Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6, 2014-02-09

Perl 6 | Reference Documentation | Rakudo

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

All times shown according to UTC.

Time Nick Message
00:01 raydiak since the (human) language part of my brain is so determined to express itself today, I wrote a sentence which I guess must have been about #perl6:
00:01 raydiak The calamitous, cacophonous calls of the crazy creatures counter-intuitively coalesces into a complex collage of cooperation, creative culture, cognitive canvases, and conceptual constructs, and even calls cosmic comprehension closer to our carbon.
00:01 virtualsue joined #perl6
00:03 Shellcat joined #perl6
00:05 timotimo http://t.h8.lv/p6bench/2014-02-09-for_improvements.html
00:05 timotimo i'll try to enlarge these graphs so that you can actually see something ...
00:08 timotimo i've managed to make it wider, but i see no way to make the individual graphs less flat
00:11 timotimo everything for-related on moarvm got a whole bunch faster
00:11 timotimo well, not really "a whole bunch" - still less than 2x faster, but it's definitely noticable in the benchmarks where for performance dominates
00:11 timotimo like empty for loops %)
00:13 timotimo oh nice. in visit_2d_indices_while_native, nqp-moarvm went from noticably slower than parrot to a tiny bit faster than parrot
00:13 dalek rakudo/nom: ba1cc33 | jnthn++ | src/Perl6/Actions.nqp:
00:13 dalek rakudo/nom: Fix an attribute access type-o.
00:13 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/ba1cc337ae
00:14 timotimo much more interesting: visit_2d_indices_loop_native sees a dramatic speed difference in parrot
00:15 timotimo but it seems like the performance of visit_2d_indices_cross has regressed significantly for parrot
00:16 jnthn That's an odd one...
00:17 jnthn Time for some rest...'night o/
00:17 timotimo i'll rest, too
00:27 BenGoldberg joined #perl6
00:27 bjz joined #perl6
00:32 bjz_ joined #perl6
00:33 LLamaRider joined #perl6
00:36 bjz joined #perl6
00:37 lue why in  nqp/src/HLL/Grammar.nqp  are (), [], <>, and {} listed twice?
00:37 lue (line 4)
00:41 xenoterracide joined #perl6
00:52 LLamaRider joined #perl6
00:52 Alula joined #perl6
00:59 mavcunha joined #perl6
01:22 nebuchad` joined #perl6
01:22 telex joined #perl6
01:25 Tene_ joined #perl6
01:32 sjn joined #perl6
01:59 FROGGS_ joined #perl6
02:00 Sqirrel_ joined #perl6
02:05 rurban joined #perl6
02:32 dalek rakudo-star-daily: 1e8686a | coke++ | log/ (5 files):
02:32 dalek rakudo-star-daily: today (automated commit)
02:32 dalek rakudo-star-daily: review: https://github.com/coke/rakudo-star-daily/commit/1e8686a7f9
02:32 dalek perl6-roast-data: 23ee590 | coke++ | / (6 files):
02:32 dalek perl6-roast-data: today (automated commit)
02:32 dalek perl6-roast-data: review: https://github.com/coke/perl6-roast-data/commit/23ee590469
02:33 simcop2387 joined #perl6
02:37 grep0r joined #perl6
02:48 rom1504 joined #perl6
02:57 SamuraiJack joined #perl6
03:00 Sqirrel joined #perl6
03:00 colomon joined #perl6
03:00 FROGGS__ joined #perl6
03:10 skids joined #perl6
03:20 grep0r joined #perl6
03:46 regreg joined #perl6
04:05 japhb__ raydiak: I just don't want to leave people waiting on me for long periods of time if I can help it.  Don't like doing it at work, don't like doing it for community projects either.
04:06 japhb__ That said, there's a chance I will be rather squeezed for tuits soon, which would exacerbate the problem, so it was extra important to get as "caught up" as I could now.
04:10 japhb__ LHF for a Rakudo hacker: Implement $*BASETIME.  May be as simple as setting 'GLOBAL::<$BASETIME> = now;' in the right place at the right time.
04:12 sctt joined #perl6
04:29 bjz joined #perl6
04:47 bjz joined #perl6
05:29 rurban_ joined #perl6
05:35 cognominal joined #perl6
05:38 bjz joined #perl6
06:21 cognominal joined #perl6
06:36 japhb_ joined #perl6
06:39 japhb joined #perl6
06:43 moritz m: say now.^name
06:43 camelia rakudo-moar ba1cc3: OUTPUT«Instant␤»
06:52 thou joined #perl6
06:58 PZt joined #perl6
07:03 timotimo o/
07:12 samebchase joined #perl6
07:20 ruoso joined #perl6
07:22 moritz \o
07:22 FROGGS__ morning
07:23 moritz good morning__ FROGGS :-)
07:25 FROGGS morning moritz :o)
07:26 moritz japhb_: if I read S28 correctly, it's supposed to be $*INITTIME, not $*BASETIME
07:45 FROGGS jnthn: for v5 I did: <statementlist=.LANG($*MAIN, 'statementlist')>
07:45 FROGGS jnthn: how do I call statementlist(1) there?
