On Mon, 31 Jul 2017, Merlin Moncure wrote:
> On Sun, Jul 30, 2017 at 8:34 PM, Steve Singer <steve@ssinger.info> wrote:
>>
>> We don't seem to describe logical replication on
>>
>> https://www.postgresql.org/docs/10/static/different-replication-solutions.html
>>
>> The attached patch adds a section.
>
> This is a good catch. Two quick observations:
>
> 1) Super pedantic point. I don't like the 'repl.' abbreviation in the
> 'most common implementation' both for the existing hs/sr and for the
> newly added logical.
Updated
>
> 2) This lingo:
> + Logical replication allows the data changes from individual tables
> + to be replicated. Logical replication doesn't require a particular server
> + to be designated as a master or a slave but allows data to flow
> in multiple
> + directions. For more information on logical replication, see
> <xref linkend="logical-replication">.
>
> Is good, but I would revise it just a bit to emphasize the
> subscription nature of logical replication to link the concepts
> expressed strongly in the main section. For example:
>
> Logical replication allows the data changes [remove: "from individual
> tables to be replicated"] to be published to subscriber nodes. Data
> can flow in any direction between nodes on a per-table basis; there is
> no concept of a master server. Conflict resolution must be handled
> completely by the application. For more information on...
>
> what do you think?
Sounds good.
I've incorporated these changes into an updated patch but I changed the
language around conflict resolution. Conflict resolution could be handled
by triggers or adding extra columns to the primary key, that wouldn't be
'completely by the application'
>
> merlin
>