Fri, 25 Sep 1992 17:21:46 +0200
|
A lot of relatively minor problems have been discovered with 1.7d and
subsequently fixed; people who reported them got a fix directly from me,
but I did not issue formal 17D-xxxx fix shipments for them as there are
simply too many. All these problems are corrected by a change to the
large (350k) LSWPLIB MODULE, which it is not possible to service in the
same way as the REXX code:
1. Because it is a single file, sites which do not use source code for
maintaining LISTSERV cannot choose which of the fixes they want to
apply and which they want to ignore. If I were to distribute a MODULE
with fixes for only problem 1 and one with fixes for only problem 2,
there would be no way to have both problems fixed at the same time
without building your own customized MODULE from the source code. With
just two fixes I could make a third MODULE available, but with a dozen
fixes this is simply not possible.
2. Each fix is a 350k replacement MODULE file, since there is no decent
patching system for VM. In addition to the network traffic, it means a
lot of your disk space is wasted, since the fix installation procedure
keeps a copy of the old MODULE (for obvious reasons).
Because of this I just include all minor fixes into the major fix
shipments I generate. When doing that, I respect the rule that regular
fix shipments must never be updated unless a *really* severe problem is
discovered (like a security exposure). This means fix number 001o will
not contain minor fixes discovered after it was issued, but the next
major fix will have them. If you install the latest fix that replaces
LSWPLIB MODULE, you will have all the minor fixes written up to then. It
may seem unsatisfactory from the philosophical point of view, but the
only way to get finer granularity is by using source-based maintenance,
which most of you don't have time for.
As far as 1.7d is concerned, all the minor problems reported on the list,
except the incorrect header line for AFD of F=MAIL files, are fixed by
17D-002D rev 2 (the one which adds support for F=VMSDUMP - this is NOT a
new revision, there is no need to re-install anything if you have it
on!). This fix has been running on many servers without problem so far,
so it is safer to install it than a "custom fix" I might prepare just for
you and which never ran anywhere.
Eric
|
|
|