Re: make3.81-beta4 mingw not failing on error?

[Paul D. Smith]

%% Eli Zaretskii <address@hidden> writes:

  >> $(DDD_FILES):
  >> @echo -- generating makefile: $@
  >> @set -e; rm -f $@;                                          \
  >> echo "$(DFILE_HDR)" > $@;                                   \
  >> $(CXX) -M $(CXXFLAGS) $< > address@hidden(PID);                      \
  >> sed 's,^.*\.o[ :]*,$(@:%.d=%.o) $@: ,g' < address@hidden(PID) >> $@;  \
  >> rm -f address@hidden(PID)

  ez> The command here is a single shell command, as evidenced by the
  ez> backslashes before each newline.  So Make doesn't regain control
  ez> until this whole multi-line command returns.

The "set -e", in a POSIX shell, turns on "exit on error" mode; in this
mode the shell will exit immediately when any command line exits with a
non-0 code.

So, on UNIX, this command would exit immediately if the $(CXX) program
exits with a non-zero code, as Matt expects.

  ez> I don't know how this worked on Cygwin for you, perhaps you used a
  ez> different shell, or perhaps the Cygwin Make interacts differently
  ez> with the shell.

I agree, though, that this must be an issue with the shell that make is
invoking: I can't see it being an issue with GNU make proper.  The only
relationship between this problem and make could be which shell is
invoked or how the shell is invoked.


Matt: can you remove the "@" prefix to this command ("set -e; ...") so
we can see what exactly is being invoked here?  Also, if you use -d or
just --debug=j you can get info on the return code the shell gives back
to make.

-- 
-------------------------------------------------------------------------------
 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