Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6-dev, 2016-05-24

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

All times shown according to UTC.

Time Nick Message
00:16 lizmat joined #perl6-dev
00:19 timotimo joined #perl6-dev
00:19 ugexe joined #perl6-dev
01:07 jdv79 joined #perl6-dev
01:07 lizmat And another Perl 6 Weekly hits the Net: https://p6weekly.wordpress.com/2016/05/24/2016-21-a-quick-one-from-houston/
01:49 ilbot3 joined #perl6-dev
01:49 Topic for #perl6-dev is now Perl 6 language and compiler development 2.0 | Logs at http://irclog.perlgeek.de/perl6-dev/today
02:24 japhb joined #perl6-dev
02:25 bisectable joined #perl6-dev
02:31 bisectable joined #perl6-dev
02:32 bisectable joined #perl6-dev
02:52 lizmat joined #perl6-dev
05:21 M-tadzik joined #perl6-dev
06:23 masak "Houston, we have a quick one" :)
06:23 masak lizmat++
06:24 sno joined #perl6-dev
06:30 masak lizmat: feel better soon!
07:52 RabidGravy joined #perl6-dev
09:54 timotimo yes, i hope you do get well :)
09:57 |Tux| joined #perl6-dev
10:02 sno moritz: did you read http://www.netbsd.org/~sno/talks/nrpm/Cross-Compiling-For-Perl-Hackers-Handout.pdf ?
10:10 |Tux| This is Rakudo version 2016.05-8-gb348f2f built on MoarVM version 2016.05
10:10 |Tux| test            19.529
10:10 |Tux| test-t          12.728
10:10 |Tux| csv-parser      33.484
10:18 moritz sno: no
10:19 sno moritz: I'd really appreciate when you do :)
10:19 moritz sno: I know. My reading list is just long
10:20 sno moritz: I'd like to work on a follow-up presentation and after that (and hopefully your and some others read/review) some patches for MoarVM, NQP & Co. configure stage :)
10:20 |Tux| sno, totally off-topic. do you agree with my https://github.com/perl5-dbi/SQL-Statement/commit/a503418a68fe3817914e57606651999031c1edbb ?
10:21 sno and it will be much easier to talk about the goal when you've read the stuff
10:22 sno |Tux|: looks reasonable - are there tests with multiple statements
10:22 * |Tux| will do a perl6 presentation on the NLUUG conference Thursday
10:22 |Tux| sno, no tests (yet)
10:22 |Tux| not by me at least
10:23 sno |Tux|: no such change without tests - you know ..., but I really appreciate work on that beast
10:23 |Tux| so much to do, so little time
10:24 * sno nods
10:24 * |Tux| promises to add two tests
10:25 * sno currently plays with router tech and a new web-project and wants Perl6 on IoT (cross compiled) before starting over and do the perl6 modules I miss ...
10:26 sno so yes, much to do - and there is another S::S issue which requires some extra affort ...
10:26 timotimo are IoT devices powerful to satisfy the memory hunger of current moarvm? :S
10:26 sno timotimo: they are
10:27 sno timotimo: they are powerful enough to satisfy the memory hunger of jvm
10:27 timotimo oh
10:27 sno and I managed running perl6-j on such an IoT device
10:27 timotimo that's impressive
10:28 sno but there are many issues in MoarVM, NQP and Rakudo configure and build stage wrt. cross-compiling, and I'd like to improve that
10:29 sno which needs some understanding in the team responsible for reviewing patches and care of infrastructure
10:29 timotimo i see
10:30 |Tux| new 88.pl failure (not a trailing 0):
10:30 |Tux| # expected: Buf.new(61,10,0,56,143,36,56,119,182,81,88,70,88,139,28,119,142,151,108,12,215)
10:30 |Tux| #      got: Buf.new(61,10,0,56,143,36,56,119,182,81,88,70,88,139,28,119,142,151,108,12,215,127)
10:30 sno http://www.netbsd.org/~sno/talks/nrpm/Cross-Compiling-For-Perl-Hackers-Handout.pdf primarily explains the goals and general pitfalls when creating configure/build/install toolchain
10:30 sno and every perl6 hacker is welcome to read it :D
10:32 timotimo i'll keep the tab open for now. right now i'm watching nine++ 's gpw talk
10:32 stmuk timotimo: that was an excellent talk
10:33 timotimo i'm not in the target audience, really, as i already know the basic awesome stuff I::P5 can do
10:33 timotimo not sure if there was anything "new" presented in that talk
10:34 |Tux| do we have longdouble support yet?
