Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6, 2014-09-14

Perl 6 | Reference Documentation | Rakudo

| Channels | #perl6 index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:14 jerrycheung joined #perl6
00:29 jepeway thanks, timotimo++, for the tiny tour
00:41 colomon joined #perl6
00:48 cognome joined #perl6
00:48 mr-foobar joined #perl6
00:52 mberends joined #perl6
00:52 cognome_ joined #perl6
01:23 FROGGS__ joined #perl6
01:57 toto joined #perl6
01:59 colomon joined #perl6
02:49 noganex joined #perl6
03:29 Px12 joined #perl6
03:31 jerrycheung joined #perl6
03:41 davido___ joined #perl6
03:45 ClarusCogitatio joined #perl6
04:00 ClarusCogitatio joined #perl6
04:09 xenoterracide joined #perl6
04:16 ClarusCogitatio joined #perl6
04:19 rindolf joined #perl6
04:24 rindolf joined #perl6
04:30 telex joined #perl6
04:42 ClarusCogitatio joined #perl6
04:57 kaare_ joined #perl6
05:43 xenoterracide joined #perl6
05:44 davido___ joined #perl6
05:54 xiaomiao joined #perl6
06:07 ClarusCogitatio joined #perl6
06:29 SamuraiJack joined #perl6
06:36 virtualsue joined #perl6
06:37 darutoko joined #perl6
06:39 ClarusCogitatio joined #perl6
06:49 davido___ joined #perl6
06:53 bjz joined #perl6
07:09 jerrycheung joined #perl6
07:23 Px12 joined #perl6
07:36 jantore joined #perl6
07:48 ClarusCogitatio joined #perl6
08:07 denis_boyun_ joined #perl6
08:17 cognome joined #perl6
08:20 spider-mario joined #perl6
08:22 Px12 joined #perl6
08:23 spider-mario joined #perl6
08:25 kurahaupo joined #perl6
08:26 jerrycheung joined #perl6
08:32 kjs_ joined #perl6
08:37 ClarusCogitatio joined #perl6
08:42 spider-mario joined #perl6
08:52 cognome joined #perl6
09:17 ClarusCogitatio joined #perl6
09:18 jerrycheung joined #perl6
09:21 slmult0 joined #perl6
09:24 ClarusCogitatio joined #perl6
09:49 virtualsue joined #perl6
09:52 cognome joined #perl6
09:52 brrt joined #perl6
09:54 _sri joined #perl6
09:58 lizmat good *, #perl6
09:58 lizmat and my has it been silent today so far
09:59 nine lizmat: shhhht
10:00 timotimo you're ruining it! =o
10:01 jnthn YEAH, BE QUIET.
10:01 * nine stomps on the floor
10:01 * lizmat whistles while walking away
10:02 FROGGS what what? somebody said something?
10:03 * FROGGS yawns
10:03 * jnthn has safely returned to The Other System. :)
10:03 FROGGS jnthn: what other system?
10:03 nine Should I a) finish my god damn master's thesis, b) continue working on Inline::Perl5 or c) go to Gamestage Expo with friends?
10:03 FROGGS a political system?
10:03 jnthn FROGGS: Yeah.
10:04 vendethiel nine: always d)
10:04 FROGGS nine: a)
10:04 jnthn FROGGS: Reference to "two systems, one country" :)
10:04 moritz m: say <thesis Inline::Perl5 Gamestage>.pick
10:04 camelia rakudo-moar 21a2a4: OUTPUT«Gamestage␤»
10:04 FROGGS lol
10:04 moritz nine: rakudo has a very clear opinion on that one :-)
10:04 nine Seems like Rakudo is not so keen on cuddling up to Perl 5 :)
10:05 * lizmat wonders how moritz rigged camelia to do that
10:05 jnthn Nor your master's thesis, apparently... :P
10:05 FROGGS jnthn: is that just NYI or a NONO? Cannot create rw-accessors for natively typed attribute '$!keepBlanks'
10:05 moritz FROGGS: NYI
10:05 FROGGS k
10:05 jnthn FROGGS: NYI, and that stuff is basically top of my todo list by now
10:06 FROGGS jnthn: is it hard to implement?
10:06 jnthn Well, it depends on a bunch of other native-related pieces.
10:06 FROGGS jnthn: in case not, I'd like to do it this sunday, but I am thankful for pointers :o)
10:06 FROGGS (not opaquepointers, mind)
10:06 FROGGS hmmm :/
10:06 jnthn It's related to native arrays and is rw on natives and a bunch of other things
10:06 moritz FROGGS: I think it needs a bit of conceptual work
10:07 FROGGS ohh
10:07 jnthn Or at least, I want to solve all of thoese with a single mechanism.
10:07 * FROGGS has no idea about that
10:07 moritz once we have rw for natives, creating the accessors that way is a SMOP
10:07 jnthn And yeah, it needs careful design, in so far as I need to be able to optimize away much of the cost of things either statically or in spesh...
10:07 jnthn Or probbably both
10:07 jnthn Not to mention coming up with something that will work ou tOK on JVM...
10:08 FROGGS okay, I leave it on your list then :o)
10:08 * moritz consults the ROADMAP again
10:09 moritz FROGGS: maybe you want to take a stab at :: and ::: in regexes? (backtracking control)
10:09 FROGGS moritz: that would mean I have to understand it first :o)
10:09 moritz FROGGS: :-)
10:09 FROGGS but yeah, I am open for these things
10:09 moritz 2 **    Rat/FatRat/Rational cleanup
10:09 moritz what does that mean?
10:10 moritz also, coercion types seem to occur twice
10:10 moritz 2 ***   new syntax/semantics for coercion (jnthn)
10:10 moritz 2 ***   coercion types
10:11 jnthn Should probably clarify that while the natives stuff is in theory top o my todo list, it's one of those Big Hard Things that needs a good chunk of time when I can concentrate, and that's hard to come by while I'm away on consulting gig.
10:12 jnthn So I'll probably continue picking off other little bits over the next couple of weeks. :)
10:13 jnthn After that I've got some free days at home when I can do Perl 6 stuff, teaching on a very familiar topic which won't be especially tiring, and then Austrian Perl Workshop and hackathon time. :)
10:15 jerrycheung joined #perl6
10:16 jaffa4 joined #perl6
10:19 brrt \o
10:19 brrt where's this ROADMAP thingy?
10:19 jnthn hi brrt
10:19 brrt (was backlogging) hi jnhtn :-)
10:20 moritz brrt: docs/ROADAMP in rakudo
10:20 vendethiel is that up-to-date?
10:20 brrt m: <study hack play>.pick
10:20 camelia rakudo-moar 21a2a4: ( no output )
10:20 moritz vendethiel: pretty much
10:20 moritz vendethiel: last updated end of July
10:20 vendethiel oh, okay :)
10:21 vendethiel moritz: for regexps, @<> could be added
10:22 timotimo brrt: that's *so* symbolic
10:22 moritz vendethiel: what would that do?
10:22 brrt what is timotimo?
10:22 timotimo brrt: <study hack play>.pick → "no output"
10:22 vendethiel moritz: what's specced. basically, it forces the capture to be an array
10:23 brrt yeah i forgot to say. and if is asked again i couldn't get an answer
10:23 brrt camelia is telling me to make up my own mind
10:23 eternaleye joined #perl6
10:23 timotimo FROGGS: maybe you'd like to do the CAPHASH removal in rakudo's Match?
10:23 vendethiel moritz: it's useful when you have, ie [ @<a>='a' | <a> ] with regex a { \d + }
10:23 vendethiel (so that $/[0] is always a list)
10:25 brrt sized/shaped arrays, is that native arrays?
10:25 cognome joined #perl6
10:26 FROGGS timotimo: how?
10:26 FROGGS timotimo: what is needed to do that?
10:27 lizmat jnthn: any suggestions wrt http://irclog.perlgeek.de/​perl6/2014-09-13#i_9352237 ?
10:28 timotimo FROGGS: well, it's either understanding the original change in the nqp version and applying the same idea on the rakudo one
10:28 timotimo or ... er ...
