home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Nebula 2
/
Nebula Two.iso
/
SourceCode
/
MiscKit1.7.1
/
MiscKitArchive.mbox
/
mbox
/
000088_kane@sonata.cc.purdue.edu_Mon Nov 1 10:21 MST 1993.msg
< prev
next >
Wrap
Internet Message Format
|
1994-10-30
|
3KB
Received: from sonata.cc.purdue.edu by alaska.et.byu.edu; Mon, 1 Nov 93 10:21:19 -0700
Return-Path: <kane@sonata.cc.purdue.edu>
Received: from concerto.cc.purdue.edu by sonata.cc.purdue.edu (5.61/Purdue_CC)
id AA20957; Mon, 1 Nov 93 12:19:20 -0500
From: kane@sonata.cc.purdue.edu
Message-Id: <9311011719.AA20957@sonata.cc.purdue.edu>
Received: by concerto.cc.purdue.edu (NX5.67d/NX3.0X)
id AA07442; Mon, 1 Nov 93 12:19:18 -0500
Date: Mon, 1 Nov 93 12:19:18 -0500
Received: by NeXT.Mailer (1.95)
Received: by NeXT Mailer (1.95)
To: yackd@alaska.et.byu.edu (Don Yacktman)
Subject: version numbering (Re: release of MiscKit 1.0 miscellanea)
Status: RO
> You know, one thing we never decided is how to choose
> whether to bump up a major or minor revision number...hmmm...
> it might be good to state some sort of policy in the charter,
> don't you think? What might you suggest? If I do a major number
> each time I add objects/resources, it'll go up too quickly, but
> if I don't then how do I distinguish between bug fix releases
> and new functionality? Maybe something like x.y.z for the number
> where z is only bug fixes, y adds functionality (new resources)
> and x is >10 new resources or something like that. Opinions?
The major.minor.maintenance form is quite common. My opinion?
Use this form, with bug fixes getting maintenance increment,
minor version getting incremented on new object additions with
maintenance reset to 1. I don't agree with your suggestion
about the "x" value: this may sound philosophical or metaphysical
("Use the force, Luke") but I think that the major number should
stay at 1 until such a time as it should be incremented to 2; I
think you (or the group) will know when it is time for that.
Another option (one which you reject, but I think would be reasonable)
is to use an y.z system, where y and z are as you indicate above.
A version number of 114.17 doesn't seem too ridiculous to me (the
info panel for Mail.app claims the version I am using to be v95).
Have you been receiving submissions at a rate which would indicate
that "y" would be over 100 in two years (assuming this enterprise
lasts that long)? (This is a rhetorical question, of course.)
A final thought on this: it would also be reasonable to "save up"
a few bug fixes or new submissions to put in a new release, rather
than a new release every week with one or two minor bugs fixes
(at the extreme). A variant on the numbering scheme above arises
out of this: the version number could be the date of release.
For instance, if you were to decide to do a release-a-month, the
version number could be 1993.11, 1993.12, 1994.1, .... or quarterly
might be 1993.4, 1994.1, 1994.2, ... Intermediate releases (for
serious bug fixes for example) could be designated with letters,
1993.4a, 1993.4b, ... The *.sources.* newsgroups work on a system
like this (but, issue/number).
There are many varieties of version numbering schemes. The important
thing with any version numbering system is that it be clearly
monotonically increasing. The rest is just convention or personal
preference. People are generally intelligent creatures who can
adapt and figure out what is going on.
Christopher