Re: ISO/IEC 10646 as Document Character Set

Gavin Nicol (gtn@ebt.com)
Thu, 4 May 95 04:14:52 EDT

>Well, it's a little late now. We can blame them for omitting the
>charset parameter, but is it fair for us to dictate to them how to
>migrate their infrastructure to a more interoperable situation?

Only if our solution is technically better, and offers some advantages
to them. It does. In addition, this is not just the Japanese (your
most lucrative foreign market?). We need the charset parameter support
everywhere, and *everyone* should support it, or be prepared for
failure.

>Many of the sites are not "customers". They are universities taking
>part in this great thing called WWW. Why should anybody consciously
>destroy the interoperability that these folks are enjoying?

Because they are the places that can most easily upgrade. Almost all
of the current Japanese sites are experimental.

>Exactly what I'm doing. But I haven't contacted the Japanese yet,
>to see what they think. That is the next step. I'm only mentioning
>my own concerns here first.

Please go ahead. I *have* talked to many people. Once I explain the
alternative to the charset parameter, and it's *global* importance,
most people agree that it should be deployed. The largest issue for
most is to make sure that clients with charset support arrive *before*
the servers go out.

>Who's the "we"? You mean this working group? Why would it be fair for
>this group to make decisions about "destroying interoperability" in
>somebody else's part of the Internet?

Why not, people in the US have been doing this for years. Besides, I'm
in Japanese territory ;-)

>What I'm saying is exactly analogous. We are now proposing to have the
>servers add the charset parameter to the content-type header. So I'm
>asking if it's necessary to "negotiate" the absence/presence of the
>charset parameter, because I'm worried that older clients may not be able
>to handle content-type headers that include the charset parameter.

As I said before. Adding such support is probably at least as hard,
maybe harder, than adding charset=xxx support. After all, at a
minimum, you only have to *parse* it.