07:50 regreg joined #perl6
08:01 FROGGS jnthn: nvm :o)
08:02 raydiak japhb: makes sense, you sound like a very busy guy. hope you clear your plate enough to spend a little time breathing :) just figured I'd ask how it's going, it's a protocol I tend to follow when a person isn't around much and then puts their affairs in order
08:07 darutoko joined #perl6
08:13 FROGGS segfault when compiling Perl5::Terms :o(  https://gist.github.com/FROGGS/a78e5ce755315e879065
08:14 FROGGS yeah, src/core/interp.c:3481 is op serialize
08:40 raydiak need to start sleeping more. starting right now. good *, * o/
08:46 araujo joined #perl6
08:51 molaf joined #perl6
08:52 nwc10 FROGGS: It struck me that it's useful for Panda to have the alternative of using an external fetching command such as curl or wget
08:53 nwc10 because the external commands are going to cope better with some things, such as SSL, some sorts of proxies, and any new or obscure prototocols that Panda doesn't yet know about
09:04 Rotwang joined #perl6
09:09 dmol joined #perl6
09:13 stevan_ joined #perl6
09:15 xenoterracide joined #perl6
09:17 stevan__ joined #perl6
09:43 Rotwang joined #perl6
09:49 FROGGS nwc10: then it might be even helpful to be able to specify the backend
09:49 berekuk joined #perl6
09:49 FROGGS maybe using an env var
10:05 Rotwang joined #perl6
10:06 virtualsue joined #perl6
10:07 Rotwang joined #perl6
10:27 masak antenoon, #perl6
10:31 FROGGS o/
10:41 bjz_ joined #perl6
10:51 SamuraiJack joined #perl6
10:58 jnthn o/
11:05 masak \o
11:07 * masak happily realizes that 'our' variables can be used inside -n and -p oneliners in Perl 6
11:08 spider-mario joined #perl6
11:09 masak and END works nicely, too!
11:10 berekuk joined #perl6
11:15 jnthn state works out nicely also in that you can use it to run the assignment only the first time.
11:17 stevan_ joined #perl6
11:19 dalek rakudo/jvmbl: 008d9f1 | jnthn++ | src/Perl6/Actions.nqp:
11:19 dalek rakudo/jvmbl: Turn on JVM bind lowering.
11:19 dalek rakudo/jvmbl:
11:19 dalek rakudo/jvmbl: Not everything works out yet, but it can run spectest with (quite a
11:19 dalek rakudo/jvmbl: lot of) regressions.
11:19 dalek rakudo/jvmbl: review: https://github.com/rakudo/rakudo/commit/008d9f1d94
11:24 dalek nqp/jvmbl: 7258671 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
11:24 dalek nqp/jvmbl: Missing deconts (for consistency with MoarVM).
11:24 dalek nqp/jvmbl: review: https://github.com/perl6/nqp/commit/7258671303
11:24 masak jnthn: ooh, yes.
11:27 kaleem joined #perl6
11:31 dalek rakudo/nom: 45bb17e | moritz++ | src/core/Instant.pm:
11:31 dalek rakudo/nom: Implement $*INITTIME for japhb++
11:31 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/45bb17e864
11:40 dalek nqp: 7584c1a | jnthn++ | tools/build/MOAR_REVISION:
11:40 dalek nqp: Get a MoarVM with buffer read/write fixes.
11:40 dalek nqp: review: https://github.com/perl6/nqp/commit/7584c1a74f
11:43 dalek rakudo/nom: 5838d5f | jnthn++ | tools/build/NQP_REVISION:
11:43 dalek rakudo/nom: Bump for fixes to unbust Buf spurt/slurp on Moar.
11:43 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/5838d5f826
11:44 jnthn That should deal with the regressions from the latest [Coke]++ daily run.
11:49 Adriaaan joined #perl6
11:52 dalek rakudo/nom: 46234b5 | moritz++ | docs/ChangeLog:
11:52 dalek rakudo/nom: add some ChangeLog entries
11:52 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/46234b5001
11:52 bjz joined #perl6
12:00 virtualsue joined #perl6
12:03 stevan__ joined #perl6
12:08 dalek roast: 91c81e6 | jnthn++ | S03-operators/assign.t:
12:08 dalek roast: Correct precedence mistake in a test.
12:08 dalek roast: review: https://github.com/perl6/roast/commit/91c81e6f78
12:33 kivutar joined #perl6
12:45 denis_boyun joined #perl6
12:52 kurahaupo_mobile joined #perl6
13:01 beastd joined #perl6
13:10 timotimo yay, progress
13:22 mavcunha joined #perl6
14:14 masak hi from... a train. \o
14:14 masak (yes, I'm a lot less active with Perl 6 these days. yes, I travel a lot more by train these days. yes, those two are related.)
14:14 masak it was so nice to be able to submit a handful of RT tickets the other day.
14:15 FROGGS hi masak :o)
14:22 hoelzro run($cmd) is *supposed* to just run a command, right?
14:23 timotimo masak: don't train rides often cause bursts of activity? :P
14:23 masak they do, sometimes.
14:23 timotimo so you only have unhelpful train rides? :(
14:24 masak but used to be I spent train rides working with Perl 6 a lot more.
14:24 masak these days, less so.
14:24 masak not this one, for example :/
14:24 hoelzro hmm
14:24 hoelzro run() is broken on Moar, I think
14:24 colomon Sum module started failing overnight.
14:24 colomon skids: ^^
14:24 masak hoelzro: gist?
14:24 masak hoelzro: also, can you trace it to a broken spectest?
14:24 hoelzro masak: https://gist.github.com/hoelzro/8899762
14:25 hoelzro I can try
14:25 psch joined #perl6
14:25 psch hi #perl6 o/
14:25 yoleaux 8 Feb 2014 14:16Z <jnthn> psch: Please be a little more careful to update test plans.
14:26 masak hi psch
14:26 psch right, test plans
14:26 psch i blame that again on being exposed to throws_like first... :)
14:27 masak hoelzro: I mean, I would almost find it surprising if moar's run() was busted and that *didn't* trigger a spectest failure.
