Re: Improving deadlock error messages

Neil Conway <neilc(at)samurai(dot)com> writes:
> On Fri, 2007-04-20 at 02:55 -0400, Tom Lane wrote:
>> I don't think you've thought of quite all of the failure cases. One
>> that's a bit pressing is that a deadlock isn't necessarily confined to
>> objects in your own database.

> I'm not sure I follow. If we conditionally acquire the locks we need and
> always fallback to just printing the numeric OIDs, ISTM we should be
> able to avoid the infinite recursion problem. (If necessary, we can
> always special-case objects outside our own database, although I'm not
> sure that's necessary.)

Maybe so, but you're going to be writing quite a lot of duplicative
code, because the existing routines you might have been thinking of
using (lsyscache.c etc) don't behave that way.

The basic objection I've got to this is that it'll introduce a lot of
complexity and fragility into a seldom-taken error-recovery path,
which is almost by definition not well enough tested already.

regards, tom lane