-17+1i␤»
BenGoldberg m: ((-1) ** .5 ).say
camelia rakudo-moar 6252fc: OUTPUT«NaN␤»
BenGoldberg m: (-1 ** .5 ).say
camelia rakudo-moar 6252fc: OUTPUT«-1␤»
ShimmerFairy m: use Test; is-approx i, <-1+0i> ** .5, 1e-6 01:21
camelia rakudo-moar 6252fc: OUTPUT«ok 1 - ␤»
ShimmerFairy close enough though :P
(for some reason not specifying 1e-6 is broken with Test, even though that's the default)
BenGoldberg m: (<-1> ** .5 ).say
camelia rakudo-moar 6252fc: OUTPUT«NaN␤»
BenGoldberg m: my $x = 2; my $y = 3; say ($x.$y); # imagine if this printed '2.3' ?
camelia rakudo-moar 6252fc: OUTPUT«Cannot find method 'postcircumfix:<( )>'␤ in block <unit> at /tmp/bzNQcKrh0m:1␤␤»
BenGoldberg m: my $x = 2; my $y = 3; say($x.$y); # imagine if this printed '2.3' ? 01:22
camelia rakudo-moar 6252fc: OUTPUT«Cannot find method 'postcircumfix:<( )>'␤ in block <unit> at /tmp/X6TQ3o0npr:1␤␤»
ShimmerFairy BenGoldberg: yeah, that's the one thing that makes me hesitant about making -1 literal; it'd make '-' the only thing that's part of a literal *and* available as an operator :) 01:24
say (<-1+0i> ** .5).reals».Rat.map({$^a + $^b\i}) 01:25
m: say (<-1+0i> ** .5).reals».Rat.map({$^a + $^b\i})
camelia rakudo-moar 6252fc: OUTPUT«(0+1i)␤»
ShimmerFairy BenGoldberg: for the record, piping the pieces of the Complex number through Rat happens to eliminate floating-point oddities ^^^ :)
BenGoldberg m: say (<-1+0i> ** .5).reals 01:35
camelia rakudo-moar 6252fc: OUTPUT«(6.12323399573677e-17 1)␤»
BenGoldberg m: say (<-1+0i> ** .5).reals>>.Rat
camelia rakudo-moar 6252fc: OUTPUT«(0 1)␤»
ShimmerFairy BenGoldberg: I can't guarantee that Num->Rat will always do that, but in this case it happened to :) 01:39
ShimmerFairy ("that" being "remove some of the oddities that can appear in binary-based floating point numbers") 01:40
BenGoldberg m: Rat(1,1).say 01:47
camelia rakudo-moar 6252fc: OUTPUT«2␤»
BenGoldberg m: Rat(1,-1).say
camelia rakudo-moar 6252fc: OUTPUT«2␤»
BenGoldberg m: Rat(1).say 01:48
camelia rakudo-moar 6252fc: OUTPUT«1␤»
BenGoldberg m: Rat(-1).sqrt.say
camelia rakudo-moar 6252fc: OUTPUT«NaN␤»
dha docs for C<while> and C<until> added. 02:03
dha You *can* git from an airplane! 02:04
geekosaur network no longer costs more than the plane tickets? 02:13
dalek ecs: bdbe964 | ShimmerFairy++ | S03-operators.pod:
Add postfix:<i> to S03

It wasn't mentioned at all before, and since it's at the methodcall level in rakudo, which may not be expected at first, I figured it was worth mentioning. It also makes postfix:<i> the only operator at that level to not start with a dot, which seems like another noteworthy detail.
zostay m: sub foo(Str() $s) { say $s }; foo(4) 03:27
yoleaux 14 Sep 2015 19:40Z <jdv79> zostay: IO::Socket::INET seems a bit too specific
camelia rakudo-moar ea2d21: OUTPUT«4␤»
zostay !tell jv79 IO::Socket::INET is too specific, but using the Raw IO is completely the wrong solution always and the alternative is to be implementation-specific, which might as well not even be in the standard 03:30
dalek p: 17b6555 | hoelzro++ | src/vm/js/nqp-runtime/deserialization.js:
Remove comments from JS backend

These are notes I was taking during my merge of the JS branch that I forgot to remove
03:31
zostay .tell jdv79 IO::Socket::INET is too specific, but using the Raw IO is completely the wrong solution always and the alternative is to be implementation-specific, which might as well not even be in the standard 03:32
yoleaux zostay: I'll pass your message to jdv79.
travis-ci NQP build failed. Rob Hoelz 'Remove comments from JS backend 03:36
travis-ci.org/perl6/nqp/builds/81121244 github.com/perl6/nqp/compare/a2c05...b6555833df
Zoffix TigerDirect-- # 20 minutes between my call and "I dunno. OK. BAI!" EVGA++ # 20 minutes between "Fix my card. Dunno what you want" and "K. Send it here. We'll fix it" 03:48
TEttinger Zoffix: good to know. I never know what companies have good customer service before it's too late 04:12
dalek ast: 3fdcdae | ShimmerFairy++ | S02-literals/allomorphic.t:
Fix tests for GLR

This test file was forgotten in the merging mess surrounding allomorphs, so it took a while for this to show up.
05:00
dalek ast: 41fa6e7 | ShimmerFairy++ | S02-literals/allomorphic.t:
Change complex number val() tests.

On second thought, it seems that stacking signs like was being tested isn't all that useful, so the tests in question have been changed to testing a different kind of complex number (one using Inf in the imaginary part).
05:17
dalek ast: 12fe957 | ShimmerFairy++ | S02-types/ (3 files):
Fix plans in a few test files

