[Click] Timestamp changes

[Eddie Kohler]

This sounds like a good idea, but we won't tackle this ourselves for 
now -- happy to accept a patch!!

E


On Dec 30, 2004, at 4:57 PM, Jonathan Day wrote:

> Oh, I doubt it's nearly as annoying as the question
> I'm about to ask. :) There's an "experimental" patch
> for Linux, PPSkit, which can be found on any of the
> standard kernel FTP/web sites:
>
> http://ftp.kernel.org/pub/linux/daemons/ntp/PPS/
>
> This provides nanosecond timer support for the Linux
> kernel, by extending what's already there. It does
> have its problems, but I imagine it's using a fairly
> standard approach to nanosecond timing.
>
> Will Click be able to take advantage of a
> nanosecond-aware/enhanced kernel (either now or in
> future)?
>
> --- Eddie Kohler <kohler at cs.ucla.edu> wrote:
>
>> Hey all,
>>
>> I have some probably annoying news.
>>
>> In the interest of supporting nanosecond-scale
>> timestamps, we've deeply changed
>> the way Click handles time values.
>>
>> In particular, the good old 'struct timeval' is
>> gone.  It was replaced with a
>> Click type, 'Timestamp', defined in
>> include/click/timestamp.hh.  Timestamp acts
>> like a 'struct timeval' unless you configured with
>> '--enable-nanotimestamp', in
>> which case it has nanosecond-scale precision.
>>
>> Timers are scheduled with Timestamps.  Packet
>> timestamps are Timestamps, not
>> 'struct timeval's.
>>
>> This was a big change and may require some time to
>> settle in.  Please feel
>> utterly free to report bugs to the list.  Element
>> authors may want to check over
>> the attached diff.
>
>
>
> 		
> __________________________________
> Do you Yahoo!?
> Take Yahoo! Mail with you! Get it on your mobile phone.
> http://mobile.yahoo.com/maildemo
> _______________________________________________
> click mailing list
> click at amsterdam.lcs.mit.edu
> https://amsterdam.lcs.mit.edu/mailman/listinfo/click