[net.unix] kim walden & make

rad@MITRE-BEDFORD.ARPA (07/13/84)

  I thought those considering the problem of a "makefile maker" might
be interested to know that the Cambridge University Modula2 system for
BSD4.1 includes the "mm2m" command, which automatically generates
makefiles for Modula2 programs.  Of course, the problem is much simpler
for Modula2 than for C, where IMPORT statements in the code make it
clear where routines are located.

  The Cambridge Modula-2 package has very nice utilities; too bad the
code generated by the compiler is, as they put it, "far from optimal".

Dick Dramstad
rad@mitre-bedford

DBrown@HI-MULTICS.ARPA (07/13/84)

  The article actually discuses the problems of making correct makefiles
in the presence of .h files including other .h files, .y files including
.h files and so forth.
  It points out that there are several classical errors made by humans
writing makefiles, such as claiming that foo.c depends on bar.h, when it
is really foo.o which depends on both foo.c and bar.h.
  The extraction of #includes is the "easy" part (albeit very language
and system dependant), and the program referred to is the one which uses
the include lists to build dependancy graphs.
  --dave