[gnutls-devel] GnuTLS | WIP: add support for 0-RTT (!775)

[Development of GNU's TLS library]

> By the way, I am currently implementing ClientHello recording (RFC 8446 8.2), as anti-replay measure.  That suggests that in some occasions (e.g. a ticket arrived out of time window), the server should reject 0-RTT but can continue with 1-RTT.  In that case, the client may want to resend the early data as a normal application data after handshake.  So rather than having a separate send function (`gnutls_record_send_early_data`), it might make more sense to reuse `gnutls_record_send2` and allow it to fallback from 0-RTT to 1-RTT.

> What do you think?

I looks easier to use, but could it be misused, e.g, to send early data when not intended? If not (e.g., because early data must be explicitly enabled), it looks better to me.

One other comment when going through the API again, is that the `gnutls_record_recv_early_data` when used either with the callback or after the handshake, it assumes that all early data are cached. So maybe we make a note of that expectation in the documentation and refer to the function that can be used to adjust the value when the default may seem excessive to a server.

-- 
Reply to this email directly or view it on GitLab: https://gitlab.com/gnutls/gnutls/merge_requests/775#note_110383881
You're receiving this email because of your account on gitlab.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.gnupg.org/pipermail/gnutls-devel/attachments/20181019/648730a7/attachment.html>