[ntp:hackers] Why are we going down this road? Question on listen-on, query-on, -I

David Mills mills at udel.edu
Fri May 29 00:57:24 UTC 2009


Sachin was tasked to do two things only. Provide a syntax-directed 
configuration capability and upgrade the NTP simulator. He was told 
whatever he did could not compromise the existing configuration command 
syntax or semantics.

The asynchronous resolver has been on the list for a long time, but this 
is not a complaint about that. Sachim's idea was to avoid the messy 
little configuration file used when DNS is slow and keep the syntax tree 
around while the asynchronous resolver is doing its thing. This makes 
the configuration process fully dynamic, but some of the commands needed 
for full dynamic configuration, like unconfigure, have not been defined 
yet. That would be a good project for a summer intern.


Harlan Stenn wrote:

>Whatever.  There is a large pile of things to fix, and async DNS is on
>that pile.  One of the GSoC projects for this year should help us make
>good progress on that, and it could happen for 4.3.0.
>Near as I can remember, you drove the parser project and when I
>mentioned we had some issues we wanted to see resolved you shut that
>input down as you didn't want any "distractions".  Please don't be
>miffed that those issues are still hanging around.
>Regardless, we are probably otherwise in violent agreement - the core of
>the parser is nost likely not an issue, and it was not what I was
>talking about.
>>Bullshit. The overwhelming gorilla in the room is the asynchronous DNS, 
>>which would allow a much more flexible configuration process, as well as 
>>dynamic host discovery. This has been a sticky issue for several years 
>>with no resolution. Fix that and I'll be much more friendly about 
>>changes in the configuration process. As it is, I supported grad 
>>students for several years to improve the configuration process. The 
>>parser and the table-driven semantics are not the problem.

More information about the hackers mailing list