>> I noticed that my pg_stat_statements is cluttered with hundreds of entries
>> like "DEALLOCATE dbdpg_p123456_7", occuring each only once.
>>
>> It seems to me that it would be more helful if these similar entries where
>> aggregated together, that is if the query "normalization" could ignore the
>> name of the descriptor.
>>
>> Any thoughts about this?
>
> You might find this relevant:
> <http://blog.endpoint.com/2014/02/perl-dbdpg-postgresql-prepared-statement.html>
Indeed. Thanks for the pointer. I had guessed who the culprit was, and the
new behavior mentioned in the blog entry may help when the new driver
version hits my debian box.
In the mean time, ISTM that progress can be achieved on pg_stat_statements
normalization as well.
--
Fabien.