Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2017-05-23

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

All times shown according to UTC.

Time Nick Message
12:22 jberger joined #pdl
17:06 chm joined #pdl
17:07 chm q
17:07 chm joined #pdl
17:09 chm Anyone interested in a PDL BoF/Meet-up during the week of YAPC::NA 2017?
17:10 mohawk jberger, woo!
17:10 mohawk chm, hello! sadly i won't be at that
17:13 chm mohawk: We'll it seems you may already have met your annual requirement at the perl toolchain events!
17:13 mohawk ha ha!
17:14 chm mohawk: With the PDL-2.018 bitrot release, I'm back to trying to get PDL::NG off the ground.
17:14 mohawk a point that came up during PTS: for those who've not met in person, i believe it's valuable to try to at least partly close that gap using a video-chat
17:14 chm mohawk: Got what I think is a viable way forward----just not critical mass in developers.
17:14 mohawk chm, for p5 or p6?
17:15 chm mohawk: My plan was to use Moo OO stuff which should be compatible with p6.
17:15 chm The XS layer would be a COS (C Object System) with C FFI compatibility.
17:15 mohawk ok
17:15 chm That should make it work for both.
17:16 mohawk i'm a recent convert to FFI
17:16 mohawk i'll plug the new, fairly minimal p5 NativeCall here :-)
17:16 chm The idea for PDL::NG was that the C/XS layer would be "equivalent" to the perl one.
17:17 sivoais I enjoyed reading both jberger's and mohawk's PTS blog posts :-)
17:17 chm That way, any code with C FFI capability could use the new functionality---even from Python or Julia!
17:17 mohawk ha ha, great! jberger's was actually informative
17:17 chm sivoais: hi!
17:17 sivoais o/
17:18 mohawk mine reflected staring for 2 days at the same few bits of EUMM
17:18 chm sivoais: Any more updates for Jupyter and PDL/Perl?
17:19 chm sivoais: I keep not getting up the learning/usage curve and it would be a good way to interoperate with my python colleagues.
17:19 chm I've enjoyed the updates from PTS as well.
17:20 sivoais chm: No major updates at this time, but that will change soon. I'm free a for a while until I start medical school in September.
17:20 chm mohawk: As an EU::MM user by default---pretty much the only thing you can count on everywhere---your contributions are appreciated.
17:21 sivoais For that reason, my Perl-tuits have mostly gone towards working on my document reader <https://project-renard.github.io/blog/posts/2017/05/16/curie-release-v0.002/>
17:21 mohawk thanks for the kind words
17:21 mohawk sivoais, for the ASperl note from chm - is there a way to make appveyor routinely build that?
17:22 sivoais It should be possible. Does that usually use ppm or would using the same cpanm build process be acceptable?
17:22 mohawk (i'm just digging into the last EUMM regression so we can hopefully get a new real release)
17:22 chm sivoais: The document reader looks cool++
17:23 mohawk sivoais, as long as it fails when the asperl build fails...
17:23 chm mohawk: Unfortunately, the ASPerl problems are, I think, in their build system details and not necessarily in PDL stuff.
17:24 chm I still is annoying that Rob can build working PPMs with no problem but the official ASPerl site doesn't.
17:24 chm I guess it is good news that they will be moving to a cpan build process... maybe that will help.
17:25 mohawk is that doc-ed anywhere?
17:25 sivoais Is it possible to coordinate with them? I've seen AS send people to YAPC before.
17:25 mohawk i was about to ask whether rob had liaised with them
17:25 mohawk great minds ;-)
17:25 sivoais :-P
17:27 sivoais these results are a bit confusing as to why they would fail under one but not the other : <https://code.activestate.com/ppm/PDL/>
17:30 sivoais We should probably focus on 5.2{2,4} Win{32,64} since those are what are available to download as the AS Community Edition
17:33 chm sivoais: Those are the ones of interest (the last 2 versions on the various platforms)
17:33 chm sivoais: at least it built on MacOSX this time. :-)
17:37 mohawk agreed
17:38 chm mohawk: nice with the NativeCall frosting for FFI::Platypus!