10:28 FROGGS hmmm
10:28 jnthn lizmat: Whatever creates a thread or gets a thread probably needs to have nqp::scwbdisable() and nqp::scwbenable() or so wrapped around it
10:29 FROGGS timotimo: I can try I think
10:29 jnthn lizmat: So the write barrier wont be tripped;
10:29 lizmat hmmm...
10:29 lizmat since yesterday, I made $*THREAD lazy as well
10:29 lizmat now it says Missing serialize REPR function for REPR MVMOSHandle
10:29 lizmat that would point to STDOUT I guess...
10:37 lizmat jnthn: making $*IN/$*OUT/$*ERR lazy, changes the error to:
10:37 lizmat Missing serialize REPR function for REPR ReentrantMutex
10:39 jnthn lizmat: I'm pretty sure those arne't a big saving, from the profiling I did...
10:39 lizmat but @*INC *is*, and that's being blocked now
10:39 jnthn Right, so we should fix what's needed for that, but not pile up more things that are going to be awkward. ;)
10:40 lizmat I'm all for that, but also out of my depth
10:40 jnthn lizmat: What branch is the busted patch in?
10:40 timotimo doesn't $*OUT get almost immediately vivified in almost every program anyway?
10:40 jnthn Or, can you push it to one?
10:40 jnthn I can look later this eveing, probbably.
10:41 lizmat you need to re-apply d05d411e968b3408420bbc7b02c1248f3dc005de
10:41 jnthn ah, ok
10:41 * jnthn will get checkout -b bug && git cherry-pick d05d411 later, then :)
10:41 jnthn Dinner now &
10:41 lizmat and then run $ perl6 -e '"bar".IO.spurt("use Test"); CompUnit.new("bar").precomp'
10:46 lizmat jnthn: thank you
10:46 lizmat timotimo: not for $perl6 -e1
10:50 zakharyas joined #perl6
10:55 jerrycheung joined #perl6
11:06 timotimo jnthn: how do you feel about me building a string equivalent of "littlebigint"?
11:06 timotimo oh
11:07 timotimo i just now remember why i shouldn't try that yet ... uthash ...
11:11 nine jnthn: why is this leaking memory? perl6 -e 'my $destroyed = 0; class Foo { method DESTROY {$destroyed++;} }; loop { Foo.new }'
11:11 nine jnthn: but this not? perl6 -e 'class Foo { method DESTROY {my $destroyed;$destroyed++} }; loop { Foo.new }'
11:11 carlin .oO( Battle of the Little BigInt )
11:13 nine timotimo: do you happen to know how I can access force_gc() from Perl 6 code? Would help me with Inline::Perl5 tests.
11:13 carlin can you do nqp::force_gc()
11:13 lizmat m: nqp::force_gc
11:13 camelia rakudo-moar 21a2a4: ( no output )
11:13 lizmat m: say nqp::force_gc
11:13 camelia rakudo-moar 21a2a4: OUTPUT«===SORRY!===␤arg expression cannot be void␤»
11:14 lizmat I guess the hookup is suboptimal still  :-)
11:14 woolfy1 joined #perl6
11:15 timotimo with nqp:: ops, the training wheels are off :P
11:16 kaare__ joined #perl6
11:17 lizmat I was wondering whether something like: $*SCHEDULER.cue: :every(.1), { nqp::force_gc } could provide something close to timely destruction
11:18 timotimo don't forget each gc run will freeze all threads for a bit
11:19 lizmat yes, I realize that
11:20 lizmat but having this, *will* mean people will use it
11:20 timotimo well ... damn ;)
11:20 timotimo people be crazy
11:21 lizmat hmmm... I just realize there is no mention of Proc::Async in the Changelog
11:21 timotimo i think forcing gc often would make our generational hypothesis b0rk
11:21 timotimo because short-lived objects are more likely to be promoted to the gen2 if we GC more often
11:21 lizmat that's something new in this month's release, isn't it?
11:21 lizmat I've never said it would be sensible  :-)
11:23 nine timotimo: thanks, seems to work
11:23 dalek rakudo/nom: a278a2a | moritz++ | docs/ROADMAP:
11:23 dalek rakudo/nom: update ROADMAP
11:23 dalek rakudo/nom:
11:23 dalek rakudo/nom: * add @<> regex feature, vendethiel++
11:23 dalek rakudo/nom: * remove duplicate mention of coercions
11:23 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/a278a2a389
11:24 timotimo nine: er, i didn't do anything %)
11:24 moritz lizmat: Proc::Async was added 2014-08-11
11:25 lizmat so it was in the previous release, but not documented as such ?
11:25 moritz lizmat: correct
11:25 moritz lizmat: I can add it if you want
11:25 nine ah
11:25 nine carlin: thanks, seems to work
11:25 lizmat moritz: was just doing that
11:25 moritz lizmat: ok, I'll let you do it
11:26 lizmat "+ added experimental support for Proc::Async, MoarVM only for now"
11:26 lizmat ok?
11:27 moritz ok
11:27 carlin nine: I think lizmat++ was more help than me
11:27 carlin :p
11:28 lizmat who? /me?  you mean timotimo ?
11:28 nine carlin: or not... while it compiles and runs, I don't see my object's destructor called.
11:28 jaffa4 hi guys, how am I supposed to install panda, I followed the instructions and I get an error message?
11:28 dalek rakudo/nom: 5de1e70 | (Elizabeth Mattijsen)++ | docs/ChangeLog:
11:28 dalek rakudo/nom: Mention Proc::Async
11:28 dalek rakudo/nom:
11:28 dalek rakudo/nom: Even though it was in the previous release already
11:28 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/5de1e707bd
11:29 nine perl6 -e 'my $destroyed = 0; class Foo { method DESTROY { $destroyed++; } }; Foo.new; nqp::force_gc(); say $destroyed;'
11:29 nine 0
11:29 FROGGS jaffa4: I think there is an issue on parrot... can you test it on moarvm?
11:29 jaffa4 I am on moarvm
11:29 moritz jaffa4: what's the error?
11:29 FROGGS jaffa4: can you no-paste the error?
11:30 nine whereas: perl6 -e 'my $destroyed = 0; class Foo { method DESTROY { $destroyed++; } }; until $destroyed { Foo.new; }; say $destroyed;'
11:30 nine 8
11:30 jaffa4 perl6 panda\bootstrap.pl -> Could not find lib in any of: ., blib, D:\rakudogit\install/languages/perl6/lib, D:\rakudogit\install/languages/nqp/lib
11:30 FROGGS hmmm
11:31 FROGGS jaffa4: both rakudo and panda are up-to-date?
11:31 jaffa4 I just cloned it into panda
11:32 FROGGS what's your perl6 --version?
11:32 jaffa4 This is perl6 version 2014.07-238-g688c289 built on MoarVM version 2014.07-426-g49cecc3
11:32 FROGGS the please upgrade rakudo
11:32 moritz jaffa4: did you 'make install' in rakudo?
11:33 FROGGS nmake install*
11:34 jaffa4 I need to look
11:39 jaffa4 I recompiled At Frame 4, Instruction 3, op 'sp_guardcontconc' has invalid number (0) of operands; needs 3.
11:41 FROGGS then your MoarVM might be too new
11:41 timotimo yeah, there was an nqp stage0 update for moarvm
11:41 timotimo recently-ish
11:41 colomon I just tried to say a list of 300 strings and got a truncated list with "…" at the end?!
11:41 FROGGS jaffa4: update MoaVM, nqp and rakudo
11:41 timotimo colomon: yeah, you're using ".gist", TimToady recently made gist give you only the gist of a list instead of all of it
11:41 timotimo you can still ~@list it
11:41 FROGGS timotimo: that's what we are talking about it seems: https://github.com/perl6/nqp/commit/77​5ac85288898bcc50af356be995c2f2b160712d
11:42 timotimo that's right
11:43 jaffa4 well, I did I think, I pull the git, and issue configure again... what version of git should I use?
