Re: Scope question

On Wed, May 5, 2010 at 10:45 AM, Phillips, Addison <addison@lab126.com> wrote:
> Hello Adam (and WG),
>
> This is my first post after becoming, with Marc, the co-chair, so I welcome you all and hope that we can make rapid and useful progress together.
>
>> Would a document describing these decisions be of interset to this
>> working group?  If so, how would that document relate to RFC 3986
>> and RFC 3987?  Would the working group be interested in such a document
>> in an informative or a normative voice?
>
> (as co-chair)
>
> The charter of this group does indeed include creating a document or documents that describe IRI syntax, parsing and comparison (to supersede RFC 3987). One of the key things this WG will need to consider is a test suite to validate processing of IRIs as defined by our documents.

I guess part of my question stems from not understanding the
distinction between RFC 3986 and RFC 3987 as well as between URLs,
URIs, IRIs, and the various other things with similar names.

>> In the course of this work, I'm going to have make some
>> exciting decisions like whether the URL processing pipeline should
>> covert schemes to lower case or whether the a # character in a
>> fragment should be represented using the # character or using the
>> characters %23.
>
> If there are differences in the handling by various vendors, documenting them could be useful in deciding what normative behavior should be. Documenting processing behaviors is also useful, but in terms of what this WG formally produces, it probably needs to live within the bounds of what is (proposed to be) normative in IRIbis (e.g. RFC 3986 Section 3.1 already says how case should be handled in scheme; the implementation details for how that is achieved are not as important as the conformance test cases themselves). To be in scope to this WG's effort, I believe your document would need to focus on normative behaviors defined by our documents. Is this what you had in mind?

RFC 3986 Section 3.1 is helpful w.r.t. the casing of the scheme.
However, it's not as clear as it could be.  For example, it says:

"documents that specify schemes must do so with lowercase letters"

It's unclear whether that's a requirement for folks who produce
documents or for folks who consume documents.  Later it says:

"An implementation should accept uppercase letters as equivalent to
lowercase in scheme names"

Leading me to believe the first requirement is for folks who produce
documents, assuming "implementation" above refers to document
consumers.

As I read the charter, we're not supposed to address issues in RFC
3986, which might place this document out of scope depending on the
division of responsibilities between RFC 3986 and RFC 3987.

Adam

Received on Wednesday, 5 May 2010 18:12:53 UTC