These were forgotten about when various duplicate tests were removed.
05:33
bartolin could it be that eval-dies-ok and eval-lives-ok from Test.pm is busted (and has been for some time)? it seems that symbols from the surrounding scope are not known: 05:43
m: use Test; my $a = 1; eval-dies-ok( '$a = 7' ) 05:44
camelia rakudo-moar ea2d21: OUTPUT«ok 1 - ␤»
bartolin m: use Test; my $a = 1; eval-lives-ok( '$a = 7' )
camelia rakudo-moar ea2d21: OUTPUT«not ok 1 - ␤␤# Failed test at /tmp/EAYzorI5Ow line 1␤# Error: Variable '$a' is not declared␤»
bartolin star-m: use Test; my $a = 1; eval-dies-ok( '$a = 7' )
camelia star-m 2015.03: OUTPUT«5===SORRY!5=== Error while compiling /tmp/lcbJJuG6xn␤Undeclared routine:␤ eval-dies-ok used at line 1. Did you mean 'eval_dies_ok', 'eval_lives_ok'?␤␤»
bartolin star-m: use Test; my $a = 1; eval_dies_ok( '$a = 7' )
camelia star-m 2015.03: OUTPUT«ok 1 - ␤»
bartolin star-m: use Test; my $a = 1; throws-like '$a = 7', Exception # works for throws-like 05:45
camelia star-m 2015.03: OUTPUT«5===SORRY!5=== Error while compiling /tmp/5NDOMQiyA1␤Undeclared routine:␤ throws-like used at line 1. Did you mean 'throws_like'?␤␤»
bartolin m: use Test; my $a = 1; throws-like '$a = 7', Exception # works for throws-like 05:46
camelia rakudo-moar ea2d21: OUTPUT« 1..2␤ not ok 1 - '$a = 7' died␤ ␤# Failed test ''$a = 7' died'␤# at /tmp/CPXtcpXSwR line 1␤ ok 2 - # SKIP Code did not die, can not check exception␤ # Looks like you failed 1 test of 2␤not ok 1 - did we throws-like Exception?…»
bartolin hmm, adding 'context => CALLER::CALLER::CALLER::' to the EVAL in eval_exception (Test.pm:484) seems to fix that. 05:56
dalek kudo/nom: 9856a43 | TimToady++ | src/core/ (2 files):
change list associative reductions to onearg
06:06
bartolin with that change to Test.pm there are about a dozen spectest failures. (maybe wrong tests) 06:27
dalek osystem: 133a138 | ugexe++ | META.list:
Update META.list
07:28
garu_ hi everyone! quick question: what's the most idiomatic way to fetch the last element of an array? Like $array[-1] in Perl 5 07:47
ShimmerFairy garu_: @array[*-1] (the * in this case means "number of elements") 07:50
mst every time I see *, I think "and I thought $_ meant lots of things" 07:51
garu_ heh
mst then again also every time I see *, I think "ooooh shiny"
"this sword is double edged, what a surprise"
dalek ecs: 3bf2eb9 | (Stefan Seifert)++ | S07-lists.pod:
Fix syntax error and add clarification in S07-lists
07:52
garu_ ShimmerFairy: thanks! Now that you mentioned it, I see it right there on the error message :)
ShimmerFairy np :) 07:53
lizmat technically, *-1 creates a code object that takes a single parameter 08:02
the slicing logic will call that with the number of elements 08:03
but you can really do anything with this
m: my @a = ^10; @a[ { say "got $^a" } ] 08:04
camelia rakudo-moar 9856a4: OUTPUT«got 10␤»
nine Should I rebase gmr onto nom before merging or merge nom into gmr? 08:40
timotimo personally, i prefer rebasing 08:43
psch hi #perl6 o/ 08:47
rebase both ways 08:48
timotimo ohai psch
psch always rebase everything!
timotimo yeah, rebase nom onte gmr!
psch and then back!
timotimo draw a pretty picture in the history graph
ShimmerFairy GitTurtle programming language
dalek ast: b88cb33 | (Stefan Seifert)++ | S (3 files):
Rename EnumMap to Map
08:59
roast: 473e246 | (Stefan Seifert)++ | S (7 files):
roast: Merge Enum into Pair
timotimo in the past i've tended to build "new" branches with "_2", "_3", .. whenever i've rebased things i expect others would already have played with 09:00
but if you're going to merge the stuff into nom anyway, there's no need for that
nine github.com/rakudo/rakudo/commit/7ee6ef9 Remove PairMap - its use case is unclear and it's untested
github.com/rakudo/rakudo/commit/0d40e00 Fix return 1, :a(2)
github.com/rakudo/rakudo/commit/0f699bb Make %hash.pairs .value writable
github.com/rakudo/rakudo/commit/0e0495e Fix passing adverbs to 'import' and 'no' statements
github.com/rakudo/rakudo/commit/5256ed6 Fix {:a(1) :b(2)} dieing due to missing value arg to Pair.new
github.com/rakudo/rakudo/commit/86536d2 Fix (:a(1) :b(2)) dieing due to missing value arg to Pair.new 09:01
github.com/rakudo/rakudo/commit/de907e5 Fix use Foo :whatever<1> losing the adverb's value
github.com/rakudo/rakudo/commit/4023a33 Make :a($i) and a => $i mutable like 'a' => $i
github.com/rakudo/rakudo/commit/2232157 Merge Enum into Pair
pink_mist nine++
ShimmerFairy The advice I've seen (and agree with): rebase unless/until you've pushed the stuff-to-rebase somewhere, at which point it saves everyone grief to just merge
nine Makes sense, yes. 09:03
ShimmerFairy Even then, I don't think merging is as bad as some people would like to make it out to be :)
(merging feels more natural as a way of saying "I am taking this feature I've been working on, and moving it into the mainline" to me, for example, and unlike rebasing explains the time warping you'll see in commit timestamps ☺) 09:05
nine Well you have to merge the feature branch into the mainline even after rebasing it. I should have used --no-ff though to get a proper merge commit. OTOH half of gmr's commits were fixes that would have been good to have even without the gmr. 09:06
nine So it wouldn't have made as much sense to revert the whole merge 09:06
nine And 2 tickets closed by the gmr :) 09:09
psch huh, IntStr has a P6OpaqueDelegateInstance in field_0 during construction..? 09:10
does that make sense..? :/
timotimo daleninek? 09:15
psch ...and P6bigint.c isn't really helpful :/ 09:21
the thing that i'm kind of hung up on right now is
timotimo you may want to look into P6Opaque instead?
to find out how these things work? 09:22
psch yesterday i had a case where REPRData.unboxIntSlot had the right hint for the field the bigint should go in
timotimo that should be the common case, no?
psch but i forgot the exact case, and the one i use now doesn't have it
i'm guessing &val and <0> do it differently somehow, still 09:23
but even if that hint is there it's still wonky, cause we want to bind_native
but IntStr has a P6opaque in field_1 that has the BigInteger in field_0 09:24
or "had", rather, in cases where it worked
but Int has a BigInteger in field_0
which i guess is not-quite-right composition 09:25
and composition being wonky would also explain why we get a delegateInstance in field_0 - and not in .delegate
well, excepting the (reasonable) explanation "psch doesn't really get it, it's fine" :P 09:26
timotimo huh 09:29
timotimo doesn't knows
psch P6opaque.c looks like i'd expect it - it checks unbox_int_slot and gets the bigint from there 09:30
timotimo yeah
psch which means the IntStr type doesn't get the right hints on jvm, but does on moar 09:32
timotimo OK, that may be a good hint perhaps 09:34
psch it's puzzling, cause seeing the fitting unboxIntSlot in the debugger is what made me rewrite my impl from "iterate the reflected fields and check their type" to "check for the slot in the REPRData" 09:36
timotimo we're supposed to have only one slot for every kind of unboxing operation, so that'd make sense 09:37
at least that's how moar does it, as you've figured out for yourself, too
psch i mean, all unbox*Slot are -1, so none get set apparently 09:39
that's with &val and <0>, so potential differences in object construction there aren't it :s
timotimo very strange
ShimmerFairy psch: have you tried seeing what happens with the older version of IntStr construction? It won't work when you want bigger-than-native integers, but does a construction process mirroring the other allomorphs still cause problems? 09:43
psch ShimmerFairy: oh, you mean without the add_I workaround? i expect that to not hit the same code path at all 09:44
ShimmerFairy: but the getBI/makeBI bits are what's wonky in the first place 09:45
ShimmerFairy psch: yes, that's what I mean. I also wonder what would happen if IntStr did nqp::create like the other types, but still did add_I to actually set the Int part. 09:46
psch: OK. I've gotten the feeling that it's the VM's fault, but admittedly the IntStr is being constructed a bit weirdly :)
psch ShimmerFairy: well, it's not the VM, it's our bigint handling :)
ShimmerFairy: nqp::create would only add another (unused) allocation, makeBI allocates as well 09:47
ShimmerFairy psch: it's the VM's fault for not making bigints easier at first implementation :P
psch ShimmerFairy: idk, the first implementation was really easy and worked :P 09:52
well, until we had IntStr that is 09:53
timotimo i'm glad you're doing it, psch 09:54
i mean, working on fixing the JVM port of rakudo
moritz psch: just wait for intInt, the new type with a native int and a bigint slot, hold separate values 50% of the time
:-)
timotimo and also the AtAt class that represents objects that have two locations in physical memory 09:58
psch and ObjAtAt that points at the ObjAt 09:59
timotimo and, of course, AtSt which points at an object's STable's memory location
_itz I assume GMR fallout might be less pain than GLR? 10:00
psch oh right
timotimo much, much, much less
probably not noticable
_itz :D
psch and then we can implement polymorphic objects that use an algorithm adapted from C3MRO to find their REPRs, and call them C3PO right 10:01
timotimo yes.
erxeto moritz: thsnka for the patch on DBIish. Now Task::Star installation completes without issues. 10:03
thanks*
timotimo sweet
moritz erxeto: you're welcome 10:11
dalek ar/panda-bootstrap: 7b2465c | moritz++ | tools/star/release-guide.pod:
Wee correction in the release guide

since there is no more PARROT_VERSION to adjust, there are only three lines to change
10:14
ar: 06f75c4 | moritz++ | tools/build/panda-state.p6:
Socket.send is now .print
10:18
dalek kudo/nom: a3962b0 | (Stefan Seifert)++ | src/core/IO/Handle.pm:
Fix IO::Handle not correctly limiting lines in batch processing

The iterator's push-all did not actually stop iteration when $!todo got down to 0. Also it did not close the file handle on reaching eof despite being asked to do so.
Fixes regression in S16-filehandles/io.t
10:22
moritz isn't it wonderful that we have tests? :-) 10:23
ShimmerFairy moritz: and so many! :) 10:24
nine If only we could have more of them. 10:27
nine The closing the file handle part is not covered by any test 10:28
cognominal Where is the S[] = ... with the upper case S documented in the synopses? 10:43
timotimo that's just the "sequential" metaop
right?
but i haven't seen it used in that way yet
ShimmerFairy timotimo: perhaps the S/// non-mutating substitution 10:44
psch S03:Sequential_Operators
cognominal I see it alright in the grammar and actions github.com/rakudo/rakudo/blob/nom/....nqp#L6803 but not in roast or the spec. I probably don't know where to look
psch vOv
design.perl6.org/S03.html#Sequential_operators
but there's also S///, as ShimmerFairy says
timotimo oh, of course
in that case it's the non-mutating substitution 10:45
psch m: $_ = "foo"; say S[o] = "O"
camelia rakudo-moar 7ee6ef: OUTPUT«fOo␤»
psch m: $_ = "foo"; say S[o] = "O"; .say 10:46
camelia rakudo-moar 7ee6ef: OUTPUT«fOo␤foo␤»
cognominal m: $_ = "foo"; say s[o] = "O"; .say
camelia rakudo-moar 7ee6ef: OUTPUT«「o」␤fOo␤»
psch S/// is .subst, s/// is .subst-mutate 10:47
cognominal ok
masak right, it's what the /r flag does in Perl 5. 10:50
moritz ARGL, why do we have both Term::ANSIColor and Terminal::ANSIColor in the ecosystem? 11:01
of course this makes a star relase harder :( 11:02
ShimmerFairy moritz: because it was suddenly a bad thing for "Term" to be a name in modules. Somehow. 11:04
moritz though Term::ColorText, Term::ProgressBar and Term::termios don't seem to be particularly bad 11:05
dalek ar: 4051620 | moritz++ | / (3 files):
Bump some versions
11:06
ar: 9c32c32 | moritz++ | modules/ (27 files):
Update submodules
ShimmerFairy moritz: yep, IIRC the reason was this "but there's a name called 'term' in the grammar!!!" or something along those lines, which I don't buy at all.
(Term::ColorText is mine, and damned if I rename it) 11:07
dalek ar: 8b1cb9f | moritz++ | / (4 files):
Use Terminal::ANSIColor instead of Term::ANSIColor

