Tom Lane wrote:
> Bruce Momjian <pgman@candle.pha.pa.us> writes:
> > Tom Lane wrote:
> >> This is actually an issue though. Row-level shared locks would be
> >> really nice to have for foreign-key handling. Right now we have to
> >> use X locks for those, and that leads to deadlocking problems for
> >> applications.
>
> > Is the plan to allow one backend to shared lock the row while others can
> > read it but not modify it, or is the idea to actually allow multiple
> > backends to record their shared status on the row?
>
> Plan? We have no plan to fix this :-(. But clearly there has to be
> some way to tell which backends hold read locks on a shared-locked row,
> else you can't tell if they've all dropped the lock or not.
I suppose we could allow one backend to mark the page with a shared lock
for primary key purposes while others read it. Does that buy us
anything?
-- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610)
359-1001+ If your life is a hard drive, | 13 Roberts Road + Christ can be your backup. | Newtown Square,
Pennsylvania19073
