[Rd] RH7.0 compilation
Peter Dalgaard BSA
21 Dec 2000 21:57:40 +0100
> Hi everyone,
> Like most people with RH7.0 Linux, I had problems with their "experimental"
> gcc-2.96 compiler. To circumvent the problem, I installed stock gcc-2.95.2
> in the /usr/local tree, so it will be picked up first as /usr/local/bin if
> before /usr/bin on my PATH. With this setup I was able to compile R-1.2.0
> straight out of the box (as well as compile Linux kernels). One problem I
> had, however, was that the Matrix package would not link, with linker
> complaining about "undefined versioned symbol name clog@@GLIBC_2.1".
> Yesterday I downloaded updated RPMs for the gcc compiler from the RH site.
> I installed them and, just to check them out, recompiled R-1.2.0 with it.
> It compiled fine and passed all the checks. Moreover, now the Matrix
> package compiles and installs fine. The only thing I see different is that
> while compiling R-1.2.0 I get a bunch of warnings like this:
> making devices.d from devices.c
> In file included from devices.c:25:../../src/include/Defn.h:73:22: warning:
> nothing can be pasted after this token
> making dynload.d from dynload.c
> and like this
> FFTab.h:2:29: warning: pasting would not give a valid preprocessing token
I think these are harmless and it should work to remove the ## in
# define DYNLOADEXT "." ## SHLIB_EXT
# define F77_QSYMBOL(x) #x ## "_"
You're not supposed to use ## to stick text strings together like
that, only tokens, but it seems to have worked with all other
compilers. (I think we analyzed this earlier and decided not to change
the code, but I forgot why. Maybe we just plain forgot to fix it.)
O__ ---- Peter Dalgaard Blegdamsvej 3
c/ /'_ --- Dept. of Biostatistics 2200 Cph. N
(*) \(*) -- University of Copenhagen Denmark Ph: (+45) 35327918
~~~~~~~~~~ - (firstname.lastname@example.org) FAX: (+45) 35327907
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: email@example.com