Re: Lintian error on- rpath: Would like an exemption

[Dirk Eddelbuettel <edd@debian.org>]

Kurt,

Thanks for taking the time to follow-up on this.
On 8 January 2010 at 22:04, Kurt Roeckx wrote:
| On Tue, Jan 05, 2010 at 07:58:36PM -0600, Dirk Eddelbuettel wrote:
| > 
| > littler (aka 'r') is a small wrapper around GNU R that Jeff Horner and I
| > released in 2006 and have been maintaining since.  It is similar to Rscript
| > (which appeared a little later as part of base R) but implemented differently
| > -- we query a lot of the environment at build time and encode this.  And we
| > also use an -rpath against R's main dynamic library.  As a consequence, we
| > start faster than either Rscript or R.
| > 
| > And I was about to make a new upstream release, but now see
| > 
| >     E: littler: binary-or-shlib-defines-rpath ./usr/bin/r
| 
| $ objdump -p /usr/bin/r |grep RPATH
|   RPATH                /usr/lib64/R/lib:/usr/lib/jvm/java-6-openjdk/jre/lib/amd64/server:/usr/lib/jvm/java-6-openjdk/jre/lib/amd64:/usr/lib/jni:/usr/lib/libfakeroot:/usr/lib64/libfakeroot:/usr/lib32/libfakeroot
| 
| And you only seem to be using libraries from /lib, /usr/lib and /usr/lib/R/lib.
| 
| I think lintian is right to complain.
| 
| Anyway, I think that /usr/lib/R/lib/libR.so really should be in
| /usr/lib/ and that you shouldn't need an rpath at all.

A few quick comments:

i)    There are several entries because I (at compile time) deconstruct the
      LD_LIBRARY_PATH set by /usr/bin/R (a shell script) used to load R from
      /usr/lib/R/bin/exec/, this uses /usr/lib/R/lib.  We could strip every
      thing but /usr/lib64/R/lib and the Java libs, but why bother

ii)   /usr/lib/R/lib (or /usr/lib64/R/lib) is needed and will not move,
      unless someone (you ?) convinces R Core to change it. (Hint: Won't
      happen).   This is how it is, this has been discussed before and you
      are kindly invited to read up on the old discussions.

iii)  /usr/bin/r, for which I am 1/2 of upstream, won't change either because
      this embedding of R is its whole point (vis-a-vis the shellscript /usr/bin/R)

My post was a false alert anyway; I had one actual bug with an 'empty' entry
and that has been fixed. 

Cheers, Dirk


-- 
Three out of two people have difficulties with fractions.