14:27 hoelzro agreed
14:27 masak psch: why? throws_like also needs to update the test plan...
14:27 hoelzro I don't run the spec tests often myself unless I'm actually working on Rakudo, though =/
14:28 timotimo hoelzro: spectests are about 2x faster on moar than on parrot :)
14:28 hoelzro indeed =)
14:29 psch masak: it does? i haven't seen through how many tests it runs per call. maybe it's better to blame S32-exceptions/misc.t then? it doesn't have a plan
14:29 hoelzro I'll run the full spec test
14:29 stevan_ joined #perl6
14:29 psch in the end of course it's more of a humorous remark, i actually realize that i have to take care about the plans
14:30 masak psch: fwiw, I've forgotten them sometimes, too.
14:30 masak psch: "doesn't have a plan" probably means "has a 'done;' at the bottom". it's either-or.
14:31 timotimo it could also mean "is broken", but that's not the case here :)
14:32 psch well, in any case a reminder can't hurt, i did forget it a few times it seems
14:39 FROGGS psch: just rerun the test file and compare the plan to the tests it runs before pushing
14:43 hoelzro the spec tests *look* fine for moar
14:43 hoelzro er, perl6-m
14:48 masak how can that be? either run() is broken and the test fails, or run() works, or the test isn't testing whether run() works.
14:48 hoelzro I know, it's really confusing...
14:48 hoelzro unless there's something about the testing environment that makes run work
14:50 FROGGS what exactly is wrong?
14:50 hoelzro FROGGS: see my gist
14:50 FROGGS hoelzro: this one? https://gist.github.com/hoelzro/8899762#file-gistfile1-txt-L2
14:51 hoelzro FROGGS: yes
14:51 FROGGS hmmm, works on my box
14:52 hoelzro I'll try pulling, I guess
14:52 hoelzro FROGGS: are you on any special branch?
14:52 FROGGS hoelzro: no, master/master/nom
14:52 FROGGS HEAD
14:52 hoelzro ok, that might have something to do with it
14:52 FROGGS just pulled 30mins ago and rebuild (and reconfigured) everything
14:53 hoelzro I'm on the openpipe branch
14:53 FROGGS ahh
14:53 FROGGS wip-openpipe got merged
14:53 hoelzro did it?
14:53 hoelzro nice
14:54 FROGGS yes, right after S29-os/system.t passed :o)
14:54 hoelzro \o/
14:54 FROGGS on windows too fwiw
14:54 FROGGS so, there is just the osx problem left I hope
14:54 hoelzro nice.
14:57 pippo joined #perl6
14:57 pippo o/ #perl6
14:57 hoelzro ahoy pippo
14:58 timotimo o/
14:58 pippo hoelzro: openpipe implemented does mean that qx/ls/ should work? On my box nope.
14:58 hoelzro pippo: it *should*
14:58 hoelzro it does on mine
14:59 pippo I mean on moarvm. Mine segfaults!
14:59 timotimo yikes!
14:59 pippo :-)
14:59 timotimo how recently did you do a full clean + rebuild?
15:00 hoelzro we had a segfault problem recently
15:00 hoelzro like, yesterday
15:00 hoelzro but it's fixed now
15:00 timotimo oh, that sounds good
15:00 lee__ FROGGS: is this the os x problem you refer to? https://gist.github.com/leedo/c451ee9c86f80807c039
15:00 lee__ trying to build latest moar
15:00 pippo My moar is up to date. I'll do a full clean + rebuild... &
15:01 FROGGS lee__: yes
15:01 lee__ ok, just making sure
15:02 hoelzro whew
15:02 hoelzro run() works for me now
15:06 pippo hoelzro: is it possible to use run like this: run('ls -alrt')?
15:07 timotimo alart!
15:07 timotimo !!DANGAR!!
15:07 hoelzro pippo: I believe you have to do run('ls', '-alrt')
15:07 hoelzro i just discovered that myself =)
15:07 timotimo run <ls -alrt> should work, too?
15:07 timotimo or would that have to be flattened?
15:08 ajr joined #perl6
15:09 pippo Ok thank you both. I'll try. Another (the last :-)) question. Is it possible to have the result of such 'ls -alrt' saved on an array of lines?
15:10 pippo shell('ls -alrt') ==> lines() ==> my @a; or something...
15:10 hoelzro pippo: you probably want qx[ls -alrt], then
15:11 colomon p: say qx[ls -alrt]
15:11 camelia rakudo-parrot 46234b: OUTPUT«qx, qqx is disallowed in restricted setting␤  in sub restricted at src/RESTRICTED.setting:2␤  in sub QX at src/RESTRICTED.setting:9␤  in block  at /tmp/wLMlsaezKk:1␤␤»
15:12 pippo j: say qx[ls -alrt];
15:12 camelia rakudo-jvm 46234b: OUTPUT«total 156␤drwxr-xr-x 13 root        root         4096 Oct  8 17:18 ..␤-rw-rw-r--  1 p6eval_eval p6eval_eval    75 Oct 23 09:26 .selected_editor␤-rw-------  1 p6eval_eval p6eval_eval   734 Oct 23 09:26 .viminfo␤-rw-rw-r--  1 p6eval_eval p6eval_eval 3…»
15:13 pippo j: qx[ls -alrt] ==> lines() ==> my @a; say @a;
15:13 camelia rakudo-jvm 46234b: OUTPUT«total 156 drwxr-xr-x 13 root        root         4096 Oct  8 17:18 .. -rw-rw-r--  1 p6eval_eval p6eval_eval    75 Oct 23 09:26 .selected_editor -rw-------  1 p6eval_eval p6eval_eval   734 Oct 23 09:26 .viminfo -rw-rw-r--  1 p6eval_eval p6eval_eval 39896 Nov…»
15:13 timotimo you'd want to .perl or maybe .elems that :P
15:13 colomon my @a = qx[ls -alrt].lines works nicely
15:13 timotimo yeah, but pippo is a friend of feed syntax
15:13 colomon there's more than one way to do it
15:14 pippo :-))
15:15 pippo Thank you very much to you all: hoelzro, timotimo and colomon.
