[LinuxPPS] LinuxPPS evolution.

[Paul]

There have been a number of topics since I raised this problem. My
opinion is:

1. We do actually need some sort of ATOM driver. I know some of you were
asking 'Why?'. There are a couple of reasons, the first being that there
are actually some atomic clocks out there and secondly that cheap GPS
modules do not provide reliable NMEA data.

2. The CAT solution looks pretty weak if you are writing a HowTo for
LinuxPPS (I'm working on one at the moment). It is just the thing to
discourage a newbie.

3. Can't the OPEN be in ppsctl or something similar so you only have to
do one thing to that port.

4. Although it would be nice to use the rest of the serial port for
something else is it necessary? There isn't much else today that uses
that port.

5. Perhaps the future is only to use the NMEA patch as general purpose,
so the NMEA patch becomes renamed the PPS patch.

6. By the way the NMEA patch is written against p2, are you sure it is
still O.K. against p4? I was getting some odd segmentation faults, but I
didn't have time to investigate further.

Paul

 
On Mon, 2007-11-12 at 16:34 -0700, clemens at dwf.com wrote:
> The LinuxPPS has evolved over the last year from something very
> different to the code in its present form.
> 
> I guess I have to ask the question why we have the /dev/pps<n> file
> at all in its current implementation.
> 
> It would seem that all the discussion that we have been having about
> parent/child files and opening this before you can do that, would
> just go away IF the ioctl that we currently do on the pps<n> file 
> was done on the 'parent' file instead.  The internal code could remain the
> same (almost) but the ioctl would be done on the parent rather than
> the child.
> 
> For things to work these two (parent / child) clearly know about one
> another so moving the ioctl to the other device (or adding a 2nd ioctl)
> would seem possible.
> 
> And we would be back to an API that everyone understands, just one
> file for our standard case.  Whether the /dev/pps<n> files (and an ioctl
> there) would be useful for some of Rodolfo's strange devices is a question
> I can't answer).
> 
> Is this crazy or possible?
>