»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.
ugexe the promise is also marked as kept 00:04
m: say Proc.new(:exitcode(42)) 00:06
camelia rakudo-moar 91212c: OUTPUT«Proc.new(in => IO::Pipe, out => IO::Pipe, err => IO::Pipe, exitcode => -1, pid => Any, signal => Any)␤»
ugexe FROGGS: github.com/rakudo/rakudo/blob/nom/...roc.pm#L14 BUILD needs :$!exitcode 00:13
for this line in Proc::Async: $!exit_promise.keep(Proc.new(:exitcode(status))) 00:14
japhb ugexe: Send a PR? The collection of people with commit rights in Rakudo is much larger (and spans many more timezones). 00:30
ugexe im testing it right now
japhb Excellent! 00:34
timotimo atweiden, raiph: you want "my Silo @silos" rather than "my Array[Silo] @silos", otherwise you're declaring an AoA 00:41
00:42 ShimmerFairy left 00:48 Ben_Goldberg joined
atweiden timotimo: intended to be multidimensional array 00:52
00:56 ][Sno][ left
tony-o m: enum Silo <ASSETS EXPENSES INCOME LIABILITIES EQUITY>; my Array[Silo] @silos; my Silo @assets = ASSETS; @silos.unshift($(@assets)); say @silos.perl 00:56
camelia rakudo-moar 91212c: OUTPUT«Array[Array[Silo]].new(Array[Silo].new(Silo::ASSETS))␤»
00:57 [Sno] joined
tony-o atweiden ^^ 00:58
ugexe nqp-m: nqp::say(nqp::istype(-1, Int));
camelia nqp-moarvm: OUTPUT«0␤»
ugexe m: use nqp; say nqp::istype(-1, Int);
camelia rakudo-moar 91212c: OUTPUT«1␤»
tony-o nqp-m: nqp::say(nqp::istype(-1, int)); 00:59
camelia nqp-moarvm: OUTPUT«0␤»
tony-o nqp-m: nqp::say(nqp::istype(-1, twitter));
camelia nqp-moarvm: OUTPUT«0␤»
tony-o nqp-m: nqp::say(nqp::istype(-1, int8)); 01:00
camelia nqp-moarvm: OUTPUT«0␤»
timotimo oh, now i see that, too 01:05
01:13 yqt left 01:16 [Sno] left, [Sno] joined 01:42 atroxaper joined 01:46 atroxaper left 01:58 ShimmerFairy joined 02:05 aborazmeh joined, aborazmeh left, aborazmeh joined 02:09 noganex_ joined 02:12 noganex left
raiph m: constant #`( s/constant/term/? ) scalar = $ = 42; say scalar; scalar--; scalar.say; 02:12
camelia rakudo-moar 91212c: OUTPUT«42␤41␤»
raiph m: constant #`( s/constant/term/? ) dict = { pi=>i, e=>now }; say dict; dict<pi>--; say dict.invert
camelia rakudo-moar 91212c: OUTPUT«e => Instant:1435371205.021879, pi => 0+1i␤Instant:1435371205.021879 => e -1+1i => pi␤»
raiph m: constant #`( s/constant/term/? ) list = [pi, e]; list[2] = True; say list; list>>--; say list; say [+] @(list)
camelia rakudo-moar 91212c: OUTPUT«3.14159265358979 2.71828182845905 True␤2.14159265358979 1.71828182845905 False␤3.85987448204884␤»
02:14 rubz joined 02:16 rubz left 02:23 sue__ joined
raiph m: constant dict =%= pi, i, e, now; say dict; dict{pi}--; say dict.invert 02:24
camelia rakudo-moar 91212c: OUTPUT«2.71828182845905 => Instant:1435371876.325275, 3.14159265358979 => 0+1i␤Instant:1435371876.325275 => 2.71828182845905 -1+1i => 3.14159265358979␤»
02:24 sue__ is now known as sue
raiph .oO ( enjoying a few rounds of sigil golf ) 02:25
m: constant scalar =$= 10; scalar += 5; say [1,2] >>+>> scalar; 02:27
camelia rakudo-moar 91212c: OUTPUT«16 17␤»
raiph m: constant list =@= 10; list += 5; # compile error? 02:28
camelia ( no output )
raiph m: my @list = 10; @list += 5; # compile error?
