Re: WIP: Barriers

On Tue, Nov 22, 2016 at 4:42 PM, Thomas Munro
<thomas(dot)munro(at)enterprisedb(dot)com> wrote:
> On Tue, Nov 1, 2016 at 5:03 PM, Thomas Munro
>> Here's a new version which is rebased and adds support for passing
>> wait_event through to pg_stat_activity.
>
> Here's a version updated for the new conditional variables interface
> which has just been committed as e8ac886c. Some comments improved.

The code here looks OK. A few thoughts:

- I'm a little unsure whether it's a good idea to remove the existing
barrier.h and immediately add a new barrier.h that does something
totally different. It's tempting to try to find another name for
these things just to avoid that. Then again, I bet there's not much
non-core code that is relying on the existing barrier.h, so maybe it's
OK. On the third hand, the risk of confusing developers may be
non-nil even if we don't confuse any code. I think I'll go commit
remove-useless-barrier-v4.patch now and see if anyone shows up to
complain...

- Should parallel bitmap heap scan be using this to decide who runs
the subplan? It looks like it would work for what is needed in that
case, and it looks like it would be simpler (and probably more
correct) than what's there now. PBMState could go away completely, I
think.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company