Re: [gpsd-dev] Major progress!

[Eric S. Raymond]

Hal Murray <address@hidden>:
> So far it looks OK on Linux.

That's good, and expected.

> I'm having troubles on NetBSD and FreeBSD.
> 
> FreeBSD gets typical lost sentence problems.  So I changed WRITE_PAD from 
> 0.001 to 0.01

Alas.  I was hoping we could zero out WRITE_PAD everywhere.  What is 
sys.platform on that system?

> Then it hung on the second or 3rd test.  It seems easy to reproduce.
> 
> top says:
>   PID USERNAME  THR PRI NICE   SIZE    RES STATE   C   TIME   WCPU COMMAND
>  9021 murray      1 103    0 16116K  8216K CPU1    0   4:48 100.00% python
> 
> This morning, NetBSD also hung.  I haven't seen that this evening.

Those sound like they might be the same problem Greg Troxel reported
from NetBSD.  He was able to pin it to "CLOSE_DELAY is gone.  Test source
termination is now deterministic."

> The latest SHM cleanup doesn't seem to work on *BSD.
>
> I don't know if that is a bug in their SHM implementation or we don't 
> understand the rules.

I don't either.  This is going to need attention from Greg Troxel or
someone else intimate with *BSD.

On the bright side, these are all almost certainly test-frame problems 
rather than issues with gpsd or the tools.  That's a good thing.

The only non-test-frame issue still pending is newdata.time not being
set in gpsmon.  I'll tackle that today.
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>