RE: (ACCEPT*) Consensus

I agree with Roy.

That said, isn't "require" a little strong? I seem to recall that it was
strongly suggested that a text/plain or test/html document be returned
with all error responses, with info about what went wrong, and (if
appropriate) links to click on, etc, to help the user rectify the
situation. The spec for 406 adds more detail about what should be in the
document when the response is 406.

After all, while it might be rude to not return such information,
nothing much will break if a server didn't. I wouldn't buy such a
server, but that's exactly what forces the implementors to do a better
job.

>----------
>From: 	Roy T. Fielding[SMTP:fielding@avron.ICS.UCI.EDU]
>Sent: 	Sunday, April 14, 1996 3:10 PM
>To: 	Koen Holtman
>Cc: 	http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
>Subject: 	Re: (ACCEPT*) Consensus 
>
>> The old 1.1-01 draft required 406 responses to contain `a list of
>> resource characteristics and locations from which the user or user
>> agent can choose the one most appropriate.'  Such a thing is not
>> required in the 416 response I defined, so I thought it best to assign
>> a new code, rather than rewrite the 406 text.
>
>That would be incorrect -- the 416 response would still require some
>message in the response as a fall-back to older clients (or simply
>as more information).
>
>What you describe is exactly what 406 is intended to do, so use it.
>Rewrite whatever text no longer applies, but do not introduce a new
>error code.  There do not exist any implementations that depend on
>draft 01's description of 406 (not that it would matter, since the
>description is essentially the same even after your changes).
>
> ...Roy T. Fielding
>    Department of Information & Computer Science   
>(fielding@ics.uci.edu)
>    University of California, Irvine, CA 92717-3425   
>fax:+1(714)824-4056
>    http://www.ics.uci.edu/~fielding/
>
>

Received on Monday, 15 April 1996 11:56:53 UTC