camelia ( no output )
raiph m: constant list =@= 10,20; say list >>+>> 5 02:29
camelia rakudo-moar 91212c: OUTPUT«15 25␤»
raiph .oO ( s/constant/a/, where 'a' introduces a term, to enable "a foo =@= ..." etc.) 02:32
02:38 rmgk_ joined, rmgk is now known as Guest30350, Guest30350 left, rmgk_ is now known as rmgk 02:39 atweiden left 02:44 rjbs left, vendethiel left
ugexe Proc::Async exitcode PR sent 02:50
raiph ugexe++ 02:51
ugexe: Do you like or dislike the sigilless idiom (see ^^)? Do you know what it's downsides are? 02:52
ugexe i like sigils, because i feel it explains more naturally while reading left to right 02:56
raiph ugexe: Me too. I think it's a mistake to leave out the sigils in serious code. 02:57
Ben_Goldberg m: my @list = 10; @list += 5; say @list
camelia rakudo-moar 91212c: OUTPUT«6␤»
02:57 colomon left 02:58 colomon joined
raiph ugexe: Unless serious code includes code for beginners. 02:58
Ben_Goldberg: great example 02:59
m: constant foo = 1,2; #`( code much later ) foo += 5; say foo # not obvious that the increment of foo is wrong 03:02
camelia rakudo-moar 91212c: OUTPUT«Cannot modify an immutable Int␤ in block <unit> at /tmp/DMcRrUSE1g:1␤␤»
raiph Heh, interesting
ugexe problem is, if you put code like that out there for beginners it will become what people think is standard. sort of like how perl5 was plagued with terrible google results like matts script archive 03:03
03:04 nys left 03:05 kaare_ joined
Ben_Goldberg m: my @x = 'Just Another Perl/Bidi Hacker,'.comb; print(chr(8237) ~ @x.shift ~ chr(8238) ~ @x.pop) while @x; 03:08
camelia rakudo-moar 91212c: OUTPUT«‭J‮,‭u‮r‭s‮e‭t‮k‭ ‮c‭A‮a‭n‮H‭o‮ ‭t‮i‭h‮d‭e‮i‭r‮B‭ ‮/‭P‮l‭e‮r»
ugexe if i have to write some ugly C-like code for whatever reason i might find it useful to have a slang to do that so I can make the code more terse
Ben_Goldberg m: my $x = +my @x = 'Just Another Perl/Bidi Hacker,'.comb; print 8237.chr, @x.shift, 8238.chr, @x.pop while @x; print 8236.chr x $x; 03:15
camelia rakudo-moar 91212c: OUTPUT«‭J‮,‭u‮r‭s‮e‭t‮k‭ ‮c‭A‮a‭n‮H‭o‮ ‭t‮i‭h‮d‭e‮i‭r‮B‭ ‮/‭P‮l‭e‮r‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬»
raiph ugexe: I meant as introductory, lies-to-children code. Create an alias for 03:16
oops
Ben_Goldberg Lies-to-children are important. If noone told you: When playing, Reversi (aka Othello), whomever has the most corners at the end of the game wins... you'd play extraordinarily poorly, at least at first. 03:19
raiph m: constant foo = 1,2; #`( code much later ) foo += 5; say foo 03:24
camelia rakudo-moar 91212c: OUTPUT«Cannot modify an immutable Int␤ in block <unit> at /tmp/NwKu2T8_p_:1␤␤»
raiph m: constant foo =@= 1,2; #`( code much later ) foo += 5; say foo
camelia rakudo-moar 91212c: OUTPUT«7␤»
raiph m: constant @foo = 1,2; #`( code much later ) foo += 5; say foo 03:25
camelia rakudo-moar 91212c: OUTPUT«5===SORRY!5=== Error while compiling /tmp/0WWo3E3sqj␤Prefix + requires an argument, but no valid term found␤at /tmp/0WWo3E3sqj:1␤------> 3 @foo = 1,2; #`( code much later ) foo +7⏏5= 5; say foo␤ expecting any of:␤ prefix␤»
raiph m: constant @foo = 1,2; #`( code much later ) @foo += 5; say @foo
camelia rakudo-moar 91212c: OUTPUT«Cannot modify an immutable Int␤ in block <unit> at /tmp/Owht8dqc0b:1␤␤»
raiph m: constant @foo = [1,2]; #`( code much later ) @foo += 5; say @foo
camelia rakudo-moar 91212c: OUTPUT«7␤»
03:49 aborazmeh left 04:12 Ben_Goldberg left 04:13 Ben_Goldberg joined 04:18 aborazmeh joined, aborazmeh left, aborazmeh joined
tony-o m: \foo = Int.new(10); foo +=5; foo.say 04:24
camelia rakudo-moar 91212c: OUTPUT«5===SORRY!5=== Error while compiling /tmp/slJwFu7BUZ␤Preceding context expects a term, but found infix = instead␤at /tmp/slJwFu7BUZ:1␤------> 3\foo =7⏏5 Int.new(10); foo +=5; foo.say␤»
tony-o m: my \foo = Int.new(10); foo +=5; foo.say
camelia rakudo-moar 91212c: OUTPUT«Default constructor for 'Int' only takes named arguments␤ in block <unit> at /tmp/Ws2FHwOh59:1␤␤»
tony-o m: my \foo = $(1,2); foo +=5; foo.say 04:36
camelia rakudo-moar 91212c: OUTPUT«Cannot assign to a readonly variable or a value␤ in block <unit> at /tmp/AKn0reHeu0:1␤␤»
