OpenStreetMap logo OpenStreetMap

Jon Schneider's Diary

Recent diary entries

Mobile phone mapping and tracking

Posted by Jon Schneider on 8 January 2008 in English.

Used some data from our Sunday morning ride to Saffron Walden.

I've done a bit of investigation into using OSM data in my phone-based tracker. It looks quite feasible but I don't know whether it would be re-inventing the wheel and whether I have better things to do.

Something I could do relatively easy is make the tracker a tool for collecting GPS data for those that don't want to buy a full GPS device.

Instead they would buy a ~£25 bluetooth GPS. Those with phones such as the Nokia N95 with integrated GPS might use an experimental variant I am tinkering with that uses the phone's own GPS.

Have a look at this

http://tracking.axismilton.ltd.uk/tracker/

Jon

Post-mapping ramblings

Posted by Jon Schneider on 2 January 2008 in English.

I have now done some of Cottenham with Donald, a few tweaks in Ealing and a start on Great Gransden.

Obviously I'd like to see how the maps are rendered. There are various quirks in this area of course.

From the users' point of view OSM has a few obvious failings.

The search is horribly horribly slow. Why is this so ? Is somebody doing this or is there a position available ?

I have read somewhere that tiles that appear as "...more OSM coming soon" mean that a tile is being re-rendered.

What is wrong with displaying old tiles until their replacements are ready ?
If that cannot be done would it be wrong for the text to say "tile being rendered" instead ?

For user-facing layers displaying this text instead of a tile is wrong.

At the time of writing c.tile.openstreetmap.org is sick . In fact a typical tile requests produce the 3 byte text "nil" still with a Content-type of image/png so that's not right.

It looks like [abcd].tile.openstreetmap.org are the same machine and this seems to be a valid technique for making browsers request more in parallel (in the absence of pipelining) as they only do a few per hostname.

More generally I am interested in the policy for encouraging http cache-ability, while propagating updates when they are available at the same time as tracking the tiles of interest.