On Thu, 5 Apr 2001, Kyle wrote:
> I've done some work on a solution to allow you to drop and rebuild your
> db from scratch relatively painlessly. It enables you to munge portions
> of the database, drop columns, undermine RI, clobber views, etc. and
> then put the whole thing back together clean, fresh and perfect.
> Pg_dump is focused on doing a drop/restore exactly. What is needed is a
> way to drop/modify/restore. That's kind of what my tool enables you to
> do (without the fear that you're going to lose some subtle detail of the
> schema in the process).
>
> It's a management tool that sits in between Postgres and your
> application. You store chunks of SQL in a relational format that
> describe your schema exactly. Then you can call those bits of SQL on
> demand to destroy or build parts of the schema selectively. Because its
> relational, you can store dependencies so the program knows which parts
> of the DB to rebuild when you clobber something.
>
> Right now, it is a handful of shell scripts and files, but we're using
> it with good success at ATI. I have created an outline of how to do it
> with a tcl/tk GUI and actually storing the SQL chunks in a Postgres
> database.
>
> I want to do some more development on it and then do a GPL release of
> it. Problem is, I don't seem to find the time to get it all done. Are
> you interested in collaborating on its development? Or do you know
> someone who is?
Post it along w/any quick notes about the architecture (about 20 minutes
of your time covering the concepts would probably save me two hours). I'd
be happy to look over it to see if I can help.
--
Joel Burton <jburton@scw.org>
Director of Information Systems, Support Center of Washington