Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6, 2014-05-12

Perl 6 | Reference Documentation | Rakudo

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

All times shown according to UTC.

Time Nick Message
00:18 rurban joined #perl6
00:40 robinsmidsrod joined #perl6
00:43 [particle] joined #perl6
00:43 awwaiid joined #perl6
00:45 BenGoldberg joined #perl6
00:49 [particle]1 joined #perl6
00:50 daniel-s__ joined #perl6
00:51 arnsholt_ joined #perl6
00:51 renormalist joined #perl6
00:51 yakudzo joined #perl6
00:51 thou joined #perl6
00:52 molaf_ joined #perl6
00:52 sunnavy joined #perl6
01:00 btyler joined #perl6
01:02 daniel-s__ joined #perl6
01:03 dayangkun joined #perl6
01:08 Su-Shee joined #perl6
01:08 telex joined #perl6
01:10 FROGGS_ joined #perl6
01:17 klapperl_ joined #perl6
01:19 woosley joined #perl6
01:21 rurban joined #perl6
01:24 daniel-s__ joined #perl6
01:28 dayangkun joined #perl6
01:29 _daniel-s__ joined #perl6
01:30 Sqirrel joined #perl6
01:35 _daniel-s__ joined #perl6
01:41 dayangkun joined #perl6
01:41 daniel-s joined #perl6
01:51 Gruber joined #perl6
01:51 ssutch_ joined #perl6
01:51 labster_ joined #perl6
01:51 mls_ joined #perl6
01:51 TimToady joined #perl6
01:51 Khisanth joined #perl6
01:51 jercos joined #perl6
01:51 simcop2387 joined #perl6
01:52 krunen joined #perl6
01:52 nwc10 joined #perl6
01:52 chenryn joined #perl6
01:52 Maddingue joined #perl6
01:52 japhb_ joined #perl6
01:52 jtpalmer_ joined #perl6
01:52 lee_ joined #perl6
01:52 huf joined #perl6
01:55 bowtie joined #perl6
01:55 grep0r joined #perl6
01:56 nhayashi joined #perl6
02:00 eternaleye joined #perl6
02:03 dayangkun joined #perl6
02:03 daxim joined #perl6
02:03 cotto joined #perl6
02:04 go|dfish joined #perl6
02:07 xenoterracide joined #perl6
02:08 araujo joined #perl6
02:08 araujo joined #perl6
02:12 daniel-s_ joined #perl6
02:23 rurban joined #perl6
02:25 lue on latest perl6-m, the test for panda all say "No subtests run"
02:26 sorear joined #perl6
03:08 aindilis joined #perl6
03:13 rurban joined #perl6
03:40 xragnar joined #perl6
03:43 grep0r joined #perl6
03:46 rurban joined #perl6
03:49 rurban1 joined #perl6
03:50 hoverboard joined #perl6
03:51 thou joined #perl6
03:55 ssutch joined #perl6
03:56 yoleaux joined #perl6
04:02 kaare_ joined #perl6
04:03 grep0r joined #perl6
04:18 hoelzro joined #perl6
04:19 atrodo joined #perl6
04:23 daniel-s_ joined #perl6
04:28 SevenWolf joined #perl6
04:28 TimToady p6: say (1, 2, 3, 4) ~~ 3..10; say (1, 2, 3, 4, 5) ~~ 1..4;
04:28 daniel-s__ joined #perl6
04:29 camelia rakudo-{parrot,jvm,moar} b8567c: OUTPUT«False␤True␤»
04:29 camelia ..niecza v24-109-g48a8de3: OUTPUT«False␤False␤»
04:29 TimToady I think those should be True False
04:30 TimToady numeric range on the right should be testing range against +LHS
04:30 anaeem1_ joined #perl6
04:32 TimToady I guess there's a bit more dwimmery in the table than that, hmm...
04:33 TimToady Any       Range     in generic range        [!after] X.min,$_,X.max (etc.)
04:34 TimToady not sure I buy that, looks like a refugee from symmetrical smartmatch
04:36 TimToady oh, it's X.min, which is the range min, not the array min
04:36 TimToady in fact, by that rule it would fail, since integers are not commensurate with parcels
04:36 TimToady but I'd argue that a numeric range should coerce the left to numeric
04:37 TimToady m: say (1,2,3,4,5) cmp 42
04:37 camelia rakudo-moar b8567c: OUTPUT«Less␤»
04:37 TimToady m: say (1,2,3,4,5) cmp 4
04:37 camelia rakudo-moar b8567c: OUTPUT«Less␤»
04:38 TimToady hmm, looks like it's doing element-by-element comparison, treating the right as a single element
04:38 TimToady m: say (1,2,3,4,5) ~~ 5
04:38 camelia rakudo-moar b8567c: OUTPUT«True␤»
04:38 TimToady m: say (1,2,3,4,5) ~~ 5..5
04:38 camelia rakudo-moar b8567c: OUTPUT«False␤»
04:38 TimToady that seems completely bogus to me
04:43 TimToady m: say +(1,2,3,4,5) ~~ 5..5
04:43 camelia rakudo-moar b8567c: OUTPUT«True␤»
04:44 TimToady that's the semantics it oughta have
04:45 TimToady not turn the right side into a list
04:50 rurban joined #perl6
05:22 kaare_ joined #perl6
05:23 raiph left #perl6
05:23 raiph joined #perl6
05:27 cooper joined #perl6
05:30 SamuraiJack_ joined #perl6
05:48 dalek specs: 6672de4 | TimToady++ | S03-operators.pod:
05:48 dalek specs: range smartmatch coerces when real or stringy
05:48 dalek specs: review: https://github.com/perl6/specs/commit/6672de406d
05:49 TimToady see also S03:3500
05:49 synopsebot Link: http://perlcabal.org/syn/S03.html#line_3500
05:52 rurban joined #perl6
06:06 jlaire joined #perl6
06:11 daniel-s__ joined #perl6
06:25 dayangkun joined #perl6
06:26 darutoko joined #perl6
06:36 sergot morning o/
06:42 FROGGS joined #perl6
06:45 zakharyas joined #perl6
06:49 kaleem joined #perl6
06:52 jnthn morning o/
06:52 FROGGS morning
06:55 rurban joined #perl6
06:57 igorsutton__ joined #perl6
07:00 yakudza joined #perl6
07:14 dmol joined #perl6
07:16 Alina-malina joined #perl6
07:19 darutoko- joined #perl6
07:29 daniel-s__ joined #perl6
07:34 berekuk joined #perl6
07:35 daniel-s__ joined #perl6
07:44 hummeleB1 joined #perl6
07:58 rurban joined #perl6
08:12 _daniel-s__ joined #perl6
08:13 scumdog joined #perl6
08:18 _daniel-s__ joined #perl6
08:24 FROGGS m: say $*OS
08:24 camelia rakudo-moar b8567c: OUTPUT«linux␤Saw 1 call to deprecated code during execution.␤================================================================================␤$*OS called at:␤  /tmp/QDDWMAQIHr, line 1␤Please use $*DISTRO.OS instead.␤-----------------------------------…»
08:24 FROGGS m: say $*DISTRO.OS
08:24 camelia rakudo-moar b8567c: OUTPUT«===SORRY!===␤Dynamic variable $*DISTRO not found␤»
08:24 FROGGS .tell lizmat does it make sense to point somebody from $*OS to $*DISTRO.OS when we do not yet have $*DISTRO?
08:24 yoleaux FROGGS: I'll pass your message to lizmat.
08:25 lizmat good *, #perl 6
08:25 yoleaux 08:24Z <FROGGS> lizmat: does it make sense to point somebody from $*OS to $*DISTRO.OS when we do not yet have $*DISTRO?