04:39 khw left 04:50 amurf joined 04:55 amurf left 04:57 tinyblak joined 05:11 [Sno] left, [Sno] joined 05:18 Ben_Goldberg left 05:19 StavroMueller joined, atroxaper joined 05:20 bjz left 05:24 atroxaper left 05:31 aborazmeh left 05:37 telex left 05:38 telex joined 05:46 bjz joined, bjz left 05:54 tinyblak left 05:56 skids left
dalek kudo/nom: 254eff9 | (Nick Logan)++ | src/core/Proc.pm:
bugfix: exitcode for Proc::Async

Proc::Async sets the exitcode via `Proc.new(:exitcode(1))` so BUILD requires a :$exitcode argument
See github.com/rakudo/rakudo/blob/nom/...nc.pm#L176		 06:04
kudo/nom: 0ff4557 | FROGGS++ | src/core/Proc.pm:
Merge pull request #449 from ugexe/patch-4

bugfix: exitcode for Proc::Async
06:15 RabidGravy joined
raiph .tell tony-o I've realized that use of {(...)} entirely eliminates the problems ("fooling around with ending a block and restarting a block or whatever ... buffering crap into a temp str to use that as the return value") 06:17
yoleaux raiph: I'll pass your message to tony-o.
raiph .tell tony-o (Also, that the say-in-say approach will end in tears) 06:18
yoleaux raiph: I'll pass your message to tony-o.
raiph .tell tony-o I've updated github.com/tony-o/perl6-template-p...e/issues/1 06:19
yoleaux raiph: I'll pass your message to tony-o.
06:30 xfix joined, xfix left, xfix joined 06:34 araujo_ joined, rindolf joined 06:35 araujo left 06:40 amurf joined 06:41 araujo__ joined 06:42 araujo_ left 06:44 amurf left 06:58 tinyblak joined 06:59 gfldex joined
dalek c: 5e2a992 | RabidGravy++ | lib/Type/IO/Path.pod:
Add link/symlink
 07:18
07:23 tinyblak left 07:25 tinyblak joined
dalek c: 0b24f07 | RabidGravy++ | lib/Type/IO/Path.pod:
Add chdir()
 07:59
08:02 darutoko joined 08:08 vendethiel joined 08:09 atroxaper joined 08:13 atroxaper left 08:22 rurban joined 08:23 xfix_ joined, xfix_ left, rurban left 08:26 xfix left 08:35 jaffa4 joined
jaffa4 hi all 08:35
How can I translate binmode? 08:36
I guess it is :bin in open
I would like to be able to set it independently
not at open
labster jaffa4: what do you mean by independently? 08:40
jaffa4 like binmode in Perl 5
labster Ah, you mean setting it on a filehandle outside of an open call. 08:42
my $fh = "foo".IO.open(); $fh.encoding('bin'); # should work, not sure if it does 08:44
hm, no test cases for this, not an encouraging sign 08:49
jaffa4: if you want binmode with a layer, like utf-8 or iso-8859-1 or windows-1252 -- all of those will work by passing it to the encoding function. I'm just not sure about actual binary binmode yet. 08:51
*encoding method
jaffa4 where is the source? 08:52
labster for this? /src/core/IO/Handle.pm, but it looks like I'll need to dig into NQP to see how it actually handles binary. 08:53
jaffa4 what is the url? 08:59
[Tux] m: IO::Handle.^methods».name.say 09:00
camelia rakudo-moar 0ff455: OUTPUT«FALLBACK gist new␤»
labster github.com/rakudo/rakudo/blob/nom/...le.pm#L572
[Tux] hmm, locally I get this: pipe open input-line-separator nl close eof get getc words lines read seek tell write opened t print print-nl slurp slurp-rest spurt copy chmod IO path flush encoding DESTROY e d f s l r w x modified accessed changed watch umask say Str gist perl <anon> <anon>
labster [Tux]: probably RESTRICTED.setting 09:01
m: say "/home/camelia/test".IO.open.^methods 09:04
camelia rakudo-moar 0ff455: OUTPUT«pipe open input-line-separator nl close eof get getc words lines read seek tell write opened t print print-nl slurp slurp-rest spurt copy chmod IO path flush encoding DESTROY e d f s l r w x modified accessed changed watch umask say Str gist perl <anon> <a…»
09:04 spintronic joined
RabidGravy looking at the code for IO::Handle it would seem that the only major thing that :bin does at that level is *not* set any encoding 09:08
jaffa4 yes, but encoding is set, how to undo it 09:09
09:09 diana_olhovik_ joined
labster read it in as ascii, and then call .encode on the string to get a Buf? not awesome. 09:09
RabidGravy read will just get the bytes anyway doesn't it? 09:10
labster yes. 09:11
m: "test".IO.open.read(20).say 09:12
camelia rakudo-moar 0ff455: OUTPUT«Buf[uint8]:0x<>␤»
labster perhaps I should try that on a bigger file :D
09:14 [Sno] left
labster I wish there was someone around who could give a more definitive answer, but early Saturday morning is not the best time for this channel. 09:14
This is something that should work and be documented, though. 09:15
And tested, of course.
