On Fri, Oct 20, 2017 at 11:47 PM, François Bissey <frp.bis...@gmail.com> wrote:
>
>
> On 21/10/2017, at 10:37, Michael Orlitzky <mich...@orlitzky.com> wrote:
>
> On 10/20/2017 11:30 AM, Nils Bruin wrote:
>
> Do we actually get real benefit from packaging R with sage? It seems to
> me R is a very popular, very mature piece of software. Isn't it easy to
> install R on any of the platforms supported by sage? If that is the
> case, wouldn't it be much easier to let sage just work with the
> system-wide R if that is available?
>
>
>
> No, yes, yes.
>
>
>
> Of course we had those conversations before. sage interfaces with R in two
> ways:
> * directly with the R command line through the pexpect interface
> * through rpy
>
> The first way can happily use system wide R.
> The second is more complicated. We have to build sage’s rpy and rpy
> links directly to libR and that’s where the potential madness comes in,
> libR links to blas/lapack. So there is a potential clash between
> system blas/lapack and sage blas/lapack in rpy.
> This may (or not) be dealt with by being careful in building rpy in sage.

This is a problem for using some other system packages in Sage as
well, so it would be good to resolve in general.

I don't have a strong opinion on whether or not Sage should include a
package for R, or if should be standard or optional.  But I do believe
that in either case it should be able to resolve an R dependency from
the system R, as should be the case for almost any other dependency
(which is a problem I'm still working on but progress is slow due to
slow progress in getting the necessary preliminaries accepted...)

-- 
You received this message because you are subscribed to the Google Groups 
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to sage-devel+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.

Reply via email to