Re: [Fedora-r-devel-list] Building RPMs from CRAN...
On Thu, 2010-01-07 at 14:46 -0500, Allen S. Rout wrote:
Pierre-Yves <pingou(a)pingoured.fr> writes on R-devel:
> One of them is R2spec [1], which I have worked on. I would also
> point you to a 'fork' of R2spec [2] which aims at building RPM more
> than spec (reason why they did not merge). R2spec aims a helping to
> generate spec files as close as possible to the final one for
> inclusion in official Fedora/EPEL repositories. Allen S. Rout have
> made a nice job converting R2spec to generate RPM directly, solving
> some of the circular dependences issues.
I appreciate your kind mention of my effort, Pierre-Yves. Thank you!
That's because I appreciate them :)
I am dissapointed that you don't see a merge in the future. I would
like to change that if I can. What would you like me to change?
I will for sure
include some of the change you have made (like using the
regular expression instead of my crappy (but working) implementation).
IMHO the problem comes from what you want to achieve and how I see
R2spec.
You want to build a build tool which is able to build R RPM in a chain
way resolving the dependencies. This is great !
But this is R2rpm not R2spec ;-)
IMHO R2spec is here to help packagers to create/generate spec file as
close as possible to the Fedora standards.
Fedora's RPM will never be generated fully automatically and be
maintained without human intervention that's against its philosophy.
So your tool is I believe really interesting if we want to set up an
external repository out of the official Fedora repository in which we
would be able to use your tool.
There are already some projects hanging around in bioinformatics which
try to do so (I have in mind the BioRPMS [1], the biolinux [2] or the
BioConductor RPM repo [3]). The problem is that most of the time they do
not respect the standards asked by Fedora.
One example which appears in your application is the %file section of
your spec. In order to be able/allow the user to do an rpm -ivh R-*.rpm
--excludedocs we marked a number of files as being documentation. Using
your spec it is not possible.
This is fine, for an external repository but not for inclusion in the
Fedora repository.
It is my hope that the R2spec code will make it possible,
eventually,
to build and maintain _all_ the CRAN packages as RPMs, without the
extensive by-hand treatment they now get. I understand that, if this
is to happen, more folks than just me need to be happy with the
code. :)
What I think we could do, is actually set up a repository of RPM for
CRAN, for Bioconductor external to the official Fedora repositories
where your tool would be used and very much appreciated.
I am part of the project on bioinformatics.org ([3]), and if you are
willing to give it a shot I believe we could sort something out. On my
side time available is really small but I believe we could find some
more people willing to help for such project.
My position has always been and is still to push packages to the
official repositories rather than creating another place and thus
dividing the forces. But I have seen the questions of the availability
of R RPMs more than once, people contacted me. I unfortunately have not
seen them coming into the Fedora packaging world, maybe it is time for
me to accept this fact and accept to push for an external repository
maybe as a transitional phase before inclusion into Fedora.
I hope I have been clear on how I see the difference between your tool
and mine and where I stand at the moment.
I would be pleased to hear from the people of this list what they think
about this problematic (also considering what Debian has done).
Anyway thanks again for your work.
Best regards,
Pierre
[1] http://informatics.umdnj.edu/BioRPMs/
[2] http://biolinux.org (down atm)
[3] http://www.bioinformatics.org/r-repo/wiki/