W3C home > Mailing lists > Public > public-html@w3.org > June 2015

Re: Proposed split of the HTML specification

From: Sam Ruby <rubys@intertwingly.net>
Date: Wed, 24 Jun 2015 07:25:59 -0400
Message-ID: <558A93C7.5030006@intertwingly.net>
To: Robin Berjon <robin@w3.org>
CC: "HTML WG (public-html@w3.org)" <public-html@w3.org>
On 06/24/2015 05:05 AM, Robin Berjon wrote:
> On 23/06/2015 17:50 , Sam Ruby wrote:
>> On 06/22/2015 05:30 PM, Travis Leithead wrote:
>>>
>>> . . . And then for implementers and the community, devise some sort
>>> of indexing service to provide equivalent functionality of WHATWG
>>> find on page and to indicate cross-link dependencies.
>>
>> Quick and dirty prototype: http://intertwingly.net/tmp/breakup/
>
> That's pretty cool! Any reason we couldn't include it directly in the
> root HTML specification?

Thanks!

I do believe it (or rather some future incarnation of it) belongs at (or 
prominently linked by) https://www.w3.org/TR/html/.

My preference is that the root HTML specification (I assume by this you 
are referring to http://darobin.github.io/breakup/specs/html/?) be 
separately and also linked from that location.  But there is no 
technical reason why it couldn't be included there, possibly in addition 
to be included on the /tr/ page itself.  Or even in every document, for 
that matter.

The same JSON files produced/used by this could also be used by the 
scripts to do WHATWG-style intra-, and inter-, document linking within 
the individual documents themselves.

  - - -

I'd also like to have a discussion as to what to link. Header elements 
that contain ids are an obvious candidate, two downsides being 
boilerplate headers (like "Normative references"), and duplication if 
the section containing the header contains definitions.

More subtle, how should I handle the use of code elements like the 
following?:

>  <ul><li><p>Authors can use the <code id="extensibility:classes">
>  <a href="/breakup/specs/html-attributes/#classes">class</a></code>
>  attribute to extend elements,

Ideally this would pick up not only the text 'class' but also the word 
'attribute' to put in the glossary.  This could be done with heuristics 
(words like attribute and element should be picked up, others 
discarded?), or with metadata (perhaps a new attribute that contains the 
text to be used in the glossary?)

Thoughts?

This code was written quickly and I can iterate quickly.  So don't be 
afraid to toss out ideas for me to try out.  Those that work out will be 
kept, others can be quickly discarded.

I'll also try to clean up my code and post it.  It currently is written 
entirely in Ruby (even the JavaScript is generated).  Once it 
stabilizes, it can either be left as is, or converted entirely to 
node.js.  (Before you vote for that option, wait until you see the Ruby 
source, it truly is more maintainable than the generated code, even 
though I took great pains to make the generated code look as if it were 
hand authored).

- Sam Ruby
Received on Wednesday, 24 June 2015 11:26:30 UTC

This archive was generated by hypermail 2.4.0 : Saturday, 9 October 2021 18:46:13 UTC