10:34 timotimo no, we don't
10:34 |Tux| :(
10:35 timotimo it requires a piece of design for moarvm, as moarvm currently doesn't deal with natives that are bigger than 64 bit
10:35 timotimo and in order to have any long doubles we'd have to have that
10:35 timotimo alternatively, we could have a C library that implements everything you might want to do with long doubles and then use that with nativecall
10:38 sno timotimo: Apr 19 17:26:30 <nine_>[Sno]: oh I wish I hadn't read your slides. This makes things...more complicated. Makes it harder now to continue living in the "if it works on your system it must be good" bubble :)
10:38 timotimo mhm
10:44 sno timotimo: beside different cpu architectures, biggest problem is the assumption of "compile once, run anywhere" in non-native code which will fail heavily when storing environment (PATH's, available 3rd party tools, ...)
10:45 timotimo ah, yes
10:46 stmuk sno: are you the maintainer of pkgsrc moarvm?
10:46 sno stmuk: nope, but maybe I work on that at summer
10:48 stmuk ok
10:48 sno timotimo: given the situation of packaging (some Debian folks are very restrictive then ...), assumption of tools (which is the right location for perl6 to invoke to run post-processing stuff, which jvm to use, which library path during compile compared to installed package)
10:49 sno stmuk: I worked last 2 years with open-embedded, not that much time for pkgsrc :(
10:49 sno but I intend to return :D
10:54 stmuk sno: I was experimenting with rakudo pkgsrc wip and it compiles with a version increase but I also noticed nqp test failures (didn't test on NetBSD)
10:55 stmuk I need to give some proper feedback
10:56 sno stmuk: pkgsrc wip is primarily maintained by wiz@netbsd.org and has another feedback channel than pkgsrc
10:57 sno I intend to fix the tooling of moar,nqp,rakudo .... and then do proper recipes for pkgsrc (native, not wip) and open-embedded
10:58 stmuk sno: I was actually using a mix of pkgsrc proper and pkgsrc wip (looked like someone was half way through doing it and was on pause)
10:58 sno since more and more tasks are in IoT/embedded market, I need full running perl6 on all those platforms before starting work with and in perl6
11:02 stmuk what arch?
11:04 sno several - last arch was ARM (not ARM64) with multi-media features, previous was an Octeon (64-bit PowerPC) with network packet processor (checksum calculation, fast routing table lookup, ...)
11:05 sno most of those tasks are C, C++ or Assembler - Perl is introduced (and currently perl5, but I want introduce Perl6 instead of introducing Perl5)
11:06 timotimo all that's left for Inline::Perl6 is that cpanm install Inline::Perl6 builds rakudo for you :P
11:08 ilmari Alien::Rakudo?
11:08 ilmari ISAGN
11:08 sno that's insane and won't work at all
11:10 stmuk "perl6 - making the insane work"
11:11 sno Nope, Inline::Perl6 or Alien::Rakudo is insane from IoT perspective
11:11 timotimo only has to work on some systems
11:25 sno have a look at https://www.linkedin.com/pulse/create-images-several-boot-devices-using-yocto-jens-rehsack to guess why such a tool is a fool for typical IoT environments
11:25 sno even moritz - it's a 5 minute read
11:26 moritz sno: please stop telling me what to do
11:29 sno moritz: I don't tell you what to do and when I generated the impression, I apologize - HL you to give you the chance to add something to your overful readlist
11:31 moritz sno: well, please stop doing that as well, it feel patronizing, especially with "even moritz"
11:33 sno moritz: wasn't meant "patronizing" - the intension was just being the opposite in a ironic way ... went wrong
11:40 sno moritz: are serious question: when I do such a talk as above and lizmat tells me to introduce it to you - shall I ignore lizmat or is there a reasonable way to inform you about the resource and ask you for feedback?
11:44 Woodi sno: go for perl5 in 2016, IMO
11:44 Woodi ...for IoT/embaded things
11:44 Woodi hi all today :)
11:45 sno Woodi: I don't understand what you're trying to tell me
11:46 Woodi you want show using perl on things like Rpi0, right ?