15:16 pippo timotimo: after full celan and compile. qx[ls] on moarvm segfaults on my machine :-((
15:16 colomon BTW, I tested it locally on moar, and it works there, too.  :)
15:16 Vlavv joined #perl6
15:16 colomon :\
15:17 FROGGS pippo: what OS and architecture?
15:17 timotimo pippo: did you --gen-moar=master or just --gen-moar?
15:17 timotimo because it *might* have installed a non-current moar
15:18 FROGGS pippo: maybe paste perl6-m --version for us?
15:18 mavcunha_ joined #perl6
15:18 pippo I am using slackware linux on an intel machine.
15:18 FROGGS gah, sometimes I hate eclipse >.<
15:18 FROGGS pippo: x64?
15:18 pippo timotimo: --gen-moar
15:19 pippo FROGGS: 32bit
15:19 FROGGS hmmmm, maybe that is the problem?
15:19 pippo Noooooo :-(((
15:20 pippo I have to say that perl6-j works though.
15:23 timotimo pippo: --gen-moar gives you, at least as far as i remember, whatever is in build/tools/MOAR_REVISION of the nqp you get
15:23 timotimo and the nqp you get is pulled from build/tools/NQP_REVISION of the rakudo
15:23 timotimo could you possibly try again with --gen-moar=master --gen-nqp=master?
15:24 pippo Thank you. I'll try... &
15:24 clkao joined #perl6
15:26 timotimo though of course 32 bits may be a problem; i'm not sure how many of our developers are actively trying moar on 32bit systems (:
15:26 timotimo :(
15:26 psch https://gist.github.com/peschwa/8900636 # because i'm sure i'd forget about plans again
15:26 psch works as pre-commit hook for roast
15:27 colomon moritz has been saying moar doesn't work on 32-bit, period.
15:27 timotimo colomon: oh, that's cute! :)
15:27 timotimo er.
15:27 timotimo psch: that's cute!
15:27 timotimo colomon: that's a little bit sad
15:28 FROGGS but when only qx{} is failing...
15:28 psch FROGGS++ # making me indirectly learn a little about git hooks
15:29 psch now back to what-i-actually-was-doing haha
15:29 colomon psch++
15:30 FROGGS :o)
15:31 timotimo animal's rights experts will be happy to know that yaks are regularly being shaved.
15:31 hoelzro should assigning values to %*ENV automatically Str them?
15:32 timotimo once we have coercive types, yeah why not
15:38 * colomon wonders how hard it would be to set up a Linux 32-bit virtual machine....
15:38 * colomon wonders how hard it would be to set up a Linux 32-bit virtual machine….
15:42 nwc10 colomon: which can't actually be an accuate conclusion, as it does built for me on x86 Linux
15:42 nwc10 OK, perl6-m does
15:42 colomon nwc10: does it run?
15:42 kivutar joined #perl6
15:42 nwc10 passes its tests
15:42 colomon moritz was saying moar built but NQP-m and Rakudo-m wouldn't
15:42 colomon interesting.
15:43 nwc10 note, x86, so it might end up being something more subtle than just "32bit"
15:43 FROGGS nwc10: and the qx{ls} example works?
15:43 * colomon wonders if he is confused or if moritz's analysis is a bit off.
15:43 timotimo well, i'd say that not running on 32 bit is less problematic than not running on 64 bit
15:43 colomon timotimo: sure, but it would be best if both worked
15:43 nwc10 I'm suspecting that there are at least 2 architecture or platform specific problems on 32 bit systems
15:43 timotimo of course
15:43 FROGGS well, not running on pippo's machine is problematic though
15:43 nwc10 that is a guess
15:44 timotimo especially since moar's memory usage is so good compared to our other backends
15:44 timotimo (still has a long way to go, i figure)
15:45 pippo timotimo: Finished compiling. Sadly --gen-moar=master does not help either.
15:45 jnthn I've heard there's a problem on Windows 32-bit. Didn't get to recreate it...
15:45 Adriaaan joined #perl6
15:45 jnthn ...but that's a "doesn't build NQP" issue, not a "feature X doesn't work" issue
15:46 timotimo pippo: :(
15:46 timotimo jnthn: should i start my rakudo-jvm benchmarks now-ish? should i compare newer nqp-jvm, too?
15:48 jnthn timotimo: Well, you could do them, may be small improvements there too... I didn't get all the way with the bind lowering thing. Not sure I will today.
15:48 pippo One strange thing is that I have discovered that "q/ls -alrt/ ==> shell()" works on moar. Does this help debugging?
15:49 pippo i.e. does 'qx' and 'shell' have code in common?
15:50 timotimo jnthn: well, i'll leave my desktop behind in a few minutes anyway, wether i do the benchmarks now or an hour later doesn't matter to me
15:50 regreg joined #perl6
15:55 timotimo jnthn: are the revisions sufficiently bumped so that i'll get the performance improvements if i just use the newest rakudo commit?
15:55 timotimo 4 hours ago it says
15:56 timotimo yup, it's up to date. good.
