Thread: Documentation on Streaming Replication interface for Logical Decoding

In section 46.3 (Streaming Replication Protocol Interface) the create command
for logical slots, CREATE_REPLICATION_SLOT, is defined as taking “options”
after the LOGICAL keyword when it in fact requires the plugin name.  The start
command is correctly listing the options but isn’t listing the required start
position.

The attached patch synchronizes these command signatures with the more detailed
descriptions in section 50.3* (Streaming Replication Protocol).

    cheers ./daniel

(*) Section number from the 9.5beta1 manual

On Tue, Nov 10, 2015 at 7:40 AM, Daniel Gustafsson <daniel@yesql.se> wrote:
> In section 46.3 (Streaming Replication Protocol Interface) the create command
> for logical slots, CREATE_REPLICATION_SLOT, is defined as taking “options”
> after the LOGICAL keyword when it in fact requires the plugin name.  The start
> command is correctly listing the options but isn’t listing the required start
> position.
>
> The attached patch synchronizes these command signatures with the more detailed
> descriptions in section 50.3* (Streaming Replication Protocol).
>
>         cheers ./daniel
>
> (*) Section number from the 9.5beta1 manual

Why do we need to maintain the semantics of those commands in two
places by the way? There is already a link to the replication protocol
page in logicaldecoding.sgml, hence it seems to me that we had better
just cite the command names, per se the attached.
--
Michael

> On 10 Nov 2015, at 05:59, Michael Paquier <michael.paquier@gmail.com> wrote:
>
> On Tue, Nov 10, 2015 at 7:40 AM, Daniel Gustafsson <daniel@yesql.se> wrote:
>> In section 46.3 (Streaming Replication Protocol Interface) the create command
>> for logical slots, CREATE_REPLICATION_SLOT, is defined as taking “options”
>> after the LOGICAL keyword when it in fact requires the plugin name.  The start
>> command is correctly listing the options but isn’t listing the required start
>> position.
>>
>> The attached patch synchronizes these command signatures with the more detailed
>> descriptions in section 50.3* (Streaming Replication Protocol).
>>
>> (*) Section number from the 9.5beta1 manual
>
> Why do we need to maintain the semantics of those commands in two
> places by the way? There is already a link to the replication protocol
> page in logicaldecoding.sgml, hence it seems to me that we had better
> just cite the command names, per se the attached.

Thats a perfectly reasonable approach as well, admittedly I don’t see much
point in the duplication.

    cheers ./daniel