W3C home > Mailing lists > Public > public-html-data-tf@w3.org > October 2011

Re: Multiple itemtypes in microdata

From: Ian Hickson <ian@hixie.ch>
Date: Tue, 18 Oct 2011 18:37:44 +0000 (UTC)
To: Jeni Tennison <jeni@jenitennison.com>
cc: public-html-data-tf@w3.org
Message-ID: <Pine.LNX.4.64.1110181825010.21128@ps20323.dreamhostps.com>
On Tue, 18 Oct 2011, Jeni Tennison wrote:
> On 18 Oct 2011, at 07:29, Ian Hickson wrote:
> > These use cases aren't very compelling. It's hard for me to understand 
> > why we would need to mark up a single item as being two incompatible 
> > types when there's not really a compelling case for there even being 
> > one type.
> 
> I suspect that publishers are marking up their data in these ways 
> because it costs them very little to do so -- certainly less than 
> supporting a separate API -- and they want to provide information for 
> *potential* consumers, not just existing ones. It's an "if we build it 
> they will come" attitude.

We know from past experience on the Web that publishing data using a 
syntax, format, or feature that is not yet processed by software used by 
the author will lead to a "poisoned the well" situation. For example, 
longdesc="" is now an unusable feature in HTML because well-meaning but 
misguided authors tried to use it to make their pages accessible, and the 
vast majority of longdesc="" attributes today have useless values, because 
when the authors added them, they did not test them.

This may seem counter-intuitive to the data comunity, but we should be 
strongly discouraging authors from publishing data until the data is 
actively used. Otherwise, we will be making these technologies unusable.

This is why I focus only on concrete use cases with real software. I'm 
eager to address use cases that require multiple item types on a single 
item, but I do not intend to add features to support multiple item types 
when there is no actual software that will use this data.

(Incidentally, how long is it expected that it will take for software to 
come along to use this data? At what date do we determine if the 
experiment with publishing such data is a success or a failure? If people 
publish data for five years and no software comes, does that mean we just 
haven't waited long enough, or can it be taken as a sign that maybe we 
should be trying a different approach? Ten years? Twenty?)

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'
Received on Tuesday, 18 October 2011 18:41:49 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 18 October 2011 18:41:50 GMT