Re: libpq changes for synchronous replication

Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> On Mon, Sep 20, 2010 at 12:17 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Personally I think this demonstrates that piggybacking replication
>> data transfer on the COPY protocol was a bad design to start with.
>> It's probably time to split them apart.

> This appears to be the only obvious unresolved issue regarding this patch:

> https://commitfest.postgresql.org/action/patch_view?id=412

> I don't have a strong personal position on whether or not we should do
> this, but it strikes me that Tom hasn't given much justification for
> why he thinks we should do this, what benefit we'd get from it, or
> what the design should look like. So I guess the question is whether
> Tom - or anyone - would like to make a case for a more serious
> protocol overhaul, or whether we should just go with the approach
> proposed here.

I was objecting to v1 of the patch. v2 seems somewhat cleaner --- it at
least avoids changing the behavior of libpq for normal COPY operation.
I'm still a bit concerned by the prospect of having to shove further
warts into the COPY data path in future, but maybe its premature to
complain about that when it hasn't happened yet.

Just in a quick scan, I don't have any objection to v2 except that the
protocol documentation is lacking.

regards, tom lane