Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6, 2014-01-08

Perl 6 | Reference Documentation | Rakudo

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

All times shown according to UTC.

Time Nick Message
00:33 lizmat joined #perl6
00:37 mavcunha joined #perl6
00:48 diakopter the backlog tonight
00:48 berekuk joined #perl6
00:53 [Coke] japhb__: you about?
00:53 japhb__ [Coke], yes.  Currently servicing three (3) requests, plus four (4) queued.
00:57 [Coke] one more: I want to clone rakudo locally, then when I need to test it, fetch updates, and clone a local copy. help on options to git appreciated.
01:02 japhb__ [Coke]: Roger that, give me a sec.
01:03 jnthn [Coke]: cd /path/to/main/repo; git fetch; cd /path/to/workspace; rm -rf the-clone; git clone /path/to/main/repo the-clone # or so
01:04 jnthn japhb__++ may be a cuter solution though :)
01:06 [Coke] ah, so you're not using a mirror or a bare copy anywhere in there?
01:08 avuserow [Coke]: you might be interested in git-new-workdir from the contrib directory of git.git: http://nuclearsquid.com/writings/git-new-workdir/
01:08 jnthn [Coke]: /path/to/main/repo can be bare
01:09 jnthn Haven't played with the mirroring stuff
01:09 jnthn Though I suspect it is gonna land in one of the $dayjob git courses at some point so I'd better figure it out :)
01:10 [Coke] I'll still with regular for now. Danke.
01:10 [Coke] *stick
01:10 jnthn pzh
01:10 jnthn ok, time or some rest :)
01:10 jnthn *for
01:10 * jnthn hopes he actually manages to sleep tonight...
01:10 jnthn &
01:13 japhb__ [Coke], I created a share directory, cloned parrot, nqp, and rakudo normally under that directory, and then use: https://gist.github.com/japhb/8309906
01:13 japhb__ My apologies for slow response ...
01:20 berekuk joined #perl6
01:23 Pleiades` joined #perl6
01:30 japhb__ [Coke], jnthn: IIRC, the reason perl6-bench does --mirror clones for the main repos had to do with working with multiple branches and detached HEADs.  Using normal clones for the main repos didn't work sanely in that case.  The (much more careful) clone and update code used in perl6-bench starts here:  https://github.com/japhb/perl6-bench/blob/master/bench#L35
01:31 japhb__ (That function and the next)
01:31 [Coke] I'll try the simplest thing and see if it works.
01:31 japhb__ er, (That function and the next couple especially -- setup, fetch, and extract )
01:35 [Coke] I wonder, if I manually put a copy of nqp/moar/parrot/whatever in place, if rakudo's configure.pl is still going to try to do a git pull
01:36 japhb__ It won't try to clone, but ISTR it will fetch and then checkout a particular rev.
01:43 berekuk joined #perl6
01:47 [Coke] working so far. Pugs is a mess, I think I need to kill it. :|
01:48 diakopter to be fair, it was working ok on feather
01:48 dalek nqp: e264d33 | diakopter++ | / (3 files):
01:48 dalek nqp: add serialization tests for moar
01:48 dalek nqp: review: https://github.com/perl6/nqp/commit/e264d33d3d
01:49 diakopter uhm.  s/serialization/continuation/
01:49 diakopter .oO( what as I working on again? )
01:52 [Coke] ah, yes. I blame diakopter, then!
01:53 [Coke] seriously, though, I am even less a haskell hacker than I am the other kinds. it needs some love.
01:54 [Coke] any particular reason that we check things out in "nqp,parrot", but "nqp,nqp/MoarVM" ?
01:54 japhb__ hugme: hug Pugs
01:54 * hugme hugs Pugs
01:55 berekuk joined #perl6
01:57 [Coke] japhb__: only problem seems to be trying to clone and checkout a branch.
01:57 [Coke] ... and Im getting kicked out. back laters.
01:59 ssutch_ joined #perl6
02:01 xinming_ joined #perl6
02:04 jnap joined #perl6
02:06 berekuk joined #perl6
02:09 Vlavv__ joined #perl6
02:09 Timbus joined #perl6
02:11 sorear joined #perl6
02:18 berekuk joined #perl6
02:21 [Coke] ... and --mirror does what I want here.
02:23 Timbus joined #perl6
02:26 japhb__ [Coke], I suspect that eventually you'll end up converging on the way that perl6-bench does it, because that was pretty much arrived at via painful debugging ... but I completely understand not wanting to jump from zero to there.
02:27 [Coke] not sure if you're also doing this, but I rebuild from scratch every day and don't install outside the build dir.
02:29 berekuk joined #perl6
02:30 [Coke] huh. rakudo.jvm checks out a tag instead of leaving nqp at master.
