R breaks if SAGE_LOCAL undefined

[sagetrac-ahd]

horatio@havelock ~ $ R
/Users/horatio/bin/R: line 212: /lib/R//etc/ldpaths: No such file or directory

This is because on line 31 of the script someone rewrote $R_HOME_DIR in terms of $SAGE_LOCAL. When SAGE_LOCAL is missing from the environment of the running shell, the R script breaks. The fix is to test for empty SAGE_LOCAL before the rewrite.

This is from Sage version 4.6.2.

Depends on #9668

CC: @jasongrout @williamstein @nexttime

Component: packages: standard

Keywords: R script SAGE_LOCAL R.sh.in spkg local/bin r-project

Reviewer: Jeroen Demeyer

Issue created by migration from https://trac.sagemath.org/ticket/10967

[sagetrac-ahd]

patch to fix

[sagetrac-ahd]

comment:1

Attachment: trac_10967_R_sage_local_fix.patch.gz

[sagetrac-ahd]

Description changed:

--- 
+++ 
@@ -3,3 +3,5 @@
 
 This is because on line 31 of the script someone rewrote $R_HOME_DIR in terms of $SAGE_LOCAL. When SAGE_LOCAL is missing from the environment of the running shell, the R script breaks. The fix is to test for empty SAGE_LOCAL before the rewrite.
 
+This is from Sage version 4.6.2.
+

[sagetrac-mvngu]

comment:2

Is this ready for review?

[sagetrac-ahd]

comment:3

Replying to @sagetrac-mvngu:

Is this ready for review?

The fix is. It's not a mercurial-generated patch and there don't seem to be any developer guidelines for shell scripts.

[sagetrac-mvngu]

comment:4

Which file in the Sage distribution are you patching? Provide the exact path to the file from SAGE_ROOT.

[sagetrac-ahd]

comment:5

$SAGE_ROOT/local/bin/R

[sagetrac-mvngu]

comment:6

Replying to @sagetrac-ahd:

$SAGE_ROOT/local/bin/R

That's a wrong file to patch in order to get any changes into the R spkg. The file SAGE_ROOT/local/bin/R is part of the R spkg, so you should patch the R spkg. Get the latest Sage development release, unpack it, and look under SAGE_ROOT/spkg/standard/ for the R spkg. For more information on patching an spkg, see this page in the Sage Developer's Guide.

[kcrisman]

comment:7

Does this actually make the problem worse? The way I read the patch, before one did check for SAGE_LOCAL and afterwards one doesn't. Is this patch backwards?

I can't do it right now, but I should be able to produce an updated R package within a few days if ahd isn't able to.

Changing component, since the problem isn't the statistics functionality but rather the way the package is bundled.