On 4/6/22 11:33, Andrew Dunstan wrote:
> On 4/6/22 11:11, Stephen Frost wrote:
>> Greetings,
>>
>> * Andrew Dunstan (andrew@dunslane.net) wrote:
>>> On 4/6/22 09:20, Andrew Dunstan wrote:
>>>> On 4/5/22 22:21, Andres Freund wrote:
>>>>> On 2022-03-27 16:53:57 -0400, Andrew Dunstan wrote:
>>>>>> I'm therefore going to commit this series
>>>>> The new jsonb_sqljson test is, on my machine, the slowest test in the main
>>>>> regression tests:
>>>>>
>>>>> 4639 ms jsonb_sqljson
>>>>> 2401 ms btree_index
>>>>> 2166 ms stats_ext
>>>>> 2027 ms alter_table
>>>>> 1616 ms triggers
>>>>> 1498 ms brin
>>>>> 1489 ms join_hash
>>>>> 1367 ms foreign_key
>>>>> 1345 ms tuplesort
>>>>> 1202 ms plpgsql
>>>>>
>>>>> Any chance the slowness isn't required slowness?
>>>> I'll take a look.
>>> I've committed a change that should reduce it substantially, but there
>>> might be more work to do.
>> All for improving the speed, but this broke the recovery tests (as
>> noticed by the buildfarm). Maybe we should add
>> --no-unlogged-table-data to those pg_dumpall runs?
>
>
> I think we should, but I think here the obvious solution is to drop the
> table when we're done with it. I'll test that.
>
>
It does work, but Tom prefers not to have the test at all, so I'll just
rip it out.
cheers
andrew
--
Andrew Dunstan
EDB: https://www.enterprisedb.com