> On 22 Mar 2018, at 22:38, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> This stuff seems reasonably non-controversial, so pushed.
Thanks!
> BTW, really the point of what I'd mentioned before was to avoid having
> dblink_res_error constructing a message out of fragments, which it's
> still doing. I'd thought perhaps we would shove the responsibility for
> mentioning the connection name out to the callers to get rid of that.
> But handling the possibility of an unnamed connection seems like it'd
> complicate the callers considerably. And as long as we don't actually
> have translation support in that module, it'd just be make-work, so
> I didn't do it.
Right, that would be a lof of complexity for no real gain until dblink is
translated.
cheers ./daniel