15:57 timotimo running the stuff now. will get back to you with results later
15:57 timotimo o/
15:59 Psyche^_ joined #perl6
16:11 hoelzro alright, my super awesome MoarVM module compatability checker is complete!
16:11 hoelzro https://gist.github.com/anonymous/8901287
16:11 hoelzro that's all of the Rakudo * modules and their build status on MoarVM
16:11 hoelzro summary.txt is what it says
16:11 hoelzro ufo works
16:12 hoelzro so someone please make good use of this information!
16:12 stevan_ joined #perl6
16:15 jnthn The URI failure looks...odd
16:16 FROGGS jnthn: I get that too when I try to make my m-summary
16:17 jnthn It's like it's finding an old CORE.setting.moarvm somehow
16:18 jnthn Or a newer one
16:18 timotimo rm -fr / and try again
16:18 colomon panda: STable conflict detected during deserialization.
16:19 FROGGS ewww
16:19 hoelzro jnthn: I had a better one last night
16:19 hoelzro I pasted it...somewhere
16:19 * colomon thinks MIME::Base64 needs to be a pure p6 implementation first and foremost.  it seems like there's one somewhere in the module, but it never gets used because the tests fail on non-pure version.
16:20 timotimo mhm
16:20 hoelzro jnthn: http://pastebin.com/HtaeMiL5
16:20 hoelzro that was from last night
16:20 hoelzro oh
16:20 hoelzro you know what
16:21 hoelzro I didn't run a make clean first
16:21 hoelzro so that probably f'ed things up
16:21 timotimo parrot andd moar had b64 decoding built in anyway, right? why not expose it as a nqp:: op for all of rakudo
16:27 thou joined #perl6
16:27 molaf_ joined #perl6
16:32 berekuk joined #perl6
16:36 virtualsue joined #perl6
16:41 nwc10 1) does MoarVM still need Base64 for anything other than obsolete serialisation formats
16:41 nwc10 2) to me it feels a bad idea to specialise stuff in C that doesn't need to be in C
16:41 timotimo d'oh, i neglected to start the benchmarks in a tmux
16:42 timotimo so i'll have to periodically check ps for the process
16:42 nwc10 that's somewhat repeating the mistakes of Perl 5
16:43 timotimo nwc10: i'd still vote for a pure-perl6 implementation of base64 of course
16:43 timotimo and i think moar only needed it for serialization stuff
16:43 timotimo but we thankfully got rid of that
16:46 nwc10 FROGGS: I believe that CPAN.pm tries each network "method" in turn until something works. It has configuration for things like "where is your wget program" but I don't see configuration options for what order to use downloaders
16:51 denis_boyun joined #perl6
16:56 kbaker joined #perl6
17:01 pippo timotimo: actually qx/ls/ works on Moar when there are a few files in the dir. Segfaults if the number of files is not a handful.
17:04 hoelzro pippo: nice catch
17:04 hoelzro I wonder if it's because of buffer size
17:04 pippo hoelzro: is it the same on your machine?
17:05 hoelzro I'll try in a second
17:05 hoelzro but I wouldn't be surprised =)
17:06 psch i notice that i am confused
17:06 zakharyas joined #perl6
17:06 psch my refactor has parity with nom except for X::TypeCheck::Binding
17:07 psch in my branch, $got and $expected are swapped
17:07 psch but according to --ll-exception the program doesn't reach anywhere where i changed anything
17:09 xenoterracide joined #perl6
17:10 xenoterracide joined #perl6
17:10 FROGGS psch: it works here even when there are only two entries
17:11 FROGGS psch: is there something in the failing dir that has non ascii unicode chars or something else that could cause this?
17:12 psch FROGGS: i don't quite follow
17:12 psch oh
17:12 psch FROGGS: you're talking to pippo
17:13 pippo FROGGS: when there are two entries it is OK. The problem occurs when there are many more.
17:16 ajr_ A user reaction to an O-O language (JS) feature, that might amuse language designers:  http://ars.userfriendly.org/cartoons/read.cgi?id=20140209&amp;tid=3804719
17:16 FROGGS psch: I am sorry, you are right of course
17:17 pippo hoelzro, FROGGS: I made a new dir and switched to it. Then "for i in $(seq 1 10 ); do touch $i; done" and afterwards run moar with qx/ls/ ==> OK;
17:18 pippo for i in $(seq 1 100); do touch $i; done ==> NOT OK (segfaults)
17:18 hoelzro pippo: I would try increasing values for dd if=/dev/zero of=/dev/stdout bs=1 count=$x
17:18 hoelzro see what happens
17:18 hoelzro I'm trying to fix an unrelated issue atm
17:18 hoelzro after I work on it, I'll look
17:19 FROGGS pippo: I have a dir with 182 entries, and I can qx{ls} it fine#
17:19 FROGGS -#
17:20 pippo Do not know what to think. :-((
17:21 FROGGS pippo: do you have gdb?
17:21 FROGGS a backtrace would help
17:21 pippo yes. But I need directions on how to use it.
17:21 FROGGS np
17:21 FROGGS you do something like this, yes? perl6-m -e 'say qx{ls}'
17:22 FROGGS do a `which perl6-m`, then cat the result
17:22 bjz joined #perl6
17:23 FROGGS replace perl6-m of the first command with what you got from cat
17:23 xenoterracide joined #perl6
17:23 FROGGS if should look like this in the end:
17:24 FROGGS gdb --args /path/to/moar ...stuff...runtime/perl6.moarvm -e 'say qx{ls}'
17:24 FROGGS when you are in gdb, type: run
17:24 FROGGS and when it crashes: bt full
17:25 pippo OK. I'll do.
