Re: Version numbering for release DECISION!

[Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>]

:Well, in FreeBSD-land, any later releases made from RELENG_5_3 would 
:indeed be called 5.3.1-RELEASE, 5.3.2-RELEASE etc.  I'm not sure what
:the system reports itself as if you cvsup to RELENG_5_3 between
:releases.  If I understand you right, every time a commit is made to
:the -RELEASE branch, the sub-version number gets bumped up.  I'm not 
:sure that's desirable.  A version number bump should be reserved for
:an "official" bugfix release, with public announcement etc.
:
:Rahul

    The fact that the commit is made on the branch itself is official enough
    from my point of view.  The idea is not to support parallel development.
    The last thing I want to do is to have multiple 'official' sub-releases
    within a branch.

    The ONLY things going into a branch would be bug fixes.

    This does mean that the sub-version could get fairly large, but who
    cares?  Remember, we aren't FreeBSD.  Our numbering scheme does not
    have to match.

    In fact, the more I think about it the more I believe that not having
    'official' sub-releases (like 1.2.1, 1.2.2) is a good thing.   I want
    our developers to *always* be focused on the HEAD of the tree.  I do
    not want our developers to have to do major release engineering on a
    branch.

					-Matt
					Matthew Dillon 
					<dillon@xxxxxxxxxxxxx>