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

Reply via email to