On Wed, Aug 23, 2000 at 02:50:51PM -0400, Uri Guttman <[EMAIL PROTECTED]> wrote:
> solaris has true aynch file i/o.
Just as linux: you clone/block/signal (i am not talking about pthreads). Its
just implemented differently. And the posix aio functions do not use this
approach and are rather slow, but this also has nothing to do with this
discussion.
Remember that the point was to use a different *event* mechanism than using
select/poll, and sigio (vav. the extesions found on modern unix kernels)
is the perfect event mechanism for low and medium load.
Aio will not help a bit with Event, sinc eEvent does not do any I/O,
rather it does event processing. And sigio+friends is exactly that (and
very similar to kqueue).
> but it uses sigio to tell you when it something happens. then you have
> to scan you pending operations to see which one it was. the idjits
> should have integrated it with poll().
Solaris can tell you which event caused the signal, so there is no need to
scan your list of operations. After all, that's the point of sigio.
--
-----==- |
----==-- _ |
---==---(_)__ __ ____ __ Marc Lehmann +--
--==---/ / _ \/ // /\ \/ / [EMAIL PROTECTED] |e|
-=====/_/_//_/\_,_/ /_/\_\ XX11-RIPE --+
The choice of a GNU generation |
|
