Re: libnettle/libhogweed WIP

[Ted Zlatanov <tzz@lifelogs.com>]

On Thu, 27 Jul 2017 18:56:08 +0300 Ted Zlatanov <tzz@lifelogs.com> wrote: 

TZ> On Wed, 26 Jul 2017 21:36:31 -0700 Paul Eggert <eggert@cs.ucla.edu> wrote: 
PE> Eli Zaretskii wrote:
>>> Can you tell what change in semantics could explain this failure?

PE> In principle, yes. But it would take some time for me to come up to speed on GnuTLS.

TZ> Thanks so much for the hint, Paul. I was able to put together a Docker
TZ> image using Ubuntu 16.04 that reproduces the problem. It happens with
TZ> the OS packages of libgnutls-dev.

TZ> I'll try to put more time into chasing this, since 16.04 is a LTS
TZ> release and just requiring a newer GnuTLS will inconvenience users that
TZ> compile there. Maybe we'll have to disable AEAD ciphers there, but I
TZ> hope to find a better solution. If you have more insight into the
TZ> specific breakage, let me know, and I'll post as soon as I have useful
TZ> information.

I'm finally back from traveling and found that this morning Paul pushed
a fix to block AEAD ciphers in 3.4.x GnuTLS, citing a fix made a year
ago in 3.5.x. Since we can't easily do more on the Emacs side, I agree
with the fix and it should resolve the problem for you and other 3.4.x
users by avoiding AEAD tests. Sorry this took so long from my side.

Thanks
Ted