09:15 spintronic left
RabidGravy there isn't an nqp op for the opposite of setencoding as far as I can tell 09:16
09:16 domidumont joined
labster QASTOperationsMAST.nqp:QAST::MASTOperations.add_core_moarop_mapping('setencoding', 'setencoding', 1); It's just sent straight down the the VM anyway. 09:17
09:17 Foxcool joined 09:18 spintronic joined 09:20 [Sno] joined 09:21 domidumont left 09:22 domidumont joined 09:25 atroxaper joined
labster night, #perl6 09:27
RabidGravy erp
Woodi hallo #perl6 09:29
RabidGravy erp
Woodi grammars kills me constantly :) gist.github.com/anonymous/bcb011dc022b76857855
09:29 atroxaper left
Woodi G1 works, G2 IMO should be simple replacement but don't work... 09:29
RabidGravy Woodi, change the "rule parts" to "token parts" 09:31
09:32 domidumont left
Woodi yay :) 09:34
RabidGravy a, er, rule of thumb is that a "rule" is the combination of some tokens
not strictly accurate but it helps me remember ;-) 09:35
Woodi so tokens as "leaves" and rules as "nodes", will try that. thanx :) 09:36
RabidGravy pretty much, it's down to the difference between the implicit flags applied to the "rule", "token" and "regex" largely 09:38
09:44 raiph left
RabidGravy right, off out now 09:50
09:51 Al-Caveman joined 09:52 diana_olhovik_ left 09:55 RabidGravy left 10:01 spider-mario joined 10:04 spintronic left 10:13 gagalicious left 10:17 amurf joined 10:22 amurf left 10:23 domidumont joined 10:35 dpk left 10:37 dpk joined 10:41 cognominal left 10:46 tinyblak left 10:47 cognominal joined, xfix joined, xfix left, xfix joined, xfix left 10:49 cognominal left, leont joined, cognominal joined 10:50 aborazmeh joined, aborazmeh left, aborazmeh joined 10:51 cognominal left 10:55 bin_005 left
lizmat good *, #perl6! 11:03
m: Proc::Status.new # where did it go? 11:04
camelia rakudo-moar 0ff455: OUTPUT«Could not find symbol '&Status'␤ in block <unit> at /tmp/_JwT2ZYuIi:1␤␤»
lizmat there's a Proc::Status class in the setting... 11:05
11:11 domidumont left
jaffa4 how would you translate this &(%linkage{$opt})($A,$B.$C) from Perl 5 11:13
?
moritz %linkage{$opt}($A, $B, $C) 11:16
jaffa4: but I doubt that what you wrote is actually valid perl 5 code
jaffa4 moritz: ... and you are right 11:17
let us approximately
moritz no, it's good to be precise when talking about syntax 11:18
11:24 Alina-malina left 11:25 Alina-malina joined
dalek kudo/nom: acaeb17 | lizmat++ | src/core/IO/Handle.pm:
Revive Proc::Status

This is a bandaid. Some recent changes made Proc::Status disappear from the world when still having code defining it. Suspect some sort of MOP messup with creating parent packages. By creating a stub for the parent package beforehand, we seem to have quelched the issue. Something that needs to be investigated further!		 11:33
lizmat interestingly, now test 13 in t/spec/S17-procasync/basic.t (line 39) fails because the returned object is a Proc rather than a Proc::Status 11:37
whereas the other tests work out ok afterward
.tell jnthn you might want to have a look at why github.com/rakudo/rakudo/commit/acaeb17897 was needed to revive Proc::Status 11:38
yoleaux lizmat: I'll pass your message to jnthn.
lizmat .tell jnthn as an added datapoint, check why test 13 in t/spec/S17-procasync/basic.t (line 39) fails 11:41
yoleaux lizmat: I'll pass your message to jnthn.
