Re: possible bug not in open items

On Fri, 2009-03-27 at 15:43 -0400, Tom Lane wrote:
> Heikki Linnakangas  writes:
> > I'm not too familiar with this code, but I think we could just enable 
> > ImmediateInterruptOK in CopyGetData().
> 
> Only if you are wanting to break things.
> 
> The reason we don't allow client read to be interrupted is the fear of
> losing protocol sync on an incomplete message.  For the SIGTERM case
> this would (probably) be okay, since we aren't going to pay any more
> attention to the client anyway, but accepting SIGINT there is right out.
> 

That's perfectly acceptable to me. I'm only concerned about the shutdown
case, and that's the only case that's in conflict with the docs.

Regards,
	Jeff Davis