11:43 FROGGS just rakudo/nom
11:44 FROGGS and then pass --gen-moar to configure
11:44 FROGGS --gen-moar implies --gen-nqp btw
11:44 jaffa4 this works perl Configure.pl --gen-moar --gen-nqp --backends=moar, right?
11:44 FROGGS right
11:45 jaffa4 that is when I get the error
11:45 FROGGS what does 'git describe' give you?
11:45 jaffa4 Minsk-339-g5de1e70
11:45 kurahaupo joined #perl6
11:46 FROGGS that fits
11:46 timotimo maybe we forgot a version bump after some op got introduced?
11:47 jaffa4 what version should I get which is ok?
11:47 FROGGS timotimo: NQP_REV looks good
11:47 timotimo and MOAR_REV, too?
11:48 FROGGS yes
11:48 timotimo hm. maybe nuke install/ from orbit?
11:48 carlin m: class Foo { method DESTROY { say 'kaboom'; } }; { my $x = Foo.new }; nqp::force_gc(); nqp::force_gc()
11:48 camelia rakudo-moar 21a2a4: ( no output )
11:48 carlin m: class Foo { method DESTROY { say 'kaboom'; } }; { my $x = Foo.new }; nqp::force_gc() xx 2;
11:48 camelia rakudo-moar 21a2a4: OUTPUT«kaboom␤»
11:49 timotimo nobody promised you timely destruction :S
11:51 kurahaupo timotimo: a long time ago we were promised timely destruction, at least for enrolled objects or types. Seems a shame that's been dropped
11:51 jaffa4 from orbit?
11:52 jaffa4 what orbit?
11:52 FROGGS jaffa4: he meant to remove the 'install' directory entirely :o)
11:55 breinbaas joined #perl6
11:57 jaffa4 yes , that is turn off, turn on
11:57 FROGGS right
11:58 FROGGS I built latest rakudo just fine
12:01 Px12 joined #perl6
12:01 jaffa4 compilation does not work with msc9
12:02 jaffa4 it works with msc10
12:02 Akagi201 joined #perl6
12:03 nine carlin: where does that difference between nqp::force_gc() xx 2 and nqp::force_gc(); nqp::force_gc() come from?!
12:04 FROGGS jaffa4: eww
12:04 brrt joined #perl6
12:05 alexghacker joined #perl6
12:07 vendethiel nine: mayb the GC *refuses* to :p
12:07 vendethiel m: class Foo { submethod DESTROY { say 'kaboom'; } }; { my $x = Foo.new }; nqp::force_gc();
12:07 camelia rakudo-moar 21a2a4: ( no output )
12:07 xragnar joined #perl6
12:07 vendethiel .oO( we c++ now )
12:08 vendethiel .o( heisendestroy )
12:11 * nine thinks force_gc is not as forceful as it likes to be
12:13 spider-mario joined #perl6
12:14 brrt nine: on moar?
12:14 jnthn Well, it certainly doesn't force a ful collection, and it's not entirely clear that the Foo instance you created is out ot scope
12:15 jnthn The optimizer can flatten away unrequired scopes when it can see we couln't possibly introspect and discover a discrepancy.
12:16 FROGGS eww
12:16 jnthn Trying to reason about what will go out of scope and become collectable when is a fools errand.
12:16 FROGGS do we have new chars for Match.gist now?
12:16 FROGGS m: say "a" ~~ /a/
12:16 camelia rakudo-moar 21a2a4: OUTPUT«「a」␤␤»
12:16 FROGGS that's.... ugly
12:17 grondilu n: say "a" ~~ /a/
12:17 camelia niecza : OUTPUT«Cannot open assembly './run/Niecza.exe': No such file or directory.␤»
12:17 jnthn FROGGS: We had the corner brackes for ages now, no?
12:17 FROGGS then it is my font >.<
12:17 FROGGS I upgraded to ubuntu 14.04 just on saturday
12:18 FROGGS lunch &
12:18 jnthn .oO( Baptized at the font of missing glyphs )
12:24 jnthn nine: Answer to questions like "difference between xx 2 and not could be all kinds, but quite easily it could just be that xx 2 re-uses registers somewhere and so leaves objects unreferenced in that case.
12:25 nine jnthn: sounds like there's no way to write reliable tests?
12:26 lizmat jnthn: fwiw, I'm now writing the spec of Proc::Async so that people have something to look at when the next release comes out and it mentions Proc::Async
12:26 timotimo that seems like a good decision
12:26 jnthn You seem to want to write reliable tests that rely on subtleties of register allocation, optimization, GC, and probably more...
12:27 timotimo sounds like you want to use the MASTCompiler directly
12:28 jnthn Well, no, because there's no real reason for the interop stuff not to work out on JVM
12:28 xragnar joined #perl6
12:28 nine I wonder if it actually works on the JVM right now. Except for DESTROY support I wouldn't know a reason why not.
12:28 timotimo right
12:28 jnthn And so even if you manage to get the stars to align "reliably" on Moar, you're probably going be juggling a different bunch of things.
12:28 lizmat jnthn: would it make sense to make Proc::Async's path parameter to be a positional, rather than a obligatory named ?
12:29 Px12 joined #perl6
12:29 lizmat jnthn: and args a slurpy ?
12:29 jnthn lizmat: Is there precednet for doing so anywhere else in the built-ins?
12:29 carlin nine: the trait_mod craziness to get NativeCall to work doesn't work on JVM
12:30 lizmat jnthn: S29:577
12:30 synopsebot Link: http://perlcabal.org/syn/S29.html#line_577
12:30 jnthn lizmat: That's hard to apply in so far as it's a sub call, not an object construction.
12:30 lizmat true
12:30 jnthn lizmat: I was more thinking of other IO related classes
12:31 lizmat I guess we need to unspec :bg in S29 and link to S17's Proc::Async ?
12:31 jnthn lizmat: Uh, yes. :)
12:32 jnthn lizmat: I'm quite confident in saying that a named arg indicating async or not is LTA API design. ;)
12:33 nine jnthn: why is this leaking memory? perl6 -e 'my $destroyed = 0; class Foo { method DESTROY {$destroyed++;} }; loop { Foo.new }'
12:33 nine jnthn: but this not? perl6 -e 'class Foo { method DESTROY {my $destroyed;$destroyed++} }; loop { Foo.new }'
12:37 dalek specs: 81994a0 | (Elizabeth Mattijsen)++ | S29-functions.pod:
12:37 dalek specs: Remove :bg parameter from shell()/run()
12:37 dalek specs:
12:37 dalek specs: Now done by Proc::Async, and can not really be intergrated seemlessly here.
12:37 dalek specs: review: https://github.com/perl6/specs/commit/81994a0b80
12:37 dalek specs: fa80c00 | (Elizabeth Mattijsen)++ | S29-functions.pod:
12:37 dalek specs: Spec the :cwd parameter for shell()/run()
12:37 dalek specs: review: https://github.com/perl6/specs/commit/fa80c00549
12:38 lizmat IO::Notification.watch_path doesn't take a named path either
12:39 jnthn nine: No idea, though I can see it is...
12:40 jnthn nine: At it's sensitive to the DESTROY being there. Odd.
12:41 jnthn lizmat: Yeah, that's a stronger indicator.
12:41 jnthn nine: Oh...
12:42 jnthn nine: Could be due to the way the destroy queue handler is invoked
12:43 xragnar joined #perl6
12:43 jnthn (As a special return handler)
12:43 * lizmat muses about a postfix:<&> for IO::Path that would do a Proc::Async.new(path,args).start  :-)
12:43 jnthn nine: Yeah, seems to be that.
12:44 jnthn nine: It's a lot less likely to affect real world code.
12:44 jnthn nine: Which actually has a call graph rather than just sitting in a hot loop.
12:45 nine jnthn: let's hope so :)
12:45 jnthn Well, that bit is easy enough to explain
12:46 jnthn The typical way to do these things is to run DESTROY on some thread. That doesn't risk leaks like this, but does spoil the thread affinity you wanted.
12:47 jnthn So if we need to run them on the same thread, we need to make that call happen at some safe point.
12:47 jnthn Not every GC safe point is also an invocation safe point, however.