11:43 AlexDani` joined 11:46 tinyblak joined 11:54 diana_olhovik joined 11:57 tinyblak left, tinyblak joined 11:59 rurban joined
lizmat looking at #125485, I guess 12:00
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=125485
lizmat m: "{ @_ }" # is just an expression of: 12:01
camelia rakudo-moar acaeb1: OUTPUT«5===SORRY!5=== Error while compiling /tmp/V4Li3mqwS1␤Placeholder variable @_ may not be used here because the surrounding block takes no signature␤at /tmp/V4Li3mqwS1:1␤------> 3"{ @_ }7⏏5" # is just an expression of:␤ expecting any of…»
lizmat m: -> { @_ } # this one
camelia rakudo-moar acaeb1: OUTPUT«5===SORRY!5=== Error while compiling /tmp/FbH2YyCuDC␤Placeholder variable '@_' cannot override existing signature␤at /tmp/FbH2YyCuDC:1␤------> 3<BOL>7⏏5-> { @_ } # this one␤»
12:10 aborazmeh left
lizmat perhaps use of @_ and $^a in blocks should be referring to the first outer code instead ? 12:10
12:13 pmurias joined
moritz but then you couldn't say reduce { $^a + $^b }, @vals 12:13
just write explicit signatures for routines that contain other blocks, and you're fine 12:14
jaffa4 Is perl6-debug-m supoosed to work?
What is the state of perl6-debug-m? 12:15
lizmat moritz: good point 12:17
well, hmmm... then perhaps only if the block is part of an interpolated string? 12:18
I can see that as something that would help p5 -> 6 migration
anyways, afk until later today&
12:23 ShimmerFairy left
jaffa4 What could help? 12:24
12:33 bjz joined, rindolf left 12:36 ShimmerFairy joined 12:45 colomon left 13:02 atroxaper joined 13:07 atroxaper left 13:09 domidumont joined 13:10 RabidGravy joined
jaffa4 I have a code which seems to fail without saying a word 13:11
13:11 Fleurety joined 13:12 spintronic joined 13:15 tinyblak left 13:17 rindolf joined 13:18 nowan joined 13:19 Fleurety left
Woodi RabidGravy++ # mixing rules with tokens works :) 13:24
13:24 Fleurety joined
Woodi about GLR: for me "tuples" sound weird too. they was used in old relational database texts so they have v. old-ish conotations... 13:27
pmurias nqp-js: say("still alive...?");
camelia nqp-js: OUTPUT«/bin/sh: node: command not found␤»
Woodi Nil is planned to disappear ? so how to get results with NULLs from db ? 13:28
13:29 araujo__ left
pmurias would just having it as DBI::NULL be a problem? 13:29
Woodi also in 1990 Mr. Codd augmented relational database theory with one more NULL, some 4th-valued-logic... 13:30
timotimo and why wouldn't a type object be a good fit?
Woodi pmurias: you get data from db, ( "J.Kowalski", male, Nil, Nil), this looks nice... 13:31
timotimo the cool thing about type objects is you still get to know what type the column has 13:32
also: oh god, NULL values are EVIL in SQL :)
Woodi timotimo: I don't know. but we have some Mu's so maybe someone already thinking about this :)
timotimo: but 'we don't know value for that' is proper answer... at least to keep shape of tables :) 13:33
13:36 sjn_phone_ joined, araujo joined, araujo left, araujo joined 13:47 AlexDani` is now known as AlexDaniel
AlexDaniel So, how can I run a shell command, get its output, and have the parameters quoted correctly? This link describes everything besides shell quoting perl6advent.wordpress.com/2014/12/...om-perl-6/ 13:50
this was a painful perl5 problem which required CPAN dependency (ShellQuote) 13:52
moritz AlexDaniel: doc.perl6.org/type/Proc::Async
13:53 sjn_phone_ left
AlexDaniel moritz: are you sure that it actually answers my question? 13:53
geekosaur um? you could do it without ShellQuote
unless your notion of how to do it is incorrect from the start, which tbh is suggested by that response 13:54
(the correct answer is *don't use the shell*)
AlexDaniel thanks, that does not help as well
moritz: you see, most people will not bother with that 13:55
moritz: look at that perl6advent link and notice this line
my @commit_results = qqx/git commit -am "$message"/.lines;
clearly a mistake
13:55 amurf joined, FROGGS[mobile] joined
AlexDaniel so I'm asking how to fix that, and you tell me that the answer is to use Proc::Async. If that's true, do you understand that we will keep seeing shell injection everywhere? 13:56
FROGGS[mobile] AlexDaniel: use: run($prog, ..., :out)
moritz AlexDaniel: I'm just saying hat Proc::Async offers capturing and no need for quoting 13:57
FROGGS[mobile] that's available since yesterday
moritz oooh
AlexDaniel m: say run 'echo' 'test'
