[R] installation : make fails (R-1.7.1 on RedHat 8.0)

Laurent Faisnel laurent.faisnel at ariase.com
Fri Sep 26 16:12:57 CEST 2003

Peter Dalgaard BSA wrote:
> Laurent Faisnel <laurent.faisnel at ariase.com> writes:
>>Peter Dalgaard BSA wrote:
>>>Laurent Faisnel <laurent.faisnel at ariase.com> writes:
>>>>I'm trying to compile R-1.7.1 from source (on a RedHat 8.0) instead of
>>>>using the binary version, as it has often been advised.
>>>>However I don't manage to find a solution to the following error which
>>>>occurs during the make procedure :
>>>>gcc -I../../../../include /usr/include/mysql  -D__NO_MATH_INLINES
>>>                            ******************
>>>>-mieee-fp  -fPIC  -g -O2 -c ansari.c -o ansari.o
>>>The stars indicate the problem. This wants to -I/usr/include/mysql if
>>>anything, otherwise gcc will believe that there are two files to
>>>compile and complain about the -o. I suspect that you entered the
>>>wrong thing during configuration.
>>Thanks for your answer, but everything is not clear to me. This
>>installation was made with the default ./configure without any
>>additional option (so if I made something wrong, it is to forget an
>>option). I checked out ./configure --help.
>>What should I add ? I don't believe there are headers or libraries in
>>unusual places on my disk. Moreover, why does the script search for
>>mysql now ? I'm interested by mysql, but this comes later with RMySQL !
>>Anyway, mysql headers *are* in /usr/include/mysql.
> That's not the problem. The problem is that the "/usr/include/mysql"
> sneaked into your configuration files somehow. This can happen by
> other means than configure options:. environment variables or
> configuration files (~/.Rconf) are the two that I can recall just now.

I fixed the problem. You gave me the decisive clue : it was a problem 
with environment variables.

I had the two following definitions :

PKG_CPPFLAGS=/usr/include/mysql   and   PKG_CPPLIBS=/usr/lib/mysql

I set both variables to "" and now make works fine.
Thanks again.

More information about the R-help mailing list