[
Lists Home |
Date Index |
Thread Index
]
tbray@textuality.com (Tim Bray) writes:
>Simon St.Laurent wrote:
>
>> Nope. I'm thoroughly frustrated with REST's refusal to let me
>> actually talk about representations, but this is a separate matter.
>
>Over the past year, I've heard a dozen different people say "but we
>want to talk about representations". OK, why not do it and see what
>you get?
I am:
http://simonstl.com/projects/vellum/
Targets that contain representation elements make life a lot easier than
squashing everything into a semi-opaque string.
> REST doesn't give you a hook to identify one, but doesn't stop you if
>you if you want to invent your own.
>
>It's a bit trickier than you might think; you could be talking about a
>very particular bag of bits, perhaps with an MD5 signature. You could
>be talking about "the HTML version of the weather forecast" which is
>distinguished from the WAP and XML versions but still time-varying. Of
>course, it's hard to talk about anything without having a way to refer
>to it, and I suspect Simon will barf if anyone suggests using URI to
>identify representations; but I think you could.
Yep. Makes me barf, pretty quickly. But I'm well aware that I could do
something like:
http://simonstl.com/rsrc/index/?content-type=application/xml+whatever=
whatever....
with appropriate escaping. Seems like a good way to create enormous
nouns, kind of like German. If you like that, fine, but yecch. I'd
rather keep my adjectives separate, but remind people that adjectives
are useful and that focusing exclusively on verbs and nouns is limiting.
>So it's certainly possible in principle. REST doesn't get in the way.
REST gets in the way because far too many people assume that a URI is a
sufficient description for "what I want" and forget that there's a whole
world of clarifications to that which don't live in the URI. It shows
up throughout the specification world. See, for instance:
http://lists.w3.org/Archives/Public/www-tag/2002Dec/0260.html
>What is this useful for? -Tim
Getting media-type-dependent fragment identifiers to function reliably,
among other things. Making content-negotiations explicit, ordinary, and
useful seems like another nice side effect.
--
Simon St.Laurent
Ring around the content, a pocket full of brackets
Errors, errors, all fall down!
http://simonstl.com -- http://monasticxml.org
|
