Eric Scheid schrieb:
On 8/7/06 9:47 AM, "James M Snell" <[EMAIL PROTECTED]> wrote:
Server implementations
MAY attempt to comply with the request.
This sounds and feels negative to me, as if we don't really believe that
PaceSlugHeader should really be supported. May I suggest instead:
"Server implementations MAY ignore the Slug header."
Again, that's something that doesn't require RFC2119 terminology. If the
purpose of the header is clearly stated, no more conformance statements
are needed.
The syntax of this header MUST conform to the augmented BNF grammar in
section 2.1 of the HTTP/1.1 specification [RFC2616]. The [TEXT] rule is
described in section 2.2 of the same document. Words of *TEXT MAY
contain characters from character sets other than [ISO88591] only when
encoded according to the rules of [RFC2047].
A simple example after this paragraph would be appreciated (in addition to
the example preceding)
For example,
POST /myblog/fotes HTTP/1.1
Host: example.org
Content- Type: image/png
Content- Length: nnnn
Slug: Drinking-coffee-at-the-caf%xx%xx
...binary data...
(with the appropriate %encoding, of course)
I say this because without the example I would have guessed that é would be
encoded as %8E ... which would be wrong :-(
Well, Eric got it wrong (it's not encoded using RFC2047), which kind of
shows that just saying "use RFC2047" isn't sufficient. Part of the
problem here is that the proposal should clearly state what level of
encoding it applies to (if this is about specifying parts of URI
segments, URI syntax is sufficient; and RFC2047 doesn't even need to be
mentioned).
Best regards, Julian
