[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Decision on R datasets

Hi Dirk,

Sorry, I only just saw this message. I forgot to ask for a CC in my
original message.

On Fri, 27 Sep 2013 20:22:01 +0000 (UTC), Dirk Eddelbuettel<edd@debian.org> wrote:

> Faheem Mitha <faheem <at> faheem.info> writes:

>> I'm sorry to hear that you will not be working on R packaging for
>> Debian any more. Unfortunately. there are very few people working
>> on R packaging in Debian. There is Dirk, of course, but few other
>> names appear consistently. In particular, there is nothing like a R
>> packaging team despite R's significant and growing importance in
>> the larger FOSS community.

> There is Don Armstrong's r-debian.debian.net which turns CRAN
> packages into Debian packages (building on two earlier cran2deb
> efforts I was involved in, once with an extremely gifted GSoC
> student).

> And there is Michael Rutter's c2d4u variant using launchpad (for

Google gives me https://launchpad.net/~marutter/+archive/c2d4u

Is this the right site?

> Both autogenerated thousands of packages.

I wasn't aware of either of these. Thanks.

> Within Debian, it is complicated. I am not sure what percentage of
> CRAN we should package.  The most important packages, yes.  But
> indiscriminately?  Not sure.  Installing in /usr/local and using R
> to update works really well too.

True, there are thousands of packages in CRAN, and packaging all of
these to Debian's exacting standards would be a Herculean task.

An issue does arise when one wants to package some software that
depends on R packages. Because there are so many R packages, it is
relatively likely some of these dependencies will not be packaged, so
that adds extra work. But I don't see a good solution to this. In
practice people seem to package these on an as-needed basis, which may
be the best one can do.

Packaging even a single R package to conform to Debian standards may
not be trivial. For example, I recently tried to package the R YAML
package, but discovered that it includes the YAML C library files in
the R sources and was linking to them. This is a fairly obvious
violation of Debian Policy. I didn't have time to sort it out, so I
still haven't finished that packaging.

                                                       Regards, Faheem

Reply to: