Re: [cp-patches] [Patch] support for direct buffers

[Michael Koch]

Am Donnerstag, 18. November 2004 14:25 schrieb Jeroen Frijters:
> Michael Koch wrote:
> > Can you provide us with numbers that prove this ?
>
> I could give some IKVM numbers, but I don't see how that would
> prove anything. It's obviously highly VM specific.
>
> Do you have any strong objection against using subclasses to
> distinguish between writable and read-only buffers? The rational is
> pretty simple, since we already pay for virtual method invocation,
> we might as well make use of it to get rid of an additional test.

Not really. It just keeps the numbers of places lower where too look 
when a bug needs to get fixed. But for speed it will be surely better 
to use different classes for read-only and read-write.

Can we agree that you commit your patch when mine is in ? Or shall I 
take your patch and build it into my patch ?


Michael
-- 
Homepage: http://www.worldforge.org/