11:46 moritz sno: it's fine if you just inform me of things that might be interesting to read
11:46 sno Woodi: nope
12:09 [Tux] joined #perl6-dev
13:39 skids joined #perl6-dev
13:43 RabidGravy joined #perl6-dev
13:47 brrt joined #perl6-dev
13:50 tomboy64 currently rakudo compile fails with this: https://bpaste.net/show/6459e774d665 (java backend)
13:50 tomboy64 this happens with 2016.05 and live
13:50 tomboy64 any suggestions?
13:51 tomboy64 i'm using icedtea 3 (openjdk 8)
13:52 timotimo isn't rakudo-j known to not 100% install at the moment?
13:53 tomboy64 well, it worked just fine (tm) with 2016.04
13:55 timotimo oh, i didn't know that
13:59 brrt maybe a precomp change that wasn't carried over to rakudo-j yet
14:06 tomboy64 :/
14:06 tomboy64 not nice that
14:08 brrt well, the weak defence is that nobody has yet promised to support and keep running rakudo-j
14:08 brrt and the reasonable reason for that is that this is a lot of work, and we have limited resources
14:09 brrt and to a large extent, the MoarVM backend is better for most users than the java backend
14:09 tomboy64 :) i'm aware of that.
14:10 tomboy64 i just spent a lot of time making the java-backend work properly on gentoo
14:10 brrt it doesn't sound undebuggable, and you might want to ask nine about it whenever he is here
14:10 brrt well, that sucks
14:10 tomboy64 will do
14:10 brrt for what it is worth, it is why we have releases :-)
14:10 tomboy64 :p
14:13 timotimo well, we've got two or maybe more people who invest time in making the java port of rakudo work with latest things
14:13 brrt how is the rakudo-m situation with gentoo?
14:13 tomboy64 works just fine
14:13 tomboy64 oh
14:13 tomboy64 well, need to rebuild
14:14 tomboy64 haven't tested 2016.05 yet
14:14 tomboy64 2016.04 was just fine
14:14 lizmat joined #perl6-dev
14:17 tomboy64 but yea, works fine with both clang and gcc.
14:18 timotimo well, can you build rakudo-j with gcj? :P
14:19 tomboy64 ugh
14:19 tomboy64 that would require me to rebuild my gcc
14:19 brrt don't make him do it timotimo :-P
14:20 brrt it's not likely an issue with icedtea anyway
14:21 tomboy64 $ perl6 -e 'say "okay"'
14:21 tomboy64 okay
14:21 tomboy64 works fine on 2016.05
14:23 lizmat_ joined #perl6-dev
14:25 timotimo does gcj even exist any more?
14:25 timotimo i thought development on that stalled like 5 years ago?
14:26 ilmari still ships with gcc 6.1
14:26 timotimo does that mean anything significant?
14:27 ilmari it's still there at least
14:28 ilmari but no mention of it (or java) in the changes pages for gcc 6, 5 or 4.9
14:28 ilmari latest news item on https://gcc.gnu.org/java/ is from 2009
14:28 ilmari seems like it's stable (in the biological sense)
14:34 tomboy64 gcj is still shipped with gcc, but not enabled per default anymore
14:34 tomboy64 it was replaced by go
14:38 ilmari debian still enables it, at least
14:39 ilmari but they tend to enable everything that can reasonably be enabled
14:39 timotimo for stability reasons
14:39 tomboy64 debian's backwards-compatibility madness is even worse than microsoft's
14:40 tomboy64 but no idea how ms handles that these days. haven't actually used their products for a decade or so.
14:42 geekosaur microsoft ripped out a lot of back-compat starting in vista and continuing into win10
14:52 moritz fwiw ocasionally I notice when Debian rolls out a change that's not backwards compatible, and I'm not eager on experiencing that more often :-)
14:52 moritz where "notice" is usually "suffer from"
14:57 brrt geekosaur: oh really... that explains something
14:57 ilmari these days if you need ancient windows you can run it in a VM, I guess
14:58 ilmari compatibility with, that is
14:59 geekosaur yep, they added WOW (WIndows-on-Windows) to allow XP stuff to run
14:59 ilmari huh, modern.ie claims «IE6 through IE11», but the dropdown only has IE8 and onwards (and no windows older than 7)
15:00 brrt hmmm
15:00 brrt i'm having to work with a legacy program now and the compatibility experience has been less than stellar...