08:25 FROGGS morning lizmat :o)
08:25 lizmat wel, maybe not  :-)
08:25 lizmat it was more a proof of concept of deprecation of a *variable* rather than a sub or method
08:26 FROGGS we should at least wire $*DISTRO.OS up soonish
08:26 lizmat yes, that was my plan for today
08:27 FROGGS lizmat++
08:27 FROGGS my plan for today is a very nice one
08:27 FROGGS can't wait for $dayjob being done
08:27 scumdog left #perl6
08:32 lizmat FROGGS++   # as a measure of support
08:43 dakkar joined #perl6
08:59 rurban joined #perl6
09:03 igorsutton__ joined #perl6
09:06 Woodi joined #perl6
09:18 pecastro joined #perl6
09:23 SamuraiJack__ joined #perl6
09:28 thou joined #perl6
09:29 timotimo http://p6weekly.wordpress.com/?p=160&amp;shareadraft=5370947573439 - feel free to point out anything i have missed
09:30 timotimo FROGGS: since last week, how much did loop-labels move forward?
09:31 nwc10 timotimo: Linux or linux? :-)
09:31 timotimo i think i'll just remove mention of linux from there
09:31 timotimo because ... what other operating system runs on those architectures anyway? :P
09:32 nwc10 timotimo: you could say "tested on Linux"
09:32 timotimo that being said, i wonder if we can run on haiku and beos
09:32 nwc10 I think that that would be helpful and accudate
09:32 nwc10 accurate
09:32 nwc10 like I can type
09:32 timotimo i mention it once at the end now
09:32 timotimo instead of three times
09:32 nwc10 better than my suggestion
09:33 lizmat timotimo: maybe mention dwarring adding more tests from advent posts ?
09:33 timotimo ah, of course!
09:36 lizmat I also fixed filtering of array/hash slices
09:36 lizmat non-existing keys would return an empty list, instead of nothing
09:36 timotimo ah, right
09:36 FROGGS timotimo: AFAICS perl6-p passes its tests, perl6-m and perl6-j need a little adjusted (right now I store the Label's mem address, which is bogus when the Label moves in GC)
09:37 timotimo ah
09:37 timotimo yes, that's not a good idea there :)
09:37 FROGGS so I need to keep the label object itself, but that should be pretty easy
09:37 timotimo aye
09:37 timotimo after that everything should work fine? :D
09:38 FROGGS MONDAY: I expect that we have labels merged until next MONDAY
09:38 FROGGS yes, I need to put in more tests though :o)
09:39 FROGGS cpan will also get a lot of attention this week, I hope that I have the PR about the indexer ready until PLPW
09:39 FROGGS so, weekend
09:39 timotimo ah, plpw is this weekend
09:39 timotimo i should probably point that out
09:39 FROGGS yeah
09:40 lizmat timotimo: and the CZPW is two days *after* PLPW
09:40 lizmat also the *first* CZPW
09:41 timotimo oh, nice
09:41 timotimo there'll be perl6 stuff on czpw, too?
09:41 lizmat jnthn, lichtkind and lizmat will be giving presentations, afaik
09:42 timotimo ah, nice
09:42 timotimo who's going to present at the plpw except tadzik?
09:42 FROGGS timotimo: http://act.yapc.eu/plpw2014/schedule
09:42 FROGGS timotimo++ # great blög
09:42 lizmat and masak
09:44 lizmat at PLPW: FROGGS, masak, jnthn, lizmat, woolfy
09:45 64MAANX81 joined #perl6
09:46 timotimo found the right links to look at
09:47 timotimo http://p6weekly.wordpress.com/?p=160&amp;shareadraft=537098b18ceae - new draft
09:48 _daniel-s__ joined #perl6
09:49 yakudza joined #perl6
09:49 nwc10 timotimo: s/monday/Monday/
09:50 lizmat did we add http://brrt-to-the-future.blogspot.de already to sixplanet ?
09:55 virtualsue joined #perl6
10:03 rurban joined #perl6
10:06 sergot timotimo: tadzik will give a talk during plpw too :)
10:06 tadzik yep, but the question was "except tadzik" :P
10:08 sergot oh, right, why so? :)
10:09 tadzik I assume timo knew about me already :P
10:09 tadzik but I don't seem to be included in the blogpost
10:09 sergot that's why I wrote about you :P
10:10 tadzik :)
10:28 sergot lizmat: regarding to our Friday hackathon, can we invite some "new" people?
10:29 lizmat depends on what they expect
10:29 lizmat if they expect to be helped with installing Perl 6
10:30 lizmat not so much, if they already have Perl 6 installed, and have tried to write code already
10:30 lizmat I think more so  :-)
10:30 sergot ok, great :)
10:31 lizmat the Perl 6 hackathon in Kiev sort of went awry in the sense that there were too many newbies
10:31 lizmat maybe we need a separate "upgrade to Perl 6" event in the future
10:32 lizmat raiph++ moritz++ for not giving up on perlmonk
10:32 lizmat s
10:33 apejens trying to install rakudo-star via homebrew failed: https://gist.github.com/omega/a2362e8e9156d3c17dad
10:35 nwc10 lizmat: agree with your assessment of Kiev's hackathon
10:38 sergot lizmat: ok, so we have place, we have people, we have everything, looking forward. :)
10:38 * lizmat as well
10:38 FROGGS apejens: do you have an nqp-j in PATH by any chance?
10:39 apejens FROGGS: nope
10:43 ssutch joined #perl6
10:45 FROGGS apejens: is there a way to tell it to build rakudo-star-2014.04 by any chance?
10:46 apejens FROGGS: well, I can edit the brew I guess and try :P I moved on to my next brew install, but will try once that is done :p
10:47 FROGGS k
10:48 FROGGS I never tried brew but I know that the chance to build for jvm backend is better for 2014.04
10:48 FROGGS apejens: ahh, what is your java version btw?
10:49 apejens 1.7.0_25 it says
10:49 apejens I guess I can skip the jvm backend as well :P
10:54 apejens trying 2014.04 now
10:58 apejens FROGGS: different error at least :P make: *** [gen/jvm/stage1/QAST.jar] Error 1
10:59 apejens forgot to run with verbose
10:59 FROGGS hmmm :/
10:59 FROGGS build for the MoarVM backend, if you don't deal with large texts or with async servers...
11:00 apejens I don't really deal with anything, was gonna try the games from https://github.com/tadzik/steroids
11:01 tadzik ha. Then you have to go moarvm :)
11:01 FROGGS apejens: then you probably want MoarVM, as it is the fastest for this
11:01 tadzik parrot may be fast enough, if you have a CPU from 2020 or so
11:01 tadzik and jvm has a dealbreaking nativecall bug
11:01 apejens ok :P
11:02 apejens it runs out of heap space btw, thats what breaks building jvm
11:02 tadzik ouch
11:02 tadzik argh, chromium on aura is so horrible
11:09 ssdfsdf joined #perl6
11:12 apejens seg fault this time
11:13 apejens ahh well, office aircon is off, time to head home
11:17 tadzik oh, segfault?
11:32 prevost joined #perl6
11:34 timotimo added tadzik and can now publish, i think
11:36 Colby` joined #perl6
11:50 kurahaupo joined #perl6
11:55 * lizmat just did a spectest with parrot: Files=816, Tests=30831, 430 wallclock secs ( 9.53 usr  4.10 sys + 2955.84 cusr 245.47 csys = 3214.94 CPU)
11:55 lizmat twice as long as Moar, twice as much CPU and 1800 tests less
11:57 dalek rakudo/nom: f26787c | (Elizabeth Mattijsen)++ | src/core/Deprecations.pm:
11:57 dalek rakudo/nom: Make it possible to obsolete variables elsewhere
11:57 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/f26787ce50
11:57 dalek rakudo/nom: c1bd713 | (Elizabeth Mattijsen)++ | README.md:
11:57 dalek rakudo/nom: Show building with Moar as first option
11:57 dalek rakudo/nom:
11:57 dalek rakudo/nom: Since we now have a Rakudo* that supports Moar, I think it is time.
