Upgrading Minor versions

[Glenn Steen]

On 03/08/07, Res <res at ausics.net> wrote:
> On Fri, 3 Aug 2007, Glenn Steen wrote:
>
> >> This is why I mentioned a couple of days ago about bringing it into line
> >> with most other software versioning,  major.minor.beta_number as in
> >> 4.63.b1, 4.63.b2 etc reaching the official release as 4.63 (as only 4.63)
> >> major. MailScanners installer does not understand at all 4.63.3-5 only
> >> 4.63.3, thereby over-writting. As Julian ignored my comments I dare say he
> >> has no intention of changing it, as no one else commented, it seems most
> >> are happy with it, and I also gather after years of observations 95% of
> >> people here use RPM base OS's so they may not be affected by the
> >> installers tarball shortfall.
> >>
> > In the RPM case the -<number> isn't really part of the packaged
> > software version number... it is more considered to be a package
> > revision number. Indeed, RPM (and other packagers) will take note of
> > that difference and act accordingly/safely.
> > So perhaps it is the "tarball install method" that need be amended,
> > more than the actual numbering scheme... Or one could make it a
> > documentation thing... Prominently (on the download page) warn to make
> > a cakcup of the /opt/MailScanner<.whatever> directory prior to
> > unpacking/installing the tarball.
>
> Yes, but wouldn't it make more sense to use the unified versioning method
> that everyone around the entire world is acustomed to?
>
That scheme was dreamed up in the linux kernel dev... and (perhaps...
It was not always so:-) suits them very well. And it has spilled over
on a lot of projects. But it is hardly the one true unified version
numbering method about.
And it's not sure to cure anything like this, where _packaging
versions_, not software versions, handling is the real issue. We/Jules
will just have to think of some more or less clever way of handling
it.

Cheers
-- 
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se