On May 11, 2007, at 3:15 PM, John Allsopp wrote:
(abbr pattern problems,
Clearly, there's a need for markup for the generic pattern of marking up a triple of data presented to humans, the microformat class and a normalized easy-to-parse representation of the data. HTML5 <time> addresses only one instance of this pattern.
The problem with using <abbr> for this pattern is that title='' is intended to be human-readable and the pattern contaminates abbreviation data, so with microformats <abbr> is now less useful for e.g. non-microformat-aware but abbr-aware screen readers.
The question that needs to be asked is: Will microformat producers and consumers be willing to migrate to a replacement of the abbr pattern if one is provided or will they continue to use abbr anyway for backwards compat?
For example, should HTML 5 define a uf-data='' attribute as a common attribute such that the value of this attribute would be considered in preference over textContent by microformat consumers? Or should HTML 5 just mitigate the damage to the title attribute by defining a boolean attribute title-is-uf='' for flagging title='' attributes not meant for human consumption?
Is it too late to get rid of this? <abbr title='uf data'>human data</abbr> Would this be accepted by the uf community? <span uf-data='uf data'>human data</span> If not, would this be backwards-compatible with uf consumers? <span title='uf data' title-is-uf>human data</span>
even with ufs no one uses profiles so HTML 5 should get rid of them ...)
We already got rid of profile='' before the W3C adopted the draft. There may be some pressure to put it back due to theoretical considerations. This is part of the Descriptivist vs. Prescriptivist debate. It looks pretty obvious that microformat consumers experience more practical benefit when they ignore profile=''.
-- Henri Sivonen [EMAIL PROTECTED] http://hsivonen.iki.fi/ _______________________________________________ microformats-discuss mailing list [email protected] http://microformats.org/mailman/listinfo/microformats-discuss
