[Rd] Downsized R configuration for flat deployment

[Gábor Csárdi]

Hi,

FWIW this recent project of mine has somewhat similar goals:
https://github.com/r-hub/r-minimal

Gabor

On Sat, Dec 21, 2019 at 2:40 PM <dmedri using gmail.com> wrote:
>
> Dear folks,
>
> I'm testing a downsized R build - in features and obviously sizes -
> for a "modern" flat deployment (eg. like python virtualenv, just to
> name one).
>
> Questions:
>
> 1) Is flat style possible?
> 2) With this setup, R and packages can be installed/updated?
> 3) The directory can be easy renamed or moved?
>
> -------------------------------------------------------------------
>
> I didn't find any official git repo, so I downloaded latest stable
> R release (3.6.2), exploring different setups of configure options.
>
> # tested in debian 10
> # $ sudo apt-get build-dep r-base
>
> $ mkdir ~/r-virtual
> $ cd R-3.6.2/
> $ ./configure --prefix=/home/dmedri/r-virtual/ \
>             --exec-prefix=/home/dmedri/r-virtual/ \
>             --without-recommended-packages \
>             --without-cairo \
>             --without-libtiff \
>             --without-jpeglib \
>             --without-x \
>             --without-aqua \
>             --without-tcltk \
>             --disable-rpath
>
> $ make && make install
>
> This minimal R could run inside a container, no desktop just cli,
> with just one graphical formats in output (png). Java support was
> leaved in. I didn't find options to disable html docs. Recommended
> packages were leaved out, for a 2nd stage.
>
> $ cd r-virtual/
> $ ./bin/R
>
> The environment was there, "yes" to question 1).
>
> I can install/update packages, obviously with required includes
> installed in the system, not fetched online -- a limit, looking at
> python pip, partially by design/choice. BTW answer to question 2)
> is "yes" for packages, "no" for the core.
>
> Without the "--disable-rpath" option, abs paths break everything
> with the easy case of "renaming r-virtual directory". IMHO this option
> should be a general default, and another "yes" to question 3).
>
> While I'm still investigating, tips are welcome...
>
> HTH
>
> Best Regards,
>
> DM
>
> ______________________________________________
> R-devel using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel