- From: Ian Hickson <ian@hixie.ch>
- Date: Thu, 8 Apr 2010 00:35:03 -0700
- To: Dan Connolly <connolly@w3.org>
- Cc: public-html@w3.org
On Tue, Apr 6, 2010 at 9:57 AM, Dan Connolly <connolly@w3.org> wrote: > This is informed by discussion with lots of people, but nobody > else has looked at it, so it's just from me. Would the following be an acceptable compromise? <p>What some specifications, in particular the HTTP and URI specifications, refer to as a <i>representation</i> is referred to in this specification as a <dfn title="">resource</dfn>.</p> Here are some specific concerns I had with the exact text you proposed: > The term resource is used to refer to what is sometimes called a representation There are two ways to interpret this: as saying that the term "resource" sometimes is used in the manner that refers to representations, or as saying that the term "resource" is always used in that manner. Since not every occurrence of the word "resource" means "representation" in the HTML5 spec, I think it's important for readers unfamiliar with any of these terms to not confuse this sentence for a definition of "resource" but merely an indication that the word "representation" is not used. > in protocol literature I think this overstates the case. Unlike "Internet media type", which truly is unambiguously used in the sense of MIME type in almost any spec that doesn't need the term "media" to mean what CSS uses it to mean, "representation" used far less commonly, and the term "resource" is used in a manner that often can be interpreted both ways. (Even in HTML5, if you want to make the case that the word "resource" is used to mean what the HTTP specs take it to mean, it's not very hard to find a number of cases where that interpretation makes sense.) So I think it's better to focus on specific specs that use "representation". > such as section 1.2.2 I try to avoid referencing specific section numbers because they change when the specs are updated, leading to extra editing work later. > of [RFC3986]. The spec's style is to not use references in sentences but to use the colloquial names of the specifications and then have the reference after the sentence. > Where that specification speaks of a URI that identifies a resource whose state is communicated via a typed byte sequence called a representation, we simply say that a URL identifies a resource, which is a typed byte sequence; the indirection is not mentioned in this specification. There's already a comment about the URI/URL terminology in the URL section; mentioning it here might just confuse matters further. It's also not clear to me that the above sentence really clarifies anything. It's not like the indirection isn't mentioned, it's mentioned very explicitly: just not in abstract terms. The HTML spec speaks of contacting servers, sending HTTP headers; it speaks of server-side scripts; it speaks of HTTP headers in the response and so forth. It's just all done explicitly, without having to reference the nebulous "resource" concept. Also, representations aren't always typed, and might not be a byte stream. (For example, the resource returned from evaluating a javascript: URL is a Unicode string.) -- Ian Hickson
Received on Thursday, 8 April 2010 07:35:36 UTC