17:38 chm mohawk: This is why having the core computation with C FFI for PDL::NG is the way to go.
17:38 mohawk ha ha, indeed
17:39 chm mohawk: not difficult to get market share if anyone can use you: NumPy, Matlab, Octave, Julia, Python, ...
17:39 mohawk part of my plan is to make an "h2nc" which will be like h2xs
17:39 mohawk indeed!
17:41 chm mohawk, sivoais: I would like to start PDL::NG (Next Gen) with code that can fallback to PDL-2.x for missing functionality.
17:41 chm That way, we could develop the new stuff while using the old.
17:42 chm If you have any ideas for what would be needed to handle that type of delegation, I'd love to hear them.
17:42 chm Email or wiki or whatever.
17:43 chm At the moment, I'm at the play around stage to see if anything works.
17:43 mohawk i'd say start with a scratch proof of concept that doesn't delegate
17:47 chm mohawk: good point, no point to delegating if you're starting with garbage....
17:50 mohawk MVP :-)
17:55 chm mohawk: Thanks for the MVP reference as well.  Getting "buy in" from other developers could be the MVP goal.
17:55 mohawk i don't quite agree
17:56 mohawk i'd suggest here the MVP is something that more or less compiles, and shows off the direction you're going in
17:56 mohawk then you'd iterate by getting buy-in from various key PDL folks on the approach
17:57 chm mohawk: sounds reasonable
17:57 mohawk embrace the "Minimum" :-)
18:02 opkick [pdl] zmughal created appveyor-activestate-perl (+1 new commit): https://git.io/vHtb8
18:02 opkick pdl/appveyor-activestate-perl 882c4c8 Zakariyya Mughal: Appveyor: Add ActivePerl configuration
18:06 mohawk #thuglife
18:07 sivoais hehe, we'll see if this one goes through
18:55 azawawi joined #pdl
18:56 azawawi hi
19:01 opkick [pdl] zmughal pushed 1 new commit to appveyor-activestate-perl: https://git.io/vHqUt
19:01 opkick pdl/appveyor-activestate-perl 33adbd0 Zakariyya Mughal: Appveyor: Need Devel::CheckLib to configure PDL
19:38 mohawk sivoais, would it not be better to do cpanm --installdeps instead of manually specifying configure deps?
19:40 sivoais mohawk: I'll add a not about this, but remember that Devel::CheckLib is an author-side CONFIGURE_REQUIRES
19:40 sivoais *note
19:40 sivoais the META.yml doesn't exist yet in the repo
19:40 mohawk you're saying CONFIGURE_REQUIRES is author-side?
19:40 sivoais It only worked on Strawberry Perl because it already has Devel::CheckLib
19:41 mohawk i would have thought cpanm would read META.json?
19:42 sivoais ah, yes, that one
19:42 sivoais Remember that in .travis.yml, we do the same thing.
19:43 sivoais Devel::CheckLib ends up in the dist that hits CPAN, but the META.json file isn't in the repo.
19:44 sivoais btw, both Strawberry Perl and AS Perl passed. Making a PR.
19:44 mohawk i would also have thought the configure deps would end up in MYMETA.*
19:44 mohawk nice work!
19:44 mohawk ahhh, this is a fault in Makefile.PL
19:46 mohawk it does the "use" and instead should eval { require } that
19:46 opkick [pdl] zmughal force-pushed appveyor-activestate-perl from 33adbd0 to e061c9b: https://git.io/vHqOa
19:46 opkick pdl/appveyor-activestate-perl e061c9b Zakariyya Mughal: Appveyor: Need Devel::CheckLib to configure PDL
19:47 opkick [pdl] zmughal opened pull request #206: Appveyor: Add ActiveState Perl build (master...appveyor-activestate-perl) https://git.io/vHqOM
19:50 mohawk sivoais, this is clearly an improvement and great work
19:50 mohawk what do you think of having a go at making the Makefile.PL runnable even when D::CL is not installed?
19:53 mohawk i reviewed
19:57 sivoais It looks like it isn't even used until EUMM hits the subdirs
20:12 opkick [pdl] coveralls commented on issue #206:
23:38 opkick [pdl] wchristian created index_ops (+4 new commits): https://git.io/vHqHq
23:38 opkick pdl/index_ops 196ed29 Chris Marshall: Start work for indxxx routines...
23:38 opkick pdl/index_ops 62bdd92 Chris Marshall: Update original indadd signature...
23:38 opkick pdl/index_ops 9944faa Chris Marshall: Add IndOps/ to the PDL build process

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