pg_dump doesn't handle CREATE FUNCTIONs well

Rainer Mager (rmager(at)vgkk(dot)com) reports a bug with a severity of 2
The lower the number the more severe it is.

Short Description
pg_dump doesn't handle CREATE FUNCTIONs well

Long Description
We have a table (table_A) that has a constraint on it that uses a custom function (func_B), which references a different table (table_C). When we do a pg_dump all of the table and function definitions are created correctly but in the wrong order.

The dump has the items in this order: table_C, table_A, func_B
The problem is that the creation of table_A requires the existence of func_B. Therefore the dump should have dumped things in this order: table_C, func_B, table_A.

This is reasonably easy to fix by hand because we only have one such situation and the trail is easy to trace. A more complicated situation could be a pain to fix by hand.

Sample Code

No file was uploaded with this report