annotate admin/notes/cpp @ 83171:09bbf2fc80da

Merged in changes from CVS trunk. Patches applied: * [e-mail]--gnu-2004/emacs--cvs-trunk--0--patch-439 Update from CVS * [e-mail]--gnu-2004/emacs--cvs-trunk--0--patch-440 Update from CVS * [e-mail]--gnu-2004/emacs--cvs-trunk--0--patch-441 Update from CVS * [e-mail]--gnu-2004/emacs--cvs-trunk--0--patch-442 Update from CVS * [e-mail]--gnu-2004/emacs--cvs-trunk--0--patch-443 Update from CVS * [e-mail]--gnu-2004/emacs--cvs-trunk--0--patch-444 Update from CVS * [e-mail]--gnu-2004/emacs--cvs-trunk--0--patch-445 Tweak permissions * [e-mail]--gnu-2004/emacs--cvs-trunk--0--patch-446 Update from CVS git-archimport-id: [e-mail]--2004/emacs--multi-tty--0--patch-211

ttn 2003-04-09

we use a C preprocesor not only in the normal compilation of .c files
into object files, but also for creating

	src/Makefile
	lib-src/Makefile

(delimited by comment "start of cpp stuff").  some cpp implementations
insert whitespace in between tokens, which explains the "$(dot)$(dot)"
and "UNEXEC_ALIAS" workarounds for unixoid systems and the bundled
tradcpp/ for vms.  [NOTE: tradcpp/ not yet checked in.]

during makefile creation, the preprocessor symbol NOT_C_CODE is defined.
this should not be defined for normal .c file compilation.

there has been discussion on eliminating this use of cpp and relying
solely on autoconf processing.  rms says to leave it be.

;;; arch-tag: d654291e-9fc8-41b7-ab0c-d3cde842a8e0