I wonder if it may be possible to make an alternate API call to allow a
timeout parameter to be passed along with the request to execute a
statement. This way, there wouldn't have to be any complex asynchronous
processing going on in the background. The server would just keep track of
the duration of the query being processed on the server and then if it
exceeds the time allowed it will terminate the request and exit with an
errcode signifying that the query timed out. At that point an exception
could be raised in the application and the user would be informed that the
query took too long to process.

This would at least be ideal for the development mode where a developer
would want to make sure that all their queries are faster than a certain
period of time so they would just set it to timeout a little past that.

In an application the user could at least be given the assurance that the
server won't stay busy too long so they won't be as inclinded to
CTRL+ALT+DELETE the app. They know if they wait at least 120 seconds that it
will come back to them...

On a web-site the user could just be told, sorry, your request timed out on
the server, please try again another time or try to use criteria that is
more specific, etc...

FWIW,
Jason Wharton
CPS - Mesa AZ
http://www.ibobjects.com

Developer's Search Site! http://developers.href.com