The following future developments are planned, depending on resources:
-
Generalised locking mechanism not based on Object Ids (OIDs).
-
This would be based on table name and Primary Key data. The table name
would be hashed to obtain another hashtable. The key fields would then be
turned into a string to be used as a hash. Obviously a key field separator
character is needed and this has to be stuffed if it appears in a key field
itself.
-
Tool for generating COBRA objects from RDBMS meta information.
-
Support for 2 phase commits.
-
Automated installation mechanism.
-
Support for object hiearchies and composition - perhaps using Joins.
-
Support for direct SQL execution.