> Magnus wrote:
> > Damn!
> > That patch missed a few things - so now it doesn't compile any more!
> > Here is a fix for the missing.
>
> I applied this patch last night --- dunno why Marc didn't do it yet,
> since as you say the system won't build without it.
>
> I still had problems building the revised backend/libpq code on HP-UX.
> I solved them by the expedient of removing the MSG_WAITALL flags from
> the recv() calls. That flag doesn't exist on HP-UX, nor on a lot of
> other systems I'll bet. So I went ahead and checked in that change.
>
> If you think MSG_WAITALL is really important for performance, you
> could offer a patch that would do something like "#ifdef MSG_WAITALL".
> My advice is not to bother though. 99% of frontend queries are going
> to arrive in a single packet, probably, so I doubt it's worth worrying
> about or introducing a possible cross-platform behavior
> difference for.
Nope, it shuold probably be Ok to do without. Definitly not worth
introducing incompatibilities. Get rid of it :-)
//Magnus