Subject: Re: cksum listing also?
To: None <>
From: None <>
List: current-users
Date: 05/18/1994 10:29:46

I write:

> > Could there be generated (as part of the weekly processing) a set
> > of files each containing a cksum listing of a major part of the NetBSD -
> > current source tree.

Any you respond:

> I see several problems with this:
> 	(1) if you generate the cksums weekly, then, on all the other
> 		days of the week, some files will look bogus.  i don't
> 		like this -- it defeats the purpose of doing the cksum
> 		to begin with.

Very true.  I chose weekly so that:

	o	They would match the source.tar files which I think are
		created each week.

	o	Would keep the load down (daily as you noted would kill the

To re-construct todays source tree both the most recent cksum listings and
the following few days `daily CVS update output' would be needed.

What about making the cksum outputs available weekly and attach to them a
README file containing something like:

	``These weekly cksum(1) listings are provided as is.  Trying
	  to maintain a NetBSD-current source tree using this information
	  is not supported.

	  It is assumed that people tracking the NetBSD-current source tree
	  are able to make IP level connections to the NetBSD-current archive
	  and can run sup(1).

	  It is also assumed that people wishing to obtain individual files
	  from the NetBSD-current source tree have some form of ftp access to
	  the archive site.

	  If you do not have access to sup(1) then these cksum listings may
	  be helpfull (but remember `they are not supported')

	  To check your NetBSD source tree against NetBSD-current you will
	  need both these cksum files _and_ any `daily CVS update output'
	  posted since the files were last generated.  Even then, because of
	  network delays, you may find that you are playing a `catch up game'
	  with the NetBSD-current sources.''

Re-phrase it (make it stronger) if you want, I think, however, it gets most of
the points across.

				regards Andrew

PS: Cksum's on the `daily CVS update output' would be nice but I should
    think that is pushing things too far :-)