> I had a deeper look at this now. The patch looks clean and applies without
> any problems, regression tests passes. However, ATRewriteTables() has a
> problem when adding columns with domains and constraints. Consider this
> small test case:
>
> CREATE TABLE bar (id INTEGER);
> CREATE OR REPLACE VIEW vbar AS SELECT * FROM bar;
> CREATE DOMAIN person AS TEXT CHECK(value IN ('haas', 'helmle'));
> ALTER TABLE bar ADD COLUMN name person;
> CREATE OR REPLACE VIEW vbar AS SELECT * FROM bar;
>
> The last command confuses ATRewriteTable(), which wants to scan the relation
> leading to this error:
> ERROR: could not open relation base/16384/16476:
>
> I see that ATRewriteTable() errors out on heap_beginscan(), since needscan
> is set to TRUE. One solution would be to teach ATRewriteTable(s) to handle
> view alteration differently in this case.
After looking at this, I think the root cause of this problem is that
ATPrepAddColumn isn't smart enough to know that when the underlying
relation is a view, there's no point in asking for a table rewrite.
Please find an updated patch that addresses this problem.
Thanks again for the review - let me know what you think of this version!
...Robert
