Обсуждение: PostgreSQL 9.1beta 1 Some domains not being able to be auto cast to their base type

In regression testing PostGIS 2.0, our topology module regression tests are
failing in PostgreSQL 9.1 beta.

We have a PostGIS ticket open for it here, but we suspect it's a PostgreSQL
9.1 bug and would like to close it out.
http://trac.osgeo.org/postgis/ticket/980

Our domain type is defined as:

CREATE DOMAIN topoelementarray
  AS integer[][]
   CONSTRAINT dimensions CHECK ((((array_upper(VALUE, 2) IS NOT NULL) AND
(array_upper(VALUE, 2) = 2)) AND (array_upper(VALUE, 3) IS NULL)));

If I try this in PostgreSQL 9.1 beta1:

SELECT array_upper(ARRAY[ARRAY[1,2], ARRAY[3,4]]::topoelementarray,1);

I get error:
function array_upper(topoelementarray, integer) does not exist

In my PostgreSQL 9.0 and 8.4 the above operation works fine.


Thanks,
Regina

On 05/24/2011 07:05 PM, Paragon Corporation wrote:
> In regression testing PostGIS 2.0, our topology module regression tests are
> failing in PostgreSQL 9.1 beta.
>
> We have a PostGIS ticket open for it here, but we suspect it's a PostgreSQL
> 9.1 bug and would like to close it out.

Do you think you might be able to figure out how to boil it down to a
test case without PostGIS? It might help show that it's a Pg bug more
clearly.

Just a quick reply because I notice you haven't had any responses.

--
Craig Ringer

On Wed, May 25, 2011 at 7:37 AM, Craig Ringer
<craig@postnewspapers.com.au> wrote:
> On 05/24/2011 07:05 PM, Paragon Corporation wrote:
>>
>> In regression testing PostGIS 2.0, our topology module regression tests
>> are
>> failing in PostgreSQL 9.1 beta.
>>
>> We have a PostGIS ticket open for it here, but we suspect it's a
>> PostgreSQL
>> 9.1 bug and would like to close it out.
>
> Do you think you might be able to figure out how to boil it down to a test
> case without PostGIS? It might help show that it's a Pg bug more clearly.
>
> Just a quick reply because I notice you haven't had any responses.


This issue is known and is going to be addressed -- see recent threads
on -hackers regarding arrays of domains.

merlin

On Wed, May 25, 2011 at 9:06 AM, Merlin Moncure <mmoncure@gmail.com> wrote:
> On Wed, May 25, 2011 at 7:37 AM, Craig Ringer
> <craig@postnewspapers.com.au> wrote:
>> On 05/24/2011 07:05 PM, Paragon Corporation wrote:
>>>
>>> In regression testing PostGIS 2.0, our topology module regression tests
>>> are
>>> failing in PostgreSQL 9.1 beta.
>>>
>>> We have a PostGIS ticket open for it here, but we suspect it's a
>>> PostgreSQL
>>> 9.1 bug and would like to close it out.
>>
>> Do you think you might be able to figure out how to boil it down to a test
>> case without PostGIS? It might help show that it's a Pg bug more clearly.
>>
>> Just a quick reply because I notice you haven't had any responses.
>
>
> This issue is known and is going to be addressed -- see recent threads
> on -hackers regarding arrays of domains.

er, domains of arrays :-).

merlin