Thread: pg_get_constraintdef failing with cache lookup error

Hi all,

pg_get_constraintdef() fails rather easily:
=# select pg_get_constraintdef(1);
ERROR:  cache lookup failed for constraint 1
STATEMENT:  select pg_get_constraintdef(1);
ERROR:  XX000: cache lookup failed for constraint 1
LOCATION:  pg_get_constraintdef_worker, ruleutils.c:1305

Now by looking at pg_get_constraintdef_worker()@ruleutils.c it is
commented that this error should never happen:
        if (!HeapTupleIsValid(tup)) /* should not happen */
                elog(ERROR, "cache lookup failed for constraint %u",
constraintId);
But it happens.
Am I missing something? Shouldn't this error be changed into something
more informative?
Regards,
--
Michael

Michael Paquier <michael.paquier@gmail.com> writes:
> pg_get_constraintdef() fails rather easily:
> =# select pg_get_constraintdef(1);
> ERROR:  cache lookup failed for constraint 1

It's expected that the caller will supply a valid pg_constraint OID.
I don't think it's necessary to go out of our way to throw a friendlier
error message if the OID isn't found.

            regards, tom lane

On Mon, Apr 27, 2015 at 11:16 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Michael Paquier <michael.paquier@gmail.com> writes:
>> pg_get_constraintdef() fails rather easily:
>> =# select pg_get_constraintdef(1);
>> ERROR:  cache lookup failed for constraint 1
>
> It's expected that the caller will supply a valid pg_constraint OID.
> I don't think it's necessary to go out of our way to throw a friendlier
> error message if the OID isn't found.

It's clearly at least a bug in the comment to complain that it can't
happen if in fact it can.  But I would argue that we should never hit
an elog just because of bad user input.

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

On Tue, Apr 28, 2015 at 3:15 AM, Robert Haas <robertmhaas@gmail.com> wrote:
> On Mon, Apr 27, 2015 at 11:16 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> Michael Paquier <michael.paquier@gmail.com> writes:
>>> pg_get_constraintdef() fails rather easily:
>>> =# select pg_get_constraintdef(1);
>>> ERROR:  cache lookup failed for constraint 1
>>
>> It's expected that the caller will supply a valid pg_constraint OID.
>> I don't think it's necessary to go out of our way to throw a friendlier
>> error message if the OID isn't found.
>
> It's clearly at least a bug in the comment to complain that it can't
> happen if in fact it can.  But I would argue that we should never hit
> an elog just because of bad user input.

Fine for me to just remove the comment, constraintdef is the only *def
function that has such one.
--
Michael