Postgres Pro
Downloads
Home   >   mailing lists

Re: Binary support for pgoutput plugin - Mailing list pgsql-hackers

From Dave Cramer
Subject Re: Binary support for pgoutput plugin
Date
Msg-id CADK3HHKo2GW=oQeVFM7cjA5GWSJmgON5moV_2pxrZsJQH4PO=g@mail.gmail.com
Whole thread Raw
In response to Re: Binary support for pgoutput plugin  (Dmitry Dolgov <9erthalion6@gmail.com>)
Responses Re: Binary support for pgoutput plugin  (Alvaro Herrera <alvherre@2ndquadrant.com>)
Re: Binary support for pgoutput plugin  (Dmitry Dolgov <9erthalion6@gmail.com>)
List pgsql-hackers
Tree view


On Fri, 8 Nov 2019 at 11:20, Dmitry Dolgov <9erthalion6@gmail.com> wrote:
> On Tue, Nov 05, 2019 at 07:16:10AM -0500, Dave Cramer wrote:
>
> See attached

    --- a/src/backend/replication/logical/worker.c
    +++ b/src/backend/replication/logical/worker.c
    @@ -1779,6 +1779,7 @@ ApplyWorkerMain(Datum main_arg)
        options.slotname = myslotname;
        options.proto.logical.proto_version = LOGICALREP_PROTO_VERSION_NUM;
        options.proto.logical.publication_names = MySubscription->publications;
    +   options.proto.logical.binary = MySubscription->binary;

I'm a bit confused, shouldn't be there also

        --- a/src/backend/catalog/pg_subscription.c
        +++ b/src/backend/catalog/pg_subscription.c
        @@ -71,6 +71,7 @@ GetSubscription(Oid subid, bool missing_ok)
                        sub->name = pstrdup(NameStr(subform->subname));
                        sub->owner = subform->subowner;
                        sub->enabled = subform->subenabled;
        +       sub->binary = subform->subbinary;

in the GetSubscription?

yes, I have added this. I will supply an updated patch later.

Now a bigger question(s). 

Previously someone mentioned that we need to confirm whether the two servers are compatible for binary or not.

Checking to make sure the two servers have the same endianness is obvious.
Sizeof int, long, float, double, timestamp (float/int) at a minimum.

this could be done in libpqrcv_startstreaming. The question I have remaining is do we fall back to text mode if needed or simply fail ?

Dave

pgsql-hackers by date:

Previous
From: Liudmila Mantrova
Date:
Subject: Re: Does 'instead of delete' trigger support modification of OLD
Next
From: Andrew Dunstan
Date:
Subject: Re: TestLib::command_fails_like enhancement