Skip to content

Commit

Permalink
removed remaining issues now resolved or closed
Browse files Browse the repository at this point in the history
  • Loading branch information
Nicholas Car committed Sep 25, 2019
1 parent 5d213a1 commit 00ac53f
Showing 1 changed file with 3 additions and 55 deletions.
58 changes: 3 additions & 55 deletions conneg-by-ap/index.html
Original file line number Diff line number Diff line change
Expand Up @@ -512,50 +512,6 @@ <h3>Profile Identification</h3>
MUST identify the <a>resource</a> by a Uniform Resource Identifier (URI) [[RFC3986]] and MUST
identify the <a>profile</a> either by a URI or a <a>token</a>.
</p>
<div class="issue" data-number="290">
<p>The use of tokens for profile identification
was a controversially discussed topic within the working group
(cf. discussion in <a href="https://github.com/w3c/dxwg/issues/290">#290</a>).</p>
<p>The proponents of tokens stressed the value of short mnemonic identifiers
for human actors using web browsers.
Since browsers usually do not easily allow request headers to be set or modified
the easiest way for human actors to negotiate for profiles
(or any other content dimension)
is to use Query String Arguments (QSAs).
When using QSAs, the profile identifier is put in the query part of the request URI
and since all content in the query part needs to be URL-encoded
this requires that several characters that are common in URIs
(e. g. '<code>:</code>' and '<code>/</code>')
be percent-encoded which is an error-prone process if done manually.
Since tokens are usually restricted to the character group <code>[A-Za-z0-9]</code>,
they have not only the advantage of being shorter than most URIs,
but also that they can be used in the query part of the request URI
without the need to percent-encode any of its characters.
Further, there are several systems already deployed
that build on standards that use tokens for profile identification.</p>
<p>The opponents of tokens questioned the use of tokens from two main angles.
One argument was that while URIs are universally unique,
tokens will introduce non-uniqueness challenges
since there is no universal registry for token/URI mappings
(nor is it likely that there will be one).
The proposed approach that servers publish token/URI mappings for the resources they serve
was seen as unsatisfactory since the scope of that mapping would not be clear:
Would it be valid for a domain, a sub-domain or only for the requested resource?
Following on that, the benefits of such a mapping was generally questioned
on the grounds that machine-driven clients would likely use
http content negotiation with the Accept-Profile and Content-Profile headers
and full profile URIs instead of tokens.
Also, since tokens are not unique,
a client would either need to know a server’s URI/token mappings <em>before</em> using the token
(in which case it could use the URI instead)
or it would need to look it up through the URI/token mapping
(in which case it would need to know the profile URI).
In each case, the client would need to know the profile URI,
which means that tokens do not make client and/or server implementation easier
since in order to conform to the specification
they need to support not only profile URIs but profile tokens, too.</p>
<p>[When this #290 has been resolved, this issue should be turned into a note also saying what the resolution was]</p>
</div>
<p>
If a URI is used for profile identification, it SHOULD be an HTTP
URI that dereferences to a description of the profile. Other kinds or URIs, e. g. URNs MAY also be used and these may no be capable of dereferencing.
Expand Down Expand Up @@ -883,21 +839,14 @@ <h3>Hypertext Transfer Protocol Headers</h3>
<code>Accept-Profile</code> and <code>Content-Profile</code> that are to be defined in an upcoming
Internet-Draft [[PROF-IETF]].
</p>
<div class="note" title="Feature at risk">
<p>
The use of HTTP headers to transport information
about which profiles the response message's content conforms to
is <strong>at risk</strong> pending the IETF standardisation work.
</p>
</div><div class="issue" data-number="678"></div>
<div class="note" title="Issue 678: POST/PUT out of scope">
<div class="note" title="POST/PUT out of scope">
<p>
POST &amp; PUT methods are considered out-of-scope for this Functional Profile. It is not clear what functions of
the proposed Abstract Model these HTTP methods would implement.
</p>
<p>
<code>Content-Profile</code> headers may be specified in POST &amp; PUT request to communicate the conformance of a
request message's content but the Working Group has no specified requirements for this.
<code>Content-Profile</code> headers may be specified in POST &amp; PUT requests to communicate the conformance of
a request message's content.
</p>
</div>
<p>
Expand Down Expand Up @@ -2106,7 +2055,6 @@ <h3>Alternate Representations Data Model - Details</h3>
</p>
<section id="altr-implementations">
<h4>Implementations</h4>
<div class="issue" data-number="1041"></div>
<p>
A Web Ontology Language [[OWL2-OVERVIEW]] implementation of the data model in <a href="#altr"></a> is given
below in <a href="#altr-owl">Code Listing 3</a>. It is also available in the file
Expand Down

0 comments on commit 00ac53f

Please sign in to comment.