Also, it would be difficult to remove existing functionality, and get the 
callers to update. E.g. deprecation of TLS_UNIQUE is not going easy for 
apps/protocols.

Cheers,

Andrei

From: TLS [mailto:[email protected]] On Behalf Of Eric Rescorla
Sent: Tuesday, October 4, 2016 8:49 PM
To: Martin Thomson <[email protected]>
Cc: [email protected]
Subject: Re: [TLS] Exporter output size

On Tue, Oct 4, 2016 at 6:32 PM, Martin Thomson 
<[email protected]<mailto:[email protected]>> wrote:
After a bunch of discussion about the consequences of having
insufficient output from various stages of the hash functions... Could
we make an amendment to TLS 1.3 to force the output size of the
exporter to be the size of the underlying hash output?  That is,
remove the length parameter.  Or is a change to the API too
disruptive?

I don't think this is a good idea. There are plenty of reasons why you would 
want to
export values != hash_len (e.g., cryptographic keys). Putting a restriction 
here just
pushes the problem around

-Ekr


_______________________________________________
TLS mailing list
[email protected]<mailto:[email protected]>
https://www.ietf.org/mailman/listinfo/tls

_______________________________________________
TLS mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tls

Reply via email to