W3C home > Mailing lists > Public > whatwg@whatwg.org > August 2010

[whatwg] Expanding the cite element

From: Ian Hickson <ian@hixie.ch>
Date: Tue, 3 Aug 2010 02:09:53 +0000 (UTC)
Message-ID: <Pine.LNX.4.64.1008030157360.7470@ps20323.dreamhostps.com>
On Wed, 5 May 2010, Simpson, Grant Leyton wrote:
>
> Is there any value in adding an "href" or "uri" or similar attribute to 
> the <cite> element to indicate a location for a work (or information 
> about the work) or, in the case of a URI, an indicator that can be used 
> as a reference programmatically?

I don't know; is there? What would the use case be?


> <q> has a "cite" attribute, so it seems to me that if we have a place to 
> link to further information in <q> it makes sense to do so in <cite>.

To be honest, <q cite> hasn't been very successful. One could argue it is 
a failed experiment. :-) If it wasn't more or less harmless, I'd probably 
be arguing to remove it.


> If it were a URI (and therefore not necessarily retrievable), it would 
> help in cases where the same work gets referenced in slightly different 
> ways:
> 
> <p>As Ashley Crandall Amos says in <cite 
> uri="http://example.com/books/crandall/linguisticmeans">Linguistic Means 
> of Determining the Dates of Old English Literary Texts</cite> ... Amos 
> also mentions in <cite 
> uri="http://example.com/books/crandall/linguisticmeans">Linguistic 
> Means</cite></p>

Works tend to have as many different URIs to reference them as titles, 
sometimes more. :-)


On Thu, 6 May 2010, Edward O'Connor wrote:
> 
> <cite uri> has a much worse fallback story than simply embedding a link 
> in <cite>.
> 
> > <p>As Ashley Crandall Amos says in <cite 
> > uri="http://example.com/books/crandall/linguisticmeans">Linguistic 
> > Means of Determining the Dates of Old English Literary Texts</cite> 
> > ... Amos also mentions in <cite 
> > uri="http://example.com/books/crandall/linguisticmeans">Linguistic 
> > Means</cite></p>
> 
> Consider how the above would work in legacy browsers, and then consider 
> how this would work in them:
> 
> <p>As Ashley Crandall Amos says in <cite><a 
> href="http://example.com/books/crandall/linguisticmeans">Linguistic 
> Means of Determining the Dates of Old English Literary Texts</a></cite> 
> ... Amos also mentions in <cite><a 
> href="http://example.com/books/crandall/linguisticmeans">Linguistic 
> Means</a></cite></p>

Using <a> does seem like it would solve this neatly without additional 
effort.


On Thu, 6 May 2010, Schalk Neethling wrote:
>
> I guess this is the 'Paving the cowpaths' situation. Is there a 
> 'standard' where most developers do as in the sample by Ted?

That's a good question; have people found this useful enough to work 
around the lack of an attribute here? If not, maybe there's no problem to 
solve in the first place.


On Fri, 7 May 2010, Simpson, Grant Leyton wrote:
> 
> 1. Referencing something in the href attributed of an <a> tag implies 
> that the URI will resolve to a URL, that is, that it will be accessible 
> on the web at that address. Not every URI is a URL, though. That's what 
> I was trying to do with a "uri" attribute for the <cite> tag is to 
> identify the resource, not necessarily link to it.

Any link could be resolved; for example the browser could have a handler 
for isbn: links that looks up the book on Amazon or Google Books.


> 2. We would have to formally define what <a> within <cite> means, 
> otherwise we would leave the pairing up for interpretation.

Until we know that this is a problem people really want solving, that's 
probably ok.


> 3. Are there instances where tags that can be used separately take on a 
> different meaning in relation to one another?  I know what <li> means in 
> relation to <ol> and <ul>, but then again, I can't really use <li> 
> outside of either of those two.

<a> wouldn't take on a different meaning, it would just add additional 
information to the page.


On Sat, 8 May 2010, Benjamin Hawkes-Lewis wrote:
>
> I'm not opposed to adding @cite to <cite> but note that when you are 
> identifying a resource rather than linking to a resource, you could use 
> microdata or RDFa.
> 
> For example:
> 
>    http://dev.w3.org/html5/md/#global-identifiers-for-items
> 
>    http://rdfa.info/wiki/Rdfa-microdata-markup-comparison#Book_markup_with_ISBN_and_description

Indeed, that may be a better solution in general.


> > 2. We would have to formally define what <a> within <cite> means, 
> > otherwise we would leave the pairing up for interpretation.
> 
> You stated that you want to "indicate a location for a work (or 
> information about the work)".
> 
> A hyperlink indicates the location of an item or information about an 
> item, and "a href" creates a hyperlink.
> 
> What could be the other interpretations of <cite><a href="...">Work 
> title</a></cite> or <a href="..."><cite>Work title</cite></a> other than 
> that the hyperlink locates the work or information about the work?

Indeed.


> > 3. Are there instances where tags that can be used separately take on 
> > a different meaning in relation to one another?  I know what <li> 
> > means in relation to <ol> and <ul>, but then again, I can't really use 
> > <li> outside of either of those two.
> 
> I think the combination of "cite" and "a" to indicate a work title and 
> the location of the work or information about the work does not involve 
> changing the meaning of either "cite" or "a". This is the markup 
> equivalent of:
> 
>    http://en.wikipedia.org/wiki/Object_composition

Right.


On Mon, 10 May 2010, Simpson, Grant Leyton wrote:
>
> I was unaware of the Microdata spec.  Now that I have seen it, I think 
> it offers a lot of power and flexibility.  I think it should adequately 
> cover the use case I was thinking of.

Excellent!


> I'm in favor of adding a non-normative note to the section of the HTML5 
> spec that discusses <cite> that demonstrates how Microdata or RDFa could 
> be used for this purpose.  There will likely be other people like me who 
> read the <cite> section of the spec and think "What? I can't actually 
> make the citation point to something?"

Do you have a suggested example of how you've ended up using this? I 
prefer to use real-world examples where possible for this kind of thing.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'
Received on Monday, 2 August 2010 19:09:53 UTC

This archive was generated by hypermail 2.4.0 : Wednesday, 22 January 2020 16:59:25 UTC