Helge Hafting wrote:
> > With poll(), it was *not a bug* for the user code to drop events; with
> > your proposed interface, it *is a bug* for the user code to drop events.
> > I'm just emphasizing this because Simon Kirby (sim@stormix.com) posted
> > incorrectly that your interface "has the same semantics as poll from
> > the event perspective".
>
> Don't worry - existing interfaces will be around for existing
> applications. I believe Linus is trying to engineer a new and better
> interface that will make a difference - (server) programs written to use
> the new stuff will show better performance use up less kernel time.
>
> The semantics will be different, so only new programs will benefit.
> However, the benefits may be so great that people will want to rewrite
> programs to get them.
It's harder to write correct programs that use edge-triggered events.
Level-triggered events are inherantly easier to use. If our high-performance
interface only supports edge-triggered events, people who write high
performance apps may well migrate to FreeBSD, where there is a lovely
high performance interface that offers either level or edge triggered
events, whichever you prefer.
> Seems Linus don't want to do this no matter what you comes up with. You
> are free to try of course...
Or free to switch to FreeBSD, if that's the only OS which has a
high-performance poll replacement that I can get my existing codebase
to work with.
- Dan
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Oct 31 2000 - 21:00:16 EST