Re: Comments on draft spec

On Sun, 2007-02-11 at 22:01 +0100, Danny Ayers wrote:
> re. http://www.w3.org/2004/01/rdxh/spec
> Revision: 1.206
[...]

CVS change summary is...

Revision 1.208  2007/02/12 19:06:01  connolly
move protocol trace from appendix to main body

clarify "profile" to "metadata profile name"
move XHTML rules after examples

change R to N in some rules so that prose
matches mechanical rules (which have been tested)

wordsmith "any other programming language"
and "XPath data model under-determined" bits (tx Danny A.)

add todo on security considerations around multiplying requests


Details follow...

> 4. Using GRDDL with valid XHTML
> 
> I'd be tempted to shift the "Stated more formally:" green boxes past
> the first example - the informal paragraph is pretty confusing itself.

OK, I moved it after the both examples. The result might be choppy.
I'm not sure I have energy to wordsmith a transition just now.

> The second & third green boxes - defining "profile" & "typed link" -
> appear to be formalisations of part of a spec cited normatively, which
> seems back-to-front - is a comment needed?

Yes; I added one.

>  (HTML4 is pretty loose on
> those point ;-)
> 
> 5. GRDDL for HTML Profiles
> 
> (seems ok apart from an @@ already marked)
> 
> 6. Transformation Algorithms
> [[
> Developers of transformations should make available representations in
> widely-supported formats.
> ]]
> Doesn't sound right. Not sure why...
> 
> [[
> While javascript, C, or any other programming language technically
> expresses the relevant information, XSLT is specifically designed to
> express XML to XML transformations and has some good safety
> characteristics.
> ]]
> =>
> [[
> While technically Javascript, C, or virtually any other programming
> language could be used to express transformations for GRDDL, XSLT is
> specifically designed to express XML to XML transformations and has
> some good safety characteristics.
> ]]

"could" suggests that this is counter-factual. I used "may".


> If this is true, it should be bolder!

The WG decided it's true in our 31 Jan decision on the
faithful-infoset issue
http://www.w3.org/2004/01/rdxh/spec#faithful-infoset
->
http://lists.w3.org/Archives/Public/public-grddl-wg/2007Feb/att-0017/31-grddl-wg-minutes-edited.html#item02


> [[
> When an information resource is represented by an XML document, the
> corresponding XPath data model is somewhat under-determined, depending
> on, for example, whether an agent elaborates inclusions, parameter
> entities, fixed and default attributes, or checks digital signatures.
> ]]
> =>
> [[
> When an information resource is represented by an XML document, the
> corresponding XPath data model may not be fully determined, depending
> on, for example, whether an agent elaborates inclusions, parameter
> entities, fixed and default attributes, or checks digital signatures.
> ]]

OK.

> This, and the material that follows "Put another way..." tends toward
> the discursive/informative, should maybe be broken out into a separate
> section, following the more normative material.

I'm not inspired with anything better just now.

> 7. Security considerations
> 
> Seems ok, except the caching section isn't really relevant - break out
> to an appendix: "Notes for Implementors"??

It's a bit of a stretch, but better to be conservative;
Spreading expired data around is a security risk.

Plus, there's more to say; I just added a todo...

<p class="ed">@@TODO: dajobe points out that GRDDL turns a request for
info from one page into several requests. Connect this to HTTP proxy
security considerations.</p>

> 7. (should be 8.) The GRDDL Vocabulary
> 
> There's already an @@, looks like it's on DanC's plate - I assume an
> RDF/XML version will follow...

well, yes, an RDF/XML version is published as well as an XHTML
version for http://www.w3.org/2003/g/data-view 
But the RDF/XML is generated via XSLT.

Dave Beckett, Jeremy Carroll, and Henry and I are having a discussion
of how GRDDL's own profile interacts with the grddl-for-xhtml-profiles
mechanism. Resolving the @@ you note in section 5 will involve
thinking about this a bit more.


> 9. References
> 
> Can this section be moved /after/ the appendices? I nearly overlooked
> the content that follows it.

Hmm... the spec should indeed stand without
the appendices. I guess the protocol trace is too important
to be an appendix. How about making it a section
after security considerations:

* GRDDL Transformations
  * Security Considerations
  * Example: A GRDDL-aware Agent protocol trace
  * The GRDDL Vocabulary
  * References

> *** Appendix: A GRDDL-aware Agent protocol trace
> 
> I like it ;-)
> 
> Could a line be added saying which tools were used to create the trace?
> (/me wonders if DanC favours telnet as a browser)

OK...

<p>HTTP trace data was collected via <a
href="http://hathawaymix.org/Software/TCPWatch">TCPWatch</a> by Shane
Hathaway. For more details, see <a
href="http://www.w3.org/2001/sw/grddl-wg/td/testlist1#http_tracing">HTTP
tracing in the GRDDL test materials</a>.</p>

> *** Appendix: Transformations for Styling versus data extraction
> 
> Can't say I like this as it stands in this position, but can't think
> how to improve.
> 
> *** Appendix: Issues
> 
> -
> 
> *** Appendix: Implementation Experience: Test Cases, Software, and Services
> 
> Maybe move this together with any other related material into a "Note
> for Implementors" appendix
> 
> *** Acknowledgements and Change History
> 
> -
> 
> 
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Monday, 12 February 2007 19:08:05 UTC