Thread: Re: [pgsql-hackers-win32] [PATCHES] fork/exec patch
Magnus Hagander wrote: > >Absolutely, but there are other signals to send, no? Or you might want >to send a signal directly to a backend (to cancel for example), as you >can do on Unix. > > > In normal operation the only thing that should be signalling a backend is the postmaster. cheers andrew
Andrew Dunstan <andrew@dunslane.net> writes: > In normal operation the only thing that should be signalling a > backend is the postmaster. Oh? What about LISTEN/NOTIFY? -Neil
Neil Conway kirjutas K, 17.12.2003 kell 00:37: > Andrew Dunstan <andrew@dunslane.net> writes: > > In normal operation the only thing that should be signalling a > > backend is the postmaster. > > Oh? What about LISTEN/NOTIFY? IIRC cancelling queries is done by making a connection to a new backend and then this backend signalls the one needing to have its query cancelled. Or is the cancelling done by postmaster without starting a new backend ? -------------- Hannu