Re: diffutils 2.8.4 bug when make -j is used

[Paul D. Smith]

%% Chris Bainbridge <address@hidden> writes:

  cb> Hi, I've attached the output of a failed run with
  cb> --debug=jobs. Hope this is some help.

Hm.  This does make it look like a bug in make, that's for sure:

  cb> Putting child 0x08079e40 (paths.h) PID 13898 on the chain.
    ...

  cb> Obtained token for child 0x0807cc38 (diff.o).
  cb> source='diff.c' object='diff.o' libtool=no \
  cb> depfile='.deps/diff.Po' tmpdepfile='.deps/diff.TPo' \
  cb> depmode=gcc3 /bin/sh ../config/depcomp \
  cb> gcc -DHAVE_CONFIG_H -I. -I. -I..  -I../lib -I../lib   -g -O2 -c `test -f 
'diff.c' || echo './'`diff.c
  cb> Putting child 0x0807cc38 (diff.o) PID 13904 on the chain.
  cb> Live child 0x0807cc38 (diff.o) PID 13904 
  cb> Live child 0x0807c1f8 (context.o) PID 13902 
  cb> Live child 0x0807aa88 (analyze.o) PID 13900 
  cb> Live child 0x08079e40 (paths.h) PID 13898 

Right here make invokes the diff.o command even though paths.h, which is
a prerequisite according to other extracts posted here, has not yet
completed.

Can you run this again, this time with full debugging (-d)?  This will
generate voluminous output, but I only need to see the output for the
"src" directory.

Another thing that might be interesting is to run make in the src
subdirectory with "-p" to see what the makefile database looks like.
Maybe the prerequisites list we think we have is not what we really
have.

Thanks.

-- 
-------------------------------------------------------------------------------
 Paul D. Smith <address@hidden>          Find some GNU make tips at:
 http://www.gnu.org                      http://make.paulandlesley.org
 "Please remain calm...I may be mad, but I am a professional." --Mad Scientist