On Tue, 2006-02-28 at 19:25 -0600, Dirk Eddelbuettel wrote:
> Hi Ross,
> 
> On 28 February 2006 at 16:37, Ross Boylan wrote:
> | Package: r-base-core
> | Version: 2.2.1-6
> | Severity: normal
> | 
> | For example,
> | $ R CMD config --cppflags
> | -I/usr/lib/R/include
> 
> I'd argue that is indeed an upstream bug as I called configure via
> 
>                   --datadir=/usr/share/R/share        \
>                   --includedir=/usr/share/R/include   \
> 
> In fact, I raised the exact same issue on Feb 20 on r-devel.
> 
> Brian Ripley responded, acknowledged the issue (and a bit more) and said it
> is fixed upstream. 
> 
> We could downgrade this and keep it open til 2.3.0 comes out, or fix it now.

I think the former path is fine, since the symlink gets me symptomatic
relief (I think; I'm getting some weird errors in libstdc++, but I don't
think they have anything to do with this).  Certainly sounds like an
upstream issue.

> 
> | $ ls /usr/lib/R
> | AUTHORS  COPYING.LIB  FAQ   RESOURCES     THANKS  etc  library  share
> | COPYING  COPYRIGHTS   NEWS  SVN-REVISION  bin     lib  modules  site-library
> | 
> | The file has moved to /usr/share/R/include.
> | 
> | It seems possible other configuration information is also behind the
> | times.  I'm a little surprised the build info wasn't automatically
> | reflected; I suppose all the games with paths during the build (i.e,
> | Debian builds in subdirectory and the relocates) account for that.
> 
> Rest assurred that we have built dozens of R / CRAN packages for Debian in
> unattended pbuilder session, and that I have added or updated > 100 CRAN /
> BioC package during recent Quantian upgrades.

Good.  I thought when R builds packages from source it uses the results
of R CMD config; apparently it doesn't, or at least doesn't depend on it
exclusively.  I noticed the problem because I'm building something that
does use R CMD config's output.

Thanks for your quick response--actually 8 days *before* the bug if you
count your earlier exchange on r-devel!  Certainly a new record :)

Ross



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to