11:57 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/c1bd713335
11:58 bbkr joined #perl6
11:58 igorsutton joined #perl6
12:02 Pleiades` joined #perl6
12:02 btyler joined #perl6
12:05 dalek rakudo/nom: 4555386 | (Elizabeth Mattijsen)++ | / (4 files):
12:05 dalek rakudo/nom: Initial implementation of $*DISTRO
12:05 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/4555386e9e
12:06 zakharyas joined #perl6
12:16 dalek rakudo/nom: 967f9c8 | (Elizabeth Mattijsen)++ | src/core/signals.pm:
12:16 dalek rakudo/nom: Simplify signal() with $*DISTRO.signals
12:16 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/967f9c8715
12:23 daniel-s joined #perl6
12:29 virtualsue joined #perl6
12:31 retupmoca tadzik: I made Task::Star work: https://github.com/tadzik/Task-Star/pull/2
12:31 ssutch joined #perl6
12:32 lizmat vendethiel: re http://irclog.perlgeek.de/perl6/2014-05-10#i_8704094 , is that the invocation to send a signal to a process?  Or is that the invocation to get the list of available signals ?
12:34 tadzik retupmoca: awesome, thanks!
12:35 dalek rakudo/nom: c4f8f02 | (Elizabeth Mattijsen)++ | src/core/distro.pm:
12:35 dalek rakudo/nom: $*DISTRO.OS->.name (as per spec), introduce is-win
12:35 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/c4f8f02bce
12:50 dalek rakudo/nom: 081ad6d | (Elizabeth Mattijsen)++ | src/core/distro.pm:
12:50 dalek rakudo/nom: Make $*DISTRO.is-win a bit better
12:50 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/081ad6d5da
12:51 rindolf joined #perl6
12:54 anaeem1_ joined #perl6
12:55 lizmat r: $*OSVER # wtf?  no spec, same as $*OS ?
12:55 camelia rakudo-jvm 455538: OUTPUT«(timeout)»
12:55 camelia ..rakudo-{parrot,moar} 455538: ( no output )
12:55 lizmat r: say $*OSVER # wtf?  no spec, same as $*OS ?
12:55 camelia rakudo-jvm 455538: OUTPUT«3.8.0-19-generic␤»
12:55 camelia ..rakudo-{parrot,moar} 455538: OUTPUT«linux␤»
12:55 lizmat r: say $*OS, $*OSVER # wtf?  no spec, same as $*OS ?
12:56 camelia rakudo-jvm 455538: OUTPUT«Linux3.8.0-19-generic␤Saw 1 call to deprecated code during execution.␤================================================================================␤$*OS called at:␤  /tmp/tmpfile, line 1␤Please use $*DISTRO.OS instead.␤-----------------------…»
12:56 camelia ..rakudo-{parrot,moar} 455538: OUTPUT«linuxlinux␤Saw 1 call to deprecated code during execution.␤================================================================================␤$*OS called at:␤  /tmp/tmpfile, line 1␤Please use $*DISTRO.OS instead.␤------------------------…»
12:56 dayangkun joined #perl6
12:57 dayangkun joined #perl6
13:02 dalek rakudo/nom: f4b62f3 | (Elizabeth Mattijsen)++ | src/core/distro.pm:
13:02 dalek rakudo/nom: Deprecate $*OSVER, introduce $*DISTRO.version
13:02 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/f4b62f367d
13:04 thou joined #perl6
13:10 geekosaur joined #perl6
13:10 * [Coke] gives up trying to get steroids to work on the mac and gets back to work.
13:10 guru joined #perl6
13:11 lizmat :w
13:12 lizmat [Coke]: does this ring a bell: my $osnames = lc any <darwin linux freebsd MSWin32 mingw msys...
13:12 lizmat it's from t/spec/S02-magicals/config.t
13:13 lizmat I was wondering whether that list actually belongs in $*VM
13:14 SamuraiJack joined #perl6
13:14 ajr joined #perl6
13:14 btyler [Coke]: yeah, the key bit for 'make install' still points to a path on tadzik's box...however! I got it going pretty quickly with minimal pain: (assuming SDL2/ttf/image installed): Configure.pl --> make -f Makefile.sdlwrapper --> make install --> ignore the error it outputs (about not being able to install steroids to your p6 lib, because it points to tadzik's box) and ./SpaceInvaders
13:21 tadzik wow, I didn't even know there's a 'make install' there :
13:21 tadzik :D
13:21 tadzik the README didn't mention it, why did you folks run it :P
13:23 [Coke] btyler: make -f Makefile.sdlwrapper dies.
13:23 [Coke] need to point it at my SDL libs.
13:24 [Coke] (need to make Configure.pl smart enough to use sdl-config, I think)
13:27 [Coke] oh. there's already a call to sdl2-config?
13:27 tadzik yes
13:28 tadzik I will gladly accept any patches to Configure.pl that can make OSX work as intended
13:28 [Coke] sdlwrapper.c:3:10: fatal error: 'SDL_ttf.h' file not found
13:28 arnsholt tadzik: You're on the HN front page! =D
13:28 [Coke] I assumed it wasn't found because it wasn't probing, but it is.
13:28 tadzik arnsholt: holy batman :D
13:28 arnsholt http://ttjjss.wordpress.com/2014/05/11/new-game-space-invaders/
13:29 tadzik its popularity is stunning
13:29 tadzik headline of perl weekly, tweeted by brian d foy, now HN
13:29 tadzik blog is hitting its all-time record in views per day
13:31 Khisanth joined #perl6
13:31 [Coke] tadzik: do I need something other than libsdl2 @2.0.3_0 and libsdl2_image @2.0.0_1 ?
13:31 treehug88 joined #perl6
13:32 molaf joined #perl6
13:33 tadzik [Coke]: sdl2_ttf
13:33 schogen joined #perl6
13:34 [Coke] oh, you mean like it says in the readme? :)
13:35 tadzik yeah, that :P
13:36 [Coke] ok, installing that broke kcachegrind, but I think fixed this issue. :)
13:37 [Coke] Can not bind non-existent attribute '$!h'
13:37 [Coke] which then  EXC_BAD_ACCESS (SIGSEGV)'d moar.
13:37 [Coke] ^^
13:37 tadzik :(
13:37 tadzik is that a recent rakudo/moar?
13:38 [Coke] This is perl6 version 2014.03.01-128-gaeda9d0 built on MoarVM version 2014.03-201-gf7ddc7e
13:38 [Coke] so, not moldy, but not fresh
13:38 [Coke] I'll try again with a fresh one later. :)
13:38 [Coke] tadzik++
13:39 rurban joined #perl6
13:40 tadzik it probably shouldn't be happening anyway
13:51 dalek roast: 3076c69 | (Elizabeth Mattijsen)++ | S02-types/deprecations.t:
13:51 dalek roast: Add test for $*OSVER deprecation
13:51 dalek roast: review: https://github.com/perl6/roast/commit/3076c69de2
13:51 dalek roast: b6f69e6 | (Elizabeth Mattijsen)++ | / (13 files):
13:51 dalek roast: s/$*OS/$*DISTRO.name where tested by Rakudo
13:51 dalek roast: review: https://github.com/perl6/roast/commit/b6f69e682e
13:51 smash joined #perl6
14:02 bluescreen10 joined #perl6
14:03 guru joined #perl6
14:03 anaeem1 joined #perl6
14:07 xenoterracide joined #perl6
14:07 thou joined #perl6
14:13 dalek rakudo/nom: 8f76741 | (Elizabeth Mattijsen)++ | src/core/distro.pm:
14:13 dalek rakudo/nom: Add Distr.(Str|gist)
14:13 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/8f767413b3
14:14 kaare_ joined #perl6
14:19 jnap joined #perl6
14:19 araujo joined #perl6
14:19 araujo joined #perl6
14:21 ajr joined #perl6
14:24 tadzik hehe. 2000 visitors on the blog and counting. I think the internet is amused that you can use Perl 6 for practical stuff. A good thing :)
14:24 lizmat tadzik++
14:26 JimmyZ https://github.com/gaorongchao/Perl6/tree/master/Using_perl6 # Chinese version :P
14:26 tadzik wow, nice :)
14:28 btyler anyone having panda bootstrap/rebootstrap fail? I've been rebootstrapping as part of my little rakudo-brew-alike script that runs every morning, but it failed testing panda, saying that it couldn't find it (and install/languages/perl6/site/bin wasn't in the search path?)
