From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-03-04 15:31:09
At 11:23 AM 3/4/2001 -0600, Ed Brey wrote:
>Performance (in the sense used in this thread) is only one application
>non-blocking I/O. The other major application is managing program flow
>control. Just as a programmer uses non-blocking sockets to avoid to
>with threads or having his application lock up while waiting for data,
>non-blocking file I/O can be just as useful. I can't tell you how many
>times I've opened a file on a wide area network, only to have something
>wrong/slow, and cursed the application/OS designers for not using
>non-blocking I/O, which would have allowed me to cancel or do something
>else while I wait.
Good point. Hopefully that will get captured for the Library's rationale.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk