Path: utzoo!mnetor!uunet!ndsuvax!ncoverby
From: ncoverby@ndsuvax.UUCP (Glen Overby)
Newsgroups: comp.os.minix
Subject: Re: SVC - Revision control for Minix
Message-ID: <667@ndsuvax.UUCP>
Date: 7 Feb 88 04:09:24 GMT
References: <1660@ea.ecn.purdue.edu> <285@ditmela.oz>
Reply-To: ncoverby@ndsuvax.UUCP (Glen Overby)
Organization: Silo Tech  Fargo, ND
Lines: 43

In article <285@ditmela.oz> worsley@ditmela.oz (Andrew Worsley) writes:
>in article <1660@ea.ecn.purdue.edu>, housel@ea.ecn.purdue.edu (Peter S. Housel) says:
>
>  If someone could volunteer to maintain an RCS tree of all the major versions
>of MINIX it would help the job of coordinating the work done on it. Otherwise
>I fear that a lot of effort will be lost as it will not be combinale with
>other changes. I am not volunteering, at least not until I start using MINIX.

I have tried doing exactly this for local use, with some success.  The
major problem is the time involved.  And, as you mention, combinability
of changes (figuring out what changes will co-exist).  It would help
greatly if everybody had a system like RCS set up with identical
revision trees.  Then new patches could be merged into the tree in their
proper place, and co-existing changes would be documented by where in
the tree the changes fell. 

Also, how would changes from this be distributed? Re-posting to Usenet
is rather wasteful, and there is no Minix moderated sources group.  And
who is to determine what goes "in" and what stays "out"? This means
somebody will have to determine things like quality of the work (bug
free), documentation, etc.  We could go the route of the two
comp.sources.{unix,misc}, where the really good stuff (documented,
passes lint, etc) goes one way, and everything else goes the other way. 
To date, it's been everybody on their own determining what changes work
by trying them out; the only real unifying force has been Andy
Tannebaum's "accepting" people's changes.  Andy has already said he
doesn't have the time required to be a moderator of sorts. 

I think the way to really make something like RCS-style maintenance
benefit us as a whole is to have a standard "tree" (revision) format
established, and place $Header lines in every file.  Then convice people
to USE it. 

I'm willing to post the non-minix-source part of my RCS files with the
header info (stuff like descriptions and "$Header"s if I can figure out
a decent way to do this (lateley, I've been moving the $Header lines to
the END of the files because they mess me up when applying new patches,
but they're a lot nicer at the beginning)).  To give credit where credit
is due, much of this "grunt work" was done by Freeman Pascal at NDSU,
not me (I just badgered him into doing it). 

I'd also like to mention that without the 'idx' utility from
comp.sources.unix I'd surely go crazy working with Minix's internals!