Re: [idn] Requirements: partial canonicalization

[Paul Hoffman / IMC <phoffman@imc.org>]

At 10:49 AM 2/2/00 -0800, ned.freed@innosoft.com wrote:
>>The last paragraph of 2.4 reads:
>
>> >If the protocol specifies a canonicalisation algorithm, a caching
>> >server should perform correctly regardless of how much (or how little)
>> >of that algorithm it has implemented.
>
>>This makes no sense to me. It forces caching servers to check the amount of
>>canonicalization performed by the requester, which pretty much allows
>>requesters to not bother to canonicalize. Thus, if forces caching servers
>>to do much more work than they do now. I propose we remove that paragraph.
>
>I agree that the paragraph is unclear, but the requirement is nevertheless
>important. Perhaps a better way to state it is that material sent to
>a cache must already be in canonical form, eliminating the need for
>canonicalization being done by the caching server.

I agree with Ned's words, which I think are almost the opposite of what's 
in the draft requirements. How about "If the protocol requires a 
canonicalization algorithm, all requests sent to a caching server must 
already be in canonical form."

>This requirement also makes sense if things are to work with DNSSEC, BTW.

Sure does!

--Paul Hoffman, Director
--Internet Mail Consortium