02:30 jnap joined #perl6
02:39 japhb [Coke]: Same here (rebuild from scratch and don't install outside build dir).  The latter is part of why the update-rakudo is somewhat baroque -- it finds the current nom HEAD, locally clones a separate copy of Rakudo for each backend into rakudo-<git revision>, then clones the dependencies in each of them, builds them, bootstraps panda and some basic modules into them, and then finally if all went well swizzles rakudo-parrot and rakudo-jvm symlinks -- which all
02:40 diakopter which all m
02:41 berekuk joined #perl6
02:41 japhb "which all my other aliases and scripts refer to."
02:52 dalek perl6-roast-data: 6bfd71b | coke++ | bin/moarborts.p5:
02:52 dalek perl6-roast-data: Add moarborts one-shot script
02:52 dalek perl6-roast-data: review: https://github.com/coke/perl6-roast-data/commit/6bfd71b8e3
02:52 dalek perl6-roast-data: 08bfae7 | coke++ | .gitignore:
02:52 dalek perl6-roast-data: ignore this file that we create with the cron job.
02:52 dalek perl6-roast-data: review: https://github.com/coke/perl6-roast-data/commit/08bfae779e
02:52 dalek perl6-roast-data: d805a37 | coke++ | / (8 files):
02:52 dalek perl6-roast-data: Keep -one- copy of each repo needed
02:52 dalek perl6-roast-data:
02:52 dalek perl6-roast-data: When building, first fetch once per repo,
02:52 dalek perl6-roast-data: then clone locally.
02:52 dalek perl6-roast-data:
02:52 dalek perl6-roast-data: japhb++ for inspiration
02:52 dalek perl6-roast-data: review: https://github.com/coke/perl6-roast-data/commit/d805a37403
02:53 [Coke] well, there's a first pass that seems to do ok. haven't done a full run yet, but all the pieces seem ok. also cleaned up a few other things.
02:54 [Coke] ... and NOW I remember the last thing I wanted to do.
02:56 diakopter :)
02:57 FROGGS[mobile] joined #perl6
02:58 berekuk joined #perl6
02:58 dalek perl6-roast-data: 4392d52 | coke++ | bin/niecza.sh:
02:58 dalek perl6-roast-data: fix niecza clone
02:58 dalek perl6-roast-data: review: https://github.com/coke/perl6-roast-data/commit/4392d52baa
02:58 dalek perl6-roast-data: f277fb6 | coke++ | bin/ (5 files):
02:58 dalek perl6-roast-data: treat roast like all the other repos.
02:58 dalek perl6-roast-data: review: https://github.com/coke/perl6-roast-data/commit/f277fb62ef
02:59 * [Coke] kicks this off now to see if it survives, and will check in later.
02:59 [Coke] ~~ japhb++
03:01 fgomez joined #perl6
03:08 overflowing[_] joined #perl6
03:09 berekuk joined #perl6
03:20 berekuk joined #perl6
03:25 fgomez joined #perl6
03:30 laidback_01_ joined #perl6
03:32 berekuk joined #perl6
03:36 laidback_01 joined #perl6
03:43 berekuk joined #perl6
03:52 [Coke] [A
03:55 berekuk joined #perl6
03:57 V_S_C joined #perl6
03:59 V_S_C @tadzik sorry, that was almost 2am my time, & I was sleepy..
04:06 berekuk joined #perl6
04:18 berekuk joined #perl6
04:20 cognominal joined #perl6
04:22 jnap joined #perl6
04:29 berekuk joined #perl6
04:35 preflex joined #perl6
04:41 berekuk joined #perl6
04:52 berekuk joined #perl6
05:03 abnorman joined #perl6
05:03 berekuk joined #perl6
05:04 Exodist joined #perl6
05:10 FROGGS joined #perl6
05:15 berekuk joined #perl6
05:20 aindilis joined #perl6
05:23 jnap joined #perl6
05:27 berekuk joined #perl6
05:31 thou joined #perl6
05:38 berekuk joined #perl6
05:42 Exodist joined #perl6
05:45 atroxaper joined #perl6
05:50 berekuk joined #perl6
06:02 berekuk joined #perl6
06:05 mcglk joined #perl6
06:05 Alina-malina joined #perl6
06:14 berekuk joined #perl6
06:21 mcglk joined #perl6
06:25 berekuk joined #perl6
06:29 grep0r joined #perl6
06:30 moritz \o
06:31 darutoko joined #perl6
06:36 raydiak o/
06:36 berekuk joined #perl6
06:38 [Sno] joined #perl6
06:48 berekuk joined #perl6
06:53 berekuk joined #perl6
06:58 ruoso joined #perl6
06:59 berekuk joined #perl6
07:01 atrodo joined #perl6
07:02 dylanwh joined #perl6
07:02 FROGGS[mobile] joined #perl6
07:08 sftp joined #perl6
07:10 berekuk joined #perl6
07:14 xinming joined #perl6
07:15 cognominal joined #perl6
07:16 walklan joined #perl6
07:17 walklan left #perl6
07:18 c1sung joined #perl6
07:20 walklan joined #perl6
07:21 walklan left #perl6
07:21 berekuk joined #perl6
07:24 jnap joined #perl6
07:33 berekuk joined #perl6
07:36 hummeleBop joined #perl6
07:39 V_S_C joined #perl6
07:44 berekuk joined #perl6
07:45 kaleem joined #perl6
07:52 FROGGS joined #perl6
07:56 berekuk joined #perl6
08:01 labster joined #perl6
08:02 zakharyas joined #perl6
08:07 berekuk joined #perl6
08:18 berekuk joined #perl6
