Re: [TLS] draft-ietf-tls-rfc4347-bis-00.txt

Hi Eric,

I would prefer something like:

- For DTLS over TCP or SCTP, which automatically fragment
   and reassemble datagrams, the upper layer protocol
   MUST NOT write any record that exceeds 2^14 byte.

The reason is that a while ago it took me some time
to understand that someone talking about small messages
really was talking about messages of 64KB. Large message
were about several MB. I'm not sure what these guys
would understand when reading "effectively infinite".

Best regards
Michael

On Oct 14, 2008, at 12:46 PM, Eric Rescorla wrote:

> At Tue, 14 Oct 2008 10:04:37 +0200,
> Robin Seggelmann wrote:
>>
>> Hello all,
>> I was just checking the draft for changes relevant to DTLS over SCTP
>> and came across the following new paragraph:
>>
>> - For DTLS over TCP or SCTP, which automatically fragment
>>   and reassemble datagrams, the upper layer protocol
>>   SHOULD be informed that the PMTU is effectively infinite.
>>
>> What does 'effectively infinite' mean? TLS limits the message size to
>> 2^14 bytes, so shouldn't this limit also apply to DTLS? If the
>> message size really is arbitrary, doesn't this affect some cipher
>> algorithms? Or should the application then ignore the announced
>> 'infinite' PMTU and limit the message size anyway?
>
>
> Yes, that's a fair point. OTOH, this is a maximum message size
> which is related to, but not identical to, the PMTU. Operationally
> the application needs to restrict itself to 2^14 and associated
> limits. I'm tempted to simply add a parenthetical "(though of
> course applications still MUST NOT write any record that exceeds
> 2^14 bytes)"
>
> -Ekr
>
>
> _______________________________________________
> TLS mailing list
> TLS@ietf.org
> https://www.ietf.org/mailman/listinfo/tls
>

_______________________________________________
TLS mailing list
TLS@ietf.org
https://www.ietf.org/mailman/listinfo/tls