[776] in Release_7.7_team
Re: Update questions
daemon@ATHENA.MIT.EDU (Richard Basch)
Thu Oct 31 08:08:31 1996
Date: Thu, 31 Oct 1996 08:08:16 -0500
To: Greg Hudson <ghudson@MIT.EDU>
Cc: release-team@MIT.EDU, epeisach@MIT.EDU
In-Reply-To: <199610310701.CAA28562@the-light-fantastic.MIT.EDU>
From: "Richard Basch" <basch@lehman.com>
On Thu, 31-October-1996, "Greg Hudson" wrote to "release-team@MIT.EDU, epeisach@MIT.EDU, probe@MIT.EDU" saying:
> Some issues about update which people might have answers to:
>
> * do_update has support for an environment variable ROOT
> specifying an alternate root for the local filesystem. Is
> this actually used by anyone? Given that in the general
> case, part of the update has to happen after a reboot, I
> don't think this functionality can really be made to work.
Ezra and I attempted to make the update and installation scripts the
same, at one point. To do that, we had to introduce the concept of ROOT
to specify a relative directory that the update scripts should work
with. The rationale for trying to consolidate update and installation
is that there had been several releases where the workstation
functionality differed simply because of differences between the install
and the update; sometimes, special actions taken by the update scripts were
not mimicked by the installation script, and if they were one, this
would no longer be the case.
> * Craig introduced code for the SGI to edit /etc/services,
> adding the stuff from /usr/athena/lib/update/services.add,
> instead of treating /etc/services like all the other
> configuration files. Does anyone know what the motivation
> was?
> * The part of do_update which tracks /srvd is inside a
> conditional on FULLCOPY being true. FULLCOPY is always set
> to true in update_ws, so it should always be true in
> do_update. This change was added in 7.0A with no meaningful
> comment from probe in the RCS log; is there any reason I
> shouldn't rip it out?
FULLCOPY is a historical artifact. Basically, at the time, the
filesystem hierarchy had just been revamped entirely (/urvd was
eliminated, and the complex link hierarchy removed). At the time, there
was also a mandate by management that we not necessarily re-sync the
workstation with /srvd for the various incremental updates (the new
layering mkserv just came out then). FULLCOPY allowed the incremental
update scripts to set the variable to avoid the re-track of the
workstation software for simple incremental releases. It was rarely
used, especially since we demonstrated that workstations could be
customized in arbitrary ways with mkserv and the customizations would be
preserved across updates. I may have used it for a couple incremental
releases that only had to do one or two tweaks. Anyway, as I said, it
allowed for fast updates and it placated management at the time.
--
Richard Basch
Sr. Developer/Analyst, DSO URL: http://web.mit.edu/basch/www/home.html
Lehman Brothers, Inc. Email: basch@lehman.com, basch@mit.edu
101 Hudson St., 38th Floor Fax: +1-201-524-5828
Jersey City, NJ 07302-3988 Voice: +1-201-524-5049
