[ntp:hackers] 4.2.8 code freeze

Dave Hart hart at ntp.org
Wed Oct 19 18:53:09 UTC 2011


On Tue, Oct 18, 2011 at 03:08, Harlan Stenn <stenn at ntp.org> wrote:
> Folks,
>
> If you know of any open bugs that you thing should be fixed before 4.2.8
> is released, please "request blocking 4.2.8" on those bug reports.
>
> Right now, I think the only things we're waiting on before starting the
> RC for 4.2.8 are:
>
> - a new release of autogen

I presume this is to ensure Eric's AutoOpts localization work is part
of the next ntp-stable from the start?

> - validation that the recent autokey changes are working as intended

Right, Dr. Mills introduced a group-selection "ident" option to
association directives such as server, peer, manycastclient, and pool,
as well as changing ntp-keygen and ntpd to support hostname at groupname
(as well as simply hostname) in autokey certificate subject (and for
self-signed, issuer) fields.  He has spelled out the intended
functionality in the Autokey documentation.  My concern is making sure
the code matches the design, since it is new for 4.2.7 and possibly
not well-tested.

I have another issue which I've been discussing with Dr. Mills for a
few weeks, about which I finally filed a bug report:

http://bugs.ntp.org/2037
Fuzzed non-interpolated clock may decrease.

I am aware getting the fuzzing right has been a delicate endeavor and
would prefer to leave it alone, but the fact is the current code does
bad things on the minority of systems lacking interpolated clocks.  I
would like to hash that out before pronouncing another stable release,
so I've requested 4.2.8 blocking on the bug.

Cheers,
Dave Hart


More information about the hackers mailing list