Horton

Thread

Commits

  1. Fix argument of pg_create_logical_replication_slot for slot name

  1. pg_create_logical_replication_slot returns text instead of name

    Masahiko Sawada <sawada.mshk@gmail.com> — 2018-07-12T10:00:48Z

    Hi,

The documentation[1] says that both pg_create_logical_replication_slot
and pg_create_physical_replication_slot returns slot_name as a name
type. But only pg_create_logical_replication_slot returns it as text
type. I think these should be united as the name type. Attached small
patch fixes it.

[1] https://www.postgresql.org/docs/devel/static/functions-admin.html#FUNCTIONS-REPLICATION

Regards,

--
Masahiko Sawada
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center

  2. Re: pg_create_logical_replication_slot returns text instead of name

    Michael Paquier <michael@paquier.xyz> — 2018-07-12T13:18:53Z

    On Thu, Jul 12, 2018 at 07:00:48PM +0900, Masahiko Sawada wrote:
> The documentation[1] says that both pg_create_logical_replication_slot
> and pg_create_physical_replication_slot returns slot_name as a name
> type. But only pg_create_logical_replication_slot returns it as text
> type. I think these should be united as the name type. Attached small
> patch fixes it.
> 
> [1] https://www.postgresql.org/docs/devel/static/functions-admin.html#FUNCTIONS-REPLICATION

That's a small thing, but I agree with you.  As far as I can see slot
names are always mapped with the name type.  I'll push that tomorrow if
there are no objections.
--
Michael

  3. Re: pg_create_logical_replication_slot returns text instead of name

    Michael Paquier <michael@paquier.xyz> — 2018-07-13T00:48:58Z

    On Thu, Jul 12, 2018 at 10:18:53PM +0900, Michael Paquier wrote:
> That's a small thing, but I agree with you.  As far as I can see slot
> names are always mapped with the name type.  I'll push that tomorrow if
> there are no objections.

Pushed, with a catalog version bump.

While double-checking things, I have noticed that pg_stat_wal_receiver
also uses text for slot names.  I am not sure if this one is worth
bothering as the docs point out the correct type, just mentioning.
--
Michael

  4. Re: pg_create_logical_replication_slot returns text instead of name

    Masahiko Sawada <sawada.mshk@gmail.com> — 2018-07-13T02:14:01Z

    On Fri, Jul 13, 2018 at 9:48 AM, Michael Paquier <michael@paquier.xyz> wrote:
> On Thu, Jul 12, 2018 at 10:18:53PM +0900, Michael Paquier wrote:
>> That's a small thing, but I agree with you.  As far as I can see slot
>> names are always mapped with the name type.  I'll push that tomorrow if
>> there are no objections.
>
> Pushed, with a catalog version bump.
>

Thank you!

> While double-checking things, I have noticed that pg_stat_wal_receiver
> also uses text for slot names.  I am not sure if this one is worth
> bothering as the docs point out the correct type, just mentioning.

Hmm, I'm also not sure about the policy of usage of name data type for
columns that show an object identifier on external servers. There is a
similar case; we have the pubname in pg_subscritpion as name type
whereas the subpublications in pg_subscription is text[] type.

Regards,

--
Masahiko Sawada
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center

  5. Re: pg_create_logical_replication_slot returns text instead of name

    Michael Paquier <michael@paquier.xyz> — 2018-07-13T02:22:11Z

    On Fri, Jul 13, 2018 at 11:14:01AM +0900, Masahiko Sawada wrote:
> Hmm, I'm also not sure about the policy of usage of name data type for
> columns that show an object identifier on external servers. There is a
> similar case; we have the pubname in pg_subscritpion as name type
> whereas the subpublications in pg_subscription is text[] type.

Let's not bother then.  In the case of the function you pointed out the
intention behind the code was clear anyway, so that's better now.
--
Michael

  6. Re: pg_create_logical_replication_slot returns text instead of name

    Masahiko Sawada <sawada.mshk@gmail.com> — 2018-07-13T02:28:01Z

    On Fri, Jul 13, 2018 at 11:22 AM, Michael Paquier <michael@paquier.xyz> wrote:
> On Fri, Jul 13, 2018 at 11:14:01AM +0900, Masahiko Sawada wrote:
>> Hmm, I'm also not sure about the policy of usage of name data type for
>> columns that show an object identifier on external servers. There is a
>> similar case; we have the pubname in pg_subscritpion as name type
>> whereas the subpublications in pg_subscription is text[] type.
>
> Let's not bother then.  In the case of the function you pointed out the
> intention behind the code was clear anyway, so that's better now.

Yeah, I agreed with you.

Regards,

--
Masahiko Sawada
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center