Re: META

Terry Allen (terry@ora.com)
Tue, 4 Jul 95 10:17:36 EDT

Roy:
| Tom Magliery said a while back:
|
| >The unfortunate problem with META is that it's currently defined as an
| >empty element. I certainly have no problem with changing it, but isn't
| >that the sort of thing that throws people into fits about "breaking
| >existing documents" and so on?
|
| People like me. META was defined as an empty element specifically
| because certain browsers were broken in their presentation of element
| content within the HEAD of a document. If that were not the case,
| than we would not have needed META, since it is much more SGML-friendly
| to define a separate element for each significant type of metainformation.
|
| The reason why we have used
| <META http-equiv="owner" content="Roy">
| instead of
| <Owner http-equiv>Roy</Owner>
| is only because pre-2.5 Mosaic and (last time I checked) Netscape
| would render "Roy" in the latter case and not in the former.

Not quite right. As Ron has remarked many times without
contradiction, you can't anticipate all the possible metadata elements
so as to get them into the DTD in the first place. In order to
fit with your argument your second example should be

<META http-equiv="owner">Roy</META>

| META exists as a halfway-house for unrespected metainformation
| (a way for authors to define metainformation without changing the DTD
| and without impacting older user agents). Respected metainformation
| (as defined by future versions of HTML) should be defined with their
| own elements.

Unfortunately that won't work, because metadata is potentially
too variable. There's nothing wrong with the META tag as a method
of semantic labelling. Future version of HTML need not define
new elements for it.

The question is really whether we are stuck forever with certain broken
behaviors of the most popular present browsers, or whether we can
modify HTML to make it more flexible by allowing content in tags in
HEAD; not whether we can define all the needed elements for meta.
We can't do the latter; it is not apparent yet from discussion whether
we can do the former.

Regards,

-- 
 Terry Allen   (terry@ora.com)