12:47 lizmat why would we need to run DESTROY on the same thread ?
12:48 jnthn lizmat: I thought nine said he needed it that way for p5 interop?
12:48 lizmat would seem to me, that having your DESTROYs done asynchronously, would be another reason for running P5 code in Perl 6
12:48 jnthn lizmat: Having refcounts decremented by another random thread sounds like a great way to segv Perl 5...
12:48 lizmat from a performance point of view
12:49 jnthn lizmat: Especially if the same p5 interpreter is running code on another thread.
12:49 lizmat maybe P5 refcounting needs to be done from a Supply.act ?
12:50 lizmat just wondering how things can be, not why they can't  :-)
12:50 jnthn Anyway to finish what I was explaining, the way we get an invocation safepoint is attaching the "run the DESTROYs" to happen at the next block exit.
12:50 jnthn And if you're in a loop { } then that block typically is inlined so there's no block exit until program exit.
12:50 jnthn Most interesting programs are not a single hot loop, so I won't worry abou this one too much :)
12:51 * jnthn has a skype call, bbiab :)
12:51 sqirrel_ joined #perl6
12:51 kjs_ joined #perl6
12:51 brrt joined #perl6
12:52 chenryn joined #perl6
12:54 virtualsue joined #perl6
13:02 xragnar joined #perl6
13:05 brrt what, stage parse took 25.6s :-o
13:05 brrt when did that happen
13:05 FROGGS O.o
13:05 timotimo half a month ago?
13:06 FROGGS probably, yeah
13:06 FROGGS there was no speed boost over night :o)
13:08 brrt and clean spectests, too
13:08 brrt wow
13:08 FROGGS are you sure?
13:09 brrt clean-for-me
13:09 brrt i.e. some might be flappy but after a few minutes of buzzing the spectests said 'all test passed'
13:10 lizmat FWIW, I've seen a lot less flappage since the libuv upgrade
13:11 Ven joined #perl6
13:12 * Ven pulls to try locally
13:17 Ven stage parse, 34.64!
13:18 Ven and `make test` seems clean as well
13:18 Ven and the make bug is gone \o/ (people doing good and doing well)++
13:19 brrt what make bug?
13:20 Ven brrt: osx make segfault
13:20 FROGGS brrt: when expanded makefile lines were very long
13:20 brrt aha
13:20 FROGGS moritz++ fixed that together with someone I think
13:20 Ven I remember it :)
13:21 brrt did you know that os x whateveritscalled doesn't come with gdb anymore?
13:22 * brrt finds that he can't really enforce say() on DESTROY either
13:23 spider-mario joined #perl6
13:24 brrt do we have an actual working example of DESTROY?
13:27 BenGoldberg joined #perl6
13:29 SHODAN joined #perl6
13:35 grondilu left #perl6
13:41 xragnar joined #perl6
13:45 nine brrt seems to always quit right before I want to answer him
13:45 FROGGS he backlogs though
13:46 * colomon is curious how we spectest DESTROY
13:47 guru joined #perl6
13:47 JimmyZ joined #perl6
13:48 JimmyZ here is a way: https://gist.github.com/jnthn/dbe4e347cc092fa8cf62, me thingks :P
13:48 JimmyZ speaking of DESTROY
13:49 nine brrt: Inline::Perl5 is successfully using DESTROY. I verified that using a test script but am struggling to put this in a proper test. The one I have now runs indefinitely on a Rakudo without DESTROY support.
13:49 JimmyZ_ joined #perl6
13:50 nine JimmyZ_: "Make some junk to caues a GC" is not reliable though. It depends very much on subtleties and GC configuration.
13:51 itz is it S17* which are flappy?
13:52 Px12 joined #perl6
13:52 lizmat itz: in the past yes, but not so much lately
13:52 JimmyZ_ nine: yeah
13:52 JimmyZ_ Does force_gc cause a GC?
13:54 nine JimmyZ: I'd say it makes it more likely ;)
13:54 JimmyZ :)
13:55 dalek rakudo/nom: 0fbd5f3 | (Elizabeth Mattijsen)++ | src/core/Kernel.pm:
13:55 dalek rakudo/nom: Add Kernel.signal to get kernel signal values
13:55 dalek rakudo/nom:
13:55 dalek rakudo/nom: Either takes a string, or Signal Enum.  Please note that we cannot reliably
13:55 dalek rakudo/nom: map the Signal Enum to Kernel signal values, as these may (and do) differ per
13:55 dalek rakudo/nom: kernel.
13:55 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/0fbd5f3815
13:56 sqirrel_ joined #perl6
13:57 flussence I just went to update all of my rakudo stuff, and it hangs on panda/rebootstrap.pl :(
13:57 lizmat flussence: spectest went ok ?
13:58 flussence I'll give that a try, gimme 7 minutes or so :)
13:58 lizmat if that fails, I suggest rm -rf install plus a reconfigure from scratch
13:59 lizmat the libuv upgrade path was not smooth  :-(
13:59 flussence oh ok, gotcha.
13:59 FROGGS ohh, yes
13:59 flussence spectest appears to have hanged also...
13:59 FROGGS that could case that hang...
14:00 dalek specs: 3536408 | (Elizabeth Mattijsen)++ | S17-concurrency.pod:
14:00 dalek specs: Initial stab at speccing Proc::Async
14:00 dalek specs: review: https://github.com/perl6/specs/commit/3536408ac9
14:00 JimmyZ just do `make realcean` in MoarVM  :P
14:00 lizmat in my experience, realclean alone was not enough
14:00 lizmat I always had to nuke the install dir
14:05 alexghacker joined #perl6
14:07 dalek specs: b36c9eb | (Elizabeth Mattijsen)++ | S28-special-names.pod:
14:07 dalek specs: Spec Kernel.signal
14:07 dalek specs: review: https://github.com/perl6/specs/commit/b36c9eb10d
14:07 * woolfy1 fears all that nuking that lizmat does...
14:08 lizmat .oO( from orbit, no less :-)
14:08 woolfy1 o0O(the house is trembling and shaking)
14:11 alexghacker joined #perl6
14:11 rindolf Hi all. Sup?
14:12 tadzik hello rindolf
14:12 FROGGS rindolf: a pull request: https://bitbucket.org/shlomif/perl-xml-l​ibxml/issue/5/possible-typo-in-test-file
14:13 flussence lizmat: I did rm -r install/, git clean -dfx in rakudo/ & rakudo/nqp & rakudo/nqp/MoarVM, and it still hangs... I just remembered there's all that 3rdparty/ stuff that git-clean didn't catch so I've cleared *that* out and I'm trying again.
14:13 lizmat flussence++ for persevering
14:14 flussence not a big deal, it takes less time to rebuild all of rakudo these days than it used to to run hello world :)
14:14 FROGGS *g*
14:15 flussence yep, these spectests look a lot healthier now...
14:16 zakharyas joined #perl6
14:16 dalek rakudo/nom: b8e0136 | (Elizabeth Mattijsen)++ | src/core/Kernel.pm:
14:16 dalek rakudo/nom: Get Kernel.signal up to spec  :-)
14:16 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/b8e0136f7c
14:16 rindolf tadzik: what's up?
14:17 tadzik things are well :)
14:17 rindolf FROGGS: the tests pass.
14:18 FROGGS rindolf: that's not really an answer
14:19 rindolf FROGGS: well, does your pull req breaks them?
14:19 rindolf FROGGS: is it worth risking a "fix" and for atypo?
14:19 FROGGS rindolf: my pull request is a question about a possible typo in your test
14:20 FROGGS rindolf: so it is possible that you're testing the wrong hting
14:20 FROGGS thing*
14:20 FROGGS rindolf: ohh, s/pull request/issue/
14:21 rindolf FROGGS: ah.
