[tomboy-list] Question: Do you try out unstable releases and then go back to stable?

[Sanford Armstrong]

(Getting Pete's thoughts on the list...my response is below)

On 4/16/07, Pete Goodall <pete at yellowhouse.org> wrote:
> On 4/16/07, Sanford Armstrong <sanfordarmstrong at gmail.com> wrote:
> > Hey everyone,
> >
> > I'm trying to get a feel for how much backward compatibility we need
> > in our notes and various related metadata.
> >
> > Have you ever tried out an unstable release (or latest svn, etc), and
> > then for some reason decided to return to using the older stable
> > release?
> >
> > Similarly, have you ever tried out a new stable series (0.6.x), and
> > then for some reason decided to return to using an older stable series
> > (like 0.4.x)?
> >
> > Do you think either behavior should be supported in Tomboy?
> >
>
> Personally I had major issues when I was going back and forth between
> unstable and stable (0.5.x and 0.4.x).  I would say if you are testing
> the unstable version, you should have backups of your critical data
> from both versions.

Yes, introducing a new feature to the note XML like bulleted lists
makes it difficult to maintain backward compatibility, but in this
case were eventually able to fix it.

> Backward compatibility can be tough.  I had all of my work notes in
> Tomboy, so if I upgrade to the next major version, I want to still be
> able to use my old notes.  However, if I create new notes in the
> unstable version it may not be reasonable for me to expect them to
> work in old Tomboy because I may be (and was) using new functionality
> that cannot be understood by the older version of Tomboy.

And also, we may change the way that some functionality works.  For
example, we are thinking of changing the way Tomboy knows if a note
should be "pinned" in the recent notes menu, and doing it in such a
way that it impacts backward compatibility.  That's actually what
inspired me to start this thread.

> Maybe there should be a policy of how far back you will support
> functionality.  For instance, any new functionality much be
> backportable one major revision.  Is that unreasonable?

Having a formal policy is a very good idea.  Going back one major
revision sounds pretty reasonable.

Thanks for your input,
Sandy