Re: [Gpsbabel-misc] "SERIAL: tcgetattr error: Input/output error" in middle of obviously working con

[Valtteri K. <rv...@ik...>]

On 9/9/07, Robert Lipe <rob...@us...> wrote:
>
> Realtime positioning with Garmin protocol over BT isn't a road heavily
> travelled; Garmin has a trainload of products that do A800 over serial or
> USB and zero that do it over BT.   Those that do it over BT use NMEA.

Oops. There's a misunderstanding there. I actually have tried NMEA
over the same system and Maemo Mapper (N800 nav software) seemed to
recognize it, but I have no real need for it. I'm trying to use garmin
protocol for waypoint download/upload, track download and route
upload. The test case is the probably easiest, waypoint download.

> I just configured a Linux (F7) system with a Garmin on serial (PL2303 - USB
> serial, actuall) and set
> breakpoints on tcgetattr.  During startup in realtime tracking mode
> we call tcsetattr() seven times during startup.  It could be argued
> that we could call it less, but it could also be argued that ioctls
> shouldn't have to be coddled just becuase they're over BT instead of
> a physical wire.

But could the error condition be gracefully handled without hurting
normal serial line use?

> presumably using -i nmea where you've said you're using -i garmin.   Theese
> are two
> different code paths to us.  Use NMEA mode on your receiver to more closely
> models Jonathan's usage and presumably, success.

Ah, but of course. So probably -i nmea just uses the serial line "as
it is"? Unfortunately, that doesn't help me anywhere, then.

-- 
Valtteri Kokkoniemi
050 3233 707 / rv...@ik...
http://www.iki.fi/rvk/