17:30 pippo FROGGS: http://nopaste.info/04748883b6.html
17:31 hoelzro neat
17:32 vendethiel `while $d %% 2 {(PrimeCandidate $n, Int $k) ...` what does that do ?
17:32 moritz the %% ?
17:33 vendethiel no, putting variable declarations after a block opening
17:33 moritz who does the ... look like?
17:33 vendethiel while body
17:33 moritz because so far it doesn't make much sense to me
17:33 vendethiel http://perl6advent.wordpress.com/2012/12/14/primal-needs/
17:33 vendethiel Gamma: Rabin-Miller test code
17:34 moritz vendethiel: looks like copy&paste gone wrong
17:34 vendethiel oh, oke
17:37 pippo_ joined #perl6
17:39 pippo_ FROGGS, hoelzro: here is the complete gdb run (i.e. from the begining): http://nopaste.info/dc5ed0e4f2.html
17:42 hoelzro thanks pippo_
17:46 vendethiel p: sub gt($n) { -> $x { $x > $n } }; my $gth = gt(100); 1, 2, 4 ... $gth;
17:46 camelia rakudo-parrot 46234b: ( no output )
17:46 vendethiel p: sub gt($n) { -> $x { $x > $n } }; my $gth = gt(100); say 1, 2, 4 ... $gth;
17:46 camelia rakudo-parrot 46234b: OUTPUT«1 2 4 8 16 32 64 128␤»
17:50 pippo_ FROGGS, hoelzro: And here is a successful run after doing: mkdir new_dir && cd new_dir && for i in $(seq 1 10); do touch $i; done: http://nopaste.info/b4127617cd.html
17:53 denis_boyun joined #perl6
18:13 virtualsue joined #perl6
18:32 FROGGS pippo_: can you reinstall moar with --debug turned on?
18:33 FROGGS this way we'd know at what line it explodes in readall_fh
18:33 FROGGS (besides several other things)
18:45 pippo_ FROGGS: OK.
18:47 pippo_ FROGGS: err... Configure does not accept --debug. How do I have to do?
18:48 FROGGS pippo_: you have to do that when configuring moarvm
18:49 FROGGS rakudo does not pass that through yet
18:49 FROGGS so you would have to install moar first, then nqp, and then rakudo, all without --gen-something
18:50 pippo_ Ok. I'll try.
18:50 FROGGS pippo_++
19:11 pippo_ FROGGS: Here is the gdb now: http://nopaste.info/3966723b4a.html
19:11 pippo_ Does this help?
19:11 FROGGS yes it does!
19:11 FROGGS pippo_: thanks a lot
19:12 pippo_ Thnks to you :-))
19:12 pippo_ Thanks to you :-))
19:12 FROGGS :o)
19:12 FROGGS I'll ping you when I have a patch
19:13 pippo_ \o/ FROGGS++
19:16 colomon pippo_++ # find the bug is as important as fixing it.
19:20 FROGGS handle = 0x0
19:20 FROGGS what on earth...
19:25 itz joined #perl6
19:32 thou joined #perl6
19:34 masak std: constant handle = 0x0
19:34 camelia std 09dda5b: OUTPUT«ok 00:00 122m␤»
19:34 masak :)
19:34 FROGGS *g*
19:35 FROGGS r: constant handle = 0x0; handle->body.encoding_type
19:35 camelia rakudo-parrot 46234b, rakudo-jvm 46234b, rakudo-moar 46234b: OUTPUT«[31m===[0mSORRY![31m===[0m Error while compiling /tmp/tmpfile�Unsupported use of -> as postfix; in Perl 6 please use either . to call a method, or whitespace to delimit a pointy block�at /tmp/tmpfile:1�---…»
19:35 FROGGS ewww
19:36 vendethiel ._.
19:36 FROGGS so we can't C while we Perl... at least that way
19:36 masak ewww? sounds pretty sane to me.
19:37 daxim joined #perl6
19:37 FROGGS sure, it is
19:47 timotimo jnthn: http://t.h8.lv/p6bench/2014-02-09-jvm_changes.html ← there's only a few benchmarks where there is a performance difference, one of which is a *big* performance regression
19:48 timotimo rc-forest-fire is the one with the big regression
19:48 timotimo and visit_2d_indices_cross regressed also
19:49 timotimo visit_2d_indices_loop_native profited a lot from the work
19:49 timotimo any_equals profitted, too
20:00 masak PerlJam: could you have a look at the copy-paste-o identified in your blog post at http://irclog.perlgeek.de/perl6/2014-02-09#i_8255984 ?
20:01 masak (kthx)
20:03 btyler joined #perl6
20:04 arlinius joined #perl6
20:11 denis_boyun__ joined #perl6
20:14 SHODAN joined #perl6
20:14 SHODAN joined #perl6
20:19 census joined #perl6
20:19 census hi labster!
20:27 labster hi census
20:28 census didn't see you in #help-census so thought i'd reach out
20:28 census hope all is well labster
20:29 labster oh, ha, forgot to rejoin after all of the internet weather.
20:29 masak census! \o/
20:29 labster hi masak!
20:35 jnthn timotimo: Oddness...
20:39 BenGoldberg joined #perl6
20:42 rindolf joined #perl6
20:45 jnthn timotimo: Hmm, we say it's a big regression, but there's two data points...
20:46 jnthn timotimo: And it's the difference between 2 times slower than fastest and 2 times slower than fastest...
20:46 jnthn timotimo: Wonder how repeatable it is.
