Bugzilla Bug 231607

>>> Completed installing grace-5.1.21-r1 into /var/tmp/portage/sci-visualization/grace-5.1.21-r1/image/

strip: x86_64-pc-linux-gnu-strip --strip-unneeded -R .comment
   usr/bin/convcal
   usr/bin/xmgrace
rsync: link_stat
"/var/tmp/portage/sci-visualization/grace-5.1.21-r1/work/grace-5.1.21/src/y.tab.c"
failed: No such file or directory (2)
rsync error: some files could not be transferred (code 23) at main.c(1040)
[sender=3.0.3]
   usr/bin/grconvert
rsync: link_stat
"/var/tmp/portage/sci-visualization/grace-5.1.21-r1/work/grace-5.1.21/src/y.tab.c"
failed: No such file or directory (2)
rsync error: some files could not be transferred (code 23) at main.c(1040)
[sender=3.0.3]
   usr/lib64/libgrace_np.a
ecompressdir: bzip2 -9 /usr/share/man

 * QA Notice: Package has poor programming practices which may compile
 *            fine but exhibit random runtime failures.
 * plotone.c:134: warning: implicit declaration of function
���mkstemp���
plotone.c:139: warning: implicit declaration of function
���fdopen���
files.c:771: warning: implicit declaration of function ���popen���
files.c:788: warning: implicit declaration of function ���pclose���
main.c:156: warning: implicit declaration of function ���srand48���
fit.c:151: warning: implicit declaration of function ���finite���
pars.yacc:743: warning: implicit declaration of function
���drand48���
iofilters.c:163: warning: implicit declaration of function
���popen���
mathstuff.c:114: warning: implicit declaration of function ���jn���
mathstuff.c:138: warning: implicit declaration of function ���yn���
mathstuff.c:162: warning: implicit declaration of function
���lrand48���
mathstuff.c:185: warning: implicit declaration of function
���drand48���
editpwin.c:784: warning: implicit declaration of function
���fdopen���
editpwin.c:784: warning: implicit declaration of function
���mkstemp���
xmgrace.c:549: warning: implicit declaration of function
���gethostname���


 * QA Notice: Package has poor programming practices which may compile
 *            but will almost certainly crash on 64bit architectures.
 * Function `fdopen' implicitly converted to pointer at plotone.c:139
Function `popen' implicitly converted to pointer at files.c:771
Function `popen' implicitly converted to pointer at iofilters.c:163
Function `mkstemp' implicitly converted to pointer at editpwin.c:784

 * 
 * ERROR: sci-visualization/grace-5.1.21-r1 failed.
 * Call stack:
 *       misc-functions.sh, line 653:  Called install_qa_check
 *       misc-functions.sh, line 349:  Called die
 * The specific snippet of code:
 *                             
alpha*|ia64*|powerpc64*|mips64*|sparc64*|x86_64*) die "this code is not 64bit
clean";;
 *  The die message:
 *   this code is not 64bit clean
 * 
 * If you need support, post the topmost build error, and the call stack if
relevant.
 * A complete build log is located at
'/var/tmp/portage/sci-visualization/grace-5.1.21-r1/temp/build.log'.
 * The ebuild environment file is located at
'/var/tmp/portage/sci-visualization/grace-5.1.21-r1/temp/environment'.
 * 
!!! install_qa_check failed; exiting.

Created an attachment (id=160210) [edit]
emerge --info

Probably some of the files mentioned in the QAs' need to be patched
with
#include <stdlib.h> (for mkstemp)
#include <stdio.h> (for popen)

(In reply to comment #2)
> Probably some of the files mentioned in the QAs' need to be patched
> with
> #include <stdlib.h> (for mkstemp)
> #include <stdio.h> (for popen)
> 

Thanks, I'll have a look!

Best,
Markus

All right, the .c files all have the proper includes as far as
I can tell. Them main issue seems to be our grace-5.1.21-stdc99.patch
in conjunction with gcc not yet fully supporting the std=c99 flag (see [1]).
Without the -std=c99 everything works fine for me using glibc-2.8.
The patch was introduced for bug #217971. Hence, we could either drop it
or find a different workaround for the glibc-2.7 issue. Unfortunately,
I don't have a box with glibc-2.7 around at the moment.


Best,
Markus




[1] http://gcc.gnu.org/gcc-4.3/c99status.html

grace compiles fine on amd and x86 without the 
grace-5.1.21-stdc99.patch using glibc-2.8*. Hence,
I suggest that we remove this patch from the ebuild.
We have to watch out though in case glibc-2.7 goes
stable since the issue that the patch was supposed to
fix may crop up again. 

@bicatali: Does that sound ok to you or is there a
reason why we should definitely keep the patch?

Thanks,
Markus

> @bicatali: Does that sound ok to you or is there a
> reason why we should definitely keep the patch?

Thanks a lot Markus for tracking that down. I commented out the patch, and took
the opportunity for a version bump (without the patch too).