User:Billinghurst/doc

DNB and Persondata
You might want to amplify what the project pages say about Persondata. So far this hasn't seemingly entered the consciousness of participants in a serious way. Just occurred to me as I was working over what was posted in the project's early days. Charles Matthews (talk) 08:50, 14 November 2009 (UTC)
 * I cannot say that I am fully around what is happening with metadata. I will see what I can find.  That really might be a good question for Magnus when you meet him. billinghurst (talk) 09:59, 14 November 2009 (UTC)
 * Spoke to Pathoschild and he said Persondata is ugly, and we should be able to do it better. I will let him add his own comments.  He said that Magnus probably would have some good input into this. billinghurst (talk) 10:20, 14 November 2009 (UTC)


 * Metadata was one of the main motivations for creating standard and  fields, with explicit text-only fields. Implementing metadata sitewide for works and authors should be fairly easy once we decide on a good format.


 * The approach taken by w:Template:Persondata is to add an HTML table with cells identified by classes. This will work now without any software changes, but it's not an elegant solution; it depends on CSS to hide it from users, confuses screen-readers, presents data in a one-dimensional format that works well for indexes but little else, is difficult to extend, and cannot contain metadata about metadata. An example of this format on Wikisource, with some simplifications for machine-only parsing, would be:


 * An idea I discussed with Billinghurst is to have XML data tucked into a CDATA comment. This is ignored by browsers and screen-readers, is very easy to machine-parse, can contain multidimensional data, and can contain any data (even images, if we really wanted to). The example below presents the same data (with added metadata), but is 17% shorter. MediaWiki strips comments before outputting to HTML, but a very simple extension could add a  or   tag (and there would be no obstacle to implementing it, since there should be no performance issues).


 * Either format could easily be output by and, although the Persondata-style table would appear at the top of the page and confuse screen-readers even more than it does on Wikipedia (that is why  is placed at the bottom of the article there). Whichever format we choose, we can set up an API that extracts the data from the page and displays it in any of various formats. — Pathoschild 11:49:21, 14 November 2009 (UTC)


 * For my sake, a link to CDATA billinghurst (talk)


 * I spoke to Duesentrieb, who says he's working on getting HTML5 microdata or RDFa into MediaWiki before the next release. Both would allow us to easily mark up the header, author, and license templates in a standard, machine-readable way. HTML5 microdata in particular looks suited to our use. Either of these would be a more ideal solution than a custom metadata format, if they're really coming. — Pathoschild 23:55:01, 15 November 2009 (UTC)


 * Excellent news P/child. Anything that a nonce like me should be reading at this point?  Or just leave it until we have a better idea of the direction?  billinghurst (talk) 02:08, 16 November 2009 (UTC)