Re: (no subject)

[Paul D. Smith]

%% Maciej Walezak <address@hidden> writes:

  mw> Thanks a lot. But the real problem is still not solved.

We can only give you answers based on the information you provide...

  mw> The rule like this
  mw> foo.h: bar.h
  mw> tells make that foo.h depends on bar.h while it is not true.
  mw> foo.h is never updated from bar.h and command-to-build-packet is
  mw> invoked again and again without any effect.

I assumed, given your rules, that the "command-to-build-packet" created
the .h file as well.

If a command takes as input various .c and .h files, and generates a .h
file, then that target .h file _does_ depend on the prerequisite .h
file.

  mw> But I must to write it this way because command-to-build creates
  mw> more files, including object files that do depend on bar.h.  I
  mw> want to avoid senseless attempts of compilation.

If one invocation of a command generated multiple files, then any file
that might cause that command to need to be re-run (even if it only
_logically_ impacts some subset of the targets) is still a prerequisite
of _all_ the targets.

If that's not what you want then you need to modify the command so that
it generates individual targets based on a more specific set of
prerequisites.

  mw> I surmise the benefit of your solution is that make does not check
  mw> for existence of all files as it knows it has created all of them
  mw> on occasion of creating first of them -am I right?

Not quite; make will still "check for the existence" of the files.  But,
yes, this method causes make to understand that a single invocation of
the command script may update all the targets listed.

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