Re: Charset issue from issue MISTAKES
Koen Holtman (koen@win.tue.nl)
Tue, 28 Jul 1998 21:19:30 +0200 (MET DST)
Larry Masinter:
>
>Koen,
>
>In http://www.ics.uci.edu/pub/ietf/http/hypermail/1998q1/0520.html
>
>you wrote about the removal of the sentence:
> The ISO-8859-1 character set can be assumed to be acceptable to all user
> agents.
>
>arguing that it was a mistake, but that you couldn't "remember that there was
>any rationale or discussion on the list for deleting it."
>
>But in http://www.findmail.com/list/http-wg/7596.html (November 26,1997)
>
>you said, in response to a suggestion to remove this sentence:
>
>> I agree, this sencence should be removed.
>
>I think the earlier rationale is still valid, and that section 14.2
>does not require any additional edits.
Hi Larry,
Thanks for refreshing my memory in this rather spectacular way. I am
apparently capable of reaching two contradictory conclusions on
whether the removed sentence is redundant or not. What it boils down
to is an interpretation issue: is this sentence just defining Accept
header semantics, or does it also say something about user agents in
general? It it says something in general, one could infer from it
that ISO-8859-1 can be used as a default charset in the entity bodies
of error responses. Having such a default is clearly a good thing, as
is for example implied by the recent `charset policy' RFC. In
reviewing that RFC however I see that it does not _require_ HTTP to
specify a default acceptable charset.
So all in all I now think that it is safe for the sentence to be left
out. I would prefer to replace it with some clear text on default
charsets and languages for error response bodies, but it is probably
too late for that.
Koen.