Re: PATCH: Reducing lock strength of trigger and foreign key DDL

On Tue, Jan 13, 2015 at 5:21 PM, Andreas Karlsson <andreas@proxel.se> wrote:
> Here is a new version of the patch with updated isolation tests and
> documentation.
Thanks.

Some more comments:
- It would be good to have as well tests for ENABLE/DISABLE TRIGGER,
the locks of those commands being reduced as well.
- Patch has some typos, like "an share".
- Documentation can be simplified. Listing the flavors of ALTER TABLE
taking ROW SHARE EXCLUSIVE in mvcc/sgml is clumsy IMO, it is better to
mention that in alter_table.sgml and have a link pointing to the page
of ALTER TABLE in mvcc.sgml.
- In ATAddForeignKeyConstraint, I think that it would be better to not
remove the comment, adding instead that ShareRowExclusiveLock is
actually safe because CREATE TRIGGER uses this level of lock.
All those things gathered give the patch attached. Andreas, if you are
fine with it I think that we could pass it to a committer.
Regards,
--
Michael