[Bioperl-l] Packaging bioperl for Fedora

[Sendu Bala]

Alex Lancaster wrote:
> Hello bioperl,
> 
> I'm new to the bioperl world, having just started a research position
> in which I need to manage a large bioperl-based codebase.  To this
> end, I'm working on packaging bioperl as an official Fedora Package

[snip]
Thanks, that would be appreciated I'm sure.


> 1) There is one dependent package: perl-SVG-Graph that has
>    questionable licensing status that currently prevents it from being
>    packaged, see:
> 
>    http://bugzilla.redhat.com/233848
[snip]
>    If there is anybody on this list who has access to this package
>    and/or knows how to contact the upstream authors to clarify the
>    license conditions?  Please let me know, or better, if you can
>    clarify/fix the license in the CPAN modules itself, then please do!

Allen should have answered; he is still around and posting to this list 
from time to time.


> 2) http://www.bioperl.org/wiki/Getting_BioPerl lists 1.4.0 as the
>    current "stable" release, but that's getting pretty old, having
>    been released back in 2003.  Do you currently recommend people use
>    1.4.0 or one of the 1.5.x series?  I'd rather package the more
>    recent version even if it's not "officially" stable because I know
>    if most bioperl people are using more recent then they'll be calls
>    to also package the newest release and doesn't seem sensible to
>    spend a lot of time packaging the older release.  If so, is there a
>    particular 1.5.x package you recommend, perhaps one which is
>    somewhere between totally-stable and bleeding-edge?  (I started on
>    1.5.2_102 since it was the latest).

The latest release is the one to go for, suitable for use by most users. 
So yes, 1.5.2_102. We don't do 'bleeding-edge' releases - for 
bleeding-edge we ask users to use CVS directly. 'Stable' in this context 
is in regard to the API, not bugs.


> 3) Lastly I see that bioperl is now split into bioperl (core),
>    bioperl-run, bioperl-db, bioperl-network etc.  Do you recommend
>    that each be packaged separately?  For example, I notice that some
>    scripts in bioperl (core) actually depends on Perl modules that are
>    provided by bioperl-run, which means that they will both be needed
>    to installed simultaneously to make rpm happy.

Ah, well that perhaps shouldn't be the case. We like to keep them 
separate since core is usable on its own and the other packages won't be 
needed by many users. Correspondingly separate rpms would be appropriate.


Cheers,
Sendu.