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 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. 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 Pablo. 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
