> FWIW, that certainly sits fine with me, but it might be possible to
> consider more general schemes. So, for the sake of concreteness, I'll
> propose a "purpose" attribute, whose value is a string defining the purpose
> (or purposes) of this particular piece of meta-info. A few things which it
> might be reasonable to have in a document are:
>
> <meta purpose="header" name="Expires" value="Tue, 04 Dec 1993...">
> <meta purpose="index" name="Type" value="service">
>
> and even
>
> <meta purpose="index,header" name="Keywords"
> value="metaindex, resource discovery, DTD syntax">
I'd have to disagree here. There could be hundreds of different
purposes for a given bit of metainfo, most of which the author would
not be aware of. Having a HEADER option makes sense for reasons of
efficiency; the others would just make it more convenient for a particular
tool. The META name should be sufficient to define the content-type of
the metainfo and, once that is known, any tool should be able to decide
on its own as to whether or not that metainfo is useful to their purpose.
Once the metainfo is in the document (where any grubby client can get
its hands on it), the author has no control over how it is used, nor
should they care.
...Roy Fielding ICS Grad Student, University of California, Irvine USA
(fielding@ics.uci.edu)
<A HREF="http://www.ics.uci.edu/dir/grad/Software/fielding">About Roy</A>