08:21 dayangkun joined #perl6
08:25 SamuraiJack_ joined #perl6
08:25 jnap joined #perl6
08:26 FROGGS[mobile] joined #perl6
08:30 berekuk joined #perl6
08:42 berekuk joined #perl6
08:47 xinming_ joined #perl6
08:50 sqirrel joined #perl6
08:53 berekuk joined #perl6
09:01 sqirrel joined #perl6
09:07 berekuk joined #perl6
09:16 berekuk joined #perl6
09:17 fhelmberger joined #perl6
09:19 darutoko joined #perl6
09:26 jnap joined #perl6
09:28 kivutar joined #perl6
09:28 berekuk joined #perl6
09:32 SamuraiJack__ joined #perl6
09:39 berekuk joined #perl6
09:40 pecastro joined #perl6
09:42 kaleem joined #perl6
09:42 dakkar joined #perl6
09:43 finanalyst joined #perl6
09:44 pecastro joined #perl6
09:51 berekuk joined #perl6
09:56 FROGGS[mobile] joined #perl6
10:02 berekuk joined #perl6
10:07 V_S_C joined #perl6
10:11 V_S_C @tadzik today I tried starting with rakudo-star-2013.09.tar.gz
10:12 V_S_C As the stack trace was too long for command prompt
10:12 tadzik did it make a difference?
10:12 tadzik uh
10:12 V_S_C I tried reinstalling panda
10:13 tadzik well, Rakudo Star should have panda included
10:13 V_S_C & got the same stack trace as yesterday with similar numbers at the end of lines
10:13 tadzik no need to install it again
10:13 V_S_C yes it did
10:13 tadzik I've never seen an error like you showed yesterday, I've no idea where it might come from
10:14 berekuk joined #perl6
10:14 V_S_C Is there some link outlining manual modules usage
10:15 tadzik you can install them with ufo
10:15 tadzik ufo should be bundled with rakudo star as well
10:16 V_S_C Maybe I can share my toolchain details so rakudo team might know
10:16 V_S_C I'm using ActiveState Perl 64 bit on Windows 2008R2
10:16 V_S_C & the Visual Studio Express Edition
10:17 spider-mario joined #perl6
10:17 V_S_C VS2013 x64 Cross Tools Command Prompt
10:17 tadzik I've absolutely zero experience with windows, sorry
10:17 tadzik but I know at least 3 people in here who might know something :)
10:18 V_S_C thats alrite
10:18 V_S_C cause I learnt scripting on bash but the resources I got are MS
10:19 V_S_C & its one more reason I'm interested in PERL
10:19 V_S_C I dont know anything about PowerShell :)
10:20 V_S_C I tried using rakudo * pre-built binary
10:21 V_S_C but it didnt have ICU
10:21 V_S_C so panda install FastCGI said no ICU installed
10:21 V_S_C thats why I built it on my own
10:22 V_S_C I know some C++
10:25 berekuk joined #perl6
10:26 jnap joined #perl6
10:26 rindolf joined #perl6
10:29 hoelzro morning #perl6
10:34 V_S_C the stack trace from the included Panda was too long.
10:35 daniel-s_ joined #perl6
10:36 berekuk joined #perl6
10:48 berekuk joined #perl6
10:54 V_S_C @tadzik kindly follow up with the 3 persons. I'll check-in before calling it a day. From my time-zone I can only stay-in on the inactive hours here.
10:55 V_S_C For now I'm starting with just copying files to the respective directories..
11:02 masak noon, #perl6
11:05 kaleem joined #perl6
11:20 salv0 joined #perl6
11:27 jnap joined #perl6
11:43 mavcunha joined #perl6
11:43 berekuk joined #perl6
11:43 usernotexists joined #perl6
11:52 AW3i joined #perl6
11:53 berekuk joined #perl6
11:59 AW3i joined #perl6
12:08 berekuk joined #perl6
12:28 jnap joined #perl6
12:29 berekuk joined #perl6
12:37 berekuk joined #perl6
12:50 berekuk joined #perl6
12:50 denis_boyun joined #perl6
13:00 AW3i joined #perl6
13:04 berekuk joined #perl6
13:08 kbaker joined #perl6
13:15 berekuk joined #perl6
13:15 moritz noon+2h, #perl6 :-)
13:21 masak n+2h, moritz
13:23 diakopter n+
13:23 tadzik n++
13:23 diakopter n++++
13:24 diakopter nhh
13:26 dalek perl6-roast-data: e8a1465 | coke++ | bin/cull:
13:26 dalek perl6-roast-data: Track log directory change
13:26 dalek perl6-roast-data: review: https://github.com/coke/perl6-roast-data/commit/e8a1465aa2
13:27 berekuk joined #perl6
13:29 jnap joined #perl6
13:46 PacoAir joined #perl6
14:00 [Coke] [A
14:07 Alina-malina joined #perl6
14:15 kivutar joined #perl6
14:36 dalek nqp: 74d93ae | (Tobias Leich)++ | t/qregex/rx_metachars:
14:36 dalek nqp: strip U+180e from \s tests (Mongolian Vowel Separator)
14:36 dalek nqp: review: https://github.com/perl6/nqp/commit/74d93aeb6d
14:39 notjack joined #perl6
14:39 jnap joined #perl6
14:43 zakharyas joined #perl6
14:43 dylanwh joined #perl6
14:43 atrodo joined #perl6
14:43 abnorman joined #perl6
14:43 preflex joined #perl6
15:04 ajr joined #perl6
15:04 btyler joined #perl6
15:07 bluescreen10 joined #perl6
15:23 masak FROGGS: so -- wait -- \x180e is in both \h and \v, but *not* in \s ?