camelia rakudo-moar acaeb1: OUTPUT«5===SORRY!5=== Error while compiling /tmp/SBoxeFPZi0␤Two terms in a row␤at /tmp/SBoxeFPZi0:1␤------> 3say run 'echo'7⏏5 'test'␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ …»
FROGGS[mobile] comma 13:58
moritz no &run in camelia either, I think
AlexDaniel m: say run 'echo', 'test'
camelia rakudo-moar acaeb1: OUTPUT«run is disallowed in restricted setting␤ in sub restricted at src/RESTRICTED.setting:1␤ in sub run at src/RESTRICTED.setting:14␤ in block <unit> at /tmp/ci9sXRssZo:1␤␤»
AlexDaniel okay I'll recompile
FROGGS[mobile] AlexDaniel: look at S32-io/pipe.t
there are examples
AlexDaniel FROGGS[mobile]: thanks 13:59
13:59 amurf left 14:02 araujo left 14:05 lichtkind joined 14:06 nys joined
moritz FROGGS[mobile]: why does sub run do a 'try $proc.spawn(@args)'? (I mean, why the 'try'?) 14:08
FROGGS[mobile] moritz: I dunno, I forgot 14:12
moritz FROGGS[mobile]: looks dangerous to me
FROGGS[mobile] then we should remove it and test 14:14
moritz I'm doing a spectest as is, and hope that I'll remember to remove it and re-test when it finishes :-) 14:15
(or maybe I should do it on hack, not on my workstation :-)
14:15 tinyblak joined
FROGGS[mobile] :o) 14:18
I guess it was about a "cmd not found" issue
ohh dang, I forgot to test run/shell on windows :S 14:19
moritz but the API is *very* nice, FROGGS[mobile]++ 14:20
and FROGGS++ too, I guess :-)
AlexDaniel $output = run('echo', $test, :out).out.get;
$output = qqx/echo $test/;
Huffmanly speaking, which one is better and why?
FROGGS[mobile] good question 14:22
14:24 diana_olhovik__ joined
AlexDaniel FROGGS[mobile]: thanks for listening, I always feel ignored when I raise questions about shell quoting 14:24
14:25 diana_olhovik left
moritz I realize the defaults are maybe still the wrong way around, but now Perl 6 has a much nicer API for capturing err/out and no need for quoting 14:25
compared to p5
AlexDaniel the typical answer is "well, it works, so what's your problem?". Well, you can inject any arbitrary shell thing into it, that's the problem...
and if it had better Huffman coding I think that the problem will fade away 14:26
well, at least it will get better
FROGGS[mobile] AlexDaniel: I'll think about a sane and safe short-cut
moritz maybe have a quoting for that preserves scalars as single items, and only splits literals on spaces? 14:27
AlexDaniel that's what I though, actually 14:28
which basically means that it should autoquote variables, but then again, what does it mean to quote things correctly?
it depends on the underlying shell
jaffa4 Is this ok grep ($_ eq $opt, @hits)?
moritz so that my $foo = "a b"; my @list = q:w:S<echo $foo> is equivalent to my @list = ('echo', $foo) 14:29
jaffa4: it doesn't do what you want
jaffa4 std: grep ($_ eq $opt, @hits)
camelia std 28329a7: OUTPUT«5===SORRY!5===␤Variable $opt is not predeclared at /tmp/7hVV_j54vO line 1:␤------> 3grep ($_ eq 7⏏5$opt, @hits)␤Variable @hits is not predeclared at /tmp/7hVV_j54vO line 1:␤------> 3grep ($_ eq $opt, 7⏏5@hits)␤Check failed␤FAILED 00:00…»
geekosaur I feel like insistence that making the naive shell quoting solution "work" is just encouraging security holes
moritz jaffa4: if $opt is a string, you can just do @hits.grep($opt) 14:30
AlexDaniel geekosaur: what do you mean?
moritz geekosaur: the point is to not need shell quoting
geekosaur yes, I understand this is not popular. the widest proof being how the Internet is rife with such security holes (mostly in the related area of SQL quoting)
moritz geekosaur: but always pass the argument list out-of-band
geekosaur moritz, but I was informed that avoiding the shell was not the correct answer
moritz geekosaur: who informed you, and it wasn't the correct answer to what question? 14:31
geekosaur because the question was shell quoting. never mind thaty the question is wrong
sigh
moritz minds that the question is wrong
geekosaur yes, I understand the point was to ask about shell quoting itself
moritz appeal to authority doesn't work well in here, sorry 14:32
geekosaur but, from a security standpoint,m the answer to that is "avoid it at all costs"
AlexDaniel yea the question was a bit wrong, you are right
geekosaur: if the answer is "use 'run'" then it is OK though
geekosaur because even when you think uyou've covrred all the cases, you get hit by shellshock *sigh*
(I mean, who expects their dhcp client is vulnerable to shell injection attack?) 14:33
moritz only those who you call "paranoid" 14:34
AlexDaniel geekosaur: so what do you recommend? Make 'run' better Huffmanly coded and that's it?
