On Thu, Nov 06, 2014 at 10:07:43PM +0000, Jeremy Harris wrote:
> > Still experimental, right? I think it still needs some work. Also
> > I've updated the ssl_dane library on github a bit, you might want
> > to pick up the new bits.
>
> I think we're too close to release to take any major changes.
> The current code has had some months soak time on a production
> system; I don't want to discard that. If you want too work
> on the code, you could always pull a git tree to do so;
> suggested patches would be welcome, particularly if new
> regression test cases are included.
You've somewhat misunderstood my generous offer. I want to help
your team get this right, and at this point I think it makes sense
to discuss the implementation interactively, where I ask some
leading questions, and the appropriate team member can walk through
the code with me and convince us both it is correct, or note any
requisite changes. An interactive design/code review if you like.
However, I am in no position to directly contribute code to Exim.
I am not familiar with the Exim coding style or its internal
interfaces. Its style is culturally alien to me. So I think a
better way to get there is channel my experience with DANE through
someone closely familiar with Exim. Do you or Todd have any cycles?
You can decide when this is to happen, if DANE remains experimental
in the next release, then we still have plenty of time. If you're
thinking about making it a mainstream feature at this point, then
we should the review rather sooner.
--
Viktor.