15:24 masak (or did you mean to remove it from \h and \v as well?)
15:24 masak actually, two things surprise me here:
15:24 masak (a) I thought \s was the union of \h and \v (but now \x180e is in the latter two but not in the former)
15:25 masak (b) I thought \h and \v were disjoint (but I see \x180e in both)
15:26 kaleem joined #perl6
15:27 notjack does Larry do SOTO anymore?
15:28 thou joined #perl6
15:29 masak yes.
15:30 notjack The latest I can find is '07 - where are the newer ones archived?
15:30 notjack or, in particular, is there one for last year?
15:33 treehug88 joined #perl6
15:33 masak yes.
15:33 masak but I don't believe they've been archived for a while.
15:34 masak archiving is subject to someone's volunteering to archive it...
15:34 denis_boyun_ joined #perl6
15:36 logie_ joined #perl6
15:38 [Coke] so... every rakudo test that is todo'd or skipped should have an RT, aye?
15:38 notjack that's sad
15:38 [Coke] don't be sad, be a volunteer!
15:38 masak notjack: so, I'll count on you recording the SOTO for 2014, then?
15:38 [Coke] then you'll be too tired to be sad. :P
15:38 [Coke] but yes, it's sad.
15:39 notjack no, I'm too sad to volunteer.
15:39 * notjack curls up in a ball
15:39 btyler I think the 2013 edition is on youtube? it was given at a yapc, right?
15:41 notjack just searched YouTube for "yapc state of the onion 2013".  Got "The Velociraptor of Christmas Future".  Not what I was looking for, but seems apt.
15:42 masak notjack: mst tends to give "state of the velociraptor" talks on YAPCs.
15:42 masak mst++, I mean.
15:43 FROGGS joined #perl6
15:43 masak notjack: at least part of the reason for him doing so is filling in the visionary/guiding-light vacuum left as Larry started focusing more on Perl 6 than Perl 5.
15:43 masak FROGGS: http://irclog.perlgeek.de/perl6/2014-01-08#i_8091728
15:44 FROGGS masak: correct
15:45 FROGGS that is why I asked (in #moarvm) how I am supposed to handle that
15:46 masak FROGGS++
15:46 grondilu joined #perl6
15:46 masak I missed that.
15:46 FROGGS jnthn said it would be okay for nqp to just make the test pass, but not for roast
15:46 FROGGS so, that thing is not whitespace anymore but has zero width
15:46 notjack picking a velociraptor as a colophon is a positive indicator that he's crazy enough to do the job
15:46 FROGGS see http://www.unicode.org/L2/L2013/13004-vowel-sep-change.pdf
15:47 benabik .oO( colomon's a velociraptor? ...  oh, coloPHon )
15:48 [Coke] for roast, I think the right answer is to either:
15:48 [Coke] 1) make the tests work for -any- version of unicode implementation in the wild might use, as long as they're "real unicode"
15:48 [Coke] 2) conditionalize the tests based on which version of unicode a given implementation is providing.
15:49 colomon benabik: pretty sure I'm too big to be a velociraptor
15:49 berekuk joined #perl6
15:49 [Coke] (perl6-roast-data)... getting those intermediate git repos setup was far less satisfying than I hoped. (it's all hidden behind a cron job that already takes forever. :P)
15:58 Psyche^ joined #perl6
16:06 * FROGGS waits for a new aborts.md to appear :o)
16:13 REPLeffect joined #perl6
16:17 Piers_ joined #perl6
16:34 [Coke] sooon.
16:36 [Coke] S29...
16:37 [Coke] S32...
16:37 FROGGS :o)
16:38 [Sno] joined #perl6
16:39 masak this is so exciting!!!
16:40 FROGGS it is
16:41 diakopter notjack: I can confirm they removed the camera from the tripod before Larry gave the soto this year at oscon
16:42 diakopter er, last year now I suppose
16:44 masak but only to get closeups of Larry, right? :)
16:45 berekuk joined #perl6
16:45 diakopter I looked around but didn't see anyone holding up a camera
16:46 masak aww :/
16:46 diakopter (and my phone was dead) :(
16:46 diakopter it was mostly a repeat of yapc though
16:47 [Coke] FROGGS: ... but now I have to head to lunch. You'll get it after I get back. :)
16:47 diakopter (both na and eu)
16:47 FROGGS [Coke]: have a nice one :o)
16:47 diakopter er. maybe I'm wrong on similarity to yapc eu/na. my memory is poor.
