Hello all, the current process of database related changes could be in my opionion somewhat improved. If some changes are made in one module that needs a database schema update, these adoptions must be made in three different scripts for all databases. Sometimes one db (dbtext) is forgotten, or errors get introduced. We've seens this quite a few times in the past. There also exist not a real database schema, one must gather the informations from the wiki and the setup scripts.
I think it would be better to generate the create table calls from a common specification, this is also the way one well know competitor handle this. ;-) A first step would be to isolate the SQL snippets from the bash scripts, and provide one file for each database (part). This snippets could easily used from the remaining script logic, no functionality should gets lost. The second step would be to generate the sql from an abstract definition (a custom format, or XML with XSL scripts), the old architecture will be remain and would be still used for database creation. If the new process produce the same result as the old files could be deleted in the third step. A good place for the schema would be the db directory, the sql snippets could be saved in scripts/sql/$DatabaseName I would like to hear your opionion on this, as this is a central part of OpenSER. If nobody objects, i plan to start to work next week on this. Best regards, Henning _______________________________________________ Devel mailing list Devel@openser.org http://openser.org/cgi-bin/mailman/listinfo/devel
