[r6rs-discuss] Stateful codecs and inefficient transcoding

From: William D Clinger <will>
Date: Wed Nov 1 01:40:20 2006

I am posting this as an individual member of the Scheme
community. I am not speaking for the R6RS editors.

John Cowan quoting me:
>> Should the Scheme community agree that this perceived
>> requirement [of mixing text and binary] should be dropped ...
>
> I believe it should be dropped: it complicates the general
> system in favor of a special case which should be handled
> by using the R6RS library for binary input and doing one's
> own decoding. It is far more important to be able to
> handle encodings in common use than it is to handle
> mixed binary/textual encoding.

To which Per Bothner added:
> Alternatively, I believe 99% of the "need" for mixing text and binary
> would be covered by support for:
> * switching from binary to text;
> * switching from UTF-8 to binary;
> * and nothing else.
>
> Also, the default/initial text encoding should be defined by
> the locale/implementation, and not required to be UTF-8.

To everyone who has participated in this discussion:
Thank you! You have made me a significantly happier
person.

I'm going to study some of the suggestions that have
been made, run some ideas past a couple of friends,
and then come back here to ask what you think of an
alternative.

Anyone can propose alternatives, of course. In this
mailing list, I'm just an individual member of the
Scheme community. I am not speaking for the R6RS
editors.

Will
Received on Wed Nov 01 2006 - 01:40:18 UTC