[R-sig-Debian] cannot find -lquadmath

[Dirk Eddelbuettel]

On 15 January 2012 at 12:11, Peter Muhlberger wrote:
| Hi Dirk:
| 
| What would commenting out  -lquadmath in  /etc/R/Makeconf do?  If a

Iff it got there because R was conf'ed that way at _its_ buildtime yet the
compiler on the system _running the Hmisc build_ (ie your box) doesn't need
it, then commenting out would remove an unnecessary constraint.

| package, such as Hmisc, requires it, wouldn't that prevent compiling
| or does the compiler then do more of a search for the library at that
| point?
| 
| I wonder if I could have prevented this problem by pulling in a newer
| version of R from backports rather than sid?

I have no idea as I don't have the issue on my systems. I don't mix and mix
between distro variants. If I need something, I tend to build a local .deb.

Dirk

| 
| Peter
| 
| On Fri, Jan 13, 2012 at 4:48 PM, Dirk Eddelbuettel <edd at debian.org> wrote:
| >
| > Hi Peter,
| >
| > On 13 January 2012 at 16:01, Peter Muhlberger wrote:
| > | Hi Dirk:
| > |
| > | I checked several months of r-sig-debian archives for a keyword
| > | associated with my problem & found nothing.  Clearly, I used the wrong
| > | keyword (probably "libquad").  Now that you mention that there was a
| > | recent discussion, I can see it, and it's helpful.  Wish I knew of a
| > | way to do a better search on the archive.
| >
| > Yes, sorry, and it was more or less timing issue. You and the other poster
| > were pretty hit at the same time.
| > |
| > | I thought my experience might be helpful because I have a fairly
| > | plain-vanilla version of Squeeze and build-essentials.  I gather from
| > | the prior discussion on Jan 12 that my problem is that I updated my
| > | copy of R using the sid archive (had to to use the sem package), so
| > | now when I build it's probably expecting a more recent compiler setup.
| >
| > Right. That "gap" between sid and squeeze is usually non-existent, and I have
| > also at times done a pick and choose.
| >
| > In this case, though, it seems that there is a compiler change and the two
| > are different.
| >
| > You could try editing /etc/R/Makeconf and just comment out -lquadmath.
| >
| > Dirk
| >
| > |  So maybe my problem is more esoteric than I imagined.  It does seem
| > | complicated to be a R user in Squeeze.
| > |
| > | It's also helpful to know that there are binaries for things like
| > | Hmisc in the archives.
| > |
| > | Thanks in general for tackling the inclusion of R in Debian!
| > |
| > | Cheers,
| > | Peter
| > |
| > | On Thu, Jan 12, 2012 at 10:21 PM, Dirk Eddelbuettel <edd at debian.org> wrote:
| > | >
| > | > On 12 January 2012 at 21:30, Peter Muhlberger wrote:
| > | > | Apologies if this has been posted about recently, but I haven't been
| > | > | on this listserve for a while.
| > | >
| > | > That is rude. This was discussed, and a fix layed out, within the last 24 hours.
| > | >
| > | > | I ran into a problem trying to compile Hmisc (required package for
| > | >
| > | > 1)  r-cran-hmisc was part of Debian for years. Why compile?
| > | >
| > | > 2)  r-cran-rms is part of Debian, the 'rms' package has replaced Hmisc /years
| > | >    ago/ upstream.
| > | >
| > | > | rms) and have found a solution that I'd like to share.  The compile
| > | > | fails with the message:
| > | > |
| > | > | /usr/bin/ld: cannot find -lquadmath
| > | > |
| > | > | libquadmath exists on my Debian Squeeze system (up-to-date, recently installed):
| > | > |
| > | > | ldconfig -p | grep libquad*
| > | > |        libquadmath.so.0 (libc6,x86-64) =>
| > | > | /usr/lib/x86_64-linux-gnu/libquadmath.so.0
| > | > |
| > | > | The problem appears to be that libquadmath.so.0 is buried in the
| > | > | directory x86_64-linux-gnu where the compiler apparently can't see it.
| > | > |
| > | > | The fix is:
| > | > |
| > | > | cd /usr/lib
| > | > | sudo ln -s /usr/lib/x86_64-linux-gnu/libquadmath.so.0 /usr/lib/libquadmath.so
| > | > |
| > | > | Now, the compile works.
| > | > |
| > | > | Whoever is setting up R or R packages for Debian should try to insure
| > | > | that users don't have to apply fixes like the one above.
| > | >
| > | > Nonsense.  Look at the compile logs of /hundreds/ of r-cran-* packages across
| > | > /more than a dozen/ architectures.  They build.
| > | >
| > | > Look eg at http://packages.qa.debian.org/r/r-cran-rms.html and click on build
| > | > log, or go to https://buildd.debian.org/status/package.php?p=r-cran-rms
| > | >
| > | > Funny how it builds on all of our machines witout intervention, eh?
| > | >
| > | > Dirk
| > | >
| > | > |
| > | > | Cheers,
| > | > | Peter
| > | > |
| > | > | _______________________________________________
| > | > | R-SIG-Debian mailing list
| > | > | R-SIG-Debian at r-project.org
| > | > | https://stat.ethz.ch/mailman/listinfo/r-sig-debian
| > | >
| > | > --
| > | > "Outside of a dog, a book is a man's best friend. Inside of a dog, it is too
| > | > dark to read." -- Groucho Marx
| >
| > --
| > "Outside of a dog, a book is a man's best friend. Inside of a dog, it is too
| > dark to read." -- Groucho Marx

-- 
"Outside of a dog, a book is a man's best friend. Inside of a dog, it is too
dark to read." -- Groucho Marx