14:28 JimmyZ moritz: http://bbs.chinaunix.net/thread-4137901-1-1.html Perl 5 to 6 Chinese Version :P
14:28 btyler potentially related, I ufo'd panda, and 'make install' dies with a segfault (latest moar/nqp/rakudo as of 30 minutes ago) on compiling lib/Panda/Builder.pm
14:28 tadzik :o
14:29 JimmyZ Using Perl 6 online one: http://bbs.chinaunix.net/thread-4132588-1-1.html
14:29 tadzik I'll goan update my rakudo, let's see if I can reproduce this
14:29 JimmyZ # Chinese verion, I mean
14:29 lizmat btyler: fwiw, I'm seeing segfaults in spectest on OS X since yesterday
14:29 btyler just let me know what info would be helpful
14:29 btyler ok
14:29 btyler also on OSX here
14:30 lizmat I'm waiting for the next merge of jnthn / timotimo 's work on Moar before really complaining  :-)
14:30 tadzik me too :)
14:30 lizmat it's not release time just yet :-)
14:30 tadzik With Steroids I'm struggling to make rakudo look slow; they're trying to speed it up, and they keep winning the race
14:31 timotimo oh no!
14:31 timotimo what did we do wrong again?
14:31 lizmat wish I could tell you
14:32 lizmat but latest Makudo isn't as stable
14:32 tadzik things don't explode for me so far, however I'm on moar from MOAR_REVISIONS
14:32 lizmat seeing segfaults at compile time
14:32 lizmat that used to be legit compile time errors
14:34 lizmat but I got the impression several things are in the pipeline from Moar again, so I'm not pursuing finding out how and what at the moment
14:35 tadzik I suppose spesh may be a bit over-eager optimizing things
14:35 dalek rakudo-star-daily: 3ff0410 | coke++ | log/ (5 files):
14:35 dalek rakudo-star-daily: today (automated commit)
14:35 dalek rakudo-star-daily: review: https://github.com/coke/rakudo-star-daily/commit/3ff04101b0
14:35 dalek perl6-roast-data: 95c25ee | coke++ | / (6 files):
14:35 dalek perl6-roast-data: today (automated commit)
14:35 dalek perl6-roast-data: review: https://github.com/coke/perl6-roast-data/commit/95c25ee170
14:36 tadzik spaceinvaders segfaulted for me, until I recompiled all the .moarvm files
14:36 tadzik then it works fine
14:36 [Coke] (still hundreds of failures in rakudo.parrot and rakudo.jvm)
14:37 [Coke] (as of yesterday)
14:37 [Coke] :(
14:37 woolfy1 tadzik ++ for http://ttjjss.wordpress.com/2014/05/11/new-game-space-invaders/    !!wow!!
14:38 tadzik :)
14:38 tadzik I think I should've maybe added some highscores, instructions and all, make it a proper game instead of an engine showcase
14:38 tadzik but it seems exciting enough :)
14:39 woolfy1 tadzik: I am quite sure that you will do that sometime in the near future...
14:39 * lizmat suddenly represses the urge to re-implement "slot"
14:39 tadzik woolfy1: surely :)
14:39 daniel-s joined #perl6
14:39 woolfy1 :-)
14:40 tadzik I'll need at least one proper, full-blown game for PLPW, I might just add more stuff to Space Invaders, looking at how famous they became today :)
14:40 woolfy1 (games are way too addictive to me, please don't make it too much fun to play)
14:40 tadzik I'll try :P
14:41 tadzik I'm wasting so much time on gaems myself that it's amazing to me how I get anything done
14:41 tadzik but, time you enjoy wasting is said to be not wasted
14:42 jnthn Is the MOAR_REVISION Rakudo more stable than the one with HEAD?
14:42 tadzik I'm on MOAR_REVISION, and I didn't notice any instability yet
14:42 jnthn ok
14:42 jnthn that corresponds to the initial spesh_trace merge; some other things landed after it.
14:43 Ulti when panda installs where are tests run from, the t/ directory?
14:43 lizmat tadzik: but you're not on OS X, right?
14:43 jnthn Would be helpful if those seeing issues could let me know if you're on HEAD or MOAR_REVISION.
14:43 * lizmat on MOAR_REVISION
14:43 jnthn lizmat: Do you have the Panda issues?
14:43 tadzik lizmat: nope
14:43 tadzik Ulti: yes
14:44 * tadzik installs Task::Star to torment rakudo a little
14:45 JimmyZ E:\OpenSource\MoarVM\install\bin\moar --libpath=src\vm\moar\stage0 src\v
14:45 JimmyZ m\moar\stage0\nqp.moarvm --bootstrap --setting=NULL --no-regex-lib --target=mbc
14:45 JimmyZ --output=gen\moar\stage1\nqpmo.moarvm gen\moar\stage1\nqpmo.nqp
14:45 JimmyZ NMAKE : fatal error U1077: “E:\OpenSource\MoarVM\install\bin\moar.EXE”: 返回代
14:45 btyler I'm on HEAD
14:45 JimmyZ 码“0xc0000005”
14:45 JimmyZ return code
14:45 jnthn Well, that certainly didn't happen for me.
14:46 JimmyZ maybe another x86 problem? :P
14:46 jnthn Maybe.
14:46 jnthn But recent if so.
14:47 daniel-s joined #perl6
14:47 jnthn Anyway, for those seeing segfaults, doing debug builds and supplying gdb backtraces might help me find it.
14:47 timotimo .o( you can ufo panda?! )
14:47 jnthn Patches to fix it even more welcome :P
14:48 JimmyZ It will be an easy fix if it's not about spech or gc :P
14:48 JimmyZ spesh
14:48 jnthn spesh isn't always hard to fix.
14:48 jnthn The last SEGV somebody found in a spesh/gc interation, I had fixed in < an hour of looking.
14:48 jnthn *interaction
14:49 btyler jnthn: ok, I'll try to get a backtrace later today. patch unlikely unless it turns out to be a one character kind of thing :)
14:49 btyler (the C chops, they are lacking...)
