Re: [gpsd-dev] Where to apply scaling

[Christian Gagneraud]

Hi Eric,

On 24/05/12 21:49, Eric S. Raymond wrote:
> Christian Gagneraud<address@hidden>:
> > Currently in driver_ais.c an offset is applied for certain message
> > fields (IMO 236 and IMO289 met/hydro message), and a scaling is done
> > conditionally in gpsd_json.c.
> > I think it would be better to apply the offset and the scaling in
> > gpsd_json.c. Do you have any objection on doing this?
>
> Unfortunately, I thought of an objection. The shm interface.
>
> Because some clients will be reading the AIS report structures
> directly from the daemon via shared memory, all *lossless* preparation
> of the data (in particular, offset applications) should be done in the
> device driver.  This will reduce the number of magic constants that
> client software has to know about.

Interpreting the AIS structures requires already lot of foreknowledge, 
there's plenty of enum values and magic constant (data not available).
Enums are not part of the API, but the NOT_AVAILABLE are (def'ed in gps.h).

What about moving these 6 offset constants in gps.h beside their 
XXX_NOT_AVAILABLE? (FID31 needs only 1 constant not 5. I have one commit 
here for FID31 fixes). And while we are here, why not adding the scaling 
factors as well? This way, clients will have all the information to go 
from raw to interpreted data.

Another consequence of doing the offset in the driver is that the gpsd' 
version of the NOT_AVAILABLE values are different from the IMO standards 
(and from the ones documented in AIVDM.txt)

Regards,
Chris

> Lossy transformations (scaling) should continue to be done optionally
> in gpsd_json.c.


--
Christian Gagneraud,
Embedded systems engineer.
Techworks Marine
1 Harbour road
Dun Laoghaire
Co. Dublin
Ireland
Tel: + 353 (0) 1 236 5990
Web: http://www.techworks.ie/