Out of date chunks in documentation

[Hal Murray]

> Is there a way that we can fix refclock testing into CI?  If you can describe
> how we'd run that test (don't worry about the platform details), I can work
> on a way to integrate it into our pipelines.  Same question for running
> testing running without refclocks. 

The short answer is no, you can't test refclocks without having actual 
hardware to test with.

The slightly longer answer is that GPSD can mostly do it.  They have captured 
a large collection of samples from various devices and have a framework to 
replay the input and compare the output against saved good data.  GPSD doesn't 
have to worry about timing.  Eric worked on something similar in ntpd for a 
while but gave up.  Ask him if you want details.


But the big picture is that the CI stuff isn't really "testing" our code.  The 
bin-test was a giant step forward.  It checks that the linking is actually 
sane.  We need a post-install version.

The kind of testing I'm thinking of is actually running it in production and 
monitoring the logs.  I don't know how to automate that.

We need to test server mode as well as client.



-- 
These are my opinions.  I hate spam.