I'm going to be away from home for a few days starting tomorrow. So should have some real world feedback including time to lock while moving etc.

I could swear the lock times were better when I first got the unit (in December) compared to current (prepatch). So I think it may well be date bug related.

5-10 mins in some cases even starting at same location on same day. I'm hoping this is sorted.

I get the impression there is potential for a huge overall improvement by addressing a number of issues rather than just 1 big fix. So hopefully they all get picked up.

This (date issue) is one.

Confirm we are not doing a cold start each time. Could this just be a date issue symptom ?

There seam to be several potential items around AGPS. e.g.

Internet based Almaniac download.
Mobile Cell derived info fed back from phone.
Other calcs done on AGPS server and fed back to N810 ?

The Almaniac sound the easiest of the AGPS ones.

Zuber

Jussik - tried your patch on my N810, and it seems to have the GPS a lot more responsive.

Firing up MAPs / MaemoMapper after travelling from Spain to the UK resulted in a decent lock (5 satellites) within less than a minute - even though the device was indoors. Previously it was taking 10 minutes or more even outside. I'll try to see how it fares when travelling back to Barcelona next week.

Great work thanks!

-- Mark

The latest version of gps-clockd logs time to first fix. This should help in evaluating the patch.

http://nitapps.com

Beautiful, the patch makes it work within 20 secs. Down from several minutes! The first attempt still required about 2 minutes (down from 6), but after that the fix is very fast.

Thanks jussik,
Antonio

Never mind. Just went out for a walk with my son and had a chance to take some measurements.

- Fix #1 - 2 minutes 57 seconds. This was with unobstructed clear sky. GPS had been off for about 3 hours prior to this fix.

- Next 3 fixes were 5-25 minutes apart. They were all in the 15-25 second range. A definite improvement.

My impression is that the gpsd patch helps with "warm" fixes -- i.e. fixes within a couple of hours of a previous fix. Cold fixes still appear to be in the 2+ minute range. My theory on this is that N810 does not keep an almanac, and a cold fix basically involves a full sky scan.

Also, if I remember correctly, hot fixes do not rely on system time, and are unaffected by the patch.

My hot fixes are 20 seconds or less, and my cold starts are less than 80 seconds, thanks to this patch. I also use gps-clockd to set my time, and that's how I time it. even being 10 miles and 45 minutes from the last fix, it only takes < 20 seconds. Before this patch, it would always take 3 minutes or more, hot, warm, or cold.

The beauty of an open source OS and community involvement shines!

Nice work all, especially Jussi!

I stand corrected thanks for the heads up and the patchfix.

I simply don't get it.

I mean, this bug showed up this year as this is a leap year. So why the GPS was slow to get a fix in 2007 ? How come that now it is faster in getting the GPS fix ?

It must have been a bug in 2007 as well. At least it works now way better than before. In addition to much shorter times for getting a fix the patched version allows you to connect to 6 or 7 sattelites instead of 3-5. I made several experiments yesterday and turning off the N810 and turning it back on after 2 hours I got a fix in 10-15 seconds. That is a huge improvement. Before that it took almost 6 minutes for me.