Обсуждение: pg_dump include/exclude fails to error

When including tables with the new pg_dump functionality, it fails to
error out if a table is missing, but only if more than one table is
specified.

E.g., if table foo exist, but not bar:

pg_dump --table bar
pg_dump: error: no matching tables were found

with file "myfilter" containing just "table bar"
pg_dump --filter myfilter
pg_dump: error: no matching tables were found

with the file "myfilter" containing both "table foo" and "table bar"
(order doesn't matter):
<no error, but dump of course only contains foo>

Not having looked into the code, but it looks to me like some variable
isn't properly reset, or perhaps there is a check for existence rather
than count?

//Magnus

> On 10 Mar 2024, at 15:22, Magnus Hagander <magnus@hagander.net> wrote:

> Not having looked into the code, but it looks to me like some variable
> isn't properly reset, or perhaps there is a check for existence rather
> than count?

Thanks for the report, I'll have a look later today when back.

--
Daniel Gustafsson

On Sun, Mar 10, 2024 at 4:51 PM Pavel Stehule <pavel.stehule@gmail.com> wrote:
>
> Hi
>
> ne 10. 3. 2024 v 15:23 odesílatel Magnus Hagander <magnus@hagander.net> napsal:
>>
>> When including tables with the new pg_dump functionality, it fails to
>> error out if a table is missing, but only if more than one table is
>> specified.
>>
>> E.g., if table foo exist, but not bar:
>>
>> pg_dump --table bar
>> pg_dump: error: no matching tables were found
>>
>> with file "myfilter" containing just "table bar"
>> pg_dump --filter myfilter
>> pg_dump: error: no matching tables were found
>>
>> with the file "myfilter" containing both "table foo" and "table bar"
>> (order doesn't matter):
>> <no error, but dump of course only contains foo>
>
>
> is not this expected behaviour (consistent with -t option)?
>
> (2024-03-10 16:48:07) postgres=# \dt
>         List of relations
> ┌────────┬──────┬───────┬───────┐
> │ Schema │ Name │ Type  │ Owner │
> ╞════════╪══════╪═══════╪═══════╡
> │ public │ foo  │ table │ pavel │
> └────────┴──────┴───────┴───────┘
> (1 row)
>
> pavel@nemesis:~/src/orafce$ /usr/local/pgsql/master/bin/pg_dump -t foo -t boo > /dev/null
> pavel@nemesis:~/src/orafce$
>
> if you want to raise error, you should to use option --strict-names.
>
> pavel@nemesis:~/src/orafce$ /usr/local/pgsql/master/bin/pg_dump -t foo -t boo --strict-names > /dev/null
> pg_dump: error: no matching tables were found for pattern "boo"

Hmpf, you're right. I guess I don't use multiple-dash-t often enough
:) So yeah, then I agree this is probably the right behaviour. Maybe
the docs for --filter deserves a specific mention about these rules
though, as it's going to be a lot more common to specify multiples
when using --filter?

--
 Magnus Hagander
 Me: https://www.hagander.net/
 Work: https://www.redpill-linpro.com/