drapeko

One lite practicable idea for organizing version control tree for database development  is presented below:

The example of the svn tree is presented below:

/svn
  /features
    /user
	  /enable
	    create_user.sql
		create_user_permissions.sql
		pkg_user.pkb
		pkg_user.pks
	  /disable
	    drop_user.sql
	    drop_user_permissions.sql
	/comments
	 ...
  /scripts
	create_user_all.php
    drop_user_all.php
	recreate_database.php

The root directory consists of “features” and “scripts” dirs:

• Features – Logical functional units; a list of features we have (for example, a user or comments)
  • Each feature can be divided into “enable” and “disable” SQL scripts
    • Each enable/disable dir consists of the list of SQL scripts

• Scripts – List of scripts (php/perl/bash etc) that aggregate the features files into one sql file. The output of the script is the SQL file that contains all the SQL statements you need to enable/disable the feature.

DB developers work result is a list of SQL and pl/bash scripts for different variations (e.g. creates for empty db, patches et.