On 8/26/05, Michael Sperber <xxxxxx@informatik.uni-tuebingen.de> wrote:
>
> > If you aren't using a systems-level language (or don't want to go through
> > the effort), then you have to base the lower-levels on an existing upper
> > level, which implies all text gets buffered twice.
>
> Well, sure, if you don't want to go to the effort, almost anything can
> be slow.  Sorry, but this kind of argument doesn't cut it with me.

And if you're not using a systems-level language?  Do we completely
scorn the Java implementations?

> Moreover, if you really want high performance, no ports layer of the
> kind described in the SRFI will do the job at all---only the primitive
> I/O can.

Yes, exactly.  But R5RS ports built on native ports can offer high performance.

> Can you show me how the C library does it?  I can't find it offhand.

See the O_DIRECT option to open(2).

--
Alex