You are right, if pgpassfile_used is true, it SHOULD be defined, I just like to be careful whenever I'm working with strings. But I guess in this scenario I can trust the caller and omit those checks.
Good.
Patch looks ok, applies, compiles & checks, and tested manually.
I've switch in the CF to "ready for committer", and we'll see what the next level thinks about it:-)
[...] I agree with those criticisms of the multi-host feature and notifying the client in case of an authentification error rather than trying other hosts seems sensible to me.
Sure. I complained about the fuzzy documentation & imprecise warning message because I stumbled upon that while testing.
But I think fixes for those should be part of different patches, as this patch's aim was only to expand the existing pgpassfile functionality to be used with a parameter.
Yes.
Moved to next commitfest with same status (ready for committer).