[R] e1071 and netpbm

[Friedrich Leisch]

[ ... taken from r-help to r-devel ... ]

R core kind of agreed to newly introduce configure capabilities to
packages such that we can handle situations like this cleanly. I.e.,
if the package contains a configur script, this is run to create a
proper Makefile. Hopefully I can then clean up the e1071 Makefile to
build the shared lib correctly on more than one platform.

fritz

>>>>> On Mon, 22 Mar 1999 14:55:03 +0100 (CET),
>>>>> Martyn Plummer (MP) wrote:

MP> [Snip: Problems installing the e1071 package on Red Hat Linux]
MP> Instead of having a single netpbm library, Red Hat has four libraries
MP> (all provided by the libgr rpm), with recursive dependencies

MP> libpnm -> libppm -> libpgm -> libpbm

MP> So, for example, libpbm stands alone but libpgm contains references to
MP> functions in libpbm.  This means that if you want to use any functions
MP> in libpnm (which we do), you must also link against the other three
MP> libraries. If you don't do this, then the compiler won't complain about
MP> undefined references when you type "R INSTALL e1071", since it is building
MP> a shared object, but you will get run time errors.

MP> I can't see the advantage of splitting the library, myself. I wrote to
MP> Red Hat about this in November but didn't get a reply.  It looks like
MP> various other Linux distributions which use the RPM packager (OpenLinux,
MP> TurboLinux, LinuxPPC, ...) use the same setup.

MP> I use the patch below when compiling on Red Hat
MP> Martyn
MP> diff -uNr e1071/src/Makefile e1071-patched/src/Makefile
MP> --- e1071/src/Makefile	Tue Nov 17 11:44:10 1998
MP> +++ e1071-patched/src/Makefile	Fri Nov 20 15:05:00 1998
MP> @@ -4,7 +4,7 @@
MP>  OBJS=pnm.o 
 
MP>  $(LIBNAME).so:  $(OBJS)
MP> -	$(LD) $(SHLIBLDFLAGS) -o $@ $(OBJS) -lnetpbm
MP> +	$(LD) $(SHLIBLDFLAGS) -o $@ $(OBJS) -lpnm -lppm -lpgm -lpbm
 
MP>  clean:
MP>  	@rm -f *.o *.so
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._