Herbert van de Sompel wrote:
> if <content src="xxx" /> is used then the
> atom metadata elements pertain to xxx
> instead of to the Atom Entry (entry/id).
> As a matter of fact, when looking at
> existing Atom use of src="xxx" there are
> cases in which the atom metadata pertains
> to xxx and there are cases in which it
> pertains to the Atom Entry (/entry/id).
In Atom the metadata always pertains to the entry it is contained in and not
anything else. Atom:title always is the title of the entry that contains it.
The difference between inline and out-of-line content is purely syntactic;
except for XML documents that have a DOCTYPE, you can encode the same entry
with inline or out-of-line content and the meaning doesn't change.
I do agree that there are some counter-intuitive situations. For example, if
the content/@src links to a PDF document, then the PDF document will often
have its own metadata (especially the title, authorship, and modification
time) encoded in it. Most applications want to think of the atom:title as
being the title of the PDF document. But, that isn't the case; the
atom:title is always the title of the entry whose content is the PDF file.
There is no fixed relationship between the metadata for the entry and the
metadata for the entry's content.
Here is an example where we avoid using out-of-line content; notice that the
entry and its content have different titles, because atom:entry/atom:title
describes the entry's title, not the entry's content's title:
<entry ...>
<title>Title of the Entry</title>
<content type='application/xhtml+xml'>
<html:html ...>
<html:head><html:title>Title of the HTML
document</html:title></html:head>
...
</html>
</content>
</entry>
- Brian
