Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site ucbcad.UUCP
Path: utzoo!linus!decvax!tektronix!ucbcad!notes
From: notes@ucbcad.UUCP
Newsgroups: net.cog-eng
Subject: man & zillions of flags... - (nf)
Message-ID: <319@ucbcad.UUCP>
Date: Thu, 6-Oct-83 02:11:17 EDT
Article-I.D.: ucbcad.319
Posted: Thu Oct  6 02:11:17 1983
Date-Received: Sun, 2-Oct-83 19:32:26 EDT
Sender: notes@ucbcad.UUCP
Organization: UC Berkeley CAD Group
Lines: 32

#N:ucbesvax:25800001:000:1371
ucbesvax!turner    Sep  5 22:30:00 1983

I have long been an advocate of standardizing UNIX man entries so as to
factilitate the following commands:

	1) makewhatflags - run by superrotorooters to update flags-files
	   entries; these could be kept in /usr/man/flags, or some such.

	2) flags <cmd> - list the flags (more'ing if need be) of the given
	   command <cmd>.  (Or, optionally, "options <cmd>".)
	
If the command given as argument didn't happen to exist, "flags" would
write to stderr: "No dice, dummy."  Or something equally appropriate. :-)

"flags <cmd> [+-][flag] . . . " would only detail (perhaps in a longer
format) the meaning of the flags given.

The main requirement is a willingness on the part of man-page writers or
maintainers to submit existing and future documentation to a common regime
with respect to command line options.  The actual format would depend
somewhat on the intelligence of the commands proposed.

If a reasonable de facto standard could be agreed upon, makewhatflags and
flags could themselves be the enforcers--anyone straying from the standard
will simply have to put up with users complaining bitterly that "flags"
doesn't work for their nifty syntax.  (This could equally be a complaint
against the "flags" command itself--there is a co-evolutionary balance
point to be struck somewhere.  Nothing new for UNIX, certainly.)

	Michael Turner (ucbvax!ucbesvax.turner)