Обсуждение: Bug #559: MACADDR type & 00:00:00:00:00:00

Michael Handler (handler@grendel.net) reports a bug with a severity of 3
The lower the number the more severe it is.

Short Description
MACADDR type & 00:00:00:00:00:00

Long Description
postgresql version 7.1.3, compiled with gcc 2.95.2 on solaris 2.6, running on solaris 2.8.

provisioning=> \d enduser
                         Table "enduser"
    Attribute    |       Type        |          Modifier
-----------------+-------------------+----------------------------
 account         | character varying | not null
 active          | boolean           | not null default 't'::bool
 isp_id          | integer           | not null
 speed_plan_id   | integer           | not null
 service_plan_id | integer           | not null
 node_id         | integer           | not null
 cm_macaddr      | macaddr           | not null
Indices: enduser_cm_macaddr_key,
         enduser_pkey

provisioning=> insert into enduser values ('2', 't', 1, 1, 1, 1, '00:00:00:00:00:00');
INSERT 20853 1
provisioning=> select account,cm_macaddr from enduser where account = '2';
 account | cm_macaddr
---------+------------
 2       |
(1 row)

provisioning=> select count(*) from enduser where cm_macaddr is null;
 count
-------
     0
(1 row)

it's not null, so shouldn't it be retrieving as '00:00:00:00:00:00'?
not that i ever expect to encounter that macaddress in the field, but
for consistency's sake...

i just tested 'ff:ff:ff:ff:ff:ff', and it works fine.

ps postgresql sends mysql running back home to mommy. you guys are my heroes.

Sample Code


No file was uploaded with this report

> MACADDR type & 00:00:00:00:00:00
> postgresql version 7.1.3, compiled with gcc 2.95.2 on solaris 2.6, running on solaris 2.8.
...
> provisioning=> select macaddr '00:00:00:00:00:00';
> it's not null, so shouldn't it be retrieving as '00:00:00:00:00:00'?
> not that i ever expect to encounter that macaddress in the field, but
> for consistency's sake...
> i just tested 'ff:ff:ff:ff:ff:ff', and it works fine.

Already fixed in the upcoming 7.2. Thanks for the report.

                   - Thomas

pgsql-bugs@postgresql.org writes:
> it's not null, so shouldn't it be retrieving as '00:00:00:00:00:00'?

Yeah, due to some brain-fade on the part of the original implementor,
the code went out of its way to special-case 00:00:00:00:00:00 as
being represented by an empty string.  This was agreed to be broken
and we've changed it for 7.2.  If you like you can drop the 7.2
version of the file mac.c into 7.1; see
http://developer.postgresql.org/cvsweb.cgi/pgsql/src/backend/utils/adt/mac.c

            regards, tom lane