W3C home > Mailing lists > Public > www-validator@w3.org > December 2012

Re: HTML5 rel=publisher?

From: Jukka K. Korpela <jkorpela@cs.tut.fi>
Date: Sun, 30 Dec 2012 13:16:58 +0200
Message-ID: <50E022AA.1020806@cs.tut.fi>
To: Thomas Spear <speeddymon@gmail.com>
CC: www-validator@w3.org
2012-12-30 2:00, Thomas Spear wrote:

> I have an HTML5 site
> I'm writing which I am linking to an accompanying Google+ page. One of
> the recommendations for the site in terms of how it appears in search
> engine results, was to use rel=publisher -- I am getting an error that
> it is not a valid IRI, so I checked through the microformats wike.

The error message is rather confusing: "Bad value publisher for 
attribute rel on element a: Not an absolute IRI. The string publisher is 
not a registered keyword or absolute URL." The idea behind this is that 
by HTML5 drafts and WHATWG HTML Living Standard pages (it has not been 
disclosed which of these is what the validator tries to check against), 
a rel attribute value component must be a keyword from a mutable set of 
allowed keywords, but for some reason, the validator reflects the idea 
that it could also be an absolute URL (called "absolute IRI" first).

So when the validator encounters a name there that it does not 
recognize, it tries to parse it as an absolute URL.

The "microformats wiki" says: "Changes to this registry may not be 
reflected in validators in real time." This should be read as saying 
that the validator uses some version of the document, and there is no 
direct way of knowing which.

The validator does not accept rel=publisher, despite its being in that 
document, simply because it uses some old version of it. As far as I can 
see, rel=publisher was added 2 October, 2012.

> Unless I'm reading this wrong, rel=publisher was dropped for _html4
> (html version four)_,

No, it was never in HTML 4. On the other hand, HTML 4 specifications 
allow any keywords there; they only describe a set of "recognized" 
keywords (most of which have actually been ignored by most software), 
without specifying any particular allowed set:
http://www.w3.org/TR/REC-html40/types.html#type-links
The list does not contain rel=publisher, but that's irrelevant in 
validation, since validation against HTML 4 accepts any string as rel value.

> however it appears to be in proposed status for
> html5.

It's in "proposed" status in the wiki page, yes. Anyone and his brother 
can propose anything there, and some people can move proposals to 
abandoned state.

> Is there anything I can do to work around that so it validates
> cleanly, without having to remove the rel=publisher bit?

No, apart from waiting for the validator to catch or switching to HTML 4.

Yucca
Received on Sunday, 30 December 2012 11:17:30 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 30 December 2012 11:17:33 GMT