Path: utzoo!mnetor!uunet!husc6!rutgers!ucla-cs!zen!ucbvax!ulysses!gsf
From: gsf@ulysses.homer.nj.att.com (Glenn Fowler[eww])
Newsgroups: comp.unix.questions
Subject: Re: Parallelism of Make
Message-ID: <3325@ulysses.homer.nj.att.com>
Date: 4 Jan 88 19:22:30 GMT
References: <206@ksr.UUCP>
Organization: AT&T Bell Laboratories, Murray Hill
Lines: 30
Summary: nmake conversion tools

In article <206@ksr.UUCP>, benson@ksr.UUCP (Benson Margulies) writes:
> I have seen two advanced replacements for make -- mk and nmake, both
> from AT&T. Both are incompatible with 4.3 and SysVr3 make. Both offer
> partial conversion tools, in the forward direction only. 

nmake is distributed with conversion tools in both directions, but
neither provides 100% accuracy:

	mkcv:	convert old makefile to nmake makefile
	mkold:	generate old makefile given nmake makefile

The output from both programs definitely looks mechanically generated.
Some constructs do not easily convert.  Most problems involve the
property that old make actions are executed one line at a time by
separate shells whereas each nmake action is executed as a block by
a single shell.  mk actions are also executed as blocks, although
variable (macro) expansions are left up to the shell.

For parallel execution of actions any variant of make must provide
hooks for mutual exclusion; there is no standard way to specify this.

For the group I work with it is almost always a win to convert
foreign makefiles to nmake before hacking.  Our 3b2 sVr3 makefiles
went from ~20 files, ~5000 lines down to 1 file, 280 lines.

    Glenn Fowler                    AT&T Bell Laboratories, Murray Hill
    {ihnp4,allegra}!ulysses!gsf                          (201)-582-2195
-- 
    Glenn Fowler                    AT&T Bell Laboratories, Murray Hill
    {ihnp4,allegra}!ulysses!gsf                          (201)-582-2195