RE: [rest-discuss] Restful URIs
- Jeff Bone wrote:
> On Feb 6, 2004, at 4:40 PM, Justin Sampson wrote:Ah, holier-than-thou...
> > The notion that we're talking about only one resource here
> > doesn't make any sense to me. These are two different
> > human-language translations of some document, which sure feel
> > like different resources to me.
> Ah, glasshoppah...
> They are the same resource because they are the sameNo, they're not the same information. The translator has added
> *information.* The two are separate *representations* of the
> same abstract *resource* --- and the resource itself is the
> thing named by the URI.
various interpretational decisions based on the appropriate
cultural contexts. The various documents may even need to relate
to each other in different ways.
> If this isn't clear, imagine this: what if you had some magicThe URI doesn't say how something is generated; different URI may
> piece of software which translated documents in any natural
> language to any other natural language on the fly. What would
> you call the translations themselves vs. the information they
> contained? The translations are representations. The
> information itself is the resource.
be served up from different files or dynamically generated in
> Similarly: a novel in five different languages is still theI can still choose from the crappy electronic translation, the
> same *novel.* Five different printings of that novel --- same
> or different languages --- are different *books.* When I speak
> of "Beggars in Spain" by Nancy Kress, I'm talking about what
> she wrote --- not my copy of it on my bookshelf.
poetic Shakespearean translation, the contemporary readable
translation, etc. What the original author wrote is one thing;
any particular translation of it is quite another. The question
is not whether they are the same "novel", but whether it is
useful and desireable to return representations of all of them
from GETs on a single URI.
- On Wednesday, February 11, 2004, at 07:55 PM, Mark Baker wrote:
> On Wed, Feb 11, 2004 at 05:56:27PM -0800, Roy T. Fielding wrote:That may be true as well, but I was referring to human factors:
>> The rationale for REST being different from RDA, aside from greater
>> flexibility of implementation, is that remote data access interfaces
>> are extremely susceptible to abuse by rogue clients.
> My hunch is that those are the same; that less implementation
> flexibility yields more predictable forms of attack. Or perhaps you
> had some more general concern in mind?
WWW server developers are far more paranoid than DBAs because they
get abused when they aren't paranoid enough. Databases, in contrast,
are sold by feature list, just like a word processor. The WWW server
(or firewall) is supposed to protect the DB from the real world.