Re: Bonobo freeze

[Dan Mueth <dan eazel com>]

On Thu, 25 Jan 2001, Michael Meeks wrote:

> On 24 Jan 2001, Maciej Stachowiak wrote:
> > Yes, we need a freeze commitment for all of Bonobo for the 1.4
> > release, as incompatible changes will not be possible for 1.4.x.   
> 
>         I anticipate that we could fully freeze the API, and commit to
> binary and source compatibility for all of bonobo by 1st Feb. However
> clearly Miguel has the final say and is currently in Switzerland.
> Furthermore, we will both be at LWE at this time, so in reality I propose
> releasing a final API reference version on the 5th or so.
> 
>         I also anticipate that there will not be a CVS branch at this
> stage, since there are plenty of things remaining to be done, particularly
> lots of in-line documentation and various misc. cleanups.

We don't need the libraries to branch/freeze before Feb 1, or possibly Feb
5.  However, after the library branch date (probably not past Feb 5, since
we are planning on GNOME 1.4test1 on Feb 15 and 10 days is not a long time
to update and test), all libraries must be branched to <library>-1-4
versions.  In theory one could argue that if a library were under so
little development that nobody would possibly want to modify it in the
next few months except to fix a bug, then it would not need to be
branched.  But even in this case I think it is good process and standard
practice to make the branch unless there was a previously release stabled
version which can be used.

For many libraries, especially Bonobo which is under heavy development and
very important to the GNOME platform, it is a very bad idea to not branch
and instead completely freeze the module in CVS HEAD except for bug fixes
for the next few months, as this will just push back GNOME 2.0. I would
much rather see these modules branched and have the HEAD versions actively
developed so we can add documentation, new features and API, code
cleanups, optimization, etc. that we will need for GNOME 2.0 and perhaps
even later 1.x point releases.

Dan