geekosaur I don't know. I feel like there is a tendency to say "people will use the simple solution so let's make it secure" and this fails when there is demonstrably no way to make it secure. But also demonstrably steering people away from the simple solution doesn't work 14:35
AlexDaniel geekosaur: good point
FROGGS[mobile]: backticks are not used? We can use backticks for that! Haha 14:37
FROGGS[mobile] AlexDaniel: with an argument list? :o)
AlexDaniel yeah :) 14:38
14:41 bjz left
dalek c: 9d2b828 | moritz++ | lib/Type/Proc.pod:
Document type Proc (and &run)
 14:44
c: d1bcc0a | moritz++ | lib/Type/IO.pod:
Fix documentation for &run in type IO
moritz FROGGS[mobile]: a short proof-read of those docs would be very much appreciated 14:45
AlexDaniel actually, we should listen to what geekosaur said 14:47
and after reading "Piping several commands is easy too." I realize that he is absolutely right
which probably means that qx should actually go away 14:49
14:49 araujo joined
geekosaur unclear actually. the problem is if you remove it, someone will re-add it (probably in an even worse way) for convenience 14:53
see "steering people away from the simple solution doesn't work"
AlexDaniel what about qwx or something? 14:55
moritz FROGGS[mobile], jnthn: it's a bit unfortunate that Proc.out and Proc::Async.stdout have a diverging naming scheme
AlexDaniel that will split stuff by spaces and pass that to run
14:55 FROGGS joined
FROGGS[mobile] moritz: aye, and I'd vote for shorter names fwiw 14:58
FROGGS moritz: I'll proofread the docs this evening
I've got currently about five kids under age 6 around me 14:59
jaffa4 This is ok, right? return (0) if $passthrough; 15:02
timotimo i'd propose "return 0 if $passthrough" instead 15:05
15:05 FROGGS left 15:10 rindolf left 15:11 shlomif joined, tinyblak left, tinyblak joined
AlexDaniel FROGGS: "with an argument list? :o)" hm, it can do the same thing < > is doing 15:12
FROGGS[mobile] I was thinking about that, but I'm not sure about that yet 15:14
15:16 skids joined
japhb Apologies that I've only skimmed, but doesn't the quoting of << >> get what most people want from "shell quoting" when they don't *actually* want a shell? 15:19
Note << >> instead of < > 15:20
AlexDaniel japhb: I'm not really following you, can you show a code example? 15:21
japhb AlexDaniel: Well, I'm not caught up with the details of FROGGS++'s new implementations, but in the old days I might have said something along the lines of run(<< echo "foo $bar" @baz <<wheee>> >>) and expected it to work safely 15:25
15:25 JimmyZ_ joined
japhb I guess that should have been @baz[] 15:26
But my point was that << >> has a lot of the DWIM that people want from shell quoting without the danger of an actual shell parsing. 15:27
15:27 nys left
jnthn evening, #perl6 15:30
yoleaux 11:38Z <lizmat> jnthn: you might want to have a look at why github.com/rakudo/rakudo/commit/acaeb17897 was needed to revive Proc::Status
11:41Z <lizmat> jnthn: as an added datapoint, check why test 13 in t/spec/S17-procasync/basic.t (line 39) fails
japhb o/ jnthn 15:31
nwc10 good UGT heresy, jnthn
DB are less late than the scandawedigian railways?
15:31 CIAvash joined
AlexDaniel $output = run('echo', $test, :out).out.get; 15:32
$output = run(<<echo $test>>, :out).out.get;
$output = qqx/echo $test/;
timotimo << >> is still not properly implemented, right?