14:49 jnthn btyler: Thanks
14:50 JimmyZ hmm, I forgot how to add debug info to moar.dll
14:50 jnthn /pdb:$@.pdb
14:55 adu joined #perl6
14:59 dalek Perlito: 55c3f9f | (Flavio S. Glock)++ | t5/01-perlito/500-regex-escape.t:
14:59 dalek Perlito: Perlito5 - add regex test
14:59 dalek Perlito: review: https://github.com/fglock/Perlito/commit/55c3f9f102
15:08 xenoterracide joined #perl6
15:13 molaf joined #perl6
15:15 dalek rakudo/nom: 65f75fb | (Elizabeth Mattijsen)++ | / (4 files):
15:15 dalek rakudo/nom: Implement $*VM as an object with methods, per spec
15:15 dalek rakudo/nom:
15:15 dalek rakudo/nom: Mark $*VM<name|config> as DEPRECATED
15:15 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/65f75fb765
15:20 hoelzro tadzik++ # space invaders/post
15:22 igorsutton joined #perl6
15:26 lizmat fwiw, I'm not seeing any segfaults in the spectest with MVM_SPESH_DISABLE=1
15:27 lizmat with spesh: Files=870, Tests=32653, 217 wallclock secs ( 8.17 usr  4.15 sys + 1351.02 cusr 158.28 csys = 1521.62 CPU)
15:27 lizmat spesh disabled: Files=870, Tests=32670, 209 wallclock secs ( 8.27 usr  3.98 sys + 1292.98 cusr 143.84 csys = 1449.07 CPU)
15:34 hoverboard joined #perl6
15:35 spider-mario joined #perl6
15:35 rurban_ joined #perl6
15:36 dmol joined #perl6
15:51 eternaleye joined #perl6
16:02 lizmat cycling&
16:02 kaleem joined #perl6
16:16 daniel-s joined #perl6
16:18 timotimo rakudo-moar seems about 3x faster at spesh_diff than rakudo-parrot, even though it does a fair amount of string joining
16:20 telex joined #perl6
16:21 timotimo p6: 1.765.fmt("% 3.2f").say
16:21 camelia rakudo-jvm 8f7674: OUTPUT«(timeout)»
16:21 camelia ..rakudo-{parrot,moar} 8f7674: OUTPUT«1.77␤»
16:21 camelia ..niecza v24-109-g48a8de3: OUTPUT«Unhandled exception: invalid format specifier␤  at /home/p6eval/niecza/lib/CORE.setting line 1403 (sprintf @ 4) ␤  at /home/p6eval/niecza/lib/CORE.setting line 750 (Cool.fmt @ 4) ␤  at /tmp/tmpfile line 1 (mainline @ 3) ␤  at /home/p6eval/niec…»
16:21 timotimo isn't that supposed to put spaces before the 1?
16:22 timotimo p6: 1.765.fmt("%03.2f").say
16:22 camelia rakudo-{parrot,jvm,moar} 8f7674, niecza v24-109-g48a8de3: OUTPUT«1.77␤»
16:22 timotimo and that's supposed to pad to the left with zeroes?
16:22 PerlJam p6: 1.765.fmt("%05.2f").say
16:23 camelia rakudo-{parrot,jvm,moar} 8f7674, niecza v24-109-g48a8de3: OUTPUT«01.77␤»
16:23 PerlJam your field width wasn't wide enough to be padded with 0
16:23 timotimo oh! that's for the whole thing, then?
16:23 PerlJam yes
16:24 timotimo that explains it, PerlJam++
16:26 Psyche^ joined #perl6
16:32 FROGGS joined #perl6
16:37 denis_boyun joined #perl6
16:37 TimToady it is not clear to me that putting "distro" in front of "os" makes "os" any less ambiguous
16:42 * TimToady is very tempted to ban "OS" like we banned "length"
16:44 TimToady if linux is the kernel, and ubuntu is the distro, what is the OS?
16:45 TimToady if it's some intermediate idea, we must define it
16:45 TimToady or drop it
16:45 masak +1
16:45 masak (hi, #perl6)
16:46 PerlJam TimToady: +1
16:46 TimToady it does not help that "linux" is terribly overloaded to mean "any distro built on top of a linux kernel"
16:47 daniel-s_ joined #perl6
16:47 Ulti GNU/Linux
16:47 dakkar do you also need ABI, libc or whatever, and so on? "x86_64-pc-linux-gnu" and the like?
16:48 TimToady GNU/Linux is neither a distro nor a kernel
16:48 geekosaur ...and then there's the question of where debian kfreebsd fits
16:48 * dakkar lacks a bit of context, speaks anyway
16:48 Ulti TimToady: no its a family of operating systems
16:48 Ulti like Microsoft Windows is
16:48 TimToady well, there's also $*ARCH, but in a sense the kernel is also a virtual machine
16:49 Ulti a giant hash of POSIX compatibility would be more useful :P
16:49 TimToady Ulti: it's a very loose family, insofar as it merely means "some subset of the tools come from GNU"
16:49 Ulti yeah
16:50 Ulti GNU only say GNU/Linux because GNU/Hurd is trying to be a thing
16:50 PerlJam Ulti: %*ALL-THE-STUFF-YOU-NORMALLY-WOULD-PROBE-FOR
16:52 Ulti I think architecture should list the instruction set too :P just in case, and have a built in ontology so that we all know what instructions we are talking about cross CPU
16:52 Ulti (maybe joking)
16:52 TimToady well, that's more like $*ARCH
16:53 TimToady but yeah, one of the things I noticed way back when I first wrote a Configure script for rn was that we needed to test each feature, not rely on names of systems
16:53 Ulti usually when people are checking this sort of thing its a more specific feature they care about rather than the actual operating system, like a lot of people assume GNU tools if the kernel is Linux which is not a given at all
16:54 TimToady right, you can have a linux that is clang and bsd tools
16:56 Ulti and OSX which changes its mind every release on what versions of things are used and where they are kept
16:56 TimToady but naming all the features is just as much of a fuzzball as trying to straighten out all the contextual names
16:57 liztormato joined #perl6
16:57 TimToady of the two, the feature-based approach is less fragile, but requires more upfront maintenance
16:58 TimToady naming systems foists the work off on the people who later discover that $*SYSNAME is too blunt an ax
16:58 Ulti also what happens in Cygwin
16:58 Ulti is that an OS?
16:59 TimToady but portability to not-yet-existent systems is enhanced by a feature-based approach
16:59 liztormato TimToady: Fwiw. I already committed $*DISTRO.name in favor of .OS
17:00 TimToady that would be a negative favor, but okay
17:01 TimToady (just saying a native English speaker would use "in favor of" the other way around)
17:01 treehug8_ joined #perl6
17:01 liztormato Hehe. Ok
17:01 TimToady but your English is still much better than my Dutch :)
17:02 liztormato I have patches ready for $*kERNEL.name and .arch
17:02 TimToady why .arch if $*ARCH is speccced?
17:03 liztormato Ah. Good question. No particular reason. Will follow spec :-)
17:03 TimToady (which is, of course, open for negotiation, as long as we're revising the whole taxonomy of contexts)
17:04 Rotwang joined #perl6
17:04 liztormato I guess my main reason was that it was easier to fetch the info on demand
17:05 TimToady of course, cpu architectures are also so very simple to classify... :/
17:05 liztormato uname -p if i recall
17:06 geekosaur ahahah
17:06 geekosaur uname values and flags are far from usefully standardized
17:06 geekosaur posix says one thing, everyone itnerprets it their own way and the result is chaos
17:06 geekosaur (moreover posix refuses to even touch some thigs that would be quite useful if standardized)
17:06 liztormato My point was that I wanted to create a framework
17:07 liztormato Food&
17:09 [Coke] .u 1F595
17:09 yoleaux No characters found
17:09 TimToady posix is in the category of names that guarantee of certain minimum levels of api support, a feature bundle, as it were
17:10 nwc10 $ uname -p
17:10 nwc10 uname: illegal option -- p
17:10 nwc10 usage: uname [-amnrsvil] [-S nodename]
17:10 TimToady the C programming interface is much richer than that, depending on what libaries are guaranteed by your distro name and your additional downloads
17:10 nwc10 (been round that maze just a couple of days ago)
17:11 liztormato nwc10: On linux an darwin inly, atm
17:12 TimToady someone should invent a URL-equivalent for feature dependencies
17:12 nwc10 $ uname -p
17:12 nwc10 unknown
17:12 nwc10 $ uname -s
17:12 nwc10 Linux
17:12 nwc10 :-)
17:13 nwc10 The "illegal option -- p" was from HP/UX
17:13 nwc10 And it says:
17:13 nwc10 $ uname -m
17:13 nwc10 9000/800
17:13 nwc10 soooo helpful.