Grammar::Debugger has changed its dependency. Because reasons.
11:10
moritz Testing modules/p6-file-directory-tree with /home/moritz/p6/star/install/bin/perl6-m... 11:13
done without supply or react in block <unit> at t/basic.t:26
.tell labster any chance you could merge github.com/labster/p6-file-directo...ree/pull/5 soonish? It would help with the Rakudo Star release 11:14
yoleaux moritz: I'll pass your message to labster.
moritz panda and Template::Mustache use JSON::Fast, while jsonrpc still uses JSON::Tiny 11:16
I guess for now star will have to ship both :(
(unless we drop jsonrpc)
jnthn wondered why the Term/Terminal bikeshed pointlessness 11:45
I only accepted the PR to stop my module give people a warning, not 'cus I thought the name change made sense. 11:46
Guest19613 Marius1984 11:54
Identificar Marius1984 11:55
timotimo you change your nickname by typing /nick Marius1984
DrForr Or more pertinently /msg nickserv help 11:56
z448 p6: say something 11:58
camelia rakudo-moar a3962b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/tmpfile␤Undeclared routine:␤ something used at line 1␤␤»
z448 p6: say 'something';
camelia rakudo-moar a3962b: OUTPUT«something␤»
psch /o\ 11:59
ShimmerFairy jnthn: that was what I recall btw. I can't fathom why 'Term' and 'term' could ever be a problem, esp. since we're not a case-insensitive language :) 12:01
timotimo i didn't even know we changed the name; i thought someone came up with a distinct Terminal::ANSIColor 12:07
ShimmerFairy nah, it's just the result of pointless bikeshedding :) 12:08
timotimo this is the perfect opportunity to model and implement "provides" 12:09
Terminal::ANSIColor provides Term::ANSIColor
jnthn & 12:14
tadzik timotimo: the idea to change the name is that Term is a thing in perl 6 12:35
as in term 'self' or so 12:36
timotimo oh
timotimo that sounds kind of sensible 12:36
timotimo puts down pitchfork and torch
ShimmerFairy tadzik: still not convincing for me. 'Term' is not a thing in standard P6, and I don't buy that someone will mistake Term::ANSIColor for implementing a term, for example. 12:37
tadzik so we might have eventually Operator::plusminus and Term::thyself or so
ShimmerFairy: it's not a matter of mistaking it's that Term may become useful for us in module space, and I think it's good to disambiguate Terminal 12:38
ShimmerFairy But it already is useful, as meaning Term[inal] :)
tadzik heh :)
tadzik maybe we should use some unicode monospace characters that read Term 12:39
ShimmerFairy (and there's no law saying we have to make sure a part of a longname has an entirely, 100% unique and singular meaning)
timotimo why don't we have a version of Term::ANSIColor that exports the colors as terms? :P
tadzik timotimo: that... is brilliant
timotimo Termi, termn, Terma, terml?
tadzik Term::inator 12:40
that'd be a module for terming
Goblin_ How to convert string into array of chars? 12:41
tadzik .comb should work
m: "foobar".comb.perl.say
camelia rakudo-moar a3962b: OUTPUT«("f", "o", "o", "b", "a", "r").Seq␤»
tadzik fsvo array
timotimo well, if you assign it to a @foo, you'll get a proper array 12:42
but oftentimes you don't need that
colomon m: my @a = “foobar”.comb; dd @a 12:43
camelia rakudo-moar a3962b: OUTPUT«Array @a = ["f", "o", "o", "b", "a", "r"]␤»
colomon indeed. :)
_itz maybe contraversial but
github.com/rakudo/star/pull/50
tomboh I have installed rakudo and moarvm on freebsd, and I'm seeing t/04-nativecall/16-rt125408.t fail - is this a known problem?
timotimo i'm not against that 12:44
tomboh I don't see anything in RT but I thought I'd ask here before submitting a new bug report
colomon _itz: I don’t think there’d be anything contraversial at all if you just added the SPW slides. (Presuming we have jnthn++’s permission.)
_itz good point we should at least ask :) 12:45
Guest95184 Hack bank BPI whit Marius1984 12:46
psch tomboh: the filename is a ticket number, RT #125408 12:47
tomboh: it's closed though, and i'm not sure if you can request reopen on RT directly
(rt.perl.org/Public/Bug/Display.html?id=125408)
_itz jnthn: can we use your SWP slides as a replacement (for the time being) for the draft book pdf in star please? 12:48
^ SPW
tomboh psch: sure, I gather the test is a regression test to ensure the fix for rt#125408 doesn't break - and I'm seeing it fail
tomboh oh, but that was only fixed on Thursday - I'll do a fresh "git pull" and retest 12:51
tomboh psch: thank you, "All tests successful." :) 12:53
pmurias (Term::ANSIColor rename)-- 12:59
pmurias there is already Term::Curses 13:03
there is always Lang::Term:: for linguistic usage 13:06
timotimo mhm 13:08
dalek ar: e933fca | moritz++ | / (3 files):
Ship JSON::Fast with Star

panda and Template::Mustache (included from Bailador) depend on it
13:10
Goblin_ How to generate all possible permutations of characters with a given minimum and maximum length? 13:44
colomon Goblin_: of a given set of characters? 13:50
Goblin_ yes
colomon Goblin_: doc.perl6.org/routine/combinations 13:51
m: say (‘a’ .. ‘f’).combinations(3..4).perl 13:52
camelia rakudo-moar a3962b: OUTPUT«(("a", "b", "c"), ("a", "b", "d"), ("a", "b", "e"), ("a", "b", "f"), ("a", "c", "d"), ("a", "c", "e"), ("a", "c", "f"), ("a", "d", "e"), ("a", "d", "f"), ("a", "e", "f"), ("b", "c", "d"), ("b", "c", "e"), ("b", "c", "f"), ("b", "d", "e"), ("b", "d", "f"), …»
colomon oh, wait, you want permutations.
m: say (‘a’ .. ‘f’).permutations(3..4).perl
camelia rakudo-moar a3962b: OUTPUT«Cannot call permutations(List: Range); none of these signatures match:␤ (List $: *%_)␤ in block <unit> at /tmp/KUHFfrwMSI:1␤␤»
colomon m: say (‘a’ .. ‘f’).permutations(3).perl
camelia rakudo-moar a3962b: OUTPUT«Cannot call permutations(List: Int); none of these signatures match:␤ (List $: *%_)␤ in block <unit> at /tmp/ID80shav9d:1␤␤»
psch m: say (‘a’ .. ‘f’).permutations().perl 13:53
camelia rakudo-moar a3962b: OUTPUT«(("a", "b", "c", "d", "e", "f"), ("a", "b", "c", "d", "f", "e"), ("a", "b", "c", "e", "d", "f"), ("a", "b", "c", "e", "f", "d"), ("a", "b", "c", "f", "d", "e"), ("a", "b", "c", "f", "e", "d"), ("a", "b", "d", "c", "e", "f"), ("a", "b", "d", "c", "f", "e"),…»
colomon m: say (‘a’ .. ‘f’).permutations().perl
camelia rakudo-moar a3962b: OUTPUT«(("a", "b", "c", "d", "e", "f"), ("a", "b", "c", "d", "f", "e"), ("a", "b", "c", "e", "d", "f"), ("a", "b", "c", "e", "f", "d"), ("a", "b", "c", "f", "d", "e"), ("a", "b", "c", "f", "e", "d"), ("a", "b", "d", "c", "e", "f"), ("a", "b", "d", "c", "f", "e"),…»
colomon hmmm… may have to play around a bit to get a combination of those two things…. 13:54
timotimo no, i think he does want combinations, not permutations
otherwise "minimum and maximum length" don't make sense
oh, but maybe no repetition is wanted? 13:55
DrForr I'd guess 'aaa', 'aab'..'fffe','ffff' for password generation?
Goblin_ yes
timotimo ah 13:58
well, that's just 'aaa'..'zzzzz' in that case
m: say ("aa".."zzz").perl 13:59
camelia rakudo-moar a3962b: OUTPUT«"aa".."zzz"␤»
timotimo m: say ("aa".."zzz").list.perl
moritz m: say join '', ('a'..'z').roll((3..5).pick)
camelia rakudo-moar a3962b: OUTPUT«("aa", "ab", "ac", "ad", "ae", "af", "ag", "ah", "ai", "aj", "ak", "al", "am", "an", "ao", "ap", "aq", "ar", "as", "at", "au", "av", "aw", "ax", "ay", "az", "ba", "bb", "bc", "bd", "be", "bf", "bg", "bh", "bi", "bj", "bk", "bl", "bm", "bn", "bo", "bp", "bq…»
rakudo-moar a3962b: OUTPUT«uiiq␤»
timotimo ah, there you can't really see it get longer
moritz huh
timotimo m: say ("aa".."ccc").list.perl
camelia rakudo-moar a3962b: OUTPUT«("aa", "ab", "ac", "ad", "ae", "af", "ag", "ah", "ai", "aj", "ak", "al", "am", "an", "ao", "ap", "aq", "ar", "as", "at", "au", "av", "aw", "ax", "ay", "az", "ba", "bb", "bc", "bd", "be", "bf", "bg", "bh", "bi", "bj", "bk", "bl", "bm", "bn", "bo", "bp", "bq…»
timotimo oh, and there it does a-z, too
psch m: say join '', ('a'..'z').roll((3..5).pick(5)) 14:02
camelia rakudo-moar a3962b: OUTPUT«gqh␤»
psch oh right
m: say join '', ('a'..'z').roll((3..5).pick)[^5]
camelia rakudo-moar a3962b: OUTPUT«quk␤»
psch m: say ('a'..'z').roll((3..5).pick)[^5]
camelia rakudo-moar a3962b: OUTPUT«(c b t t)␤»
psch m: say ('a'..'zzzz').roll((3..5).pick)[^5]
timeout \o/ 14:03
camelia rakudo-moar a3962b: OUTPUT«(timeout)» 14:03
psch well, yeah, to be expected...
moritz it needs to generate the whole list to even know the propability to pick one 14:19
DrForr Isn't there an old interview question showing how to get around that? Something like "Choose an item from an arbitrarily long list with equal probability to each entry?" Something like the equivalent of for(@x){rand(1/$.) and return $_} (though $. doesn't work for arrays, and it's perl5.) 14:21
lucs DrForr: I believe you need to know when you've reached the end of that arbitrarily long list for that technique to work. 14:24
DrForr Entirely possible, it's been a long time since I've had to deal with that interview question. 14:25
dalek ar: f86f204 | (Steve Mynott)++ | docs/ (2 files):
Replace draft book (not currently actively maintained) by jnthn's Swiss
14:28
ar: a738fc9 | moritz++ | docs/ (2 files):
Merge pull request #50 from stmuk/master

