W3C home > Mailing lists > Public > www-validator@w3.org > March 2013

Re: error: Bad value prettyPhoto for attribute rel on element a - bushurenprijs.nl

From: Jukka K. Korpela <jkorpela@cs.tut.fi>
Date: Fri, 15 Mar 2013 11:59:22 +0200
Message-ID: <5142F0FA.2060404@cs.tut.fi>
To: Philip TAYLOR <P.Taylor@Rhul.Ac.Uk>
CC: Astrid de Vries <info@bushurenprijs.nl>, www-validator@w3.org
2013-03-12 15:39, Philip TAYLOR wrote:

> Astrid de Vries wrote:
>> validating: http://www.bushurenprijs.nl
>>
>> Error:
>>
>> Line 345, Column 65: Bad value prettyPhoto for attribute rel on element
>> a: Not an absolute IRI. The string prettyphoto is not a registered
>> keyword or absolute URL.
>> …="links"><a href="#hoewerkthet" rel="prettyPhoto">&raquo; Meer
>> uitleg</a><br />
>>
>> Can anyone help me?
>
> Er, replace "rel=" with a more appropriate attribute ?
> Given that "rel" is intended to indicate the relationship
> of the linked resource to the document from which it is
> linked, it is very unclear to me what sort of relationship
> is intended to be inferred from "prettyPhoto".

The page does not currently contain any code like the one mentioned in 
the error message excerpt. Presumably the page has been considerably edited.

Anyway, the problem with rel="prettyPhoto", from the perspective of 
HTML5 CR and relatives, is that there is no registered keyword 
prettyPhoto. Unlike previous HTML specs, HTML5 restricts rel attribute 
values to a set of keywords (or space-separated combinations of them):
http://www.w3.org/TR/html5/links.html#linkTypes

New keywords can in principle be registered as indicated in the CR. 
However, changes to the registry have not been reflected in the 
validator except perhaps after a long delay.

There is no requirement on mnemonicity on the keyword names. That is, 
the name need not be descriptive in any way. It is the registered 
definition that is supposed to explain the meaning, not the name. For 
practical reason, suggestive names are preferable, of course.

The odd thing here is that the validator says
"Not an absolute IRI. The string prettyphoto is not a registered keyword 
or absolute URL."
Apart from the abbreviation "IRI" being cryptic to most people, this 
sounds very confusing, as if it suggested that in addition to defined 
keywords, any URL could be used as a token in a rel attribute value. 
This might reflect some older draft or idea, but I cannot see anything 
about this in the HTML5 CR or in the WHATWG Living HTML du jour.

Validator.nu (which is expected to be better up-to-date) is even more 
misleading. It says:

Error: Bad value [name used] for attribute rel on element a: Not an 
absolute IRI. The string [name used] is not a registered keyword or 
absolute URL.

[line number and column information]

[source code extract]

Syntax of absolute IRI:
     An absolute URL. For example: http://example.org/hello, but not 
/hello. Spaces should be escaped as %20.

So it does not even refer to the wiki registry.

(I tried to report this bug at Bugzilla, but it's rather slow and buggy, 
with Gateway Time-outs etc.)

Yucca
Received on Friday, 15 March 2013 10:04:12 GMT

This archive was generated by hypermail 2.3.1 : Friday, 15 March 2013 10:04:24 GMT