[Asterisk-Dev] Re: ztdummy? is it necessary?

[Dan Austin]

> Based on the initial response, I should probably clarify what I'm
asking. 
> I know that some applications, as asterisk is developed now, require a
> zaptel timing source.  However, is this requirement necessary?  Would
> certain platforms, if asterisk was written to accept it, be able to
handle
> everything fine without the zaptel timing requirement?  My understand
of
> the issue is that older versions of Linux had an inaccurate real-time
> clock.  I also understand that this has since been fixed?  And that
BSD
> does not have the issue?  So does this extra driver really need to be
> required on all platforms?  Or just the ones with the broken RTC?

> Basically, I'm not asking if asterisk as it is today requires ztdummy.

> I'm asking if the requirement is necessary.
Yes and no.  Recent advances in Linux (RT patches, Hi-Res timers) that
are approaching a merge to the mainline kernel raise the possibility
that timing could be migrated to the core of Asterisk.

On the other hand, if you have a hardware timer, you might prefer to
use it, in which case making ztdummy use these new features is still
a better way to improve timing and allow for flexibility.

Lastly, even if these improvements make it in to the mainline Linux
kernel, will there be a portable way to implement the same functionality
on other operating systems?

So to answer a theoratical question with a theoretical answer:
It is possible that ztdummy will not be necessary in the relatively
near future (on Linux), but my guess is that it will be maintained and
improved instead of eliminated.

Dan
> Thanks!
> -JD-