Replace draft book by SWP slides
moritz anyone want to write a release announcment for the 2015.09 star release? 14:55
Skarsnik Hello. How do I get the type of an attribute? (getting it from MyClass.^attributes) $attr.WHAT give me.. Attribute 15:05
moritz Skarsnik: it seems that's not exposed yet 15:06
psch m: class A { has Int $!foo }; A.^attributes[0].type.say 15:07
camelia rakudo-moar a3962b: OUTPUT«(Int)␤»
moritz m: say Atribute.^can('type')
camelia rakudo-moar a3962b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/PyptEZTlbG␤Undeclared name:␤ Atribute used at line 1. Did you mean 'Attribute'?␤␤»
moritz m: say Attribute.^can('type')
camelia rakudo-moar a3962b: OUTPUT«(<anon>)␤»
Skarsnik oh there is a .type?
moritz seems like it
m: say Int.^attributes[0].type
camelia rakudo-moar a3962b: OUTPUT«Method 'gist' not found for invocant of class 'bigint'␤ in block <unit> at /tmp/IAxlHLc7NV:1␤␤»
moritz ah, bootstrapping 15:08
psch yeah, that's bigint $!value
masak some 'gist' methods still missing.
moritz m: say Enum[C.^attributes[0].type
camelia rakudo-moar a3962b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/5QL6R9_krV␤Bogus postfix␤at /tmp/5QL6R9_krV:1␤------> 3say Enum7⏏5.^attributes[0].type␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ …»
moritz m: say Enum.^attributes[0].type
camelia rakudo-moar a3962b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/gleQBHt7Ye␤Undeclared name:␤ Enum used at line 1. Did you mean 'num'?␤␤»
masak m: say Int.^attributes[0].type.^name
camelia rakudo-moar a3962b: OUTPUT«bigint␤»
psch m: say Int.^attributes[0].^attributes[2].type.^name 15:10
camelia rakudo-moar a3962b: OUTPUT«Mu␤»
psch that's $!descriptor, iirc :P
m: say Int.^attributes[0].^attributes[2].name
camelia rakudo-moar a3962b: OUTPUT«$!box_target␤»
psch ah, now
-w
m: say Int.^attributes[0].^attributes>>.name
camelia rakudo-moar a3962b: OUTPUT«($!name $!type $!box_target $!package $!inlined)␤»
psch shrugs 15:11
moritz m: say Attribute.^methods(:local)>>.name
camelia rakudo-moar a3962b: OUTPUT«(<anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> <anon> compose apply_handles get_value set_value container has-accessor readonly package inlined WHY set_why Str gist)␤»
Skarsnik I got a weird error too: I have sub plain_attr(Hash $opt, Attribute @tab). If I write plain_attr({}, MyClass.^attributes) I got a type check error, but if I write the Attribute @tab= MyClass.^attributes and call the sub with the @tab it work 15:12
dalek c: 99aef50 | moritz++ | lib/Type/Attribute.pod:
Document Attribute.type
15:13
moritz Skarsnik: I'll have to write an FAQ entry for that one, I fear :(
psch m: my Attribute @foo; say Int.^attributes.WHAT; say @foo.WHAT
camelia rakudo-moar a3962b: OUTPUT«(List)␤(Array[Attribute])␤»
psch m: say List ~~ Array[Attribute].new 15:14
camelia rakudo-moar a3962b: OUTPUT«False␤»
psch m: say Array[Attribute].new ~~ List
camelia rakudo-moar a3962b: OUTPUT«True␤»
moritz Skarsnik: if you have a paramter typed 'Attribute @thing', then it'll only accept arguments of type Positional[Attribute] (or subtypes thereof, such as List[Attribute]
Skarsnik Type check failed in binding @tab; expected 'Positional[Attribute]' but got 'Parcel' 15:15
psch makes me wonder if we want CORE to return narrowly typed lists...
moritz Skarsnik: the difference is that *assignment* to an array is coercive, so it doesn't matter what the right-hand side is typed
psch probably not, even ignoring performance
moritz Skarsnik: oh, and your rakudo is too old
Skarsnik I don't know. I install perl6 with a script from perl6.org like 2 month ago ~~ 15:16
psch hm, is parameterized type coercion design..? 15:17
moritz psch: no
psch m: Array[Int](@(1, 2, 3)) # somewhat like this
camelia rakudo-moar a3962b: OUTPUT«Type check failed in assignment to ''; expected 'Int' but got 'List'␤ in block <unit> at /tmp/Nt3naZeWF6:1␤␤»
moritz psch: see www.nntp.perl.org/group/perl.perl6....g2155.html for my rant
Skarsnik it's pretty... unatural? unexpected? x) 15:18
moritz Skarsnik: I know; assignment to arrays is coercive, so people are surprised with (signature-)binding to arrays isn't 15:19
Skarsnik It's more for me an Array is an ..Array. why would it be weird subtype that I have to pay attention? 15:21
psch Skarsnik: a (workable but clumsy) comparison is Java generics
Skarsnik: an ArrayList<Object> can be assigned an ArrayList<Integer>, but an ArrayList<Integer> can't be assigned an ArrayList<Object> 15:22
Skarsnik: and any literal array in perl6 is pretty much an Array[Any] 15:23
moritz: i think i had read your rant once already... i have to admit the approach mused about seems tempting, but i can't see the bits of the Binder i know work good with it - not to speak of those i don't... 15:24
so, yeah, "if you want to have typed arrays in your signatures, declare the arrays you want to pass as typed as well" 15:25
m: sub f(Int @a) { say [+] @a }; my Int @b = 1, 2, 3; f @b
camelia rakudo-moar a3962b: OUTPUT«6␤»
psch m: sub f(Int @a) { say [+] @a }; my @b = 1, 2, 3; f @b
camelia rakudo-moar a3962b: OUTPUT«Type check failed in binding @a; expected 'Positional[Int]' but got 'Array'␤ in sub f at /tmp/fkBHZ2t0_L:1␤ in block <unit> at /tmp/fkBHZ2t0_L:1␤␤»
Skarsnik but if the core class does not give typed thing. It make the whole type stuff useless 15:26
psch m: sub f(@a where all(@a) ~~ Int) { say [+] @a }; my @b = 1, 2, 3; f @b
camelia rakudo-moar a3962b: OUTPUT«6␤»
psch Skarsnik: i don't agree that it makes "the whole type stuff" useless 15:28
moritz psch: I wrote that before I knew that the PositionalBindFailover trick was possible
Skarsnik it make it a struggle
moritz Skarsnik: only if you overdo it 15:29
TimToady by putting the type on the receiving end and not the sending end, you're just violating postel's law
moritz Skarsnik: it takes some time to find the sweet spot between too much and too little type-constraining
Skarsnik but it's the core fault if ^attributes is not typed, no mine 15:30
moritz Skarsnik: it's not documented to return a typed list, so there is no fault.
psch Skarsnik: i think postel's law is bad advice
oh
err
TimToady then :P 15:31
i guess that should read as "we don't type the list, so you're breaking things if you want it constrained"?
Skarsnik The issue is: Ok perl6 offer type. let's type some stuff to have the compilator be able to do some check for us and avoid writing error. but it does not really work like this at all. or when it's work you have to learn weird subtility that make sens if you know most of the internal (That how It feel for me using type) 15:34
moritz Skarsnik: you don't need to know any internals to know when to expect a typed list; reading the docs is enough 15:36
masak I agree with Skarsnik's sentiment, if not his rationale. 15:37
to the point where I've concluded for the time being that typed arrays are not of interest to me.
TimToady well, we probably need a couple of things we don't have: 1) an easy way to narrow a list, and 2) and easy way to type the values of a list 15:38
masak well, hm. not really that extreme.
I did end up liking typed array attributes.
like `has Item @.items`
masak that's probably thanks to lizmat++, who implemented it so that they actually typecheck stuff. 15:38
gfldex r: my @a = [Mu]; @a.perl.say; for @a -> $i:D {} 15:40
camelia rakudo-jvm a3962b: OUTPUT«===SORRY!===␤Can not invoke this object␤»
..rakudo-moar a3962b: OUTPUT«===SORRY!===␤Cannot invoke this object␤»
gfldex r: my @a = [Mu]; for @a -> $i:D {} 15:41
moritz doesn't even know how -> $i:D { } parses
camelia rakudo-jvm a3962b: OUTPUT«===SORRY!===␤Can not invoke this object␤»
..rakudo-moar a3962b: OUTPUT«===SORRY!===␤Cannot invoke this object␤»
gfldex std: for @a -> $i:D {}
camelia std 28329a7: OUTPUT«5===SORRY!5===␤Variable @a is not predeclared at /tmp/wsIb5I8jdj line 1:␤------> 3for 7⏏5@a -> $i:D {}␤Potential difficulties:␤ $i is declared but not used at /tmp/wsIb5I8jdj line 1:␤------> 3for @a -> 7⏏5$i:D {}␤Check failed␤FAILED…»
gfldex std: my @a; for @a -> $i:D {}
camelia std 28329a7: OUTPUT«Potential difficulties:␤ $i is declared but not used at /tmp/cADPaJUhBO line 1:␤------> 3my @a; for @a -> 7⏏5$i:D {}␤ok 00:00 140m␤»
gfldex std: my @a; for @a -> $i:D { $i.say } 15:42
camelia std 28329a7: OUTPUT«ok 00:00 140m␤»
gfldex even if what I do is silly, I still feel entitled to a proper error message :) 15:43
pink_mist what's the differences between std:, p6:, m:, r:, and whatever others there are?
moritz pink_mist: they run the code through different compilers
pink_mist: std is the standard grammar, which only does syntax checks
psch m: my @a = [Mu]; for @a -> :($i:D) {}
camelia rakudo-moar a3962b: OUTPUT«===SORRY!===␤Cannot invoke this object␤»
gfldex m: - rakudo on moarvm (without jvm), r: rukudo with both moarvm and jvm (two outputs)
camelia rakudo-moar a3962b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/a7DxspFEgl␤Missing block (taken by some undeclared routine?)␤at /tmp/a7DxspFEgl:1␤------> 3- rakudo on moarvm (without jvm7⏏5), r: rukudo with both moarvm and jvm (t␤ expecting any of:␤ …»
moritz pink_mist: m is rakudo with the moarvm backend 15:44
pink_mist: j is rakudo with the jvm backend
moritz pink_mist: p6 runs both j and m 15:44
JimmyZ_ github.com/rakudo/rakudo/blob/nom/...2015.09.md
pink_mist moritz: ah, thanks! =)
psch p6 probably would run other implementations, if some come up again
JimmyZ_ Stefan Seifert is niner
pink_mist so p6 is the same as r?
JimmyZ_ they are the same one😊 15:45
psch idly wonders what happened to the Perl 6 compiler in C++ that was hinted at some time ago...
TimToady p6 used to run niecza too till it monorotted
psch or was is C99, i forget
TimToady nothing surprising :) 15:46
vendethiel psch: arakne? ded
oops, arane. github.com/BizarreCake/Arane
moritz perlito: say 42 15:47
psch yeah, it seemed somewhat ambitious at the time already... :) 15:52
jnthn masak: Your finding that typed arrays work most nicely for you on attributes is a nice reflection of the fact that types are most naturally used in Perl 6 when you are trying to keep your own things internally consistent, rather than trying to export a strongly typed view to a language where people mostly expect to easily plug things together without much regard for such things. 15:53
moritz is always skeptical when announcements are made before working code is there
jnthn: which is contrary to the common opinions that interfaces/APIs are the place where APIs matter most 15:54
jnthn We've seen across the industry that when a language tries to make all interfaces strongly typed, people end up reaching for dynamically typed things to loosen coupling ("let's use web services between these two Java things!")
moritz: APIs are where APIs matter most? :) 15:55
Oh...was one of those meant to be types?
moritz s:last/APIs/types/
TimToady m: sub foo(Rat() *@rats) { say @rats }; foo 1,2,3 15:58
camelia rakudo-moar a3962b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/ldl7JjnzYi␤Slurpy positionals with type constraints are not supported.␤at /tmp/ldl7JjnzYi:1␤------> 3sub foo(Rat() *@rats7⏏5) { say @rats }; foo 1,2,3␤ expecting any of:␤ constraint␤»
jnthn suspects that one should work eventually
Skarsnik hm, how I can test if soemthing is an Array? (or more exactly something that act like an Array) since a Array attribute is a Positional 15:58
TimToady why do you want to know? 15:59
moritz Skarsnik: fwiw the @ sigil type-checks against Positional
TimToady introspection is usually a code smell 15:59
moritz (and type checking is usually done with ~~, so "if $thing ~~ Positional { ... }") 16:00
nine m: class Bar::Foo { has $.foo-hoo; has $.bar; }; sub foo() { my @arr = 1, 2, 3; return Bar::Foo.new(bar => "2", foo-hoo => @arr); }; say foo().foo-hoo; 16:04
camelia rakudo-moar a3962b: OUTPUT«[1 2 3]␤»
nine Does anyone have an idea why this works flawlessly, while a simple return Panda::Ecosystem.new(statefile => "$pandadir/state", projectsfile => "$pandadir/projects.json", extra-statefiles => @extra-statefiles); fails with "Default constructor for 'Panda::Ecosystem' only takes named arguments"? 16:05
Skarsnik I can't find a concret example, but if I write soemthing stupid like print_stuff($a) {say 'fancyint' if $a ~~ Int; say 'fancyarray' if $a ~~ Array;} and pass it an array attribute from a class, does it fell back to Array?
dalek ast: f55aa8d | jnthn++ | S17-supply/syntax.t:
Tests for RT #126089.
psch m: sub f($a) { say $a.perl }; my $b = "foo"; my @c = ^5; f $b; f @c 16:08
camelia rakudo-moar a3962b: OUTPUT«"foo"␤[0, 1, 2, 3, 4]␤»
psch m: sub f($a) { say $a ~~ Positional }; my $b = "foo"; my @c = ^5; f $b; f @c
camelia rakudo-moar a3962b: OUTPUT«False␤True␤»
nine Oh, setting MVM_SPESH_DISABLE=1 or MVM_JIT_DISABLE=1 fixes the panda problem! 16:09
moritz oh, a JIT bug 16:10
vendethiel psch: well, if not ambitious, at least bound to be caught cheating "early on" 16:11
nine moritz: do you think it's an actual JIT bug or rather just something that has to be changed in the JIT, too after changing Perl6::Actions? 16:13
jnthn Anyone know why method Capture() {...} in List doing something other than just making a Capture whose positional bits are the List? 16:15
Well, the list elements
It's trying to do something with named args which I'm sure isn't how it's meant to work, and is also causing a bug.
jnthn m: (0 => 1, 2 => 3).Capture 16:16
camelia rakudo-moar a3962b: OUTPUT«This type cannot unbox to a native string␤ in block <unit> at /tmp/_ZcSpt6bLK:1␤␤»
jnthn That of note
m: (a => 1, b => 3).Capture.perl.say
camelia rakudo-moar a3962b: OUTPUT«\(:a(1), :b(3))␤»
jnthn m: (a => 1, b => 3).Capture.list.say
camelia rakudo-moar a3962b: OUTPUT«()␤»
nine jnthn: List.Capture is a weird mix of old List's .Capture and Parcel's 16:17
jnthn nine: Ah, OK
That explains something, thanks.
I think it wants to be a good bit simpler. I'm not sure what it'll break :)
nine jnthn: the named args handling is for "Fix binding a List to a Signature not recognizing Pairs as named args"
jnthn I don't know that it should... 16:18
What wanted it to, ooc?
nine Fixes :(:t($t), :m($m)) := (t => 1, m => 'a');
jnthn I *thought* you'd have to write that one \(t => 1, m => 'a') 16:19
nine In the next life I'm gonna add info about the fixed spec tests to every commit message but right now I'm glad that I spent as much time writing them as I did :)
jnthn That is, with a Capture literal
jnthn m: my Any $a is default(3) 16:20
camelia rakudo-moar a3962b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/9OPR5WiAl0␤Default value '3' will never bind to a parameter of type Any␤at /tmp/9OPR5WiAl0:1␤------> 3my Any $a is default(3)7⏏5<EOL>␤ expecting any of:␤ constraint␤»
jnthn has a fix for that one
moritz also note that it says "parmater" in the error message
TimToady m: gist.github.com/TimToady/003e9c5c953de6014864 16:21
camelia rakudo-moar a3962b: OUTPUT«1..8␤1..4␤1..8␤»
jnthn moritz: Yeah, I noticed that too...
TimToady so...why does the middle one produce 1..4, and the last one 1..8?
jnthn TimToady: Probably something overlooked when implementing + 16:22
TimToady: Try with --optimize=off
nine m: sub foo() { return :a(1), :b(2); }; my ($a, $b); :(:b($b), :a($a)) := foo(); say $a, $b;
camelia rakudo-moar a3962b: OUTPUT«12␤»
TimToady simply adding a statement shouldn't change the value of args 16:23
yup, it's the optimizer
jnthn TimToady: Well, the big tip-off is we only consider single statement things in the static inliner 16:24
TimToady: I'll bet the compile-time bind analyzer was never taught about +
TimToady: So it thinks it's looking at a totally ordinary positional arg
TimToady I did some stuff in there, but maybe not enuff
jnthn Apparently... :) 16:25
jnthn m: use Test; lives-ok { EVAL 'my Any $a is default(3)' }; 16:26
camelia rakudo-moar a3962b: OUTPUT«not ok 1 - ␤␤# Failed test at /tmp/iZbk4tcPUk line 1␤# Default value '3' will never bind to a parameter of type Any␤»
vendethiel jnthn: should the static inliner be taught that unused values can be removed? 16:27
m: 1; # like this warning already knows
camelia rakudo-moar a3962b: OUTPUT«WARNINGS:␤Useless use of constant integer 1 in sink context (line 1)␤»
moritz that's not a variable :-)
vendethiel variable? 16:27
jnthn vendethiel: I thought it already was 16:28
moritz vendethiel: oh, I misread
vendethiel ;)
psch gist.github.com/peschwa/30639506d16f937f393d # help o.o
vendethiel jnthn: maybe there's a discount today, and you uncovered two bugs instead of one
dalek kudo/nom: a5ef61e | jnthn++ | src/core/Variable.pm:
Fix inverted type test in `is default(...)`.
16:30
ast: fbd627c | jnthn++ | S02-names/is_default.t:
Test for RT #126104.
jnthn 16 to go before we knock a page off RT :) 16:32
moritz the R* release I'm preparing... is this our Christmas Preview? 16:33
jnthn Note that RT #126108 is about the bad error, so we can address it wiht that.
zefram++ files well-formed/analyzed bug reports 16:35
bartolin yeah, zefram++ ( but there are so many of them /o\ ) 16:36
mprelude Anyone using AnyEvent::Twitter::Stream? If so, what Twitter lib do you use for... tweeting, rting, etc? Net::Twitter? 16:38
Wrong chan :D
jnthn m: sub MAIN() { map { say 'hi' }, ^2 } 16:43
camelia ( no output )
moritz jnthn: I'm looking into that one right now 16:44
dalek kudo/nom: f7c2a72 | moritz++ | src/core/Main.pm:
RT #126029: sink MAIN (and USAGE)
16:45
jnthn I have a patch locally
oh, damm 16:46
moritz sorry :(
jnthn Well, I have a test? :) 16:46
moritz jnthn: push that, then :-)
dalek ast: a1d59e2 | jnthn++ | S06-other/main.t:
Test for RT #126029.
16:48
jnthn Went for a test that didn't end up needing to sleep :) 16:48
Your patch looks like mine except I didn't think of USAGE
moritz jnthn: ok, so it wasn't totally sunk (sic) effort :-) 16:49
jnthn ;P
dalek ast: 1db8547 | usev6++ | S03-feeds/basic.t:
Remove dubious test

