> Discussion of moving PDL web site to static github > - Seems reasonable but needs support > - We didn't have any volunteers the last time this was proposed > - With maintainer it would be nice to have more timely web support
Just to correct this point, not only did I volunteer to do this, I did it and it is running at http://pdlporters.github.io/ . There were only a few minor nits, first that metacpan's documentation (from which the static site pulls pod) doesn't render all the pod in PDL for reasons being discussed on another thread. Second the news is now out of date, though that's easy enough to correct. Third, I did port much, but not all, of the wiki to github wiki, perhaps someone could check it and port the remainder? Joel On Sun, Dec 21, 2014 at 9:26 AM, Chris Marshall <devel.chm...@gmail.com> wrote: > This is a brief note on a frustration I have trying to encourage and > coordinate PDL development efforts. > > A persistent difficulty is that there is a sort of impedance mismatch > between users of #pdl and perldl or pdl-porters list users in terms of > expectations and how problems might be posed and solved. Unfortunately, > there are some pretty interesting points being discussed but to date, no > one has figured out a way to bring those ideas into perldl or pdl-porters > for the record. I've tried reviewing the logs when I am able to and > responding asynchronously. That didn't work because the difference is more > one of expectations between the user approaches rather than just the > discussions themselves. > > I'm not able to participate regularly in #pdl but I did try to join #pdl > more often with the hopes of improving communication around the Aug 2014 > timeframe. The outcome wasn't positive in the sense that I burned several > days of PDL time trying to communicate with little PDL development to show > and no idea about how better to get #pdl ideas into PDL development. I'm > happy to have some time this winter to get back to PDL development. I > would be happier if we could figure out how to keep development and > contributors moving constructively forward more consistently. > > Here are some of the topics from the last couple of weeks that were > interesting to me and some thoughts > > * A new perl array extension Array::APX > - It appears to be perl only > - It might be useful that PDL::Tiny support a perl-only version (same > sugar, no XS) > * Inline::ExtUtils::Depends support for PDL > - Still working on the git clean up > - It'll be in PDL-2.008 or sooner with help > * PDL release schedule > - We're on an about 1 release per year > - Mostly developer and test/verification limited > - PDL is *used* for research and stability is important > * PDL3 == second system > - PDL3 was a term to separate PDL-2.x architecture from > a refactored implementation > - For stability, the idea was to use PDL3 to reimplement the PDL core > - Allow support for PDL-2.x > - Ideally PDL3 core could be used instead of PDL/Basic/Core stuff > - Trying to avoid Numpy->Numeric->NumPy mess that python went through > * PDL::Tiny > - POGL (Perl OpenGL) needs a minimal but powerful array capability > - Other modules reinvent computation tools > - Gives a chance for direct work on the new PDL3 core capability > - Setting up github for fast iteration and my own education > - not coupled to PDL-2.x development except desire interoperable > * RPerl for PDL3 > - ? > * Support for row major and col major ordering in PDL3 > - PDL3 (hence PDL::Tiny) has improved type and data layout support > - Specific use cases on what is desired are needed > - It should be up to the user which data ordering is used > - To what level can this be supported? > * Use of Roles in PDL3 / PDL::Tiny > - Yes! > - Allow better structure for PDL > - PDL3 as a whole <-> like many PDL-2.x capabilities > - But can look at by role/component > - Elegant, general, compatible with perl5 oo and perl6 oo ideas > * Discussion of *.pd -> *.xs.PL > - *.pd was the original file name for PP XS extensions > - given the issues with http://search.cpan.org and http://metacpan.org > changing to *.xs.PL seems to make sense. > - Might be worth looking at perldl and pdl-porters archives for > previous > discussions and decision process on the topic > * Plan for PDL3 > - Start kernel out > - Build in types, data, other requirements > - Moo[se] seems elegant and powerful way to express > - Performance is king so we need to monitor > - idea use a C-OO framework to implement PDL3 OO at the C layer > - Call perl from C > - Call C from perl > - PDL3 methods could be called from both/either > - Allow interoperate with both C and C++ levels of libraries > - Enlightenment Object model is possible candidate > - Need to develop specification > - Add support for Moo roles and method modifiers > - Seems easier to start here then to roll our own > * Discussion of moving PDL web site to static github > - Seems reasonable but needs support > - We didn't have any volunteers the last time this was proposed > - With maintainer it would be nice to have more timely web support > > As you can see, there is a lot going on of interest that would be nice to > have input into our development work. I welcome any ideas or further > discussion. Please feel free to join #pdl for discussion but *please* take > the time to collect the good ideas and bring them in for more formal > discussion. I hate to see good ideas going to waste but, frankly, this > pulling gold out of the stream bed by hand is something that should be > handled by the #pdl discussion participants themselves. > > Suggestions to make that possible, VERY welcome, > Chris > > > * > > _______________________________________________ > PDL-porters mailing list > PDL-porters@jach.hawaii.edu > http://mailman.jach.hawaii.edu/mailman/listinfo/pdl-porters > >
_______________________________________________ PDL-porters mailing list PDL-porters@jach.hawaii.edu http://mailman.jach.hawaii.edu/mailman/listinfo/pdl-porters
