Handle preset R_PROFILE or R_HOME variables

[qed777]

A preset R_PROFILE variable can cause some R packages to fail to build/install and interfere with doctests.

Georg Grafendorfer reported this problem on sage-release.

For R_HOME, this yields a doctest failure if set:

**********************************************************************
File "devel/sage/sage/misc/interpreter.py", line 187, in sage.misc.interpreter.SageInteractiveShell.system_raw
Failed example:
    shell.system_raw('R --version')
Expected:
    R version ...
Got:
    WARNING: ignoring environment value of R_HOME
    R version 2.15.2 (2012-10-26) -- "Trick or Treat"
    Copyright (C) 2012 The R Foundation for Statistical Computing
    ISBN 3-900051-07-0
    Platform: x86_64-unknown-linux-gnu (64-bit)
    <BLANKLINE>
    R is free software and comes with ABSOLUTELY NO WARRANTY.
    You are welcome to redistribute it under the terms of the
    GNU General Public License versions 2 or 3.
    For more information about these matters see
    http://www.gnu.org/licenses/.
    <BLANKLINE>
**********************************************************************

Apply attachment: 9847_unset_R.patch to SAGE_ROOT.

CC: @sagetrac-ggrafendorfer @kcrisman

Component: build

Author: Jeroen Demeyer

Reviewer: Leif Leonhardy, Karl-Dieter Crisman

Merged: sage-5.9.beta2

Issue created by migration from https://trac.sagemath.org/ticket/9847

[qed777]

comment:1

From around line 180 of SAGE_LOCAL/bin/sage-env:

if [ -z "$RHOME" ]; then
    RHOME="$SAGE_LOCAL/lib/R" && export RHOME
fi

Should we add a similar test for R_PROFILE?

[nexttime]

comment:2

I think we should perhaps

unset R_PROFILE

in R's spkg-install, too.

[kcrisman]

comment:3

Apparently this might be addressed by a solution to #9906.

[nexttime]

comment:4

Replying to @nexttime:

I think we should perhaps

unset R_PROFILE

in R's spkg-install, too.

#9906 (R 2.10.1.p5) does this now.

[nexttime]

Dependencies: #9906

[kcrisman]

comment:5

Thanks. Trac isn't sending emails, as you know, so I just saw this. (Well, six hours isn't bad.) I'll take a look at #9906 then.

[kcrisman]

Reviewer: Leif Leonhardy, Karl-Dieter Crisman

[kcrisman]

comment:6

Sorry I didn't come back to this. It is correct that the spkgs at #9906 do this in the spkg scripts.

So I can give this positive review as long as Leif promises that if I or some other reviewer find out that the fix at #9906 doesn't solve the type of problem above, that it will get addressed there and no longer here!

Sound good? ;-) But I don't think that will be necessary.

[nexttime]

comment:7

Replying to @kcrisman:

Sorry I didn't come back to this. It is correct that the spkgs at #9906 do this in the spkg scripts.

So I can give this positive review as long as Leif promises that if I or some other reviewer find out that the fix at #9906 doesn't solve the type of problem above, that it will get addressed there and no longer here!

In that case, you should also give this ticket positive review, such that the release manager will close it, with "resolution: duplicate".

If issues with R_PROFILE should (re)arise unrelated to (the installation of) the R spkg, i.e., which would have to be fixed elsewhere, we can still reopen this ticket.

(IIRC some of the problems showed up during doctesting, but because R_PROFILE had been set during installation, so are now fixed by the new spkg. If there needs to be done more w.r.t. this -- which I don't think -- this can be addressed at #9906 as Karl-Dieter suggests.)

Note that this ticket already has #9906 as its dependency.

[kcrisman]

comment:8

So I can give this positive review as long as Leif promises that if I or some other reviewer find out that the fix at #9906 doesn't solve the type of problem above, that it will get addressed there and no longer here!

In that case, you should also give this ticket positive review, such that the release manager will close it, with "resolution: duplicate".

Correct. I just wanted to make sure :) Great.

(IIRC some of the problems showed up during doctesting, but because R_PROFILE had been set during installation, so are now fixed by the new spkg. If there needs to be done more w.r.t. this -- which I don't think -- this can be addressed at #9906 as Karl-Dieter suggests.)

Yeah, I'm pretty sure that was the issue, or more precisely that R_PROFILE was not (re)set.

Note that this ticket already has #9906 as its dependency.

Well, that wouldn't be necessary with a duplicate/wontfix one.

See you at #9906, eventually (when I get time to do all the many tests on that).