14:28 pat_js joined #perl6
14:33 dalek roast: 665ee24 | (Elizabeth Mattijsen)++ | S32-str/lines.t:
14:33 dalek roast: Add tests for Str.lines(:eager)
14:33 dalek roast: review: https://github.com/perl6/roast/commit/665ee2487d
14:33 dalek roast: dd8a067 | (Elizabeth Mattijsen)++ | S02-magicals/KERNEL.t:
14:33 dalek roast: Add tests for Kernel.signals/signal
14:33 dalek roast: review: https://github.com/perl6/roast/commit/dd8a067453
14:38 dalek specs: 9103f96 | (Elizabeth Mattijsen)++ | S (2 files):
14:38 dalek specs: Also allow bare int's to be passed to .signal
14:38 dalek specs: review: https://github.com/perl6/specs/commit/9103f961d2
14:40 dalek rakudo/nom: 945d1ed | (Elizabeth Mattijsen)++ | src/core/Kernel.pm:
14:40 dalek rakudo/nom: Again, bring Kernel.signal up to spec
14:40 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/945d1ed12e
14:43 dalek roast: 4be29f3 | (Elizabeth Mattijsen)++ | S02-magicals/KERNEL.t:
14:43 dalek roast: Add Kernel.signal Int tests
14:43 dalek roast: review: https://github.com/perl6/roast/commit/4be29f3c95
14:44 kaare_ joined #perl6
14:48 MilkmanDan joined #perl6
14:55 xenoterracide joined #perl6
14:58 jaffa4 joined #perl6
15:07 jnthn lizmat++ # spec and test work
15:10 lizmat cycling&
15:10 jaffa4 bootstraping panda, cannot find shell command ...Could not find Shell::Command i
15:10 H2O1 joined #perl6
15:17 H2O1 left #perl6
15:20 alexghacker1 joined #perl6
15:21 rlilja joined #perl6
15:24 gtodd joined #perl6
15:24 mj41 joined #perl6
15:29 darutoko- joined #perl6
15:31 labster_ joined #perl6
15:31 alexghacker joined #perl6
15:32 chenryn_ joined #perl6
15:33 Alina-malina joined #perl6
15:44 Px12 joined #perl6
16:02 denis_boyun_ joined #perl6
16:15 pmurias joined #perl6
16:19 pmurias nine: the hanging test issue you could solve with a timeout
16:40 telex joined #perl6
16:43 ClarusCogitatio joined #perl6
16:46 xinming joined #perl6
16:55 jepeway joined #perl6
16:58 ggoebel1111117 joined #perl6
17:05 denis_boyun___ joined #perl6
17:16 nine pmurias: I know. But then the test wouldn't be 100 % reliable anymore. But I'm gonna do it anyway. No need to bother the users right at the start.
17:34 Ven joined #perl6
17:37 davido___ joined #perl6
17:38 BizarreCake joined #perl6
17:44 zakharyas joined #perl6
17:47 Sqirrel joined #perl6
17:48 slmult0 joined #perl6
17:48 dalek Inline-Perl5: 8cadf89 | nine++ | lib/Inline/Perl5.pm6:
17:48 dalek Inline-Perl5: Save even more method calls in destructor
17:48 dalek Inline-Perl5: review: https://github.com/niner/Inl​ine-Perl5/commit/8cadf89d87
17:48 dalek Inline-Perl5: 5c89aae | nine++ | p5helper.c:
17:48 dalek Inline-Perl5: Set perl_destruct_level to sequaky clean as suggested by perlembed
17:48 dalek Inline-Perl5: review: https://github.com/niner/Inl​ine-Perl5/commit/5c89aae9fe
17:48 dalek Inline-Perl5: cc683da | nine++ | / (11 files):
17:48 dalek Inline-Perl5: Rename Perl 5 interpreter variable in tests and examples to $p5
17:48 dalek Inline-Perl5:
17:48 dalek Inline-Perl5: Suggested by abraxxa++
17:48 dalek Inline-Perl5: review: https://github.com/niner/Inl​ine-Perl5/commit/cc683dae98
17:48 dalek Inline-Perl5: 4884c94 | nine++ | t/p6_object_destructor.t:
17:48 dalek Inline-Perl5: Add timeout to p6_object_destructor.t
17:48 dalek Inline-Perl5:
17:48 dalek Inline-Perl5: Destructors are only available in rakudo git. No need to punish
17:48 dalek Inline-Perl5: users trying to install the module.
17:48 dalek Inline-Perl5:
17:48 dalek Inline-Perl5: Suggested by pmurias++
17:48 dalek Inline-Perl5: review: https://github.com/niner/Inl​ine-Perl5/commit/4884c94d4b
18:12 guru joined #perl6
18:15 dwarring joined #perl6
18:19 Rotwang joined #perl6
18:25 timotimo hmm, so few individuals committed this week
18:26 timotimo apart from nines excellent work and liz's, the weekly might look a wee bit bleak
18:30 FROGGS :o(
18:31 FROGGS I implemented cglobals!
18:31 skids joined #perl6
18:31 FROGGS and did the libuv upgrade
18:34 lizmat FROGGS++  !!
18:35 lizmat and Travis support ?
18:36 raiph joined #perl6
18:37 lizmat and I think TimToady's reify fixes made the spectest run about 5% faster
18:37 pmurias cglobals?
18:38 FROGGS pmurias: C globals
18:38 pmurias for NativeCall?
18:38 FROGGS lizmat: and that travis stuff, aye
18:38 FROGGS pmurias: right
18:38 FROGGS it awaits jnthn's approval, but I think it is all right
18:39 vendethiel first of all: anything that uses nativecall is not nqp-js compitable
18:39 vendethiel s/ita/ati/
18:40 FROGGS vendethiel: why?
18:40 vendethiel FROGGS: ... because js things can't use c libs?
18:40 FROGGS is there no way to add that support?
18:40 vendethiel no
18:40 FROGGS :/
18:40 vendethiel not in a lifetime
18:41 vendethiel also, Inline::P5 is incompatible with js.
18:41 vendethiel so, let's be wary of doing everything that way :-)
18:41 FROGGS so the node ppl also have no FFI?
18:42 vendethiel FROGGS: there are the node ppl, and there are the browser ppl
18:42 FROGGS vendethiel: what about node then?
18:42 vendethiel FROGGS: node has node-ffi
18:42 FROGGS see :o)
18:42 vendethiel see what?
18:43 vendethiel why would you want to compile your perl6 to js ... To run it server-side?
18:43 vendethiel seems kinda counter-productive
18:43 FROGGS nqp-js will be able to support Native on node
18:43 FROGGS that's what it currently does
18:43 vendethiel but then why not only run it with rakudo-moar?
18:43 vendethiel to use some npm lib? that seems *far* less likely that using it on the client side.
18:43 FROGGS maybe you hava a node infrastructure? I dunno
18:44 vendethiel I dunno either. So, I don't think it's very useful
18:44 vendethiel I've seen it, people compiling scala to js to run it under node; but it seems like one hundredth of the use cases
18:45 FROGGS well, for *me* perl6-j is not useful either, but there seems to be some interest
18:45 vendethiel FROGGS: but the other option would be to code in java. Which you don't want to
18:45 vendethiel so tht's not thje same
18:45 ClarusCogitatio joined #perl6
18:45 FROGGS that might be true :o)
18:46 vendethiel what I'm saying is that, in more than 99% percents of the cases, you just run your code on moar rather than compile it to js to run it under node...
18:46 vendethiel "99% percents" right.
18:46 FROGGS yeah
18:46 _sri joined #perl6
18:47 rindolf FROGGS: i think i'll close your issue.
18:48 FROGGS rindolf: okay
18:48 rindolf FROGGS: sorry.
18:48 FROGGS rindolf: np
18:50 FROGGS rindolf: 'as per irc' - you've not really said why it is invalid :o)
18:55 rindolf FROGGS: yes, i'm being lazy.
18:55 rindolf Laziness, impatience and Hubris.
18:57 lizmat .oO( if rindolf ever complains of lack of feedback on his projects, we now know why )
18:58 rindolf lizmat: :-)
19:02 pmurias vendethiel: what started the Inline::P5/js discussion?
19:02 vendethiel pmurias: me being worried :)
19:02 pmurias about what?
19:02 Ven about people going full-C implems or full-P5 implems
19:02 dalek roast: 90fa2c0 | (Elizabeth Mattijsen)++ | S22-package-format/local.t:
19:02 dalek roast: Make sure we can precompile code with -use-
19:02 dalek roast: review: https://github.com/perl6/roast/commit/90fa2c0a9d
19:02 Ven and then making us unable to have multiple implems
19:03 pmurias you mean implementing core parts of Rakudo in C or P5?
19:04 Ven pmurias: not even core
19:05 pmurias re Native call on rakudo-js, node has ffi
19:05 pmurias and in the browser we have enscripten
19:05 pmurias * emscripten
19:06 Ven uh.. let's not :(
19:07 Ven seems heavy and unreliable to require llvm to emscripten of nativecalled library to just use them
19:07 BizarreCake joined #perl6
19:08 pmurias seems better then using an emscriptened perl5 interpreter to run a module ;)
19:08 Ven true :D.
19:08 * lizmat just realizes she specced stuff with underscores again and nobody complained
19:08 zakharyas joined #perl6
19:08 Ven but worse than just having a p6 version :)
19:09 dolmen joined #perl6
19:09 colomon lizmat: haven't noticed it yet, but you can consider this a complaint from me.  ;)
19:09 FROGGS lizmat: I hereby complain too! :P
19:10 flussence argh, some of my code's failing on 32-bit
19:19 dalek specs: a729e2b | (Elizabeth Mattijsen)++ | S17-concurrency.pod:
19:19 dalek specs: Don't spec new stuff with _ anymore
19:19 dalek specs: review: https://github.com/perl6/specs/commit/a729e2ba91
19:19 dalek specs: a6e8af2 | (Elizabeth Mattijsen)++ | S17-concurrency.pod:
19:19 dalek specs: Taking jnthn's ++ as an ok for positionals
19:19 dalek specs: review: https://github.com/perl6/specs/commit/a6e8af2b2a
19:24 lizmat FWIW, I consider using - instead of _ in identifiers as a part of eating your own dogfood
19:26 pmurias Unknown operator precedence specification "%comma" at line 28, near "$quoted~''
19:27 pmurias what does that mean?
19:27 pmurias caused by ./nqp-m --module-path gen/moar/stage2/ src/vm/js/QAST/Compiler.nqp
19:28 FROGGS that it cannot resolv %comma in method O
19:29 moritz pmurias: looks like something's wrong with the operator precedence parser, or with who an operator is set up
19:29 FROGGS hmmm, %comma is set up here: https://github.com/pmurias/nqp-js/b​lob/master/src/NQP/Grammar.nqp#L718
19:30 pmurias -e 'say(1+1)' trigger that too
19:30 dalek rakudo-star-daily: 46e5eb4 | coke++ | log/ (15 files):
19:30 dalek rakudo-star-daily: today (automated commit)
19:30 dalek rakudo-star-daily: review: https://github.com/coke/rakudo​-star-daily/commit/46e5eb40f2
19:30 dalek rakudo-star-daily: 2342a20 | coke++ | log/ (15 files):
19:30 dalek rakudo-star-daily: today (automated commit)
19:30 dalek rakudo-star-daily: review: https://github.com/coke/rakudo​-star-daily/commit/2342a20107
19:30 dalek perl6-roast-data: afedf31 | coke++ | / (6 files):
19:30 dalek perl6-roast-data: today (automated commit)
19:30 dalek perl6-roast-data: review: https://github.com/coke/perl6​-roast-data/commit/afedf31dcd
19:30 FROGGS pmurias: we are talking about nqp code, right?
19:30 pmurias * triggers
19:30 [Coke] niecza is borked on host08. rakudos seem to be ok.
19:30 pmurias ./nqp-m --module-path gen/moar/stage2/ -e 'say(1+1)'
19:31 FROGGS pmurias: is it possible that it does not run the INIT block in the link I posted?
19:31 FROGGS pmurias: you could test that by commenting line 705 and 721
19:34 [Coke] clean spectests? nope.
19:34 [Coke] (for brrt in backscroll)
19:35 FROGGS [Coke]: I was curious also
19:35 kjs_ joined #perl6
19:36 [Coke] rakudo.jvm is failing 7677 tests on host08, which could be a host08 thing.
19:36 [Coke] moar is unaffacted by the move. parrot is failing 331 tests today up from 271
19:36 dalek specs: 4df4368 | (Elizabeth Mattijsen)++ | S17-concurrency.pod:
19:36 dalek specs: Spec Str.IO.watch -> IO::Notification.watch_path
19:36 dalek specs: review: https://github.com/perl6/specs/commit/4df4368e50
19:37 dalek nqp-js: d08a287 | (Pawel Murias)++ | / (9 files):
19:37 dalek nqp-js: Migrate from parrot to moar.
19:37 dalek nqp-js: review: https://github.com/pmurias​/nqp-js/commit/d08a287135
19:37 FROGGS lizmat: watch_path or watch-path?
19:37 lizmat well, it has been implemented as watch_path for a while now
19:37 pmurias FROGGS: it seems that some strange things were happening and just tweaking things made them disappear
19:37 lizmat so we will need a deprecation cycle for that
19:38 pmurias moarvm++ # make js-test runs much much faster on moar!
19:38 lizmat but the last time I changed things like that in the spec, I upset jnthn quite a lot, so I'm reluctant to go back on that
19:38 lizmat most people will be using the shortcut anyway
19:38 nine So we have a healthy mix of - and _ in method names?
19:38 lizmat unhealthy, I would say  :-(
19:39 pmurias we have to decide on one or the other
19:39 lizmat I hope we can get clarity at the Perl 6 hackathon after the APW in a few weeks
19:40 lizmat at least we have a deprecation path that can be easily put in place
19:40 nine I really can't see the advantage of even allowing - in names anyway. We'll see foobar, foo_bar, foo-bar and fooBar in real life code. Probably all of them in larger code bases.
19:40 Ven nine: - ftw :)
19:40 lizmat nine: no need to be inconsistent from the beginning  :-)
19:42 [Coke] niecza says: Got a SIGABRT while executing native code. This usually indicates
19:42 [Coke] a fatal error in the mono runtime or one of the native libraries
19:42 [Coke] used by your application.
19:42 dalek rakudo/nom: 14cb952 | (Elizabeth Mattijsen)++ | / (3 files):
19:42 dalek rakudo/nom: Add Str.IO.watch, remove need for code stubs
19:42 dalek rakudo/nom:
19:42 dalek rakudo/nom: This will probably need a reconfigure to work.
19:42 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/14cb9528c7
19:42 [Coke] diakopter: what's the difference between host07 & host08?
19:42 nine lizmat: it's only the beginning if you're thinking of green field development with no existing libraries. I'm more thinking of existing code bases. So we'll have at least foo-bar from Perl 6 core, foo_bar from your existing code base if you stuck to Perl Best Practices, foobar because it's inevitable and fooBar if part of your code touches standard interfaces like the DOM.
19:43 Ven nine: no, no. We'll just get stuff built on top of the DOM ;)
19:43 Ven not raw calls :p
19:43 nine sure
19:43 lizmat if P5 had had '-' in identifiers, I'm pretty sure it would have been "-" instead of "_"
19:43 lizmat in PBP
19:44 [Coke] hurm. mono is gone from 2.X to 3.X, bet that's most of it. :|
19:44 nine lizmat: well it didn't. And now we have millions of lines of code using "_" and fingers trained to type it. And quite frankly I'm not even sure if I like "-" more than "_".
19:44 [Coke] seen sorear?
19:44 [Coke] .seen sorear?
19:44 yoleaux I haven't seen sorear? around.
19:44 [Coke] .seen sorear
19:44 yoleaux I haven't seen sorear around.
19:45 nine PBP's argument for "_" is that is looks almost like the space character that would be there in plain English. What's the argument for "-"?
19:45 lue nine: lowercase _ , at least on my US keyboard :) .
19:45 Ven nine: it's easier to type (and easier for me to read)
19:45 pmurias nine: it-is-more-pretty-than-_-and-easier-to-type
19:45 Ven (easier to type both on qwerty and on azerty)
19:46 FROGGS and on qwertz
19:46 dalek specs: c00c203 | (Elizabeth Mattijsen)++ | S99-glossary.pod:
19:46 dalek specs: Add PBP lemma
19:46 dalek specs: review: https://github.com/perl6/specs/commit/c00c2033ec
19:47 lizmat FROGGS: so you're saying we shouldn't use "-" ?
19:47 lue nine: fwiw, I used to not think much of - as a character for names, but I've grown to like it.
19:47 ClarusCogitatio joined #perl6
19:48 FROGGS lizmat: yes, I'd vote for - instead of _ in methods and subs, but _ in all caps constants
19:48 nine Beauty is in the eye of the beholder. And for some perceived beauty we get an even worse mix of naming conventions.
19:49 FROGGS there is only one way to mi it :o)
19:49 lue to me, foo-bar looks nicer than foo_bar (though not by much), and FOO_BAR looks better than FOO-BAR (by a fair bit).
19:49 dwarring er ++ "-"
19:49 lue (it also lets me not have to change whether or not I'm pressing down the shift key while typing a name :P)
19:50 pmurias we could enforce a naming convention but it's seem a bit unperlish (and usefull)
19:50 nine Nice arguments for a completely new language with no existing code. But I'd really like to change the latter. And then it will just worsen the mixed conventions problem. And that in a language with lousy IDE support.
19:52 [Coke] anyone else getting this:
19:52 nine We all like to laugh about PHP's mix of conventions in it's 10000s of builtin functions. But in Perl we already have the same because these functions are in different modules by different authors. And now we're making it even worse.
19:52 [Coke] S05-transliteration/trans.rakudo.moar aborted 56 test(s)
19:53 lue nine: different people doing things differently is unavoidable and, in my opinion, not worth the effort to prevent.
19:53 nine This really feels like the first change of the language that's not been thought through entirly.
19:53 nine And I say that even though I actually _like_ the possibility to use - in identifiers. I just fear it will cost more than it gains us.
19:54 skids r: use Test; for (0 .. 20) -> $e { given (1) { is 1, 0, "$e"; } } # SEGV on rakudo-m.  Also other weird (IO/timing related) errors with Test, but only when 'given' used.
19:54 camelia rakudo-moar 945d1e: OUTPUT«(signal )not ok 1 - 0␤␤# Failed test '0'␤# at /tmp/tmpfile line 1␤# expected: '0'␤#      got: '1'␤not ok 2 - 1␤␤# Failed test '1'␤# at /tmp/tmpfile line 1␤# expected: '0'␤#      got: '1'␤not ok 3 - 2␤␤# Failed test '2'␤# at /t…»
19:54 camelia ..rakudo-parrot 945d1e: OUTPUT«not ok 1 - 0␤␤# Failed test '0'␤# at /tmp/tmpfile line 1␤# expected: '0'␤#      got: '1'␤not ok 2 - 1␤␤# Failed test '1'␤# at /tmp/tmpfile line 1␤# expected: '0'␤#      got: '1'␤not ok 3 - 2␤␤# Failed test '2'␤# at /tmp/tmpf…»
19:54 camelia ..rakudo-jvm 945d1e: OUTPUT«Can't call method "syswrite" on an undefined value at /home/p6eval/jvm-rakudo/eval-client.pl line 32.␤»
19:55 lue - has been a part of identifiers for a long time now (as long as I've been using Perl 6, at least), the only thing changing is more consistency in the spelling of names in the core.
19:55 flussence I've been wanting to ask something related for a while: I've seen module titles in the ecosystem that use «::» and ones that use «-», and can't decide which is "better".
19:55 lizmat flussence: that is another issue
19:56 lizmat :: implies a hierarchy in packages
19:56 flussence I'm of half a mind to change mine to that because the vast majority use it already...
19:57 lue I think - instead of :: happens in situations where a colon is either disallowed or discouraged as part of a name.
20:02 colomon lue: module names can be any old string, so far as I know
20:02 dalek rakudo/nom: b41dda9 | (Elizabeth Mattijsen)++ | src/core/Proc/Async.pm:
20:02 dalek rakudo/nom: Fix copynpasto/thinko in chars/bytes logic
20:02 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/b41dda9b6a
20:03 lue colomon: yeah, I was thinking more of filesystems and such.
20:03 colomon flussence: yes, I think you should probably change yours
20:04 colomon the oddballs are Perl6-Math-Quaternion Test-ClientServer Test-Corpus Text-Levenshtein Text-Tabs-Wrap io-prompt
20:05 kurahaupo joined #perl6
20:06 flussence Text-Tabs-Wrap is the way it is for hysterical raisins, the others I have are easily fixed though, I'll change those
20:07 xragnar joined #perl6
20:08 FROGGS can somebody put me the output of this in a gist? perl6 -e '"a" ~~ /(a)/; say $/.DUMP'
20:10 brrt joined #perl6
20:10 colomon FROGGS: on it
20:11 FROGGS thank you :o)
20:11 colomon https://gist.github.com/co​lomon/c0c2aec35685b2fbafcd # does that look right?
20:11 colomon my rakudo may be lagging by a few days
20:11 FROGGS yes, it does :o)
20:11 FROGGS that doesnt matter
20:12 dalek specs: f175d77 | (Elizabeth Mattijsen)++ | S17-concurrency.pod:
20:12 dalek specs: Allow :%ENV to be passed to Proc::Async.new
20:12 dalek specs:
20:12 dalek specs: Seems useful to be able to run a process "cleanly" or with whatever environment
20:12 dalek specs: you want, without having to set up a lexical %*ENV.
20:12 dalek specs: review: https://github.com/perl6/specs/commit/f175d775bb
20:12 FROGGS it is just that I am hacking on that right now
20:12 colomon FROGGS: glad to help
20:12 * colomon 's server is down because the battery backup went kablooey
20:14 * colomon didn't mean to be random with that last comment, but the server is usually where he builds the latest rakudo first; on his laptop, builds risk messing with stuff I am using for $work.
20:15 FROGGS ahh
20:32 xenoterracide joined #perl6
20:51 nine Oh and how on earth is one supposed to call a perl6-method from Perl 5 code?
20:52 nine This is going to be ugly...
20:52 lizmat nine: I wouldn't mind if we're not going there
20:52 lizmat this is about providing access to Perl 5 libraries from Perl 6
20:52 lizmat not the other way around
20:54 nine if one is trying to upgrade a larfe codebase piece by piece, the other way round is going to be just as important
20:55 nine And upgrading is the only way I can see me using P6 anytime soon. Otherwise I'd have been using it for years already...
20:56 timotimo pass callables to p5 from p6 code?
20:57 dalek rakudo/nom: ef8387f | (Elizabeth Mattijsen)++ | src/core/Proc/Async.pm:
20:57 dalek rakudo/nom: Streamline Supply generating code
20:57 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/ef8387f002
20:58 nine This little cosmetic change may bring quite a lot of ugliness with it. I just hope thst I'm going to be wrong about this.
20:59 lizmat nine: what little cosmetic change are you talking about ?
20:59 nine - as new standard in identifyers
21:00 lizmat ugliness as in broken code ?
21:00 lizmat if so, that's why we can deprecate
21:01 nine As in workarounfs needed for something as simple as a method call
21:03 nine Well maybe I'll see it in a more positive light in the morning. good night!
21:03 lizmat good night, nine!
21:03 lizmat thanks for all the Inline::Perl5 work!
21:07 timotimo yup
21:07 dalek rakudo/nom: 2a0c0dd | (Elizabeth Mattijsen)++ | src/core/Proc/Async.pm:
21:07 dalek rakudo/nom: Implement :%ENV for Proc::Async.start, as per spec
21:07 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/2a0c0ddfbf
21:25 dalek nqp: fe64f00 | (Elizabeth Mattijsen)++ | docs/ops.markdown:
21:25 dalek nqp: Add doc stub for nqp::spawnprocasync
21:25 dalek nqp: review: https://github.com/perl6/nqp/commit/fe64f00719
21:25 * lizmat hopes jnthn will fill in the stub
21:29 lizmat m: my $a=Proc::Async.new(:path<echo>,:args<$SHELL>); $a.stdout_chars.tap(&say); await $a.start  # expected something else then $SHELL
21:29 camelia rakudo-moar b41dda: OUTPUT«$SHELL␤␤»
21:30 lizmat hmmm... should I even be able to do this on Camelia?
