[ntp:hackers] Protocol specification modification for MS-SNTP

Martin Burnicki martin.burnicki at meinberg.de
Fri Jul 10 07:27:17 UTC 2009


Harlan Stenn wrote:
> Folks,
> I just wanted to mention that from my POV this discusison seems to point
> out that:
> - ntp needs to be robust and support a bunch of things.
> - different people will want to disable/enable various non-default
>   'configure' choices to meet their local policy needs.
> Assuming the above is true, it means:
> - we need to be sure that NTP *can* handle the needs of as much of its
>   userbase as possible
> - we must strive to make implementing these non-default 'configure'
>   choices as clear and painless as possible
> and what we are discussing are the pros/cons (or costs/benefits) of the
> various default choices we offer in 'configure' (and sometimes the
> ntp.conf file).

Agreed. Please keep in mind most people who *use* NTP don't *build* it
by themselves. That's why I'm voting to include most commonly used
features in the default build configuration, and if they need to be
enabled/disabled provide a way to do this at runtime.

For example, see the openssl support which is included by default simply
if the openssl headers are found at compile time.

What do you think how many people are using autokey, compared to the
number of people who don't use it? Most people who just want to have
their machines running with the "right time" simply use some pool
servers for which autokey wouldn't work, anyway.

Following the argumentation not to include support for MS auth by
default, maybe openssl support should also be excluded by default in
order to yield smaller binaries, so those who want to use autokey can
twiddle with the build configuration options and compile binaries
matching their specific preferences.

I want to make explicitely clear that the above statement is
hypothetical, it's *not* my real opinion.

Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont

More information about the hackers mailing list