16:48 diakopter notjack: http://www.youtube.com/watch?v=tTbVPb-NhLU
16:50 arnsholt Hmm. The Catalan numbers grow O(n!), don't they?
16:55 arnsholt And I guess n^(n-2) is O(n^n), but what is the relation of that to O(n!)?
16:55 arnsholt masak: You happen to know this off-hand?
16:56 masak hm.
16:57 masak feels like n^n would win, but I'm not sure.
16:57 arnsholt Yeah, I'm not sure either
16:57 arnsholt Doesn't matter though
16:58 arnsholt Both of them are bad enuogh that I can say that enumerating that many options is intractable =)
16:58 benabik I think 2^n < n! < n^n
16:58 masak well, now you've made me interested :)
16:58 masak what benabik said.
16:59 masak &
16:59 arnsholt Well, the Catalan numbers I get as the number of possible parse trees for a sentence when using a CFG
17:00 benabik 2! = 2, 2^2 = 4.  n! = n (n-1)!, n^n = n (n-1)^n = n (n-1) (n-1)^(n-1)
17:00 arnsholt n^(n-2) is the number of parses using a different notion of syntax, where the words are nodes and the parse is a directed tree over the nodes
17:01 arnsholt Right, right. So n^n grows faster
17:16 notjack n^n grows faster than n! and 2^n because n^n ~~ n*n*n*n ... and n! ~~ n*(n-1)*(n-2)... 2^n ~~ 2*2*2*2
17:17 berekuk joined #perl6
17:17 mohij joined #perl6
17:18 arnsholt notjack: Durr. That is of course the simplest argument. Cheers!
17:18 arnsholt notjack++
17:36 notjack how would one express n^n / n! in point-free p6?
17:37 * moritz doesn't even know what point-free means
17:37 notjack the OEIS says the quotient grows as exp(pi * (2*n + 1/3) )
17:38 notjack without variables
17:38 berekuk joined #perl6
17:38 moritz but parameters are allowed?
17:38 moritz $N ** $N / ([+] .. $N)
17:38 notjack parameters are just bound varialbes
17:38 moritz $N ** $N / ([+] 1 .. $N)
17:39 locsmif joined #perl6
17:39 moritz parameters are also read-only by default
17:39 notjack oh yeah, p6 doesn't have ! ... so that it can have a bunchb of tutorials bragging how easy it is to implement ! in p6 ;)
17:40 notjack what's the perlbot's name?
17:40 moritz though of course the implementation of reduction contains variables
17:40 moritz p: given 8 { say $_ ** $_ / ([*] 1 .. $_) }
17:40 camelia rakudo-parrot 8a486a: OUTPUT«416.101587␤»
17:40 notjack r: sub {^n ** ^n / ([+] 1..^n)} 10
17:40 camelia rakudo-jvm 8a486a: OUTPUT«[31m===[0mSORRY![31m===[0m Error while compiling /tmp/KPt2piUns9�HyperWhatever (**) not yet implemented. Sorry. �at /tmp/KPt2piUns9:1�------> [32msub {^n **[33m�[31m ^n / ([+] 1..^n)} 10[0m�»
17:40 camelia ..rakudo-parrot 8a486a: OUTPUT«[31m===[0mSORRY![31m===[0m Error while compiling /tmp/zl0NwDqTgz�HyperWhatever (**) not yet implemented. Sorry. �at /tmp/zl0NwDqTgz:1�------> [32msub {^n **[33m�[31m ^n / ([+] 1..^n)} 10[0m�»
17:40 notjack oh, given is neat
17:41 notjack how is that different from a sub invocation?
17:41 notjack I guess you don't actually write sub...
17:41 moritz correct
17:41 notjack a subliminal sub
17:41 moritz it's just a block
17:42 moritz so it's not named, and immediately invoked, and transpared to return
17:43 notjack r: given chance {$_/2;}
17:43 camelia rakudo-parrot 8a486a: OUTPUT«[31m===[0mSORRY![31m===[0m Error while compiling /tmp/QK5ZSMKlwvâ�¤Missing blockâ�¤at /tmp/QK5ZSMKlwv:1â�¤------> [32mgiven chance {$_/2;}[33mâ��[31m<EOL>[0mâ�¤    expecting any of:â�¤        postfixâ�¤        statement endâ�¤        statement mo…»
17:43 camelia ..rakudo-jvm 8a486a: OUTPUT«[31m===[0mSORRY![31m===[0m Error while compiling /tmp/j8Sb8t1mvyâ�¤Missing blockâ�¤at /tmp/j8Sb8t1mvy:1â�¤------> [32mgiven chance {$_/2;}[33mâ��[31m<EOL>[0mâ�¤    expecting any of:â�¤        postfixâ�¤        statement endâ�¤        statement modif…»
17:43 notjack perl won't give me half a chance :(
17:43 masak if I'm allowed to use $_, I can write it in point-free.