21:30 timotimo lizmat: Proc::Async is supposed to go through the shell?
21:30 lizmat not sure now
21:30 timotimo and no, it should probably be disabled in the restricted setting ... which is weaksauce anyway
21:31 timotimo i'd actually love to bounce ideas back and forth with someone about building a secure execution thingie based on spesh
21:31 eternaleye joined #perl6
21:32 cognome r: say $*KERNEL.WHAT
21:32 camelia rakudo-moar 2a0c0d: OUTPUT«Cannot look up attributes in a type object␤  in method name at src/gen/m-CORE.setting:21480␤  in method gist at src/gen/m-CORE.setting:21333␤  in sub say at src/gen/m-CORE.setting:15106␤  in block <unit> at /tmp/tmpfile:1␤␤»
21:32 camelia ..rakudo-jvm 2a0c0d: OUTPUT«Can't call method "syswrite" on an undefined value at /home/p6eval/jvm-rakudo/eval-client.pl line 32.␤»
21:32 camelia ..rakudo-parrot 2a0c0d: OUTPUT«Cannot look up attributes in a type object␤  in method name at gen/parrot/CORE.setting:19117␤  in method gist at gen/parrot/CORE.setting:18970␤  in sub say at gen/parrot/CORE.setting:14981␤  in sub say at gen/parrot/CORE.setting:14977␤  in bloc…»
21:32 lizmat cognome: that's odd
21:32 cognome I am odd too
21:32 cognome ... and old
21:34 dalek rakudo/nom: fbef9c1 | (Elizabeth Mattijsen)++ | src/RESTRICTED.setting:
21:34 dalek rakudo/nom: Make Proc::Async not work in restricted setting
21:34 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/fbef9c127a
21:35 raiph joined #perl6
21:37 dalek rakudo/nom: f882052 | (Elizabeth Mattijsen)++ | src/core/Proc/Async.pm:
21:37 dalek rakudo/nom: Better default handling for Proc::Async's :ENV
21:37 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/f882052c98
21:37 dalek rakudo/nom: 8dd2409 | (Elizabeth Mattijsen)++ | src/core/Proc/Async.pm:
21:37 dalek rakudo/nom: say is a print, why not in the async case as well?
21:37 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/8dd2409a14
21:38 lizmat timotimo: does tainting ring a bell with you ?
21:38 timotimo yes
21:38 timotimo the pypy people had a tainting objectspace for some time
21:38 lizmat P5 has tainting
21:38 timotimo oh?
21:38 lizmat although not many people use it in production
21:38 timotimo based on TIE? :)
21:39 lizmat nope
21:40 timotimo oh, it's internal to core p5?
21:40 timotimo as in ... implemented in C?
21:40 lizmat yes
21:40 lizmat https://metacpan.org/pod/distr​ibution/PodSimplify/newrun.pod, look for -t and -T
21:40 lizmat oops
21:41 lizmat https://metacpan.org/pod/dist​ribution/perl/pod/perlrun.pod   # proper URL
21:42 lizmat it's also still specced in S19:122
21:42 synopsebot Link: http://perlcabal.org/syn/S19.html#line_122
21:44 timotimo hm.
21:44 timotimo not sure if it's exactly what i'm looking for
21:45 lizmat well, it's part of what you want, I think
21:46 timotimo maybe
21:46 timotimo i thought along the lines of not allowing a user to specify nqp:: ops, but "indirect" calling would be okay, but that doesn't sound so sensible any more ...
21:48 cognome joined #perl6
21:55 lizmat m: Systemic.new.WHAT   # cognome's bug golfed down
21:55 camelia rakudo-moar 2a0c0d: OUTPUT«P6opaque: no such attribute '$!name'␤  in submethod BUILD at src/gen/m-CORE.setting:21328␤  in method BUILDALL at src/gen/m-CORE.setting:954␤  in method bless at src/gen/m-CORE.setting:937␤  in method new at src/gen/m-CORE.setting:921␤  in any  a…»
21:55 lizmat p: Systemic.new.WHAT   # cognome's bug golfed down
21:55 camelia rakudo-parrot 2a0c0d: ( no output )
21:56 lizmat p: say Systemic.new.WHAT
21:56 camelia rakudo-parrot 2a0c0d: OUTPUT«Cannot look up attributes in a type object␤  in method name at gen/parrot/CORE.setting:3194␤  in method gist at gen/parrot/CORE.setting:18970␤  in sub say at gen/parrot/CORE.setting:14981␤  in sub say at gen/parrot/CORE.setting:14977␤  in block…»
21:56 lizmat huh?  only if I do a say() ?
21:57 sjn \o
21:57 lizmat sjn: /o
21:57 * sjn just came for a Java conference, having spent a LOT of time talking about Perl6 :D
21:58 timotimo cool :)
21:58 sjn *tired*
21:59 sjn managed to shoot down almost all the critique though, and (hopefully) convinced at least a couple people that Perl6 is worth investigating today
21:59 lizmat sjn++
21:59 sjn Also, came to think of a new "pitch"
21:59 sjn well. "angle" for promoting Perl6
22:02 sjn "Year after year, there's been cool stuff presented about Perl6 at conferences all over the place. THAT STUFF IS STILL COOL; And now it also has all the new cool stuff since then, to *multiply* with it's coolness" </hype>
22:02 cognome lizmat, to thanks me for giving you work, you can do even more work and merge my pull request :)    https://github.com/rakudo/rakudo/pull/279
22:04 sjn maybe a bit over the top, but at least I got some of the guys raise their eyebrows :)
22:05 dalek rakudo/nom: 8cb5647 | (Stéphane Payrard)++ | src/core/Grammar.pm:
22:05 dalek rakudo/nom: Grammar.parse :args support
22:05 dalek rakudo/nom:
22:05 dalek rakudo/nom: See https://github.com/perl6/specs/commit/67579a1cf7
22:05 dalek rakudo/nom: https://github.com/perl6/roast/commit/1bc8a5fee6
22:05 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/8cb5647693
22:05 dalek rakudo/nom: 351c122 | lizmat++ | src/core/Grammar.pm:
22:05 dalek rakudo/nom: Merge pull request #279 from cognominal/patch-1
22:05 dalek rakudo/nom:
22:05 dalek rakudo/nom: Grammar.parse :args support
22:05 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/351c122531
22:10 lizmat cognome: alas, no passing TODO tests  :-(
22:10 * lizmat has committed enough for today
22:10 lizmat so sleep&
22:10 cognome lizmat: ???
22:11 lizmat cognome: I was hoping the new :args feature would have had TODO tests in roast
22:11 cognome I think it is in fudged tests.
22:11 cognome my bad.
22:11 lizmat you mean, as skipped?  then it is time to unskip them  :-)
22:11 lizmat good night all!
22:11 cognome yes, skipped
22:12 cognome g'nigh
22:13 cognome night
22:23 bjz joined #perl6
22:25 timotimo gnite lizmat!
22:38 exixt joined #perl6
22:39 araujo joined #perl6
22:39 araujo joined #perl6
22:41 skids left #perl6
22:55 chenryn joined #perl6
22:58 SamuraiJack joined #perl6
23:29 gtodd joined #perl6
23:29 xiaomiao joined #perl6
23:35 ingy anyone know where fglock++ hangs out on irc?
23:37 BenGoldberg joined #perl6
23:39 nebuchadnezzar joined #perl6
23:44 tadzik internal booking channels, perhaps :P
23:58 ingy joined #perl6
23:59 [Coke] .seen fglock
23:59 yoleaux I saw fglock 23 May 2014 13:45Z in #perl6: <fglock> something like   x( :scalar )  I think

| Channels | #perl6 index | Today | | Search | Google Search | Plain-Text | summary

Perl 6 | Reference Documentation | Rakudo