S06 currently says:
   Because feeds are defined as lazy pipes, a chain of
   feeds may not begin and end with the same array without
   some kind of eager sequence point. That is, this isn't
   guaranteed to work:
   @data <== grep { $_ % 2 } <== @data;
   [...]
   Conjecture: if the cloning process eagerly duplicates @data,
   it could be forced to work. Not clear if this is desirable,
   since ordinary clones just clone the container, not the value.
  'Is not guaranteed to work' sounds rather weak and does not
justify to test for the code given to die.
Also, the code currently lives and the test only succeeded because @data wasn't visible within 'eval-dies-ok'.
16:54
vendethiel rip dalek?
bartolin oops. but it looks like he survived. dalek++ 16:55
vendethiel dalek++ :D
vendethiel is always happy to see MAIN fixes
jnthn m: while 0 { my $_ } 16:57
camelia rakudo-moar a5ef61: OUTPUT«===SORRY!===␤Cannot find method 'value'␤»
jnthn has figured that one out
moritz what should an empty block return? ()? 16:59
jnthn moritz: I'd assumed Empty (which is the empty Slip) 17:00
vendethiel m: my \a = ().Slip; say (1, \a, 2); 17:01
camelia rakudo-moar a5ef61: OUTPUT«(1 \(()) 2)␤»
dalek kudo/nom: a328aab | jnthn++ | src/Perl6/Actions.nqp:
Fix compiler explosion in `while 0 { my $_ }`.
17:02
ast: 406c27b | jnthn++ | S04-statements/while.t:
Test for RT #125876.
17:03
dalek ecs: 3f0277c | moritz++ | S04-control.pod:
S04: Fix pre-GLR-ism