17:44 notjack you just did, didn't you?
17:44 notjack or rather moritz++ did
17:44 masak oh, yes.
17:44 masak moritz++
17:44 * masak fades out again
17:44 notjack I was thinking more along the lines of a fork in J, a-la ^~ % !
17:46 * moritz doesn't J
17:47 notjack http://www.jsoftware.com/zippy.gif
17:48 notjack ^~ means n^n (in general, infix_operator~ x means x infix_operator x)
17:48 moritz :-)
17:48 notjack the fork then means (^~ x) % (! x)
17:48 notjack except you don't need the x....
17:49 moritz that kinda appeals to me
17:50 notjack I'm sure it can be done rather simply with a grammar
17:50 notjack J's parser table only contains 9 rules and 2 exceptions
17:51 notjack I've made the attempt a couple times over the years as p6 has matured
17:51 notjack but never made much progress
17:51 notjack How's p6's current performance over arrays?
17:52 moritz rakudo's: not good
17:52 moritz niecza's is much better
17:52 SamuraiJack__ joined #perl6
17:52 timotimo playing with nqp-level arrays should be much faster, though, so if you implement your J compiler in nqp, maye it'll have acceptable speed
17:52 timotimo well, the compiler doesn't have to be in nqp, but the runtime could. or directly in QAST
17:52 notjack s/compiler/interpreter/
17:53 berekuk joined #perl6
17:53 moritz yes, NQP is much faster
17:53 notjack I think it'd be sig. work to get acceptable performance, because I'd have to reimplement arrays
17:54 timotimo oh?
17:54 timotimo is nqp::list not good for your purposes?
17:54 notjack J's fundamental datatype is [number_of_dimensions,[array of dimension lengths],[ravel of value array]]
17:54 timotimo a ravel is a flattened array?
17:54 notjack i.e. it's rectilinear/orthotopic
17:55 timotimo i don't know either of those words :D
17:55 notjack J doesn't do the sub-array or ragged-array thing
17:55 notjack uh, rectangular... but more so
17:55 notjack yes, a ravel is a flattened array
17:55 timotimo in my imagination, something that's more so than a rectangle is a square :\
17:55 notjack haha
17:56 notjack it's an N-dimensional rectangle
17:56 timotimo ah
17:56 timotimo so a line, rectangle, block, ...
17:56 notjack sub-lists must all have the same length
17:56 notjack as must their sub-lists
17:56 notjack yes
17:56 timotimo what's the english mathematical term for the block shape again?
17:56 notjack prism?
17:56 timotimo hm, not quite what i thought of
17:56 grondilu parallelipiped
17:56 timotimo unless you limit it to "a prism of a rectangle"
17:57 notjack I thought that was a sleeping walrus...
17:57 timotimo grondilu: that doesn't have to have 90 degree angles in the corners, does it?
17:57 timotimo i thought the parallelpiped is the 3d version of a parallelogram
17:57 grondilu timotimo: oops indeed
17:57 timotimo :)
17:57 notjack orthotope
17:57 notjack adjective: orthotopic
17:58 timotimo ah, so that's exactly the thing you said before
17:58 * timotimo only had german math class
17:58 notjack yes; I thought you meant specifically 3D
17:58 notjack (line, rectangle, 3d-prism-block-thing, 4d-thing)
17:58 timotimo ah!
17:59 SamuraiJack joined #perl6
17:59 timotimo "Cuboid", says wikipedia
17:59 zakharyas joined #perl6
18:00 grondilu yep, found it as well.  It's also called "rectangular parallelepiped".  I was not far then.
18:01 notjack that sounds like a very uncomfortable walrus
18:01 notjack shades of bonsai kitten
18:02 timotimo yup :)
18:04 notjack I think "bonsai walrus" is the first thing I've googled in a decade that didn't result in an exact image of what I was googling for
18:05 berekuk joined #perl6
18:13 mcglk joined #perl6
18:18 berekuk joined #perl6
18:22 beastd joined #perl6
18:27 jnap joined #perl6
18:34 sqirrel joined #perl6
18:40 labster joined #perl6
18:42 hoverboard joined #perl6
18:47 treehug88 joined #perl6
18:50 pecastro joined #perl6
18:51 berekuk joined #perl6
18:57 berekuk joined #perl6
19:05 masak interesting. http://jlongster.com/Stop-Writing-JavaScript-Compilers--Make-Macros-Instead
19:06 * masak likes the way to define macros
19:06 * masak (slightly) dislikes the solution to hygiene
19:08 segomos_ masak: thanks for the link
19:16 ajr joined #perl6
19:19 * PerlJam reads to find out how hygiene is accomplished in js
19:25 dmol joined #perl6
19:35 moritz masak: interesting link, thanks for sharing
19:39 ssutch joined #perl6
19:55 Rotwang joined #perl6
20:07 masak there's a HN discussion too: https://news.ycombinator.com/item?id=7025261
20:15 abnorman joined #perl6
20:19 hoelzro joined #perl6
20:26 LordVorp joined #perl6
20:30 hoverboard left #perl6
20:32 mohij joined #perl6
20:38 * masak finds himself wanting a hybrid of 'rename' and 'git mv'... :/
20:39 moritz masak: patch rename to accept an option for a different rename command
20:39 moritz masak: so that you can call rename -c 'git mv' 's/a/X/' *
20:40 dalek rakudo/moar-support: 9333da3 | jnthn++ | src/core/GatherIter.pm:
20:40 dalek rakudo/moar-support: Make gather properly lazy, using continuations.
