[LinuxPPS] http://lkml.org/lkml/2008/4/11/42

[Rodolfo Giometti]

On Fri, Apr 11, 2008 at 02:56:21PM +0200, Bernhard Schiffner wrote:
> 
> Rodolfo,
> there are varios aspects here:
> 
> 1.) We conclude that there is nothing better to do the pps-job than 
> dedicated hardware with direct available lines for interrupt.
> (The PTB does so for their public ntp-servers.)

I just said that the *best* would be a dedicated hardware, and that we
should consider the serial solution as a compromise between costs and
accuracy. :)

> 2.) The area of interest of pps is IMHO a timing better than the use of NTP 
> over network permits (sub-ms).

I agree.

> 3.) You need not only the pps-API but a proof of pudding too. 
> __ And this is done almost exclusively on normal PC's with no direct 
> available lines for interrupt.__
> 
> So let us do our best to keep these  (PC-) interfaces as long and as good 
> as we can.

I completely agree with you, but let me state the fact that by using a
serial connection for PPS you can *not* pretend very good
accouracy. Another issue is trying to do our best in order to resolve
this problem and improve serial connections.

> 4.) (Regarding Alan's suggestions:) 
> Perhaps he can show a place better suited for data collection. But until 

I hope so.

> this happens or perhaps even after this, it is better to have or at least 
> to know about different solutions.

I agree.

> 
> 5.) (Future:)
> Split API / source ?

What do you mean by that? The actual PPS core is functional, Alan has
some doubts about serial clients...

Ciao,

Rodolfo

-- 

GNU/Linux Solutions                  e-mail:    giometti at enneenne.com
Linux Device Driver                             giometti at linux.it
Embedded Systems                     phone:	+39 349 2432127
UNIX programming                     skype:     rodolfo.giometti