Subject Re: [Firebird-Architect] Vulcan statement cancel operation
Author Dimitry Sibiryakov
On 9 Feb 2006 at 11:35, Dmitry Yemanov wrote:

>> I am confused as to why a statement cancel isn't equally useful for
>> a remote scenario as it is for the embedded case.
>
>We use blocking sockets and every API call causes the client to
>perform a synchronous wait (for the server response) on this socket.
>So we cannot deliver a cancellation call via this port.

This is not quite true. We are using non-blocking sockets on server
side and even if on the client side we have a thread blocked on
receiving from socket nothing can prevent an other thread from sendng
into the socket. The only problem that this cancellation packet must
have no response (in order not to confuse main receiving procedure)
and thus we can't tell "cancelling" thread whether the cancel
operation has been successfull.

--
SY, Dimitry Sibiryakov.