Re: New Version Notification for draft-tbray-http-legally-restricted-status-00.txt

In an ideal world, I'd start a new 600-level status series, for "3rd 
party", but strongly suspect that's infeasible due to impact on existing 
implementations. A 400-level code (451 or otherwise) is probably the 
most practical way of doing this.

Ross

On 13/06/2012 16:16, Tim Bray wrote:
> Yeah, if you look at RFC2616 sections 10.4 (4xx) and 10.5 (5xx), you 
> can make a good case that it doesn’t belong in either.   I just don’t 
> believe you can make a bullet-proof argument for either range.
>
> The reasons I currently favor 4xx are:
>
> - Lots of client software simply tries to sweep 5xx errors under the 
> carpet (“Something went terribly wrong upstream, don’t bother your 
> pretty little head”)
> - I liked the value 451
>
> -T
>
> On Tue, Jun 12, 2012 at 10:35 PM, Roy T. Fielding <fielding@gbiv.com 
> <mailto:fielding@gbiv.com>> wrote:
>
>     On Jun 12, 2012, at 9:34 AM, Tim Bray wrote:
>
>     > Aaaaaaaand, it turns out MNot was right; I checked with an
>     expert, and 451 is heavily used for “redirect” in the Msft
>     ecosystem, notably including HotMail’s hundreds of millions of
>     users.  Consider it “4xx” (which I would still argue for as
>     opposed to 5xx).  -T
>
>     4xx indicates an error by the user or user agent.  I don't see
>     any reason (aside from literary) that would justify using a 4xx
>     code for this.  5xx is typically used for non-authoritative
>     responses or server-imposed limitations -- a status that might
>     be different if the user agent chose a different intermediary
>     or tried again later.  Hence, 5xx makes more sense here.
>
>     ....Roy
>
>