True Nokia could, however there's no point us discussing what Nokia *could* do. They tend not to release information about this kind of thing ahead of release - only Maemo itself.

*We* could do this if we knew the format of supl.nokia.com: so far I've not been able to find any information on the protocol other than it being a H-SPL service. More details welcome :-)

http://cms.openmobilealliance.org/te...20070615-A.pdf

and other odds and ends on that site.

historical almanac data: http://www.arinc.com/gps/gpshistory/

It does look like it's updated every day.

So we could just feed it they way we want... cool

Yeah, an open source provider would be great, but, then, who would host the database ?
Unless we think of a locally stored small database of user locations.
That may very well work.

I can second that - still takes forever with updated gpsd.

I was talking about something much simpler than full-blown A-GPS. If we simply figured out the format of nvd_data and inserted new ephemeris from the network regularly, it seems that cold starts (3+ minutes) would become warm starts (~15 seconds as established in this thread). You don't need to talk to the chip directly as nvd_data is already fed to the chip by gpsdriver. You also don't need to know precise location. There were reports in this thread of people driving several miles and still getting a warm fix in seconds. Defaulting to last known location should work in most cases.

Here is a good experiment to try if anyone has 2 N810s.

1) Get a fix on both units so that nvd_data is initialized.
2) Wait a day
3) Get a fix on one of the units. This will take a few minutes.
4) Edit nvd_data on the second unit. Copy the B 2 section from the first unit.
5) Start GPS on the second unit.

If I am correct, the second unit should obtain a lock in about 15 seconds.

I am starting do doubt if anything from nvd_data is fed back to the GPS chip by gpsd. Here is what I did: I refreshed location in the Control Panel -> GPS Location with external LD-3W module, upon which nvd_data and gps_last_saved_report got updated. It took about 1 second (I have good view to the sky). Few minutes after that I switched to internal GPS and tried to refresh location in the Control Panel -> GPS Location. So far it has been searching for more than 10 minutes without getting a lock. Could somebody repeat this experiment?

gps_last_saved_report is saved by gpsd, so that would be updated. nvd_data should probably only be saved by gpsdriver, so I'm suprised that it was updated. Did the contents change?

Another option would be to wait for some time (so the ephereris & almanac data in the nvd_data file are old - so this will be a sky search situation), save that nvd_data file somewhere, get a lock using the internal GPS, then test the time to lock using one or other of the files. If there is a difference, then the data must be being used.

We had a quick look at some nvd_data on irc on Thursday afternoon (iirc) and it looks like the nvd_data file is filled up in sections (we theorised that this was as more satellites became visible).

Assuming this is almanac data, it should be valid for at least one day (I believe they update the alamac data daily to improve accuracy, even though it is still valid in the old form - if anyone knows for sure please pipe up), therefore it would be interesting to connect, note the satellites that were used/seen/etc., then wait for a little while (until more satellites are visible in the sky (not sure what the best tool is to track their locations - any thoughts) then connect again and compare the contents of the two fixes' nvd_data files. I'm assuming that blocks of data will match. If so, we're on the way to working out what the format of the files is....

Any help/ideas welcome

Lardman seems to be right - nvd_data timestamp changes when switching from external to internal GPS (and vice-versa). The contents of the B1 section has not changed by switching to external GPS module.
Now I have two files corresponding to different satellite sets and indeed there are matching parts in nvd_data. Here are both files with channels lists as displayed by maemo-mapper. Hopefully it will help somebody who is more familiar with GPS technology than I to figure out the nvd_data structure.

We don't think it is either -- osso-gpsd maintainer has already stated that gpsd is not part of that loop in any way. I'm also using the gps mostly with gypsy, so gpsd is not running at all but nvd_data still gets updated.

nvd_data does not get updated when external gps is used, Iäve tested this already.