Path: utzoo!attcan!utgpu!watmath!iuvax!mailrus!sharkey!itivax!scs
From: scs@itivax.iti.org (Steve Simmons)
Newsgroups: news.software.b
Subject: Re: Patch dates or Patch Numbers
Message-ID: <2876@itivax.iti.org>
Date: 18 Aug 89 12:51:00 GMT
References: <1989Aug9.164003.20669@utzoo.uucp> <6717@dayton.UUCP> <1989Aug17.171000.23302@utzoo.uucp>
Organization: Industrial Technology Institute, Ann Arbor, MI.
Lines: 46

henry@utzoo.uucp (Henry Spencer) writes:

>In article <6717@dayton.UUCP> jad@dayton.UUCP (J. Deters) writes:
>>A correctly used patch system will need numbers.  Not want.  Need.
>>If you want to know where your program is in relation to the patch you have,
>>a system of Release.Revision.PatchLevel is almost mandatory...

>Assuming that you are talking about one of those pieces of, uh, software
>that is constantly changing, with bales of new features and swarms of
>new bugs regularly showing up.  My opinion of such mushware is very low
>indeed; Geoff's opinion is unprintable.

>. . . we see no fundamental reason
>why a well-written piece of software cannot stabilize fairly quickly and
>then remain essentially unaltered unless and until it is replaced by a
>complete rewrite on radically different principles.  This is the model
>we are working towards:  software, not mushware.  We may not achieve it,
>but we are determined to try.  (Indeed, we are not interested in trying 
>the alternative.)

Sorry for the long requote, but given the response I'm about to make it
seemed only fair.

Henry, your response in no way addresses the request.  Deters (and I,
in seperate mail) have both stated that a patch numbering system in
less ambiguous than your date-oriented system.  While I salute your
efforts towards quality software (and would even say you're succeeding)
your responses to this issue have (a) not addressed it and (b) verged
on . . . ah . . . emotional.  I realize you are not trying to be insulting,
but a response laden with terms like "mushware" and "unprintable" would
lead the casual reader to assume you are applying those terms to the
feature (patch numbering) being requested.

The main virtues of patch numbering are twofold:  First, they give an
absolutely unambiguous way of telling you how many patches must be
obtained.  By contract, patch dating can lead one into a cycle of
not being quite sure one has the previous patches (one gets the june 17
patch only to discover you now need the june 2 patch.  get that, only to
find you now need the april 1 patch.  i realize you don't put out a lot
of patches.  your posting/patching frequency is irrelevant to the quality
of number vs. date as a methodology).  Second, patch numbers are easier
to remember.  One simple number, rather than a date.
-- 
Steve Simmons		          scs@vax3.iti.org
Industrial Technology Institute     Ann Arbor, MI.
"Velveeta -- the Spam of Cheeses!" -- Uncle Bonsai