isn't it wonderful how much simpler this passage becomes through the GLR?
17:05
xinming_ Anyone here knows what is the best template engine for perl6 for now? 17:06
Or, if there is a template engine that is is mostly close to perl6 language itself.
That'll be good.
jnthn Well, there's a few Saturday patches. Should probably get back to resting myself from the busy last few days... :)
jnthn also needs to blog soon... :) 17:07
o/
dalek ast: 05eed96 | usev6++ | S02-types/subset.t:
Replace test using 'eval-lives-ok' with more specific test

Also avoid conflict with other subsets used in test file.
17:23
moritz m: say (EVAL ().Slip.perl).^name 17:24
camelia rakudo-moar a328aa: OUTPUT«List␤»
moritz Slip.perl is RONG
if I change empty blocks to return Empty instead of Nil, surprisingly few spectests fail 17:25
dalek ast: c48ef39 | usev6++ | S04-phasers/end.t:
Unfudge/rewrite some tests for END phaser

  * use 'lives-ok' with EVAL instead of 'eval-lives-ok'
  * one 'todo' test only failed because $a was not visible within 'eval-lives-ok'
  * the second 'todo' test needed a rewrite with 'is_run'
18:14
lizmat nine: thanks for a3962b05480453c36fc4897a , specifically the thinko in line 413 18:46
DrForr waves to lizmat.
lizmat however, I don't think the fix in line 416 is correct
lizmat waves back to DrForr!
DrForr Must be getting better, back on IRC :) 18:47
lizmat nine: as we've done all lines that we wanted, we indicated the handle should be closed, so even if there are more lines in the file, we don't want them (ever) I would think
:close to lines means: close when you're done
DrForr: good to see you online again, you *must* be getting better! :-) 18:48
or you have some contraption that allows you to type / see your screen while laying down :-)
either way, an improvement over before :-)
DrForr Both :) 18:49
moritz imagines a transparent mattress with a screen under it 18:50
moritz m: say slip().perl 18:54
camelia rakudo-moar a328aa: OUTPUT«()␤»
moritz locally I've added multi method perl(Slip:D: |) { callsame() ~ '.Slip' } 18:55
and now I get ().Slip.Slip as the answer :/
that's a bit too slip(er)y for me 18:57
same when I change it to self.List::perl(|c) ~ '.Slip'; 18:59
dalek kudo/nom: 032dd7f | TimToady++ | src/Perl6/Actions.nqp:
suppress bogus inlining of +args
TimToady jnthn: took all morning, but I found the bug 19:00
it wasn't something I neglected to do, but something I just did rwong
ShimmerFairy TimToady: I can't place my finger on it, but something about that return statement seems almost... repetitive. :P 19:02
TimToady yeah, well, it's nqp 19:05
grondilu nqp: say([||] 1, 0); 19:12
pink_mist that seems to have worked less than well 19:13
grondilu no reduce metaoperator in nqp anyway. Nvm
camelia nqp-moarvm: OUTPUT«Confused at line 2, near "say([||] 1"␤ at gen/moar/stage2/NQPHLL.nqp:519 (/home/camelia/rakudo-inst-2/share/nqp/lib/NQPHLL.moarvm:panic:105)␤ from gen/moar/stage2/NQP.nqp:921 (/home/camelia/rakudo-inst-2/share/nqp/lib/nqp.moarvm:comp_unit:872)␤ from gen/…» 19:15
..nqp-parrot: OUTPUT«Can't exec "./rakudo-inst/bin/nqp-p": No such file or directory at lib/EvalbotExecuter.pm line 193.␤exec (./rakudo-inst/bin/nqp-p /tmp/tmpfile) failed: No such file or directory␤Server error occurred! Closing Link: ns1.niner.name (Quit: camelia)␤Lost connect…»
..nqp-jvm: OUTPUT«#␤# There is insufficient memory for the Java Runtime Environment to continue.␤# pthread_getattr_np␤# An error report file with more information is saved as:␤# /tmp/jvm-13006/hs_error.log␤»
pink_mist so jvm ran out of memory from that? 0_o 19:16
ShimmerFairy nah, it's because nqp-p is still something the bot tries to do that it left for a moment. 19:20
psch nqp-j: say("hi") 19:27
camelia nqp-jvm: OUTPUT«(signal ABRT)#␤# There is insufficient memory for the Java Runtime Environment to continue.␤# pthread_getattr_np␤# An error report file with more information is saved as:␤# /tmp/jvm-17024/hs_error.log␤»
psch nqp-j is always out of memory... :)
ShimmerFairy ah. nqp-p didn't help though :P 19:29
lizmat hmmm... S04-phasers/pre-post.t fails for me with "Lexical '$_' already declared" at compile time 19:32
TimToady always fakes me out, I think my test run is done because my fan spools down, but it's just all the S17 stuff twiddling its thumbs 19:36
ShimmerFairy I've noticed some tests with memory allocation problems on my machine, I'll go report them against MVM now that I'm reminded. 19:39
lizmat perhaps not running the spectest alphabetically could work ?
TimToady: ^^^
lizmat is there a reason why Str doesn't have a 'spurt' method ? 19:43
"foo".spurt($filename) # for easy writing a string to a file ?
pink_mist good idea 19:44
ShimmerFairy lizmat: give IO back the slurp method, and I think there's more argument for it :P
timotimo well, it could go at the end of a long chain of methods
those are kinda common
lizmat "t/spectest.data".IO.lines(:!chomp).pick(*).join.spurt("wop") # my use case
jdv79 isn't spurt more related to the IO thing than the content
yoleaux 03:32Z <zostay> jdv79: IO::Socket::INET is too specific, but using the Raw IO is completely the wrong solution always and the alternative is to be implementation-specific, which might as well not even be in the standard
dalek kudo/nom: 039de8f | TimToady++ | src/core/Any-iterable-methods.pm:
minmax now one-arg
19:45
kudo/nom: 83d7248 | TimToady++ | src/core/metaops.pm:
reductions now one-arg
dalek ast: 8cd1edb | TimToady++ | / (2 files):
reductions are now under one-arg rule
19:45
jdv79 zostay: not sure i agree but ok
ShimmerFairy I never _really_ understood the rename of slurp to slurp-rest for the method form, and it still feels like a bad thing to have happened.
dalek kudo/nom: eeed6bb | lizmat++ | src/core/IO/Handle.pm:
lines(:close) means close the handle, always
19:46
jdv79 uh, what is slurp-rest?
TimToady read-to-end 19:47
but don't close
slurp implies closing 19:48
jdv79 so slurp still exists?
i haven't updated in a while
ShimmerFairy TimToady: while I do understand slurp-rest being "from current point to end" (and apparently the not-closing thing), it felt like an inconsistency was made when .slurp was renamed, and &slurp was left alone 19:49
(even though they do behave different, AIUI)
TimToady it's a foolish inconsistency
er, however you say that...
very bikesheddy, as far as I am concerned 19:50
jdv79 "rest" doesn't jump out at me as not closing. and as for from cur pos to end - that's the sort of thing that's implied with a lot of IO operations, no?
seems wonky but ok
why not paramaterize that sort of stuff more clearly
ShimmerFairy TimToady: yeah, admittedly it's not that bad, I just still see it from when the change was made, and it felt odd to not be able to do $string.slurp like you could slurp($string) anymore
TimToady if you want to be consistent, rename .slurp to .slurp-all-and-close, but that's a bit long
lizmat well, since I've given up anything IO related design wise, I have not much interest in participating in further bikeshedding this 19:51
so whatever you guys decide, is fine by me
TimToady has bigger fish to fry 19:52
timotimo bon appetit
TimToady like whether listop map should be onearg instead of flattening
jdv79 lizmat: so newio is dead in the water? 19:52
ShimmerFairy lizmat: you may be able to do .&spurt at the end of that method call chain (depends on how the args are ordered) 19:53
lizmat jdv79: am afraid so: at some point I may revisit the branch and see if I can salvage some from it
timotimo m: say &spurt.^signature
camelia rakudo-moar 032dd7: OUTPUT«Method 'signature' not found for invocant of class 'Perl6::Metamodel::ClassHOW'␤ in block <unit> at /tmp/kXnsaZpflO:1␤␤»
timotimo m: say &spurt.signature 19:54
camelia rakudo-moar 032dd7: OUTPUT«(|)␤»
lizmat but I've gotten so much flak from it, that I got completely de-funned there
timotimo m: say &spurt.candidates
camelia rakudo-moar 032dd7: OUTPUT«(sub spurt (|) { #`(Sub|57331736) ... })␤»
timotimo hum.
ShimmerFairy lizmat: does the newio branch of specs give a good overview of what your goals were? I wasn't around when you were working on it, so I don't know anything about it.
jdv79 can we not muster a new torch bearer to bring it home?!
lizmat newio turned out to be -NoFun to me
ShimmerFairy timotimo: if 'sub spurt' has a first argument of the Str to spurt, then you'd be able to use $string.&spurt(...)
timotimo aye 19:55
lizmat &spurt has the filename as the first param
jdv79 lizmat: is it worth completing at this point?
timotimo i wanted to find out
obviously we'll need a &trups for this
lizmat ShimmerFairy: there's also a newio branch in specs 19:56
in which I described what I wanted to achieve / largely achieved
ShimmerFairy lizmat: that's the branch I asked about :) . I'll be sure to check that out when I have the chance.
dalek ast: 8e7418f | TimToady++ | S03-metaops/reduce.t:
untodo [=:=] et al. that now work after onearg
20:01
TimToady multi sub map(&code, Whatever) { (1..Inf).map(&code) } <-- who ordered that? 20:04
and why 1-based?
ShimmerFairy TimToady: I think we decided that @list.map(*) was useless not too long ago, the sub form should probably go too :) 20:05
oh, that's not quite the same thing. But still.
m: say *.map({say $_ }) 20:06
camelia rakudo-moar 032dd7: OUTPUT«WhateverCode.new␤»
ShimmerFairy ^ should do that, as far as I'm concerned
TimToady m: say map * + 0, * 20:07
camelia rakudo-moar 032dd7: OUTPUT«(...)␤»
TimToady m: .say for map * + 0, *
camelia rakudo-moar 032dd7: OUTPUT«(timeout)1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤11␤12␤13␤14␤15␤16␤17␤18␤19␤20␤21␤22␤23␤24␤25␤26␤27␤28␤29␤30␤31␤32␤33␤34␤35␤36␤37␤38␤39␤40␤41␤42␤43␤44␤45␤46␤47␤48␤49␤50␤51␤5…»
TimToady that's just kinda bizarre 20:08
ShimmerFairy I agree. I don't see why map should have a shortcut for what's basically an infinite loop.
TimToady seems like it would be much more straightforward to say 1..Inf when you mean that 20:10
TimToady doubtless it will come back around to being my fault somehow :) 20:12
lizmat TimToady: if anybody's to blame for that one, that would be me
lizmat and as to why 1-based, I would not know the answer to that :-( 20:12
I think it dates from a time where a gather loop {} did not work in the settings (is what I vaguely remember) 20:13
afaics, it can be killed with fire...
jdv79 where can i find out more about this quote "as usual, the presence of a named option is a design smell" 20:14
lizmat
.oO( who switched on the zefram bot? :-)
timotimo downloaded more zefRAM 20:17
jdv79 i don't understand that quote 20:19
lizmat jdv79: fwiw, I would like to see an explanation of that quote as well :-)
labster moritz: I merged the File::Directory::Tree change. 20:22
yoleaux 11:14Z <moritz> labster: any chance you could merge github.com/labster/p6-file-directo...ree/pull/5 soonish? It would help with the Rakudo Star release
jdv79 esp since slurp-rest has adverbs or named options or whatever you want to call them for other stuff. wut. 20:26
i am starting to imagine why newio was unfun. i glossed over it while it was going on so didn't notice. 20:27
ShimmerFairy jdv79: where did you get that quote from? 20:32
dalek kudo/nom: a98ce40 | TimToady++ | src/core/Mu.pm:
Add .self as an identity method

