Re: How to version data models

This has nothing to do with db theory, but...

JAL wrote:
> I am a new modeller, picking up work from another data modeller who
> works with the most current model from the data repository.
>
> There are no branches or versions from one release to the next. The
> modeller is always working with the physical model which is a copy of
> the last committed production model. The revison number is the only
> way to select the production data model if changes are required while
> in production.
>
> For example:
> Current Production = 1.0 (stored in repository)
> Next Scheduled Release = 1.1 (not in repository, not checked in until
> goes live in production)
> Future Release is 1.2 = (not in repository, stored locally, will
> replace in repository when goes live)
>

This next sentence describes a broken process: your model should be controlled before it reaches production.
> After each release to production, the data model is committed to a
> repository with a new revision number.

This seems strange:
 > The previous production copy,
> which becomes the newest release, is moved from a local working
> directory to the repository directory and committed thereby creating
> the new revision number. There is no way to tell which release for the
> data model.
>
> There must be a solution that allows changes to the physical production
> data model for the next release without affecting the current
> production model in the repository.
There is. It seems that either your version control software is broken or you're not using it correctly.

>
> If 1.0 is committed, changes are made to 1.1 and this copy is
> committed, there may be a change required to 1.0 that must be then
> synced one way to 1.1.
This occurs so frequently that version control systems have features that facilitate this.

Do research.

>
> Appreciate any input.
>
Received on Fri Jan 12 2007 - 10:27:30 CET