Re: NIO charsets attacked by gang of rabid platypi

[Mark Wielaard]

Hi Sven,

On Tue, 2005-04-19 at 15:00 +0200, Sven de Marothy wrote:
> I'd like everyone's opinion on whether it's OK yet to get rid of the
> now-unused gnu.java.io character encoding/decoding stuff?

I think this is fine. Nobody except our own library code should have
been using that. And we shouldn't keep dead code around.

> Also, I'd like to know which NIO charset provider is preferred, now that
> we have two in CVS. (pure java and iconv-based) They complement
> eachother pretty well, iconv being somewhat faster in the actual
> conversion, but with more overhead for invocation (=slower on short
> strings, faster on long ones). 
> 
> Should we allow this to be set as a compile option, as a system
> property, or just use both (in which order, in that case?).

I would try to avoid adding new compilation/configure options. Best
would be if the runtime could set a system property that indicated which
and in which order to use the providers gnu.classpath.nio.provider or
something. Can this be done early enough in the bootstrap process?

I don't have a strong opinion on the default order. But I would go with
using iconv first and fall back to the gnu.java.nio.charset.Provider (or
some other user supplied provider) when an encoding isn't available with
the platform native iconv.

Whatever you decide make sure you add a little note to the NEWS file.

Cheers,

Mark

signature.asc
Description: This is a digitally signed message part