17:15 nwc10 The joyful bit of POSIX I discovered today was hcreate and friends: http://pubs.opengroup.org/onlinepubs/009695299/functions/hcreate.html
17:15 nwc10 not threadsafe
17:16 nwc10 nothing is sane. :-(
17:23 geekosaur yes, ancient chunks of system v libc that got standardized for no good reason
17:24 geekosaur I am currently quite down on POSIX; turns out they took some stuff they standardized early on for the shell spec and dropped it, so scripts written "to POSIX spec" are no longer guaranteed to work on "POSIX compliant" systems
17:24 liztormato_ joined #perl6
17:25 geekosaur way to state openly that you're all about getting as many vendors signed on as possible, at the expense of being in any way useful, dear POSIX
17:26 liztormato__ joined #perl6
17:27 liztormato joined #perl6
17:42 vendethiel o/
17:43 vendethiel lizmat: I'm not sure what taskkill sends, I think it just crashes the PID
17:45 FROGGS when I need something about distro/os/whatever it is mostly about:
17:45 FROGGS a) is it unixoid or windowsish?
17:46 FROGGS b) what is the distribution and version? (like openbsd 5.5, to go insane with -pthread)
17:46 FROGGS c) is it a 32 or 64 bit machine
17:46 FROGGS maybe we need our very own model of all these things
17:47 FROGGS so that we only have properties that are really are unique properties, and not things that almost mean something but are off to 10 to 20%
17:47 liztormato That's what I'm going for
17:48 FROGGS so I think $*DIST.name and .ver and perhaps .auth is the easiest one
17:48 FROGGS the others are more questionable
17:49 liztormato I saw a lot of adhoc knowledge in the spectest that I wanted to gather at a central place
17:50 jnap joined #perl6
17:54 liztormato More cycling&
17:54 liztormato left #perl6
18:00 berekuk joined #perl6
18:00 rindolf joined #perl6
18:04 kPhone joined #perl6
18:06 kPhone Seen au
18:07 timo joined #perl6
18:07 test joined #perl6
18:07 kPhone seen camelia
18:09 isBEKaml joined #perl6
18:10 berekuk joined #perl6
18:14 treehug88 joined #perl6
18:17 FROGGS .seen kPhone
18:17 yoleaux I saw kPhone 18:07Z in #perl6: <kPhone> seen camelia
18:18 Guest86162 left #perl6
18:23 retupmoca panda on moarvm looks broken
18:23 FROGGS :/
18:24 retupmoca boostrap gives a lot of 'No subtests run'
18:25 retupmoca MVM_SPESH_DISABLE bootstrap gives a few 'Attempt to read past end of string heap (index 9109510)' as well
18:25 retupmoca jnthn, timotimo: ^
18:25 timotimo oh shit.
18:26 retupmoca This is perl6 version 2014.04-230-g65f75fb built on MoarVM version 2014.04-129-g024b994
18:27 brrt joined #perl6
18:28 tadzik is that the MOAR_REVISION revision?
18:29 retupmoca HEAD
18:30 retupmoca I build on master/master/nom normally
18:37 btyler likewise, same issue as retupmoca (mentioned it this morning, haven't had time to get a debug moar/gdb going)
18:46 lizmat TimToady: I can't find $*ARCH specced anywhere?
18:47 dalek rakudo/nom: 7e5cf89 | (Elizabeth Mattijsen)++ | src/core/ (2 files):
18:47 dalek rakudo/nom: .version -> .ver for *$VM, $*DISTRO
18:47 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/7e5cf897c0
18:47 dalek rakudo/nom: 26ddce2 | (Elizabeth Mattijsen)++ | / (4 files):
18:47 dalek rakudo/nom: NEW! $*KERNEL name,ver,release,hardware,arch
18:47 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/26ddce2363
18:47 dwarring joined #perl6
18:48 lizmat please note that at this stage, $*KERNEL is an empty framework waiting for specific OS dependent info to be implemented
18:48 jnap joined #perl6
18:50 rurban1 joined #perl6
18:52 REPLeffect joined #perl6
18:53 timotimo i'm still looking for feedback for adding a trait "is subcommand" that tells the autogenerated main to put an argument that's decorated with it before all nameds in the usage message
18:53 timotimo as in: multi MAIN("foobar" is subcommand, $arg, $arg2, :$foo, :$bar)
18:53 timotimo or maybe multi MAIN(...) is subcommand("foobar") or something like that
18:54 PerlJam timotimo: what sort of feedback?  (+1 from me btw)
18:54 timotimo or only "is subcommand" to make the first positional be "the subcommand"
18:54 moritz timotimo: IMHO that stretches the signature analogy a bit too far
18:54 timotimo feedback as in: +1, -1, suggestions for what it really should be called and where it should be put
18:54 moritz timotimo: in the case of command lines, there are typically arguments for the command and the subcommand
18:54 moritz s/arguments/options/
18:55 timotimo oh, hm.
18:55 moritz timotimo: but in signatures, all named args share the same namesapce
18:55 moritz git --git-dir=/tmp/foo log -p
18:55 timotimo i'd just like to have a way to make the usage of p6bench less stupid
18:56 moritz write a command line parsing module that supports subcommands?
18:56 spider-mario joined #perl6
18:56 timotimo mhh
18:57 timotimo but that'd be *work*! :(
18:57 moritz or if you need many args, write config files instead
18:57 slavik joined #perl6
18:59 rurban joined #perl6
18:59 timotimo i was going to write a TUI frontend at some point to ease usage
18:59 timotimo since you have to throw around revision numbers and stuff all the time
19:02 rurban1 joined #perl6
19:05 hoelzro .
19:05 hoelzro oops
19:10 igorsutton joined #perl6
19:14 hoverboard joined #perl6
19:16 timotimo .~
19:18 dalek perl6-roast-data: c2362cc | coke++ | / (6 files):
19:18 dalek perl6-roast-data: today (automated commit)
19:18 dalek perl6-roast-data: review: https://github.com/coke/perl6-roast-data/commit/c2362cc8d1
19:19 [Coke] java now has nearly -300- failures.
19:19 [Coke] moar more than doubled up to 32.
19:19 [Coke] parrot holding steady around 200
19:21 [Coke] moar/java are failing in S17. parrot is still failing in S32-temporal
19:21 telex joined #perl6
19:26 [Coke] yay, found the root cause of most of the java failures...
19:28 dalek rakudo/nom: dd8ff2f | (Elizabeth Mattijsen)++ | src/core/kernel.pm:
19:28 dalek rakudo/nom: Introduce $*KERNEL.bits (either 32 or 64 for now)
19:28 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/dd8ff2f2bb
19:28 dalek rakudo/nom: e2c9ba5 | (Elizabeth Mattijsen)++ | src/core/ (2 files):
19:28 dalek rakudo/nom: Add some intro commentaries
19:28 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/e2c9ba583f
19:28 dalek rakudo/nom: e282938 | (Elizabeth Mattijsen)++ | src/core/kernel.pm:
19:28 dalek rakudo/nom: $*KERNEL.bits probably better depends on .hardware
19:28 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/e2829387a2
19:31 * lizmat hopes [Coke] is working on a patch
19:31 molaf joined #perl6
19:32 FROGGS [Coke]: do tell!
19:35 dalek specs: ec4ff99 | (Elizabeth Mattijsen)++ | S28-special-names.pod:
19:35 dalek specs: We're not going to have a %*INC
19:35 dalek specs: review: https://github.com/perl6/specs/commit/ec4ff99570
19:36 * masak .oO( throw it down the inc-well )
19:39 moritz is that a quantum inc well?
