Ben Bucksch wrote:
>
> TenThumbs wrote:
>
> > RFC2396 seems happy with "|" and only labels "[" as "unwise."
>
> 2.4.3. Excluded US-ASCII Characters
> ...
>
> unwise = "{" | "}" | "|" | "\" | "^" | "[" | "]" | "`"
>
> Data corresponding to excluded characters must be escaped in order to
> be properly represented within a URI.
RFC2616 section 3.2.1 specifically adopts portions of RFc2396, not the
entire thing. RFC2616 also has this to say:
Characters other than those in the "reserved" and "unsafe" sets (see
RFC 2396 [42]) are equivalent to their ""%" HEX HEX" encoding.
which is interesting because the definition "unsafe" does not exist in
RFC2396. It does exist in RFC1945 (HTTP/1.0) where "|" and "[" are valid.
This leads to the conclusion that the URL is valid HTTP/1.0 according to
RFC1945, is an invalid URI according to RFC2396, and may or may not be valid
HTTP/1.1 according to RFC2616. Since I'm sure HTTP/1.0 wants to be
reasonably backwards comaptible with HTTP/1.0, I continue to think this is a
Mozilla bug.
--
Sun
2000-12-11 18:09:27.035544 TDB (JD 2451890.256563)
X = -0.004797428, Y = -0.004504044, Z = -0.001779830
X' = 0.000008004, Y' = -0.000003801, Z' = -0.000001839
