Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2014-01-09

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

All times shown according to UTC.

Time Nick Message
00:00 abenson Hi Chris
00:00 chm Hi Andrew, nice to "meet"
00:00 abenson Likewise!
00:01 chm I'm the current PDL release manager and am trying in general to improve robustness and portability
00:01 chm Do you have any experience with git/sf.net/other perl development?
00:02 abenson Not a whole lot - basically just hacking around with PDL::IO::HDF5 to make it do what I want
00:02 chm Ok, just trying to set the scene.
00:02 chm I got pinged today by John XXX (cant remember but from Miami) about HDF5 progress...
00:03 chm I figured it was time to get moving----amazing how much easier to do when there is a live user  :-)
00:03 abenson Me too - he seems eager to have the additional functionality of my copy of PDL::IO::HDF5 incorporated
00:03 abenson Agreed!
00:03 chm I can tell.  Do you have an sf.net user account?
00:04 abenson I think I may have set one up - let me check
00:05 chm That will let me add you as a developer for PDL::IO::HDF5.
00:05 abenson ok - yes, i have one and i juts logged in
00:05 chm In the meantime, I am fine with submitting/applying your patches.
00:06 abenson sounds good
00:06 chm I'm very conversant with the PDL and CPAN module release stuff and have a general
00:06 chm knowledge of HDF/HDF5 features but haven't really used them which means I haven't come uop
00:06 chm the learning curve yet.
00:06 abenson ok. i'
00:06 chm You know how it is being too busy...
00:07 abenson i've used hdf5 extensively so have good knowledge of how it all works
00:07 abenson yep!
00:07 chm Great, I was hoping that you could provide the HDF5 expertise.
00:07 chm Have you done much with PDL/what is your familiarity there?
00:08 abenson I've used PDL a lot for the past ~3-4 years, so I'm quite familiar with it. But, I haven't dug into the code at all.
00:09 chm Which version do you work with?
00:09 abenson Currently I'm using 2.7.0 if I remember correctly
00:09 chm Is that 2.007?
00:10 chm (We changed to single float numbering because it was a mess before...)
00:10 chm PDL-2.007 is the current stable PDL release.
00:10 abenson Maybe - let me check that
00:11 abenson Yep - definitel 2.007
00:11 chm Great, that means you're at the current version so nothing planned will be confusing because of lack of support...
00:12 abenson Good!
00:12 chm Let me summarize the general plan on my end and then you can comment on it.
00:12 abenson ok
00:12 chm Also, please email me your sf.net username and I'll add you as a developer to the git repo
00:13 abenson ok - sending you an email now
00:13 chm The current plan is to wrap up the PDL-2.x release sequence by fixing the remaining issues with the 64bit index support.
00:14 chm I also am working on a number of Alien modules to simplify the dependency process for PDL modules (and others)
00:14 chm Specific the PDL::IO::HDF5, this means that the PDL::IO::HDF5 would list Alien::HDF5 as its CPAN dependency
00:14 abenson ok - understood
00:14 chm and then the Alien::HDF5 module would check for an installation, give the option of building/installing, and then
00:15 chm finishes with the configuration needed to build the PDL::IO::HDF5 module.
00:15 chm Right now, you kind of have to know your way around perl, Makefile.PL, hdf5,...to get it working
00:15 chm unless you have a completely default location...
00:16 chm On your side, I'm hoping to add the additional features and, I hope, clean up some test problems, ... so that the next official release is
00:16 chm easy to install, featureful, and reasonably stable.
00:17 chm Does that sound reasonable?
00:17 abenson yes - that sounds good to me.
00:17 chm Let me see if I can add you to the PDL project at sf.net, one moment...
00:18 abenson ok
00:19 chm Well, sf.net says you have developer status now.  That means you should have rw access to the git repositories
00:19 chm pdl-io-hdf5 is one of them
00:20 abenson ok
00:20 chm I guess the next steps for PDL::IO::HDF5 are to merge in your new functionaltiy.
00:20 abenson great - i see pdl-io-hdf5 on sf.net now
00:21 chm Do you have tests written for the features?  That is a critical part of robustness and stability.
00:21 abenson for some - but not all. that's probably the first task then.
00:21 chm That is fine, just something we'll want to flesh out before an official release.
00:22 chm This will be particularly important when the PDL3 work starts in earnest.  The plan is for
00:22 chm compatibiliy with the PDL-2.x but the only way to check (easily) is via the test suites.
00:23 chm Since a big part of the PDL3 work is new, general type support that should permit full HDF5 bindings,
00:23 chm it would be good to track this going forward.  :-)
00:23 abenson definitely!
00:23 chm Well, it seems like the next step would be for you to send me some patches against the latest git
00:24 chm The directions on the git page for PDL have directions that you should be able to cut and paste to
00:24 abenson yes - sounds good. hopefully i'll have some time this week to do that.
00:24 chm clone the repository.  BTW, what is your perl platform (OS/hardware)?
00:25 abenson I'm using Linux (Fedora 20 and RHEL5) x86_64
00:26 chm Ok, then the 64bit support would be good for you as well.  Does HDF5 support 64bit data?
00:26 abenson yes, it does. one of the features of HDF5 is transparent conversion between 32 and 64-bit
00:27 chm My push for 64bit indexing ties in with large data file manipulations which are much simpler if you can just memory map the data
00:28 chm I hope to finish the basic code work remaining for the 64bit index support and hope for a PDL-2.008 release
00:28 abenson that would be very useful - i've run into a few problems where the lack of 64bit indexing is an issue
00:28 chm with that (and whatever else makes the time window) by the end of January (February-ish for sure, depends on the weather...)
00:29 chm Me as well---just yesterday in fact and I was wishing I had not taken the break from the last code spurt.
00:29 chm They say you usually have to write one to throw away---software-wise.
00:30 abenson that sounds about right!
00:30 chm That covers the topics I can think of.  Anything on your end?
00:30 abenson No - I think that's good for now. I'll hopefully submit those patches in the next day or two
00:31 chm I'll look for them.  Great chatting.
00:31 abenson Sounds good - bye!
00:31 chm (And thanks for volunteering to help with PDL!)
00:31 chm o/
00:31 abenson left #pdl
00:40 jberger_ abenson++ chm++
00:46 jberger__ joined #pdl
00:48 jberger joined #pdl
05:34 webart1 joined #pdl
10:01 lungching joined #pdl
23:51 jberger joined #pdl

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