[cvsnt] Synchronize source code with database?

[keith d. zimmerman]

Siegfried Heintze wrote:
> I'm using perl and a Microsoft Access Database to implement a web site. (Not
> my choice -- but it pays the bills).
> 
> I just added some new columns to the 10MB database. Now I'm using CVSNT to
> store the perl code but not the Microsoft access database.
> 
> Should I add the Microsoft Access database to the repository? Since we don't
> have any customers yet, there is not much interesting in the database except
> the metadata. I really don't want to store a new copy of the Microsoft
> Access database every time I commit. If I add the database with the -kb
> option, will it store a new copy of the database every time I commit the
> sandbox (assuming the database has been modified -- which is extremely
> likely)?
> 
> I believe so.

yes, you are correct

> 
> This is a problem because if I ever change some field names or table names
> in the database and need to role back, I won't have an old copy of the
> database to match the code!
> 
> Well, this problem is not unique to Microsoft Access. What do other folks
> do?

We have a "blank" data copy which we only use when we add/remove fields. 
  This then is the one stored in cvs.  For testing and development work, 
we use a copy of the blank data.

-kz