Re: Lock contention, docs vs. reality

Roland Turner <raz.cbfgterfdy.bet@raz.cx> writes:
> I'm working with 7.4, but the 8.2 docs[1] have the same apparent error:
> ROW SHARE

>         Conflicts with the EXCLUSIVE and ACCESS EXCLUSIVE lock modes.

> If that conflict list were correct, then ROW SHARE wouldn't conflict
> with itself, much less with ROW EXCLUSIVE (required to prevent
> INSERT/UPDATE/DELETE); commonsense dictates that it should conflict with
> both, and experiment demonstrates that it actually does so.

A moment's examination of the source code (see the conflict tables in
src/backend/storage/lmgr/lock.c) shows that the documentation claim is
correct.  Please provide your experiment to the contrary.

            regards, tom lane