Re: [gpsd-dev] driver_nmea2000.c

[address@hidden]

Hello,

with the last commit, the "normal" nmea2000 behaviour is back again, but not if 
i let the deamon listen on can and a serial port.

And sometimes, a second start of "gpsmon -n" give no data. 
gpsd is then in state

0 S  2018 32751  3006  0  80   0 -  6985 hrtime pts/1    00:00:00 gpsd

and can not terminated by "crtl-c", only by a "kill -9".

Note the pts/1...

Reinhard

-----Original-Nachricht-----
Betreff: Re: [gpsd-dev] driver_nmea2000.c
Datum: Mon, 26 Jan 2015 13:16:00 +0100
Von: "Eric S. Raymond" <address@hidden>
An: "address@hidden" <address@hidden>

address@hidden <address@hidden>:
> The problem mentioned in the previous mail was introduced with commit
> 
> a2cabfcd89d8ac45a60fb35847258567c3450560
> 
> "Introduce an adaptive delay into the main loop to prevent select buzzing."
> 
> CAN deliver many small data packages with short time intervals between them.

Ah, this tells me what the right fix is.  The NMEA2000 driver needs to declare 
a source type that gpsd recognizes it shouldn't delay on.  Working on that 
now...
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>