Re: [BUG?] strange behavior in ALTER TABLE ... RENAME TO on inherited columns

2010/1/31 KaiGai Kohei <kaigai@ak.jp.nec.com>:
> The attached patch modified find_all_inheritors() to return the list of
> expected inhcount, if List * pointer is given. And, it focuses on only
> the bugs in renameatt() case.

I have cleaned up and simplified this patch.  Attached is the version
I intend to commit.  Changes:

- make find_all_inheritors return the list of OIDs, as previously,
instead of using an out parameter
- undo some useless variable renamings
- undo some useless whitespace changes
- rework comments
- fix regression tests to avoid using the same alias twice in the same query
- add an ORDER BY clause to the regression tests so that they pass on my machine
- improve the names of some of the new variables
- instead of adding an additional argument to renameatt(), just
replace the existing 'bool recursing' with 'int expected_parents'.
This allows merging the two versions of the "cannot rename inherited
column" message together, which seems like a reasonably good idea to
me, though if someone has a better idea that's fine.  I didn't think
the one additional word added to the message provided enough clarity
to make it worth creating another translatable string.

> Also, the ALTER COLUMN TYPE case should be also fixed in the 9.1 release
> (or 9.0.1?).

If the fix is something we could commit for 9.0.1, then we ought to do
it now before 9.0 is released.  If you want to submit a follow-on
patch to address ALTER COLUMN TYPE once this is committed, then please
do so.

...Robert