15:02 geekosaur yes, people have found WoW to be lacking
15:04 * brrt dinner &
15:07 geekosaur (Microsoft recommends running real XP in a VM for stuff that doesn't work. This isn't just a sales ploy; you're running unstable and insecure code that they very, very badly wanted to rip out and not have to put back in)
15:21 dalek star: 122c0c9 | (Zoffix Znet)++ | tools/star/release-guide.pod:
15:21 dalek star: Point out where R* is listed in Rakudo Perl wiki
15:21 dalek star: review: https://github.com/rakudo/star/commit/122c0c990b
15:21 dalek star: c5676a2 | (Steve Mynott)++ | tools/star/release-guide.pod:
15:21 dalek star: Merge pull request #71 from zoffixznet/patch-1
15:21 dalek star:
15:21 dalek star: Point out where R* is listed in Rakudo Perl wiki
15:21 dalek star: review: https://github.com/rakudo/star/commit/c5676a2069
15:34 cognominal joined #perl6-dev
15:35 cognominal me suis réinscrit sur paris@mongueurs.net, j'avais disparu de la liste mi 2015 pour des raisons qui m'échappent. J'étais étonné de ne pas voir passer les mails à propos de la réuinions de Mercredi
16:05 AlexDaniel joined #perl6-dev
16:17 nine tomboy64: yes, the recent round of improvements to the precompilation code exposed an issue on rakudo-j.
16:18 tomboy64 nine: how big of a probem is that?
16:18 nine tomboy64: it has taken me a month or longer to fix the same issue on rakudo-m and the fix still needs to be ported to the jvm version.
16:18 tomboy64 ew
16:18 nine It hopefully won't take another month, but it's not exactly straight forward either
16:18 tomboy64 okay
16:18 tomboy64 well, i deactivated java for 2016.05 for now
16:19 nine tomboy64: the up side is that 2016.05 should be much more packaging friendly :)
16:19 tomboy64 would be nice if you dropped me a line if it's supposed to be working again
16:20 tomboy64 heh :} well, it's downloading the source and compiling. so once the package exists there's not much to do apart from copy+paste the original build-script and test whether stuff works.
16:21 tomboy64 nine: but out of curiosity, why exactly do you think it is?
16:26 nine tomboy64: it's now possible to build modules with an unpriviledged user in a separate directory, package the result and unpack the files to the file system in the intended directory with rakudo being able to use the precompiled files generated in the process.
16:26 nine tomboy64: though granted, that's much less usefull if you expect your users to wait for compilation during installation anyway...
16:27 tomboy64 :D indeed
16:49 jdv79 the abs paths are not embedded anymore?
16:50 geekosaur embedded paths are relative to the CUR, I think
16:50 geekosaur so you can move the repository around and things still work as long as it can find itself
16:50 nine jdv79: they're gone
16:50 nine what geekosaur said :)
16:56 [Coke] pmurias - how to build rakudo-js for testing? "git checkout js; perl Configure.pl --backends=js" doesn't work.
16:56 [Coke] s/testing/hacking/
17:47 sno joined #perl6-dev
18:13 jdv79 i wish i could find myself some days
18:14 jdv79 nine:  excellent news!
18:49 dalek rakudo/nom: 5444cc9 | (Zoffix Znet)++ | docs/release_guide.pod:
18:49 dalek rakudo/nom: Fix typo
18:49 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/5444cc9a07
18:49 dalek rakudo/nom: 19e91e8 | lizmat++ | docs/release_guide.pod:
18:49 dalek rakudo/nom: Merge pull request #776 from zoffixznet/patch-2
18:49 dalek rakudo/nom:
18:49 dalek rakudo/nom: Fix typo
18:49 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/19e91e889c
18:50 lizmat joined #perl6-dev
19:54 lizmat joined #perl6-dev
20:19 travis-ci joined #perl6-dev
20:19 travis-ci Rakudo build failed. lizmat 'Merge pull request #776 from zoffixznet/patch-2
20:19 travis-ci https://travis-ci.org/rakudo/rakudo/builds/132646406 https://github.com/rakudo/rakudo/compare/b348f2f02147...19e91e889c51
20:19 travis-ci left #perl6-dev
20:31 Ven joined #perl6-dev
21:04 Ven joined #perl6-dev
21:37 AlexDaniel joined #perl6-dev
22:20 skids joined #perl6-dev

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