20:40 dalek rakudo/moar-support:
20:40 dalek rakudo/moar-support: This eliminates the cheating implementation we've used so far.
20:40 dalek rakudo/moar-support: review: https://github.com/rakudo/rakudo/commit/9333da3346
20:43 [Coke] nice.
20:44 [Coke] From reading that it looks like the fix was "do stuff in nqp". :)
20:45 dalek nqp: 624dae2 | jnthn++ | t/moar/01-continuations.t:
20:45 dalek nqp: Todo two continuation tests on Moar.
20:45 dalek nqp:
20:45 dalek nqp: Would be good to look into them at some point, but they don't cover
20:45 dalek nqp: things that Rakudo depends on for gather/take.
20:45 dalek nqp: review: https://github.com/perl6/nqp/commit/624dae2b50
20:45 jnthn [Coke]: Well, it was also "implement continuations in Moar" :)
20:45 jnthn Which is what I've worked on last night / today.
20:49 raydiak r: my $t = now; for 1..1e4 {}; say now - $t;
20:49 camelia rakudo-jvm 8a486a: OUTPUT«0.798␤»
20:49 camelia ..rakudo-parrot 8a486a: OUTPUT«2.6250796␤»
20:50 raydiak wish my machine performed like that :P
20:52 raydiak anyone know what kind of box the eval bot runs on?
20:53 moritz raydiak: dual core box from digital ocean
20:54 moritz 2GHz 64bit CPU
20:55 raydiak so it's not a VPS?
20:55 moritz /proc/cpuinfo says: model name      : QEMU Virtual CPU version 1.0
20:55 raydiak mine is DO also, but the $5/mo plan, and I get about 20% of that performance :)
20:56 FROGGS my timings: :j(1.465), :r(1.2681967), :m(20.565718) # Intel(R) Core(TM) i5-3320M CPU @ 2.60GHz
20:56 raydiak thanks guys...data points are exactly what I was looking for :)
20:57 moritz 21s on moar on my laptop
20:57 * moritz doesn't have perl-j or -p built atm
20:57 FROGGS awww, shame on you! :P
20:58 jnthn Timing stuff on Rakudo on Moar at the moment is going to yield mostly awful results thanks to the binder stuff being slow-pathed...
20:58 FROGGS sure sure
20:58 jnthn That's the first thing I'll work on once we get spectest in shape. :)
20:58 moritz masak: if you want to patch rename, it's in this distribution: https://metacpan.org/pod/release/RMBARKER/File-Rename-0.20/rename.PL
20:58 raydiak I was wondering what was making the new Pray preview I'm working on so slow...lo and behold, it's not the string concat on JVM, but the time required to run the loop itself
20:59 raydiak (even if it's empty)
20:59 FROGGS I hope nobody thinks that this is The High Performance Build of MoarVM (TM)
20:59 moritz raydiak: try  loop (my int $x = 0; $x < $limit; $x = $x + 1) { ... }  instead of for (1..$limit) { }
21:00 moritz raydiak: should be 5x or 10x faster
21:01 raydiak moritz++: you've saved the day!
21:01 moritz raydiak: don't take my word, measure :-)
21:02 raydiak moritz: did before I responded :)
21:03 [Coke] jnthn: spectest in shape: what percentage counts for that?
21:04 jnthn [Coke]: > 99% :)
21:05 jnthn bbi20
21:11 [Coke] jnthn: I assume that it won't count if I switch r.jvm to use the eval server and it dies again. :)
21:14 ivanshmakov joined #perl6
21:17 dayangkun joined #perl6
21:19 panchiniak joined #perl6
21:28 pecastro joined #perl6
21:33 kivutar joined #perl6
21:35 jnthn [Coke]: That's cheating :P
21:35 [Coke] figured. :)
21:36 [Coke] tonight, I'll see if it's magically working, but won't switch it back if not.
21:36 [Coke] is there a theoretical improvement we can make to startup time?
21:36 jnthn Now it doesn't have to compile Test.pm every time, it *may* be better...
21:37 jnthn I'd imagine there's things we can do, but I'd need to go profiling to know where the big costs are.
