[Rd] compile fails with x86_64-alpine-linux-uclibc-gcc
hello,
i am trying to build R on alpine/gentoo with uclibc and it fails with the
following
error msg:
connections.c: In function 'Rconn_fgetc':
connections.c:3184:11: error: expected identifier before '(' token
connections.c:3186:15: error: expected identifier before '(' token
make[3]: *** [connections.o] Error 1
here is the whole build log :http://bpaste.net/show/39955/
any idea?
i've *walked* back the versions down to 2.14.1 and i still get exact same
error msg.
there are some patches applied but none of them touches connections.c, i
think. in any case i can provide the patches if need be.
thanks
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] compile fails with x86_64-alpine-linux-uclibc-gcc
On Mon, Aug 13, 2012 at 10:49:26AM -0400, Simon Urbanek wrote:
> I suspect that your runtime/libc is defining fgetc as a macro which breaks
> any code that uses it as an identifier. Ideally, your runtime should be fixed
> to use a proper function, but you could probably work around it with
> something like
>
> static char * fix_fgets(char *s, int n, FILE *stream) { return fgets(s, n,
> stream); }
> #undef fgets
> static char * fgets(char *s, int n, FILE *stream) { return fix_fgets(s, n,
> stream); }
>
> Cheers,
> Simon
like this? :http://bpaste.net/show/40047/
if so, then i get this err msg:
connections.c:385:15: error: static declaration of 'fgets' follows non-static
declaration
/usr/include/stdio.h:544:14: note: previous declaration of 'fgets' was here
connections.c: In function 'Rconn_fgetc':
connections.c:3192:11: error: expected identifier before '(' token
connections.c:3194:15: error: expected identifier before '(' token
and here is how it's declared in stdio.h on my system:
/* Get a newline-terminated string of finite length from STREAM.
This function is a possible cancellation point and therefore not
marked with __THROW. */
extern char *fgets (char *__restrict __s, int __n, FILE *__restrict __stream)
__wur;
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] compile fails with x86_64-alpine-linux-uclibc-gcc
On Tue, Aug 14, 2012 at 01:48:44PM +0200, peter dalgaard wrote: > > Perhaps I'm confused, but how did either of you expect to fix a problem with > fgetc by modifying fgets??? > > -- > Peter Dalgaard, Professor > Center for Statistics, Copenhagen Business School > Solbjerg Plads 3, 2000 Frederiksberg, Denmark > Phone: (+45)38153501 > Email: pd@cbs.dk Priv: pda...@gmail.com i'm not well versed in C. do know how to fix this? or, as i gather, this is not something trivial and it has to be fixed uclibc upstream? __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] compile fails with x86_64-alpine-linux-uclibc-gcc
On Tue, Aug 14, 2012 at 10:32:22AM -0400, Simon Urbanek wrote:
>
> On Aug 14, 2012, at 10:19 AM, nobody wrote:
>
> > On Tue, Aug 14, 2012 at 01:48:44PM +0200, peter dalgaard wrote:
> >>
> >> Perhaps I'm confused, but how did either of you expect to fix a problem
> >> with fgetc by modifying fgets???
> >>
> >> --
> >> Peter Dalgaard, Professor
> >> Center for Statistics, Copenhagen Business School
> >> Solbjerg Plads 3, 2000 Frederiksberg, Denmark
> >> Phone: (+45)38153501
> >> Email: pd@cbs.dk Priv: pda...@gmail.com
> >
> > i'm not well versed in C. do know how to fix this? or, as i gather, this
> > is not something trivial and it has to be fixed uclibc upstream?
> >
>
> What Peter is saying that my fix was useless because I was talking about
> fgetc but fixed fgets: P
> So the fix attempt should have been
>
> static int fix_fgetc(FILE *stream) { return fgetc(stream); }
> #undef fgetc
> static int fgetc(FILE *stream) { return fix_fgetc(stream); }
>
> It should still be fixed upstream because I don't think POSIX allows fgetc to
> be a macro.
>
> Cheers,
> Simon
thanks, 's:^static:extern:' and it works, well sort of. now compilation
gets stuck (waited 3 hours) here:
make[6]: Entering directory
`/dev/shm/portage/dev-lang/R-2.15.1/work/R-2.15.1/src/library/tools/src'
mkdir -p -- ../../../../library/tools/libs
make[6]: Leaving directory
`/dev/shm/portage/dev-lang/R-2.15.1/work/R-2.15.1/src/library/tools/src'
make[5]: Leaving directory
`/dev/shm/portage/dev-lang/R-2.15.1/work/R-2.15.1/src/library/tools/src'
make[4]: Leaving directory
`/dev/shm/portage/dev-lang/R-2.15.1/work/R-2.15.1/src/library/tools'
it's apparently running:
/dev/shm/portage/dev-lang/R-2.15.1/work/R-2.15.1/bin/exec/R --vinalla --slave
with 100% cpu usage! i don't what it's doing to be honest :) any idea?
p.s.: i'm attaching the patch to this for the curious, hope it helps.
--- src/main/connections.c.orig 2012-08-14 11:47:47.105203896 -0400
+++ src/main/connections.c 2012-08-14 11:50:49.103928599 -0400
@@ -378,6 +378,16 @@
return res;
}
+#ifdef __UCLIBC__
+ extern int fix_fgetc(FILE *stream) {
+ return fgetc(stream);
+ }
+ #undef fgetc
+ extern int fgetc(FILE *stream) {
+ return fix_fgetc(stream);
+ }
+#endif
+
int dummy_fgetc(Rconnection con)
{
int c;
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
