Thank you very much for your suggestions.
On 5/7/06, James Holderness <[EMAIL PROTECTED]> wrote: <snip>
but what if they just want to do something simple? They can't just include an atom:title element,"
</snip> Well, that case, you don't return response body at all. "HTTP/1.x 500 Internal Server Error" is nice and simple (with a Title).
I don't think atom:generator is valid as a child of atom:entry. You could use atom:author as the sole representation of the generator, but that seems like a bit of a hack. And you can't even leave out the author if you don't like the idea since it's a required element according to RFC4287.
oops, I'm sorry. fixed. <?xml version="1.0"> <entry xmlns="http://www.w3.org/2005/Atom";> <title>Internal Server Error</title> <author> <name>Foo Publisher 1.0</name> <uri>http://www.example.com/FooPublisher</uri> </author> <content type="text">couldn't connect to the DB server.</content> <id>urn:uuid:1225c695-cfb8-4ebb-aaaa-80da344efa6a</id> <updated>2003-12-13T18:30:02Z</updated> </entry> I think this is simple enough.. and better than invent yet another new format.
> <id>urn:uuid:1225c695-cfb8-4ebb-aaaa-80da344efa6a</id> > <updated>2003-12-13T18:30:02Z</updated> What purpose do these serve? Seems to me that they're also just there because RFC4287 requires them to be.
I could keep an error log locally just like other atom entries, and I could use atom:id as a log file name or primary key for an error. (note: I don' use RDBMS so my APP client depends on atom:id everywhere.) Then the users(who don't know anything about..anything) could send the error log file directly to me(a client developer) or a server developer. And <updated>? well a server dev could make use of this datetime along with the Apache server log.
Bottom line, there's way too much unnecessary cruft being forced into the response because of RFC4287. If it was possible to ignore those requirements, toss out a couple of elements, and make several others optional, then maybe you'd have something. I very much doubt the WG would approve that though. Regards James
Thank you really. I kinda expected this. That's why it is "SHOULD" not "MUST". I'm fine with "RECOMMENDED" as well. "it MAY also return a response body, which, if provided, SHOULD be an Atom Entry Document" best -Toru
