[gnutls-help] 0-length handshake fragments with DTLS

[Nikos Mavrogiannopoulos]

On Tue, Sep 2, 2014 at 3:17 PM, Manuel Pégourié-Gonnard
<mpg at polarssl.org> wrote:
> Hi,
>
> I just noticed a behaviour that, while not buggy at all, is suboptimal, with the
> way handshake messages are fragmented in DTLS to respect the MTU: when the
> length of a handshake message is an exact multiple of the maximum allowed
> fragment length (according to the MTU and the current output tranformation), an
> additional (unneeded) fragment with length 0 and offset = the length of the
> message is sent.
> Again, I'm not claiming this is a bug, and I don't think it can cause interop
> issues with other correct implementations, but it would look cleaner IMO (and
> slightly more efficient) to avoid sending this message.

It doesn't look like a feature either. Is there some way to easily
reproduce that?

regards,
Nikos