Re: _AC_PROG_F77_V (in fortran.m4) on AIX 5

[Akim Demaille]

Hi Steven,

Could you please review this patch?  Thanks in advance!

 > Hi, 

 > For xlf on AIX, the verbose flag -V is, I think, preferable
 > to -v.  Unf. the _AC_PROG_F77_V macro currently tests for flags
 > in the opposite order and settles for -v.  The unfortunate side-effect
 > is that AC_F77_LIBRARY_LDFLAGS can't extract as much information as it 
 > should to correctly link F77 and C++ code using xlC in some cases
 > (namely mine).

 > Here's an excerpt from the xlf manpage
 >       -V        Displays verbose information on the compiler's progress in
 >                  shell-executable format.
 >       -v        Displays verbose information on the compiler's progress.
 

 > Here's a sample to illustrate the difference in output between the two
 > verbosity flags:

 > % echo '        END' > test.f
 > % xlf -v test.f
 >      /usr/bin/xlf_r -F:xlf_r -v test.f
 >      exec: 
 > /usr/lpp/xlf/bin/xlfentry(/usr/lpp/xlf/bin/xlfentry,test.f,/usr/tmp/F8eTnhib,/usr/tmp/F8eTnhibF.lst,xlfsmsg.cat,xlfmsg.cat,test.f,32,THREADED,NOZEROSIZE,SAVE,ALIAS(intptr),POSITION(appendold),XLF90(noautodealloc,nosignedzero),XLF77(intarg,intxor,persistent,noleadzero,gedit77,noblankpad,oldboz,softeof),NOLM,FULLPATH,WSTREAMS(/usr/tmp/F8eTnhiah1,/usr/tmp/F8eTnhiab1,/usr/tmp/F8eTnhias1),DEFMSG(/usr/lpp/xlf/bin/default_msg),-I/usr/lpp/xlf/include_32,NULL)
 >  
 >      ** _main   === End of Compilation 1 ===
 >      1501-510  Compilation successful for file test.f.
 >      exec: 
 > /bin/ld(ld,-b32,/lib/crt0_r.o,-bh:4,-bpT:0x10000000,-bpD:0x20000000,/usr/tmp/F8eTnhib0,-lxlf90,-lxlopt,-lxlf,-lxlomp_ser,-lpthreads,-lm,-lc,NULL)
 >  |
 >      exec: 
 > /bin/awk(/bin/awk,BEGIN{a=0;b=0;s=""};{if(($0!~/address@hidden/)&&($0!~/.*-34[45].*loadmap.*/||b==1||a==0)){b=1;print
 >  $0;}else{a=1;if((s=="")&& $0 ~ /.*-34[45].*loadmap.*/){s=$0}}};END{if(b==1 
 > &&  s!=""){print s}},NULL) 
 >      unlink: test.o

 > % xlf -V test.f 
 >      /usr/lpp/xlf/bin/xlfentry test.f /usr/tmp/F8a2b8ab 
 > /usr/tmp/F8a2b8abF.lst xlfsmsg.cat xlfmsg.cat test.f 32 THREADED NOZEROSIZE 
 > SAVE "ALIAS(intptr)" "POSITION(appendold)" 
 > "XLF90(noautodealloc,nosignedzero)" 
 > "XLF77(intarg,intxor,persistent,noleadzero,gedit77,noblankpad,oldboz,softeof)"
 >  NOLM FULLPATH 
 > "WSTREAMS(/usr/tmp/F8a2b8aah1,/usr/tmp/F8a2b8aab1,/usr/tmp/F8a2b8aas1)" 
 > "DEFMSG(/usr/lpp/xlf/bin/default_msg)" -I/usr/lpp/xlf/include_32 
 >      ** _main   === End of Compilation 1 ===
 >      1501-510  Compilation successful for file test.f.
 >      /bin/ld -b32 /lib/crt0_r.o -bh:4 -bpT:0x10000000 -bpD:0x20000000 
 > /usr/tmp/F8a2b8ab0 -lxlf90 -lxlopt -lxlf -lxlomp_ser -lpthreads -lm -lc  |
 >      /bin/awk 
 > 'BEGIN{a=0;b=0;s=""};{if(($0!~/address@hidden/)&&($0!~/.*-34[45].*loadmap.*/||b==1||a==0)){b=1;print
 >  $0;}else{a=1;if((s=="")&& $0 ~ /.*-34[45].*loadmap.*/){s=$0}}};END{if(b==1 
 > && s!=""){print s}}' 
 >      rm test.o

 > Note that in -v, the library arguments are comma separated, not space 
 > separated.
 > A diff from current CVS is attached with the oneliner fix.  

 > Questions/comments are welcome, but I'm not subscribed to the 
 > autoconf-patches
 > mailing list.  Please let me know if/when the change makes it into an 
 > autoconf
 > release.

 > Regards, 

 > Gary

 > ------------------------------------------------------------------------
 >   Gary Kumfert, Ph.D.                                    <address@hidden>
 >   Center for Applied Scientific Computing          phone: 925-424-2580
 >   Lawrence Livermore National Laboratory           fax:   925-424-2477
 >   P.O. Box 808,    L-365
 >   Livermore, CA 94551-0808


 > Index: lib/autoconf/fortran.m4
 > ===================================================================
 > RCS file: /cvsroot/autoconf/autoconf/lib/autoconf/fortran.m4,v
 > retrieving revision 1.179
 > diff -r1.179 fortran.m4
 > 567c567
 > < for ac_verb in -v -verbose --verbose -V -\#\#\#; do
 > ---
 >> for ac_verb in -V -v -verbose --verbose -\#\#\#; do