15:32 CIAvash left, CIAvash joined
AlexDaniel japhb: you see, it does make it easier 15:32
japhb: maybe just a bit
japhb AlexDaniel: Of course, you can use real guillemots, but the keyboard I am on is less than awesome for entering Unicode ... 15:33
timotimo « »
japhb And then it gets more readable
AlexDaniel oh right
and it is actually one character shorter then 15:34
15:34 tinyblak left
AlexDaniel still not shorter than qqx... 15:34
ugexe i look forwarding to writing my first run(<<$cmd $var>>.list>>.value>>)
15:34 araujo left
japhb I find I often "spend" that minor savings by adding an extra space before the first item and after the last. 15:34
Since Windows doesn't actually have multiple arguments, how is run() implemented there? 15:35
dalek kudo-star-daily: ff22ac7 | coke++ | log/ (2 files):
today (automated commit)
 15:36
rl6-roast-data: 25f8ebd | coke++ | / (9 files):
today (automated commit)
japhb (For precision, I mean that at the C runtime level, processes receive a single giant string, rather than pre-split command and args as in the *nix world -- and then they do their own command string parsing)
ugexe im not sure what the answer is, but you use %* in powershell
FROGGS[mobile] japhb: libuv handles it in some way 15:37
japhb fears that implementation
15:37 lichtkind_ joined
FROGGS[mobile] we pass it a char ** 15:37
japhb Good for libuv for at least presenting the better API to its callers, I suppose. But man, I hope they've implemented something that all the C runtimes will handle correctly .... 15:38
15:38 nys joined 15:39 lichtkind left
jnthn I suspect they've had more resources to think it through than we'd have been able to give to it if we'd solved it ourselves... :) 15:41
Plus it's presumably been exercised plenty thanks to node.js using libuv. 15:42
japhb True!
skids also looked at better commandline parsing (emulating legacy syntaxes) a couple weeks ago. 15:51
15:52 Ven joined
skids There wer a couple blockers for what I wanted to do. 15:52
The primary one was that Cursor doesn't work right yet from inside rules.
15:53 shlomif left
skids So there was no way to build something like a <bag($bag)> subrule that would match strings from a bag anywhere in a grammar, but only as many times as they were in the bag. 15:53
Also I had to punt and rejoin ARGS with '\0' separators, for lack of matching against structural data. 15:54
15:56 tinyblak joined
TimToady is home again, and finds the air at sea level is too thick now... :) 15:57
16:00 Ven left
jnthn I think the structural matching stuff is very post-6.0 :) Not sure what "Cursor doesn't work right yet" means, though... 16:03
skids You don't get a current .pos. 16:04
Or a current partial match.
jnthn Depends.
japhb Welcome back to the lowlands, TimToady!
jnthn You should if in, say, a code block 16:05
skids You don't though.
jnthn m: 'a' ~~ /(.) { say $/[0] } /
camelia rakudo-moar acaeb1: OUTPUT«｢a｣␤»
skids On the first time, sometimes you do. It is not reliable. 16:06
jnthn It'd help if you actualy showed code.
*actually
skids digs around in his stuff. 16:07
mst skids: you're currently telling one of the authors you know his code better than him, without showing any examples
skids: this is ... unlikely to help anybody
jnthn On the <bag($bag)> thing, where did you plan to keep the "how many we saw" state?
skids Dynamic probably.
jnthn In TOP?
skids Likely defined in a custom .parse. 16:08
mst perl5 has a rule for regexp questions where we ask for input, expected output, and what code you have so far 16:09
generally for fiddly stuff an example of that form is the only way to clearly communicate
so maybe that would be the best next step?
16:10 estrabd joined, khw joined
skids mst: I actually have to be somewhere right now, so I can't do that right now. Was looking for a quick example but looks like I was doing it on cli, so I'll have to go see if it's still in my history. 16:11
mst skids: ok, just, "you don't though" isn't desperately easy to diagnose :) 16:12
TimToady pictures jnthn as a desperado... 16:13
mst if I was a gun fighter with his eyesight I'd be pretty desperate too
vendethiel *g* 16:14
jnthn m: grammar G { method parse(*@, :$bag) { my %*BAGGED := $bag.BagHash; my $*BAG-RX = /| @($bag.keys)/; callsame; }; token TOP { <.bag>+ }; token bag { (<{ $*BAG-RX }>) <?{ %*BAGGED{~$0}-- }> } }; say G.parse('aabbc', bag => bag <c b a a b >); say G.parse('aaabbc', bag => bag <c b a a b >); 16:15
camelia rakudo-moar acaeb1: OUTPUT«｢aabbc｣␤(Any)␤»
jnthn There's a quickly thrown together solution for the problem (or one like the problem) I think you're wanting to solve, btw. 16:16
oh, now I look at it it can be a tad simpler too 16:17
m: grammar G { method parse(*@, :$bag) { my %*BAGGED := $bag.BagHash; my $*BAG-RX = /| @($bag.keys)/; callsame; }; token TOP { <.bag>+ }; token bag { ($*BAG-RX) <?{ %*BAGGED{~$0}-- }> } }; say G.parse('aabbc', bag => bag <c b a a b >); say G.parse('aaabbc', bag => bag <c b a a b >);
camelia rakudo-moar acaeb1: OUTPUT«｢aabbc｣␤(Any)␤»
jnthn a thing of type Regex already interpolates proper