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

Reply via email to