So people aren't so tempted to hardwire * as an identity map function.
20:34
kudo/nom: d9c21e9 | TimToady++ | src/core/Any-iterable-methods.pm:
simplify listop map dispatch to a single one-arg
kudo/nom: 244b7b4 | TimToady++ | src/core/Str.pm:
listop map no longer flattens
btyler_ hi folks. in the category of "evil but useful tricks", can a perl6 role add methods to the consuming class/object in a dynamic way? here's an example in perl 5: gist.github.com/kanatohodets/aff43...igarole-pm 20:35
that is, run code to generate a bunch of methods
jdv79 i dug into the slurp-rest hits from irc log 20:37
masak jdv79: trying to understand the quote, I think it's about over-configuring, and I also think I have an example. 20:38
jdv79 ShimmerFairy: irclog.perlgeek.de/perl6/2014-11-07#i_9622385
ShimmerFairy jdv79: in my experience, it's only a big problem if it ends up a poor reimplementation of the type system; that's why we ended up with NFC, NFD, etc. types for normalization, instead of :nfc adverbs on everything :) 20:39
ShimmerFairy (though that's surely not the only time it's an issue to use nameds, just the only time I've personally encountered it being problematic) 20:40
TimToady we had a long history of trying to customize operators like range using adverbs, and it almost always meant there was some other better way to do it 20:41
masak jdv79: .roll used to be .pick(:replace)
jdv79: but having it as a separate method is clearly better
jdv79: as an example of operators with adverbs, there's :by on infix:<..> -- but it turned out to be vastly better to have that kind of stepping magic as a separate operator infix:<...> 20:42
TimToady in linguistic terms, having options often means you're trying to stretch a metaphor too far
jdv79 tahnks 20:54
*thanks ^H
jdv79 I'll just restate a bit and then drop it. As someone who just learned of slurp-rest i was immediately confused by the name and needed docs. if it was just slurp with a close param with sensible default(s) I probably would not need to look it up. but i'm guessing i'm missing some things that would make that rename make more sense. 21:01
masak jdv79: all I remember is that at some point we got more discriminating, and started considering a full-file slurp to be different from a slurp on a filehandle that had already been partially consumed. 21:06
jdv79: implementing the latter as a flag/named may or may not have been considered, I don't know.
lizmat vaguely remembers it being considered, but that may be confabulation 21:09
masak the one I *would* sign up on as being a definitely design/code smell, is when the API ends up looking like :!someflag 21:14
that is, if the default of the named is True, and you have to pass in False to override the default 21:15
lizmat lines(:!chomp) anyone ? 21:15
dalek ast/return_outside_routine: 38fdf08 | usev6++ | S04-statements/return.t:
Use 'run' to test for returning outside of routine

We can't test for 'Attempt to return outside of any Routine' with our toolchain from Test.pm or Test::Util.
21:17
lizmat masak: would you consider lines(:!chomp) a code smell ?
masak lizmat: yes.
TimToady lines(:pristine)
masak keep in mind, though, that "code smell" does not mean "destroy! destroy!" 21:18
it means we should keep on the lookout for a nicer factoring
lizmat I guess it came from hysterical raisins
lizmat I mean, in p5 you would almost *always* do a .chomp on whatever you got 21:18
masak yes, that's what I thought when you just showed it, that the term 'chomp' is historical
lizmat lines(:no-chomp) ? 21:19
masak it just moves the bump around the waterbed :)
lizmat I mean, that would just change the spelling :-)
masak yep
jdv79 no, its a much larger bump cause now you have the possibilitiy of a double negative which can be highly confusing 21:20
masak TimToady++ is on the right track with :pristine, but 'pristine' doesn't immediately ring true/simple/fitting 21:21
jnthn btyler_: Didn't see anyone answer your roles question, so: yes, any code you write in the role body will run at role composition time, and you can use the generic ::?CLASS to at stuff
*add
btyler_ oh, yay. so ::?CLASS.^add_method presumably
leont The nice thing about chomp is that it will actually well "does it do this action"
btyler_ I'll give that a shot, thanks!
leont :pristine is its own kind of negative, by meaning "don't do some action")
masak m: role B { say "run at role composition time: {::?CLASS}" }; class C does B {}; class D does B {}
camelia rakudo-moar 244b7b: OUTPUT«===SORRY!===␤Cannot find method 'collisions'␤»
masak ...hm. 21:22
btyler_ yeah, I was getting that error and assuming it was just me doing something silly
jnthn m: role R { ::?CLASS.^add_method(<foo bar baz>.pick, method () { }) }; class C1 does R { }; class C2 does R { }; say C1.^methods; say C2.^methods;
camelia rakudo-moar 244b7b: OUTPUT«(<anon>)␤(<anon>)␤»
jnthn hah :)
masak jnthn: are we doing something silly?
jnthn m: role R { ::?CLASS.^add_method(<foo bar baz>.pick, method () { }) }; class C1 does R { }; class C2 does R { }; say C1.^method_table; say C2.^method_table; 21:22
camelia rakudo-moar 244b7b: OUTPUT«foo => <anon>␤baz => <anon>␤»
jnthn m: role R { ::?CLASS.^add_method(<foo bar baz>.pick, method () { }) }; class C1 does R { }; class C2 does R { }; say C1.^method_table; say C2.^method_table;
camelia rakudo-moar 244b7b: OUTPUT«foo => <anon>␤foo => <anon>␤»
jnthn There you go :)
masak jnthn: what about the 'collisions' thing above?
jnthn masak: Expecting .gist to work before the target class is composed, then running onto the bug already in RT where somehow instead of reporting an error correctly we complain about collisions. 21:23
masak aha.
TimToady wonders, looking at sort, why signature (&cmp, +values) is not considered a better match than (+values)
jnthn m: multi m($a, *@b) { say 1 }; multi m(*@b) { say 2 }; m(1); m(1, 2); 21:24
camelia rakudo-moar 244b7b: OUTPUT«Ambiguous call to 'm'; these signatures all match:␤:($a, *@b)␤:(*@b)␤ in block <unit> at /tmp/lvM0TEeofx:1␤␤»
jnthn TimToady: You want the first to win in that case?
TimToady if it doesn't break something else or slow things down, seems like a good idea 21:25
jnthn I can't tell you if it'll break anything because I already surpassed the point where I can understand the arity-based sorting rules in multi-dispatch. :/
TimToady in some sense the first parameter is binding tighter by definition
jnthn Probably 'cus I made a bunch of reasonable-sounding reqeusts work and now I don't understand the result. :) 21:26
TimToady me too :P
jnthn We probably need to simplify things a bit :) 21:27
My gut feeling is "yeah, that should work", but given my following my gut already got us the current impl... :) 21:28
dalek kudo/nom: c1af489 | lizmat++ | src/core/Iterator.pm:
count-only doesn't take any params
21:28
kudo/nom: 8158cb5 | lizmat++ | src/core/IO/Handle.pm:
Re-implement lines(:count) using iter.count-only
kudo/nom: c5407d2 | lizmat++ | src/core/Seq.pm:
Make Seq.elems call iterator.count-only

