I would like to work on the domain casting problem. I have spent
sometime in order to understand how this whole domain handling works
when it comes to casting and I think I understand why this cannot be
fixed in isolation as Tom has described in:
http://archives.postgresql.org/pgsql-hackers/2006-05/msg00190.php
Perhaps I am way off but I am starting to think we need to handle
domains in more organized and consistent way in order to avoid bugs like
this.
First I would like to know how PG's code looked like without the
domains. I went searching in the release notes and I found the following
regarding the domains:
7.3 : Add domain support (Rod)
7.3.3 : Fix planner's selectivity estimation functions to handle domains
properly
7.4 : Add check constraints for domains (Rod): Improve automatic type casting for domains (Rod, Tom)
7.4.12 : Properly check DOMAIN constraints for UNKNOWN parameters in
prepared statements (Neil)
8.0.1 : Make ALTER TABLE ADD COLUMN enforce domain constraints in all
cases
8.0.7 : Properly check DOMAIN constraints for UNKNOWN parameters in
prepared statements (Neil)
8.1.4 : Properly check DOMAIN constraints for UNKNOWN parameters in
prepared statements (Neil)
I need to go back and see where and how the domains are handled in a
global sense. Then I hope I can gather enough information to be able to
submit a coherent proposal.
If you have any thoughts you would like to share about this, please let
me know.
Regards,
Gevik.