19:39 [Coke] S17-supply/categorize.t and S17-supply/classify.t have an invalid fudge line of:
19:40 [Coke] #?rakudo.jvm todo 3 "D: doesn't work in signatures"
19:40 [Coke] ^^ not even needed, since java passes all those tests.
19:40 [Coke] that brings java in line with moar.
19:40 [Coke] if someone could fix those up, I'd appreciate it.
19:41 lizmat [Coke]: will do
19:42 lizmat TimToady: re S28:137, shouldn't that be a role rather than a class ?
19:42 synopsebot Link: http://perlcabal.org/syn/S28.html#line_137
19:44 TimToady probably
19:44 [Coke] (the fudge line is invalid because it's supposed to be "3 todo" not "todo 3", btw.)
19:44 TimToady though Application is kind of a lousy name for it
19:45 TimToady it's more like an uniquely identified entity
19:45 TimToady http://rosettacode.org/wiki/Order_disjoint_list_items#Perl_6
19:46 TimToady this RC task made me sad that $bag{$key}-- blows up if the key isn't in the bag
19:46 TimToady seems like it would be nicer if -- would just do nothing and return false if the key isn't in a set or bag
19:46 vendethiel that seems surprising to me
19:47 vendethiel m: my %h; say %h<a>
19:47 camelia rakudo-moar 26ddce: OUTPUT«(Any)␤»
19:47 vendethiel mmh
19:47 lizmat vendethiel: bags are only hashes to some level
19:48 TimToady in particular this line: (bag{$_} && bag{$_}-- ?? N.shift !! $_ for M)
19:48 lizmat TimToady: S02:1697 needs changing then
19:48 synopsebot Link: http://perlcabal.org/syn/S02.html#line_1697
19:50 TimToady m: my $b = <a b c>.BagHash; say $b<d>-- // 0
19:50 camelia rakudo-moar 26ddce: OUTPUT«===SORRY!===␤Could not find symbol '&Return'␤»
19:50 lizmat yes, that's odd, looking at that now
19:51 TimToady p6: my $b = <a b c>.BagHash; say $b<d>-- // 0
19:51 camelia rakudo-parrot 26ddce: OUTPUT«Could not find symbol '&Return'␤  in method <anon> at gen/parrot/CORE.setting:12684␤  in any  at gen/parrot/Metamodel.nqp:2710␤  in any find_method_fallback at gen/parrot/Metamodel.nqp:2698␤  in any find_method at gen/parrot/Metamodel.nqp:949␤ …»
19:51 camelia ..rakudo-moar 26ddce: OUTPUT«===SORRY!===␤Could not find symbol '&Return'␤»
19:51 camelia ..rakudo-jvm 26ddce: OUTPUT«Could not find symbol '&Return'␤  in method <anon> at gen/jvm/CORE.setting:12661␤  in any  at gen/jvm/Metamodel.nqp:2627␤  in any find_method_fallback at gen/jvm/Metamodel.nqp:2626␤  in any find_method at gen/jvm/Metamodel.nqp:948␤  in any  at gen…»
19:51 camelia ..niecza v24-109-g48a8de3: OUTPUT«0␤»
19:51 TimToady niecza++
19:52 TimToady rakudobug--
19:53 lizmat but but, niecza is *not* according to spec
19:53 lizmat or is it?
19:53 TimToady it returns a failure, which is undefined, I imagine
19:54 TimToady failures do not throw exception when evaluated in defined or boolean context
19:54 lizmat n: my $b = <a b c>.BagHash; say ($b<d>--).WHAT
19:54 TimToady in fact, I ought to just be able to evaluate the result in the ??!!
19:54 camelia niecza v24-109-g48a8de3: OUTPUT«(Int)␤»
19:55 TimToady should perhaps be Nil, but whatevs
19:55 anaeem1__ joined #perl6
19:57 TimToady anyway, that solution is already spectacularly short compared to the other languages (except J, of course), but it's shame it has to do the lookup twice currently
19:57 TimToady well, and the error message in question is way LTA
19:57 lizmat that's because there is an error in the error message
20:00 dalek specs: 058fc44 | TimToady++ | S02-bits.pod:
20:00 dalek specs: s/Failure/undefined/ for decrement on sets/bags
20:00 dalek specs: review: https://github.com/perl6/specs/commit/058fc44f8a
20:00 TimToady failure is overkill for something that will be used frequently for set/bag membership
20:00 lizmat TimToady: would Nil work also?
20:01 TimToady Nil tends to screw up the type system a bit
20:01 lizmat so just Int then?
20:01 TimToady well, Bool on sets
20:02 lizmat indeed
20:02 rindolf joined #perl6
20:02 TimToady m: my %a; say %a<nonesuch>
20:02 camelia rakudo-moar 26ddce: OUTPUT«(Any)␤»
20:02 TimToady that seems consistent
20:03 lizmat hmmm on SetHash , -- would give False, I guess?
20:03 lizmat not Bool?
20:05 TimToady well, False is the default when subscripting, but I think Bool indicates the attempt to decrement from False, which is more undefined than False
20:06 TimToady similarly, there's a difference between removing the last element from a bag and trying to remove one that isn't there
20:07 TimToady so an undefined value is more appropriate than 0 (the default) would be, in the case of an active decrement
20:07 TimToady a mere failed subscript would return 0 or False, of course
20:08 TimToady with postfix -- it doesn't much matter, but maybe it matters more with prefix --
20:08 TimToady where a 0 means you just removed the last one, and (Int) means there wasn't one to remove in the first place
20:10 lizmat ok
20:10 TimToady so you could reasonably write: my $remaining = --$bag{$key} // die "There isn't a $key in that bag!"
20:11 TimToady oh goodness, lunch forgot to happen
20:11 berekuk joined #perl6
20:12 masak m: my $b = False; $b--; say $b
20:12 camelia rakudo-moar 26ddce: OUTPUT«False␤»
20:12 masak m: say False.pred
20:12 camelia rakudo-moar 26ddce: OUTPUT«False␤»
20:18 TimToady yes, well, there is thta
20:18 TimToady *that
20:19 TimToady I suppose one could think of bags as counted in a set of integers that cannot go negative
20:20 lizmat uints?
20:20 lizmat Natural?
20:20 TimToady not suggesting it has to be implemented that way
20:22 lizmat ok
20:30 IllvilJa joined #perl6
20:38 molaf joined #perl6
20:39 lizmat grr... Proxy doesn't seem to be implemented correctly
20:39 virtualsue joined #perl6
20:40 ssutch joined #perl6
20:44 lue anyone else unable to bootstrap panda lately?
20:57 btyler lue: yeah, I think timotimo++ identified the offending lines in moar
21:03 lizmat https://gist.github.com/lizmat/7f7ecc7fdbd4152c9932  Proxy.FETCH being called *EIGHT* times for a simple access ??
21:03 lizmat jnthn: ^^^  huh?
21:05 jnthn lizmat: Well, you've probably hit the worst case there...
21:05 jnthn for one the multi-dispatcher won't have got its caches primed yet, I guess...
21:05 jnthn So the ++ may well be hitting that
21:05 jnthn Oh...
21:05 jnthn Worse
21:06 jnthn Proxy can't hit the cache.
21:06 lizmat it's very easy to have it do it many, many times more
21:06 jnthn Because the cache can't run code.
21:07 lizmat I'm not sure how Proxy is supposed to work, as it is definitely not up to spec
21:07 jnthn So I suspect it's going to the proto, then through the multi-dispatcher, etc.
21:07 lizmat but why would it call Fetch everytime ?