20:47 timotimo agreed :)
20:48 timotimo it's a pain to run these benchmarks; but they are already run twice
20:48 timotimo did you see the other regression, though?
20:48 timotimo 49x slower instead of 30x slower
20:49 jnthn The cross one? Yes, that looks odd.
20:51 jnthn It does have a while loop in...
20:58 stevan_ joined #perl6
21:02 dalek nqp: a10bc21 | (Rob Hoelz)++ | t/moar/02-pipes.t:
21:02 dalek nqp: Add openpipe test for MoarVM
21:02 dalek nqp: review: https://github.com/perl6/nqp/commit/a10bc21936
21:02 dalek nqp: 7b1491f | (Tobias Leich)++ | t/moar/02-pipes.t:
21:02 dalek nqp: Merge pull request #156 from hoelzro/master
21:02 dalek nqp:
21:02 dalek nqp: Add openpipe test for MoarVM
21:02 dalek nqp: review: https://github.com/perl6/nqp/commit/7b1491f08a
21:08 timotimo still no commit bit, rob? :)
21:08 FROGGS this was a few days old
21:08 timotimo ah
21:08 FROGGS dunno if he has a nqp commit bit yet
21:09 hoelzro timotimo: I don't think so
21:09 hoelzro er
21:09 hoelzro I think I have an NQP one as of yesterday
21:10 hoelzro still waiting on Rakudo, though
21:10 hoelzro and I sent my CLA in like 2 weeks ago
21:16 kivutar joined #perl6
21:22 Myk267 joined #perl6
21:25 DarthGandalf joined #perl6
21:30 timotimo you should poke somebody
21:41 DarthGandalf joined #perl6
21:48 * hoelzro pokes pmichaud
21:50 hoelzro I believe Karen said it was in his hands =)
21:50 sjn <- wonders if urbandictionary.com's definitions of "poke" are true
21:52 BenGoldberg .ud poke
21:52 BenGoldberg .ping
21:52 yoleaux There is no ping command; nor can this be construed as a response.
21:52 kst` joined #perl6
21:53 lue I finally got around to making single-line comments highlight in my major mode :) https://github.com/lue/p6mode
21:55 thou joined #perl6
21:58 Ben_Goldberg joined #perl6
21:59 Ben_Goldberg .thesaurus poke
21:59 yoleaux poke — synonyms: bag, sack, pocket, gore, search, grope, thrust, jog, dawdler, project
22:13 kaare_ joined #perl6
22:14 btyler joined #perl6
22:14 masak 'night, #perl6
22:14 psch g'night masak
22:16 psch https://github.com/peschwa/rakudo/tree/typed_ex_refactor # in case anyone wants to give me some advice what to do differently
22:17 psch the way World and Optimizer build exceptions is quite different in a few spots, which leaves a bit outside of the role
22:17 psch additionally, i eventually want to pull the two loops after the call to optimize() in the optimizer into the role as well, but without typed warnings this doesn't really seem that useful to me yet
22:18 dalek rakudo-star-daily: 2b7aff3 | coke++ | log/ (5 files):
22:18 dalek rakudo-star-daily: today (automated commit)
22:18 dalek rakudo-star-daily: review: https://github.com/coke/rakudo-star-daily/commit/2b7aff3110
22:18 dalek perl6-roast-data: 00ff797 | coke++ | / (6 files):
22:18 dalek perl6-roast-data: today (automated commit)
22:18 dalek perl6-roast-data: review: https://github.com/coke/perl6-roast-data/commit/00ff797209
22:18 psch err, not after optimize(), after visit_block(), i.e. when the exceptions actually get thrown and warnings printed
22:19 virtualsue joined #perl6
22:23 psch uh, i think i messed up my git workflow again in that branch :/
22:25 colomon joined #perl6
22:26 psch fixed, looks sensible now
22:26 psch wrt the makefiles i'd definitely like some advice - it compiles but i'm not sure if i handled dependencies as per convention
22:35 [Coke] moar is now at 99.51% of java.
22:35 [Coke] m: say "moar is at {28820*100/28879}% of parrot"
22:35 camelia rakudo-moar 46234b: OUTPUT«moar is at 99.795699% of parrot␤»
22:39 pippo_ night perl6
22:44 raydiak I have read that the test percentages don't mean much because of uneven test coverage on multiple axii. Has it been considered to weight the value of different tests or groups thereof, and account for skips, to produce a more realistic "adjusted completion percentage"?  Would such a thing be useful?
22:45 jnthn No, there's nothing "clever" done in that regard.
22:46 jnthn Trying to cram so much info into a single number doesn't tend to work out too well, which is why there's things like the features page.
22:47 hoelzro so this code calls a routine twice, which catches the exception the first time, but not the second: https://gist.github.com/anonymous/8907215
22:47 hoelzro am I missing something here?
22:49 Ben_Goldberg p6: https://gist.github.com/anonymous/8907215
22:49 camelia niecza v24-109-g48a8de3: OUTPUT«here␤here␤»
22:49 camelia ..rakudo-jvm 46234b, rakudo-moar 46234b: OUTPUT«here␤there␤  in sub failing-routine at /tmp/tmpfile:9␤  in block  at /tmp/tmpfile:13␤␤»
22:49 camelia ..rakudo-parrot 46234b: OUTPUT«here␤there␤  in sub failing-routine at /tmp/tmpfile:5␤  in block  at /tmp/tmpfile:13␤␤»
22:50 raydiak jnthn: then why do we talk enthusiastically about today's pass percentage? it must meaningfully cram something into that one number if it makes people happy
22:51 jnthn raydiak: It is an indicator of "how many tests do we need to make pass to claim spectest parity with X"
22:51 jnthn raydiak: So it *is* a measure, from a dev point of view, of "how far to go". But only sort of. :)
22:52 jnthn From a marketing point of view, a number like 99.x% also sounds nice. And you don't need nuance to market, just nice sounding things. :P
22:54 jnthn hoelzro: It never makes it to the second call
22:55 jnthn hoelzro: If you don't have a successful smartmatch inside the CATCH it continues
22:55 jnthn continues throwing, that is
22:55 hoelzro I see..
