- From: Graham Klyne <gk@ninebynine.org>
- Date: Sat, 06 Dec 2014 09:07:37 +0000
- To: Sam Ruby <rubys@intertwingly.net>
- CC: Julian Reschke <julian.reschke@gmx.de>, Bjoern Hoehrmann <derhoermi@gmx.net>, public-ietf-w3c@w3.org
On 06/12/2014 08:10, Julian Reschke wrote: > On 2014-12-06 01:31, Bjoern Hoehrmann wrote: >> Goals >> >> The URL standard takes the following approach towards making >> URLs fully interoperable: >> * Align RFC 3986 and RFC 3987 with contemporary >> implementations and obsolete them in the process. (E.g. >> spaces, other "illegal" code points, query encoding, >> equality, canonicalization, are all concepts not entirely >> shared, or defined.) URL parsing needs to become as solid >> as HTML parsing. [34][RFC3986] [35][RFC3987] > > I think obsoleting RFC 3986 is a non-goal; *until* it is demonstrated that > there's actually a problem with that spec (at which point bug reports should be > sent). +1 I've been using RFC3986 in my work (as developer) for many years now, and find it for the most part to be clear and unambiguous for my needs. I also liberally add references to RFC 3986 to my code where I feel it helps explain what I'm doing. For me, there's no particular benefit in having a different spec that also explains how to parse a string to a URL/URI, and I don't really want to have to absorb and reference a different spec for describing what constitutes a valid URI/URL string and how to resolve a relative reference to obtain such a string. (I'm OK with the idea that there is a spec for processing strings into valid URI/URL structures, for use by HTML parsers and other applications that require similar functionality. I see that as the primary function of the new URL specification.) I have a concern (but I can't tell if it's actually a problem) that re-addressing the notion of relative references, which the new URL spec appears to do, will invalidate some of the assumptions in my code that are based on RFC 3986 (e.g. a relative reference being defined syntactically by the absence of a scheme, and use of the mechanism defined in RFC 3986 for resolving relative references to absolute.) Finally, I think there are other reasonable ways of getting from an input string to a valid URI/URL. In my current work, I'm making extensive use of CURIEs (http://www.w3.org/TR/curie/) in JSON, with the goal that these can be resolved to proper URIs according to the context in which they are used. So I have a problem with there being just "one true way" of resolving an intended resource reference to a URI/URL. In summary, it would, for me, be useful to keep the definition of a URI/URL separate from details of how to process URI/URL related artifacts. I would see this as a natural division of labour between RFC3986 and the new URL specification. #g -- PS: on the naming thing of URI vs URL: I have found there are places when it is useful to actually distinguish between URI and URL strings, even within a single data structure. The difference is based on context and intended use rather than any intrinsic difference in the values. Often, the URI and URL for a resource may be the same string, but sometimes I find a need to have different values (e.g. for identifier strings that I want to remain stable when a resource is moved to a different location).
Received on Saturday, 6 December 2014 09:08:13 UTC