Re: a few URI/href issues captured with test cases

On Thu, 2009-05-21 at 13:33 -0400, John Kemp wrote:
> After reading what I've written, my general feedback seems to be that  
> your examples are interesting, appear that they might be relevant, but  
> could probably be better placed into some context. I've attempted to  
> provide (hopefully not too oversimplified) in this email the context  
> in which I feel your examples make sense. Does that make sense to you?

yes...

> Summarizing my (basic) understanding:
> 
> * Links are good, and a basic feature of the Web
> * However, links are used in different contexts (for example, a link  
> in an HTML href is then used to make an HTTP request, specifically in  
> the case of an HTML form submission)
> ... and the characters, character set and encoding used in one context  
> may not be appropriate in another context
> * Some specifications defer to 3986 for URI encoding rules. 3986  
> defers to scheme specifications in particular with regard to "reserved  
> characters". None of the relevant specifications say anything about  
> the use of IRIs in links (correct?)

Well, none of the specs that I cited so far...


> Your examples appear to indicate:
> 
> i) That a space is not allowed in the path component of an HTTP  
> request, but a space should be escaped as %20 in HTML, as specified by  
> RFC3986

That doesn't sound quite right. The point is space works OK in
an href attribute but not on a GET request line, so the %20
trick makes up the difference.

> ii) That a colon in the path creates a link which is not useful  
> outside of the context of the document within which it appears (at  
> least, I _think_ that's what you mean here?)

no. I'll try in another draft.

> iii) That URIs only allow US ASCII characters per RFC3986

Yes.

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
gpg D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Thursday, 21 May 2009 18:27:40 UTC