Thanks Chris, About PGPLOT issue, I really don't use it, I just installed it because was in the recommended part of http://pdl.perl.org/?page=install.
On Wed, Dec 31, 2014 at 3:48 PM, Chris Marshall <[email protected]> wrote: > On Mon, Dec 22, 2014 at 1:06 AM, Pablo marin-garcia < > [email protected]> wrote: > >> Hello Chris, >> >> I have take a look at the the failure matrix of Module::Compile[1] in >> the automatic testing and it is a bit worrying for having this module as a >> dependency. Probably the poor score is because the undefined dependency on >> Module::Install but anyway too much RED in there. It would be nice to add a >> note in the PDL install page about it, just in case. >> >> [1] http://matrix.cpantesters.org/?dist=Module-Compile%200.34 >> > > This problem was reported and has been addressed in the latest version of > Module::Compile. Thanks mohawk and ingy. > > >> One of the reason I use R instead PDL in production software that I need >> to distribute, even for simple things that would be more natural to use >> PDL, is because is easier to install R and bioconductor as not root than >> PDL. Nowadays things has been improved a lot, and I think that for next >> release we can help to keep improving it if we make a 'install hackaton' >> and see how much success we have with a simple 'cpanm PDL' in different >> environments. Many of the issues are not with PDL itself but its >> dependencies, so having a list of common pitfalls and its workarounds would >> be helpful. >> > > I'm sorry to hear about your difficulties. PDL builds out of the box on > the platforms I have access to but the external dependencies are still a > work in progress since different platforms have different requirements. If > you have a specific set of problems with distributing PDL, please do work > with the perldl lists to resolve the problem and open a ticket for the > issue on our bug tracker at http://sourceforge.net/p/pdl/bugs/ > > >> >> For example a recent 'issue' I have was having perl module for PGPLOT >> installed: >> http://stackoverflow.com/questions/27585591/how-to-avoid-cpanm-stopping-installation-waiting-for-a-prompt-that-is-shown-only >> > > I can't help with PGPLOT as I am unable to build it on cygwin. There are > other plotting options for PDL which may simpler to provide in a > distribution: PDL::Graphics::Gnuplot, PDL::Graphics::Prima come to mind. > The latest version of PLplot now builds out of the box on cygwin so I plan > to update PDL::Graphics::PLplot as time permits. > > Regards, > Chris > > >> >> >> On Mon, Dec 22, 2014 at 1:44 AM, Chris Marshall <[email protected]> >> wrote: >> >>> Thanks for the feedback, Pablo. The warning about another PDL in your >>> path is because you can get test failures if the new install has a >>> different configuration than the previous install. In my experience so >>> far, that has meant false failures in that the new install is good for its >>> part. At some point it might make sense to "hide" the previous install for >>> testing but that hasn't happened yet. The warning is an intermediate step. >>> >>> I see you've found bug ticket #365 regarding the recent Module::Compile >>> regression. This is definitely one of the things to be fixed for the >>> PDL-2.008 release this January. >>> >>> --Chris >>> >>> >>> On Sun, Dec 21, 2014 at 6:49 PM, Pablo marin-garcia < >>> [email protected]> wrote: >>> >>>> Hello Chris and PDLers >>>> >>>> Just as you are talking about a new release and we would need to >>>> update/install it, I would like to talk about a recent issue I had >>>> installing PDL. >>>> >>>> I had 'issues' with some perl dependencies with installing cpanm >>>> PGPLOT and cpanm Module::Compile but I will I report it in a blog entry >>>> soon so I could help to improve the documentation. >>>> >>>> Here I would like to ask a very specific question: If you have already >>>> a PDL installed and you install a new version a warning appears: >>>> >>>> -------------------------------------------------- >>>> Unpacking PDL-2.007.tar.gz >>>> Entering PDL-2.007 >>>> Checking configure dependencies from META.yml >>>> Checking if you have Devel::CheckLib 1.01 ... Yes (1.02) >>>> Configuring PDL-2.007 >>>> Running Makefile.PL >>>> >>>> +------- WARNING ------------------------------------------------ >>>> | PDL version 2.4.11 was detected in your @INC >>>> | PDL may not build/test correctly with another PDL in @INC! >>>> +---------------------------------------------------------------- >>>> >>>> ------------------------------------ >>>> >>>> This mean that I need manually to remove my previous PDL installtion >>>> first? >>>> How do I remove my old vesion of @INC?, you mean I need to rm -rf it?. >>>> Could this message be improved and hints how to avoid/fix it be added? >>>> >>>> Anyway, as this is a warning seems not to impede the the installation: >>>> >>>> -------------- >>>> Successfully installed PDL-2.007 (upgraded from 2.4.11) >>>> Installing >>>> ~/perl5/perlbrew/perls/perl-5.16.0/lib/site_perl/5.16.0/x86_64-linux/.meta/PDL-2.007/install.json >>>> Installing >>>> ~/perl5/perlbrew/perls/perl-5.16.0/lib/site_perl/5.16.0/x86_64-linux/.meta/PDL-2.007/MYMETA.json >>>> 1 distribution installed >>>> ----------------- >>>> >>>> Note: a dependency of PDL is Module::Compile, but this does not get >>>> installed with cpanm because one of its dependencies it is not stated in >>>> the package dependencies: Module::Install. This is a known bug >>>> https://github.com/ingydotnet/module-compile-pm/issues/1 but in my >>>> case was more cryptic than the ones reported: >>>> >>>> | PERL_DL_NONLAZY=1 ~/perl5/perlbrew/perls/perl-5.16.0/bin/perl >>>> "-MExtUtils::Command::MM" "-e" "test_harness(0, 'blib/lib', 'blib/arch')" >>>> t/*.t >>>> | t/000-compile-modules.t .. 1/? >>>> | # Failed test 'use Module::Compile;' >>>> | # at t/000-compile-modules.t line 11. >>>> | # Tried to use 'Module::Compile'. >>>> | # Error: Can't open (eval 6) for input: ###<<<<<<======= >>>> | # No such file or directory at >>>> ~/.cpanm/work/1419122233.15164/Module-Compile-0.34/blib/lib/Module/Compile.pm >>>> line 201. >>>> | # BEGIN failed--compilation aborted at (eval 6) line 2. >>>> | >>>> | # Failed test 'use Module::Optimize;' >>>> | # at t/000-compile-modules.t line 11. >>>> | # Tried to use 'Module::Optimize'. >>>> | # Error: Can't open (eval 7) for input: >>>> | # No such file or directory at >>>> ~/.cpanm/work/1419122233.15164/Module-Compile-0.34/blib/lib/Module/Compile.pm >>>> line 201. >>>> | # BEGIN failed--compilation aborted at (eval 7) line 2. >>>> | # Looks like you failed 2 tests of 5. >>>> >>>> In my errors it was not mention what was the problem, looking at bugs >>>> opened I saw that the a missed dependency was the culprit Module::Install. >>>> I installed it and all went fine. I mention this here just in case you >>>> want to add this as a NOTE in the web page for PDL installation. >>>> >>>> Pablo >>>> >>>> >>>> >>>> >>>> On Sat, Dec 20, 2014 at 9:41 PM, Chris Marshall <[email protected] >>>> > wrote: >>>> >>>>> All- >>>>> >>>>> I've just reviewed the status of the bug and feature request trackers >>>>> at the PDL site on sf.net: >>>>> >>>>> http://sourceforge.net/p/pdl/bugs/ >>>>> http://sourceforge.net/p/pdl/feature-requests/ >>>>> >>>>> I found a number of bug tickets that had already been fixed and >>>>> determined that there are a fair number of tickets that are either low >>>>> hanging fruit or already fixed but need to be verified to close. >>>>> >>>>> I had hoped that the 64bit support could have been finished by now but >>>>> lack of developers to debug the final issue and time for me to get back to >>>>> it, has it still in limbo. We have a number of nice improvements in the >>>>> hopper already and it has been a little more than a year since PDL-2.007. >>>>> >>>>> I propose we make a bug fix/clean up/... release of PDL-2.008 for >>>>> January 2015---regardless of whether the remaining 64bit index support is >>>>> finished. I encourage all users and developers to take this opportunity >>>>> to >>>>> take a look at the outstanding bugs and close them. There are a number of >>>>> feature requests that relate to some items such as improved PDL docs >>>>> handling that could also be addressed for this release. >>>>> >>>>> And, of course, I would love some assistance to finish the >>>>> implementation debugging for the PDL-2.x 64bit index support. :-) >>>>> >>>>> As always, thoughts and suggestions welcome! >>>>> Chris >>>>> >>>>> >>>>> >>>>> >>>>> _______________________________________________ >>>>> Perldl mailing list >>>>> [email protected] >>>>> http://mailman.jach.hawaii.edu/mailman/listinfo/perldl >>>>> >>>>> >>>> >>> >> >
_______________________________________________ Perldl mailing list [email protected] http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