22:55 hoelzro hmm
22:56 hoelzro I'll tweak my example a bit
22:56 jnthn Typical trick is to put a default { ... } in the CATCH if you really do want to catch all the things.
22:56 jnthn Given we've a typed exception system, generally you've *some* idea of what you're looking for in most situations.
22:57 raydiak jnthn: thank you, I think I understand now :)
22:57 hoelzro jnthn: how about this example, then? https://gist.github.com/anonymous/8907361
22:58 jnthn p6: https://gist.github.com/anonymous/8907361
22:58 camelia rakudo-parrot 46234b, rakudo-jvm 46234b, niecza v24-109-g48a8de3: OUTPUT«1␤here␤2␤here␤3␤»
22:58 camelia ..rakudo-moar 46234b: OUTPUT«1␤here␤2␤there␤  in sub failing-routine at /tmp/tmpfile:12␤  in block  at /tmp/tmpfile:19␤␤»
22:58 hoelzro ah, it's a moar problem then.
22:58 jnthn Moar bug.
22:58 hoelzro =/
22:58 jnthn They exist...
22:58 jnthn That's a weird one. Hm.
22:59 jnthn I wonder if the lexotic somehow breaks the active handler chain fixup...
23:00 Ben_Goldberg How's this for a strange moar bug:
23:00 Ben_Goldberg m: pi.round(.1**25).say
23:00 camelia rakudo-moar 46234b: OUTPUT«-9.22337203685478e-07␤»
23:00 jnthn Ben_Goldberg: Dunno. I know power.t has failing tests we didn't triage yet...no idea if that's related.
23:01 dmol joined #perl6
23:01 jnthn hoelzro: Ugh, I think I see it.
23:01 jnthn hoelzro: Please can you ticket it?
23:01 hoelzro sure
23:01 jnthn That really wants a spectest too...
23:02 jnthn I can probably fix it, though tomorrow.
23:02 jnthn *fix it without too much trouble
23:02 jnthn m: say .1**25
23:02 hoelzro jnthn: that would be great
23:05 camelia rakudo-moar 46234b: OUTPUT«1e-25␤»
23:05 hoelzro jnthn: https://rt.perl.org/Public/Bug/Display.html?id=121213
23:05 jnthn kthx
23:06 hoelzro thanks in advance for looking at/fixing it =)
23:06 jnthn .oO( active handlers bit me when doing continuations too...wonder if I mis-designed this bit... )
23:14 jnthn jmake
23:14 jnthn gah
23:16 skids r: sub foo { for 0..10 { $_ } }; foo().say;
23:16 skids hrm.
23:17 skids 8-/
23:19 jnthn skids: Ah. That loops is always in sink context now.
23:19 jnthn skids: Statement level oens always are these days, as folks got way too many performance bugs the eager way, and just bugs the lazy way.
23:19 timotimo so if you want the list you need "do for ..."?
23:19 jnthn do for works.
23:19 jnthn lazy for and eager for also work
23:20 jnthn And are better documentation of what you want.
23:20 timotimo gotta tram now
23:20 skids jnthn++ timotimo++ thanks.  That was what broke Sum last night.
23:22 tadzik uh-oh
23:23 tadzik I remember those problems
23:23 tadzik these were the things I was panic-fixing with putting '1;' after  every loop in my code
23:23 tadzik and I wasn't the only one
23:23 jnthn tadzik: Yes, you don't have to do *that* any more. :)
23:24 tadzik skids does, apparently :/
23:24 tadzik what happens to his for loop by default? It disappears? Why?
23:24 jnthn Yeah. Well, now you only have to do it if you do a for and want the results of the body to be returned.
23:24 tadzik ah!
23:24 jnthn It runs his for loop, it just doesn't collect the results.
23:24 tadzik oh
23:24 tadzik that makes sense :)
23:24 jnthn So for is the imperative construct most of us expect. :)
23:24 tadzik that I like
23:25 tadzik right :)
23:25 camelia joined #perl6
23:26 pippo joined #perl6
23:26 pippo m: nqp::getenvhash()
23:27 pippo m: say 1
23:27 pippo p: say 1
23:27 pippo n: say 1;
23:29 raydiak joined #perl6
23:30 * lue wonders how long it will take before he stops working on his elisp STD clone out of agony. :P
23:30 jnthn lue: Did you implement an LTMer in elisp yet? :D
23:31 lue hah hah nope :) but my perl6 major mode is in all likelihood going to become 90% a perl6/std clone, which is what I'm talking about.
23:33 lue Perl 6 does quite a bit of stuff that emacs' standard major mode tools aren't quite set up to handle, as you might imagine :)
23:35 * lue hopes that (e|si)mulating the braiding of languages in Perl 6 won't be too hard when it comes up.
23:35 kst joined #perl6
23:42 camelia joined #perl6
23:47 jnthn 'night, #perl6
23:47 lue ♞ jnthn o/
23:48 FROGGS joined #perl6
23:53 raydiak joined #perl6

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

Perl 6 | Reference Documentation | Rakudo