RE: "How to Compare URIs": %xx for non-reserved characters

At 08:17 03/02/18 -0800, Larry Masinter wrote:

>Sorry, I sent my last message without thinking....
>Of course "." in hex is "%2e" and not "%0a", and
>
>http://lists.w3.org/Archives/Public/www-tag/2003Feb/0232%2ehtml
>
>works as well as
>
>http://lists.w3.org/Archives/Public/www-tag/2003Feb/0232.html
>
>I'm still not certain that all URI processing software
>in HTTP servers or clients handles hex-equivalence for
>non-reserved characters, but I'll have to look harder
>for examples.

I'm looking very much forward to more reports from you and
others. Even if we find an odd case where equivalences such
as the above are not maintained, it would probably help a
lot if they were, in the true sense of the 'uniform' U in
URI. And such cases might safely be declared as a bug.

Contrary to namespace equivalence, where all known implementations
work codepoint-by-codepoint, and where a TAG finding saying
that this is wrong would risk to produce more confusion than
architectural consistency, a TAG finding stating that
*for purposes of resolution/retrieval*, ~ == %7e == %7E
would definitely confirm (or improve if this is still needed)
an important point of architectural consistency. And it would
also help IRIs a lot.


Regards,    Martin.

Received on Wednesday, 19 February 2003 13:33:50 UTC