21:43 berekuk joined #perl6
21:44 [Coke] it's really not much better (probably drowned out over sheer # of spec tests)
21:44 [Coke] s/not much/ ... more like it doesn't feel much better, I'm just eyeballing it as the log scrolls by
21:48 berekuk joined #perl6
21:50 panchiniak left #perl6
21:50 itz I did notice calculating PI was an order of magnitude slower on moar than parrot :)
21:50 [Coke] *sadface*
21:54 jnthn Not surprising at this point
21:55 FROGGS umm, status.github.com says "We're investigating reports of outages across GitHub and Gist."
21:58 masak it's a bloody outage!
21:58 FROGGS I know taht!
21:59 treehug88 joined #perl6
22:02 nwc10 jnthn: why isn't it surprising that Moar is slower than Parrot for calculating Pi? It's not obvious to me.
22:03 masak Parrot has been calculating pi for years :)
22:04 jnthn nwc10: Every signature bind goes through a torturous slow path at the moment.
22:04 itz 1. make it work 2. make it work fast I suppose
22:04 jnthn nwc10: Since the sig binder is used for everything, and on MoarVM it was only really sane to write it in NQP
22:05 jnthn nwc10: So basically, we're interpreting a hot-path primitive doing lots of work, even for the simple cases.
22:05 masak (with no JIT)
22:06 jnthn nwc10: I suspect we'd be a couple of weeks behind where we are now if I'd not gone this "get stuff done first" route...
22:06 jnthn *done
22:06 nwc10 OK. Thanks. That was a much more comprehensive answer than I expected
22:06 jnthn Anyway, it's the first thing I'll tend to once we're in the 99% spectest range.
22:06 nwc10 and I agree
22:07 jnthn Also, MoarVM itself has had relatively little profiling. :)
22:08 * jnthn is looking forward to the performance work, but would much rather do it with spectest looking good, to find mistakes :)
22:08 masak jnthn++ # GTD-first development
22:08 nwc10 I think you're right to concentrace on getting the spectest sane first
22:08 nwc10 so that others can work on Star
22:08 nwc10 while you work on speed
22:08 nwc10 or curry and Imperial stout, or whatever is important.
22:09 * nwc10 assumes that creating the Imperial Stout is actually outsourced.
22:09 jnthn Yes, I'm hopeful that things like Panda, sockets, etc. will be picked up by others.
22:13 FROGGS perl -E 'say Foo->isa("Bar"); say "alive"' # alive <-- that is ridiculous
22:26 masak FROGGS: what is?
22:27 masak FROGGS: the delegation to UNIVERSAL?
22:29 berekuk joined #perl6
22:31 rurban I got: fork no memory error at github before they changed the response. So a process went rogue caused by their DNS outage
22:32 rurban But should be faster than the PAUSE repair (harddisc exchange)
22:36 masak ouch.
22:46 jnthn Seems github is back :)
22:47 kaare_ joined #perl6
22:47 kaare_ joined #perl6
22:47 masak website, yes.
22:47 masak repository storage, no.
22:48 zakharyas joined #perl6
22:51 jnthn ah
22:51 * jnthn blug: http://6guts.wordpress.com/2014/01/08/a-rakudo-on-moarvm-update/
22:51 masak jnthn++
22:52 japhb Ditto that.  :-)
22:52 * jnthn happily notes that "segmentation fault" does not show up in the latest https://gist.github.com/coke/8250608/
22:52 * jnthn hopes it stays that way :)
22:54 * masak .oO( just catch SIGSEGV and handle it somehow ) :P
23:00 masak 'night, #perl6
23:02 japhb o/ masak
23:06 cognominal joined #perl6
23:09 [Coke] jnthn++ blug
23:09 [Coke] rakudo.moar,2014-01-08,93.40%,9a340b1,26576,1750,603,1440,30368,28495,
23:10 japhb Anyone want to take bets on when 95% and 99% will be crossed?
23:13 ssutch joined #perl6
23:14 jnthn r: say (26576 / 93.4 * 95) - 26576
23:14 camelia rakudo-parrot 8a486a, rakudo-jvm 8a486a: OUTPUT«455.263383␤»
23:18 jnthn I think some of the "killed due to slow" ones will be better tomorrow; the gather/take cheat reified 1000 elements whether they were needed or not, which made some tests that only needed a few things take ages.
23:25 [Coke] sweet. once I test this jvm eval server stuff, I'll do another run.
23:26 jnthn I got stuff to do tomorrow, so will be away until late afternoon, I expect
23:29 krunen joined #perl6
23:30 [Coke] NOOOOOO
23:30 [Coke] jnthn, don't you love us? ;)
23:31 BenGoldberg joined #perl6
23:31 jnthn :P
23:32 jnthn Well, I can at least say that I'd *rather* be here than doing tomorrow's errands. :)
23:33 [Coke] hugme: hug jnthn
23:33 * hugme hugs jnthn
23:34 [Coke] hugme: hug hugs, hugging.
23:34 * hugme hugs hugs,
23:41 jnthn Time for some rest...'night o/
23:41 japhb__ o/ jnthn!
23:42 grep0r joined #perl6

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

Perl 6 | Reference Documentation | Rakudo