21:07 jnthn The problem with Proxy is this: decont is generally cheap and side-effect free
21:07 jnthn How can it *not* call fetch every time?
21:07 brrt left #perl6
21:08 jnthn If you're passing a containerized thing off to a bunch of things that need the value within it, then each one is going to have to obtain the value
21:08 jnthn That's done by fetch
21:09 dalek roast: 63a9cb3 | (Elizabeth Mattijsen)++ | S17-supply/c (2 files):
21:09 dalek roast: Unfudge faulty fudges, colomon++
21:09 dalek roast: review: https://github.com/perl6/roast/commit/63a9cb3e1c
21:10 lizmat sorry, I'm not seeing it
21:11 lizmat even $a<b> does 8 FETCHes, without doing an increment or decrement or whatever
21:16 igorsutton joined #perl6
21:17 jnthn lizmat: Made the best way to grok it is to take a look at how often decont shows up in the bytecode. Every time something needs to do a type check on a value in a container, there's a decont. Every time you call a method on something in a container, there's at decont. Every time the thing is put into self, there's a decont to get it and bind it into self. Just from those second two, calling a method on something will always need to decont at least twice.
21:19 lizmat all I can say is: wow, Proxy is *really* expensive
21:19 jnthn And say almost certainly is calling print, which may well in turn call a lower level thing.
21:19 jnthn Yeah, it will be.
21:19 lizmat so, is there another way to implement ++ and -- on BagHashes ?
21:19 lizmat (and SetHashes and MixHashes) ?
21:19 jnthn Well, given I don't know what it's meant to do... :)
21:20 lizmat basically the same as on a hash
21:20 masak as someone who reaches for Proxy sometimes, I'm sort of assuming/expecting it to be expensive. so is tie-ing in Perl 5, I guess.
21:20 masak 'night, #perl6
21:20 jnthn Yeah, Proxy is really rather hard to optimize...
21:20 colomon colomon?++
21:20 colomon lizmat: ?
21:20 lizmat in Perl 5, FETCH is only called once when needed
21:21 lizmat Oops, I meant [Coke]++
21:22 jnthn You're acting like we're calling it just for the hell of it :/
21:22 jnthn Please, take some time to understand the container model.
21:23 lizmat I'm just very unhappily surprised
21:23 lizmat I guess I should  :-)
21:23 jnthn As for a solution, maybe if it's like how hashes work, then the WHENCE mechanism hashes uses - which is far cheaper - could help.
21:23 lizmat anyway, the way Proxy currently works, I cannot make it return Int / Bool on -- on a non-existing key
21:24 jnthn About the Panda issue: I have it reproduced here.
21:26 jnthn Unfortunately, it's not proving easy to hunt down when things got broken... :(
21:32 rurban joined #perl6
21:37 daniel-s__ joined #perl6
21:41 _daniel-s__ joined #perl6
21:46 rurban joined #perl6
21:54 lizmat jnthn: how about this idea: have BagHash return a special BagValue object
21:55 lizmat and create candidates for postcircumfix ++ and -- for BagValue objects ?
21:59 jnthn lizmat: Sounds workable
22:01 dalek rakudo/nom: 3583767 | (Elizabeth Mattijsen)++ | src/core/BagHash.pm:
22:01 dalek rakudo/nom: Quick fix for change at S02:1698
22:01 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/3583767516
22:01 synopsebot Link: http://perlcabal.org/syn/S02.html#line_1698
22:01 lizmat TimToady: RC example should work with this
22:02 jnthn Panda appears not to work all the way back at 29c5c9d9 :(
22:03 lizmat that's in moar / nqp ?  not in rakudo or?
22:04 jnthn That's a moar sha1
22:04 jnthn Indicating the point of the spesh_trace merge
22:05 jnthn Thing is it breaks with MVM_SPESH_DISABLE=1
22:06 btyler jnthn: that'd correspond to my fuzzy check -- I do a master/master/nom/panda rebootstrap most weekday mornings EST, and nothing jumped out at me on friday
22:06 btyler (as in, rebuild everything from head and then panda rebootstrap)
22:07 jnthn It's going to take some effort to work out exactly what is to blame.
22:07 jnthn Since I'll also have to recreate NQP builds at the appropriate commits also
22:07 jnthn And I'm too tired to do that tonight.
22:08 btyler is that something that could be automated? I ran into the same problem trying to bisect a few minutes ago. seems like a script could keep things in sync as you moved forward/back on whichever axis (moar/nqp/rakudo) is interesting
22:09 btyler I was just going by rough date, though
22:11 jnthn Well, building d6a554f in Moar against an NQP pre-lexopts merge would be one itnersting data point
22:12 jnthn As would building abaec70 in Moar with where lexopts was before it was sync'd with master of NQP I guess
22:13 jnthn 2395d29 would also be an interesting one to try.
22:13 jnthn Anyway, I should sleep...really tired :(
22:18 _daniel-s__ joined #perl6
22:20 moistcherry joined #perl6
22:23 lizmat gnight jnthn!  and thanks for all the help!   jnthn++
22:24 dalek roast: 90be507 | (David Warring david.warring@gmail.com)++ | integration/advent2013-day19.t:
22:24 dalek roast: removed cruft
22:24 dalek roast: review: https://github.com/perl6/roast/commit/90be50739a
22:41 skids joined #perl6
22:46 _daniel-s__ joined #perl6
22:47 rurban joined #perl6
22:49 rurban1 joined #perl6
22:59 xenoterracide joined #perl6
23:11 btyler joined #perl6
23:23 BenGoldberg joined #perl6
23:25 guru joined #perl6
23:30 TimToady lizmat: I would think the lists of available signals would be determined by the kernel, not by the distro
23:30 lizmat so $*KERNEL.signals instead of $*DISTRO.signals ?
23:31 TimToady Ubuntu has no opinion on signals, but the syscall to Linux sure does
23:32 lizmat BTW, making BagHash<unknown_key>-- return Int, is proving to be quite a lot of extra code  :-(
23:32 lizmat I'm not sure it is all worth it
23:33 rurban joined #perl6
23:33 hoverboard joined #perl6
23:33 xinming_ joined #perl6
23:34 TimToady maybe there's a simpler way to do it
23:34 lizmat https://gist.github.com/lizmat/80b8c5b3a02f8a742c0c   work so far
23:36 TimToady m: say Int.pred
23:36 camelia rakudo-moar 358376: OUTPUT«Invocant requires an instance, but a type object was passed␤  in method Bridge at src/gen/m-CORE.setting:4495␤  in sub infix:<-> at src/gen/m-CORE.setting:4394␤  in method pred at src/gen/m-CORE.setting:4038␤  in block  at /tmp/ruYZXLkgL3:1␤␤»
23:38 * TimToady wonders what the fallout of defining .succ/.pred on type objects to be transparent
23:39 TimToady ...would be
23:39 lizmat Any.pred( self ) ?
23:39 TimToady it's an idea
23:40 TimToady dunno if it's a good one
23:40 lizmat will try that tomorrow maybe
23:40 lizmat will be on the road most of tomorrow
23:40 TimToady safe drivin'
23:42 dalek rakudo/nom: 9fa1db5 | (Elizabeth Mattijsen)++ | src/core/ (3 files):
23:42 dalek rakudo/nom: $*KERNEL.signals instead of $*DISTRO.signals
23:42 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/9fa1db5aea
23:43 lizmat and with that, I've reached my commit limit for today
23:43 lizmat gnight, #perl6!
23:51 colomon o/
23:53 colomon $*VM<config> called at:
23:53 colomon lib/NativeCall.pm6, lines 116,120
23:53 colomon Please use $*VM.config instead.
23:53 colomon hmmm...
23:54 _daniel-s__ joined #perl6

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

Perl 6 | Reference Documentation | Rakudo