Re: URL internationalization!

Francois Yergeau writes:
>Those of us who have long used, and even written, browsers, know that they
>are used very often to submit forms.  We also know that these forms can
>contain arbitrary characters, for instance when one types one's name.  We
>also know that upon submission the content of the form is encoded in an
>URL, in a contraption called the query part.  We also know that servers
>have problems dealing with such URLs, because the character encoding
>outside ASCII is undefined by the current Proposed Standard.  Hence we have

Irrelevant. It is a trivial operation to encode in the URL used in the
FORM action, or within one or more of the form fields, the character set
and encoding used by the form or by any individual data field.  Since the
server controls that data, there is never any ambiguity regarding the
character set in the subsequent URL.  The only problem that existed,
and still exists today, is that user agents don't have any understanding
of the character set, period, and just submit data in whatever octets
are currently being used for browser display.  That problem has absolutely
nothing to do with URLs -- it is a general issue of interoperability
while submitting forms (via URL or message entity) -- and it would not
be solved by mandating UTF-8 because those old browsers don't use UTF-8.

>concluded long ago that internationalized, non-ASCII URLs are an
>unavoidable requirement of a *World Wide* Web, and some of us are trying to
>do something to end the sorry state of affairs where an exagerated
>"universal typeability" requirement, roundly ignored in current practice,
>takes precedence over basic functionality in the standards.

The transcribability requiremenst of URLs are not roundly ignored in
current practice -- if they were ignored, I wouldn't be able to access large
portions of the Web that are truly international only because those
restrictions are in place.  If you obey the restrictions in the URL spec,
you are guaranteed interoperability with all WWW systems.  If you don't,
then you are not guaranteed to make it as far as a network request before
your application software pukes, rolls-over, and dies.

The difference between writing an Internet standard and inventing an
ideal world is precisely that interoperability requirements are always
more important than idle speculation.

 ...Roy T. Fielding
    Department of Information & Computer Science    ([email protected])
    University of California, Irvine, CA 92697-3425    fax:+1(714)824-4056
    http://www.ics.uci.edu/~fielding/