Re: PROPOSAL: i24 Requiring Allow in 405 Responses

Henrik Nordstrom wrote:
> ...
> For security reasons the following may also be implied "Things may
> change in the future if my configuration or the resource state is
> changed to allow other methods or significant properties of your request
> changes making you gain more access rights than you have today.", but I
> am not convinced the specs really means the latter. But makes sense from
> ...

I don't think so. If the authorized principal lacks permissions, the 
right response is 403 (Forbidden) or 404 (Not found, if the server 
doesn't want to reveal whether the resource exists). I don't see why it 
would ever be 405, except if the author of the server didn't understand 
the difference between 403 and 405 (and I think this is something where 
our spec could be clearer).

> So here is my suggestion about resolving the contents of Accept:
> 
> Does it help to get readers more aligned if "Allow" is rewritten to
> include advertised in the definition?
> 
>      [...] lists the set of methods advertised as supported ...

Works for me. Would we still remove the SHOULD-level requirement for the 
client?

BR, Julian

Received on Tuesday, 25 March 2008 13:01:53 UTC