Provide a shortcut when we want an iterator to just find out the number of elems and nothing else (think "words".IO.lines.elems)
kudo/nom: ffa8ea2 | lizmat++ | src/core/array_slice.pm:
Prevent [*-1] from consuming the iterator too soon
kudo/nom: 8c3980f | TimToady++ | src/core/Any-iterable-methods.pm:
attempt to de-flatten sort
21:30
lizmat jnthn: you may want to look at c5407d2 and ffa8ea2
jnthn ffa8ea2 concerns me a little, yeah
Like, I don't get why it's needed 21:31
lizmat jnthn: it used to do the .cache inside Seq.elems before
jnthn ah
oh, it's AT-POS we forward, we don't have an alternative set of postcircumfix candidates
lizmat otherwise the .elems will consume the iterator before we can get at the underlying seq
jnthn Yeah
OK, now I understand why it's needed 21:32
lizmat but it should allow for all other iterators to provide a fast shortcut for just counting
it makes "words".IO.lines.elems about 2.5 times as fast
TimToady only about 167 more *@ and **@ slurpies to audit in the core, sigh... 21:33
lizmat jnthn: and no spectest breakage
TimToady plus the places with hand-rolled one-arg that could use +args instead 21:35
BenGoldberg TimToady, on that sort patch, why not let the dispatcher do the work? Something like: multi sub sort(Callable \cmp, +values) { values.sort(cmp) }; multi sub sort(+values) { values.sort }; 21:36
jnthn lizmat: I'm fine with it for now, I think :) 21:37
mattp__ is there any 6guts / p6weekly coming any time soon? its been a while :)
jnthn mattp__: 6guts updates will resume shortly; I wasn't too well for a while :( 21:38
(and spent what energies I had on getting stuff done rather than blogging) 21:39
TimToady BenGoldberg: doesn't work; that's what jnthn++ and I were just discussing
BenGoldberg reads the irclog... 21:43
What about some sort of: multi sub sort(AnyThingOtherThan(Callable) \not_a_cmp, +values) { ... }?
mattp__ jnthn: given the past tense glad youre feeling better atleast :) 21:44
jnthn mattp__: Well, better in the "improved" sense than the "totally well" sense, but it's progress. :)
jnthn wants to get back to writing weekly reports, anyways. 21:45
lizmat mattp__: writing a P6W has become a daunting task 21:48
timotimo seems to have worked on it: not sure of the status right now
timotimo oh 21:53
that's right
timotimo last status: 21:53
wordpress was like "blogpost, OM NOM NOM GONE!" 21:54
timo was like: *flips table*
BenGoldberg Speaking of things I'm seeing in the irc log before I joined today... IIRC, p5 started with a chop operator, which removed and returned the last byte of a string, regardless of what that byte was. The idea of chopping off the end of string kinda makes english-language sense. chomp was added shortly after, to remove only the current input record separator; unlike chop, it was idempotent; but since it's task was so similar, it was
given a similar-but-cuter name.
lizmat FWIW, I think chomp logic should really live in the VM 21:55
as an attribute of the PIO
and it should be able to handle other line-endings as well :-)
BenGoldberg To super-super-summarize, I think maybe p6's version of chomp should be $:retain-record-separator or $:remove-record-separator 21:56
lizmat but I'm pretty sure jnthn agrees and has it somewhere down on his list :-)
jdv79 timotimo: you trust a browser as a text editor? 21:57
jnthn suspects once he's done the multi-dim array stuff, I/O related bits will be on his plate
lizmat wishes jnthn much strength
lizmat feels we need a bit of a GIOR
jnthn IO is one of those things I've continually hoped someone more experienced than me would take on... 21:59
timotimo jdv79: it has a recovery function
jnthn Concurrency and NFG both used to be in that category, though :P
pink_mist jnthn: maybe those more experienced than you are experienced enough that they hope the same thing :P 22:00
jnthn pink_mist: Probably, yeah. It's a thorny area. :)
Thankfully, it's 2015, and the list of platforms you have to care about supporting to achieve >99% market penetration is really short :) 22:01
uh, market penetration ain't the phrase I wanted...but anwyays :)
lizmat well, if someone would actually look at what I wrote in the newio branch of specs 22:04
and we could reach some consensus about that, then maybe I could be convinced to take it up again
TimToady gee, when I take flatttening off the any/or/one/none, the flattening tests flunk, who'd'a thunk?
lizmat gets a thunking feeling 22:05
jnthn lizmat: I think there's a bunch of lower-level things too
lizmat well, yes, like input-line-sep and output-line-sep
jnthn lizmat: But yeah, looking at what you did in newio would certainly be part of what I want to do.
lizmat most important thing was really getting rid of IO::Spec, I think 22:06
and $*SPEC
using / as a dir separator basically works everywhere
I don't see Perl 6 running on VMS / EBCDIC any time soon 22:07
as long as open can take a "raw" argument for opening files, I think any exotic file systems can export their own logic where needed
geekosaur <USR.LOCAL.BIN>PERL6.EXE :p (tops20 4 lyfe) 22:08
TimToady so, how much code actually depends on junction listops flattening, ka na... 22:09
geekosaur well, probably more like SYS:<USR.LOCAL.BIN>PERL6.EXE
leont Windows sometimes cares about the directory separator, but not that often 22:11
timotimo did we get somewhere with regards to the action class/instance not residing in a dynamic variable? 22:12
masak 'night, #perl6
timotimo gnite masak
timotimo or did we get an approximate measurement of how that'd improve things? 22:14
TimToady I don't think anyone has worked on that 22:22
we'd probably get a bigger performance win from a better dynvar cache in general; getting rid of $*ACTIONS is more of a correctness thing to my mind 22:23
jnthn It's one of those things on my "stuff to do it I want something fun to do" :)
TimToady for my part, I'm concentrating on language definition right now, not performance
timotimo oh, when we call into "foreign" grammars, the ACTIONS may wrongly be "derived"? 22:24
TimToady the current language should be completely encapsulated in the cursor, because then it's portable outside the current dynamic scope
timotimo mhm 22:25
jdv79 is Pod::To::HTML failing tests on install for anyone else?
TimToady currently I'm trying to bias our listops more toward one-arg instead of flattening, except when a good argument can be made for flattening by default 22:26
right now, it's kinda 50/50 whether something flattens or not; I'd rather have the exception list be all on one side 22:27
lizmat takes a long nap 22:28
TimToady so at the moment, junctional listops are on the block 22:31
_itz 22:33
timotimo 22:34
_itz Always Winter, Never Christmas </cslewis> :P
BenGoldberg Unless you're in the southern hemisphere ;) 22:35
Mans Tress 22:37
Mans Title hack 22:38
Titulo hack01
jnthn On the "defining what will 6.christmas be" effort, I decided to make a list of things in the design docs that I'm pretty sure we're *not* going to have in 6.christmas. gist.github.com/jnthn/040f4502899d39b2cbb4 22:41
jnthn (Partly so if TimToady++ feels we really must explore one or more of these sooner rather than later, I know to have it in mind.) 22:43
jnthn Time for some sleep o/ 22:45
TimToady o/ 22:47
dalek kudo/nom: f8ee5f6 | TimToady++ | src/core/Junction.pm:
onearg any/all/one/none listops
22:49
ast: 445fb73 | TimToady++ | S03-junctions/ (2 files):
onearg any/all/one/none
_itz sue-- # activating zefram bug finding bit 22:55
^ bot
geekosaur makes note not to start using p6 actively... 22:57
<-- human fuzzer
vendethiel m: say so any (1, 2) eqv any($[1, 2]) 23:10
camelia rakudo-moar 8c3980: OUTPUT«False␤»
vendethiel TimToady: shouldn't there be such a test, with only one itemized arg? 23:11
mst geekosaur: pretty sure I'll end up doing the same thing ;) 23:26
TimToady m: say so any (1, 2) eqv any(1,2))
camelia rakudo-moar f8ee5f: OUTPUT«5===SORRY!5=== Error while compiling /tmp/if1wmkRaWf␤Unexpected closing bracket␤at /tmp/if1wmkRaWf:1␤------> 3say so any (1, 2) eqv any(1,2)7⏏5)␤»
TimToady m: say so any (1, 2) eqv any(1,2)
camelia rakudo-moar f8ee5f: OUTPUT«False␤»
TimToady m: say so any (1, 2) eqv 2
camelia rakudo-moar f8ee5f: OUTPUT«False␤»
TimToady m: say any (1,2) 23:27
camelia rakudo-moar f8ee5f: OUTPUT«any(1, 2)␤»
TimToady m: say so any (1, 2) == 2 23:28
camelia rakudo-moar f8ee5f: OUTPUT«True␤»
TimToady m: say so any (1, 2) === 2
camelia rakudo-moar f8ee5f: OUTPUT«False␤»
TimToady gah, looks like someone borked === and eqv on junctions :(
TimToady m: say so any(1, 2) === 2 23:33
camelia rakudo-moar f8ee5f: OUTPUT«True␤»
TimToady or...not...hmm...
dalek kudo-star-daily: 0b00557 | coke++ | log/ (2 files):
today (automated commit)
23:44
kudo-star-daily: 87e5547 | coke++ | log/ (10 files):
today (automated commit)
rl6-roast-data: 2ef68c2 | coke++ | / (9 files):
today (automated commit)
[Coke] waves from the highway and starts backlogging. 23:46
I agree with Ovid; We already make allowances for sloppy types with IntStr. Why not a RatInt. 23:49
(Sorry, with CurtisPoe++)
geekosaur yes, that's part of my cognitive dissonance with that issue 23:50
labster m: my role foo { }; say foo.new() 23:59
camelia rakudo-moar f8ee5f: OUTPUT«foo.new␤»
labster roles can have instances?