[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ERROR verb





On 19 Jun 2004, at 15:01, Bill de hÓra wrote:


Robert Sayre wrote:
Quoting Asbjørn Ulsberg <asbjorn@xxxxxxxxxxxxxx>:
On Sat, 19 Jun 2004 12:15:53 +0200, Henry Story <henry.story@xxxxxxxxxxx>
wrote:



Could one not generalize and create an ERROR header in addition to GET,
POST

I've always liked this idea. Let's use a verb, though. Maybe "COMPLAIN".

ERR is the verb.

+1



Adding new method to HTTP (or any app protocol) is a last resort imo. Again imo, to ask for a new verb is to say both that the deployed web doesn't support Atom needs and that the deployed web should accept the burden of a new verb for Atom needs*. While we seem to be noodling on a couple of new verbs, are we at that point yet?

I agree. The problem is that we seem to have come across a problem not just for Atom, but for the whole XML community. I think I saw it mentioned as a problem in Fielding's PhD thesis on REST, but could not find the passage.


It would be interesting to make a PACE and get some feedback from the web community or XML community at large to see if perhaps someone is working on it, or has worked on it and found some good reasons not to pursue. Either of these could be very informative.

Note also that it does in fact do nearly the right thing with current web servers: they log an error message. Currently they "wrongly" log the error message that the method does not exist, but at least someone gets notified... So perhaps dissemination of this solution is not so difficult at all.

This should not be an Atom only thing. The problem is larger than Atom. But if it is why not tackle it, instead of inventing another ad hoc Atom only solution?

Henry


cheers Bill

* all aside from adoption questions, which are significant. Claim: any syndication technology that does not require new HTTP verbs has an immediate adoption advantage other one that does, irrespective of the engineering merits.