Fujii Masao <masao.fujii@oss.nttdata.com> writes:
> Regarding 0001 patch, on second thought, to me, it seems odd to expose
> a function that doesn't have anything to directly do with PostgreSQL
> as a libpq function. The function simply calls poll() on the socket
> with POLLRDHUP if it is supported. While it is certainly convenient to
> have this function, I'm not sure that it fits within the scope of libpq.
> Thought?
Yeah, that does not seem great, partly because the semantics would be
platform-dependent. I don't think we want that to become part of
libpq's API.
regards, tom lane