RFC: Spec versioning

Fred Drake fdrake at gmail.com
Thu Mar 24 14:49:07 EET 2005

On Thu, 24 Mar 2005 11:06:21 +0000, Mark McLoughlin <markmc at redhat.com> wrote:
>         (That seems perfectly natural to me, but I know others were totally
> insistant that 2.10 made no sense and we should have bumped to 3.0 -
> perhaps this is a locale/cultural thing?)

There are two primary camps:

- The "version numbers are a single real number" camp thinks 2.1 and
2.10 are the same thing.  This is perhaps more common in the Windows
word, though the lines are not clearly drawn along platform lines. 
For these people, version numbers are as much a marketing tool as
anything, and can (at best) describe a single sequence of releases.

- The "version numbers are a sequence of numbers" camp knows better. 
These are the people who actually expect to extract semantic
information (as you've described) from the number.  The numbers are
useful for programmers and sys admins (once you understand all the
various camps the software maintainers are in).  There is an
undestanding among this camp that knowing how a piece of software
derives from past versions is important, since it let's them know
which versions may introduce new features and which are strictly
maintenance releases.

Pretty clealy, I'm in the latter camp for software.  I'm less
convinced that this is an important model for specifications, but
would not object to its application.


Fred L. Drake, Jr.    <fdrake at gmail.com>

More information about the xdg mailing list