[Openstreetmap] cache of editor applet background?

[Mikel Maron]

We really need to get tiles invalidated by changes in the database.
The basic idea is to have a script running periodically, requesting all changes since the last run.
>From each lat/long pair (from gpx uploads or streets), a list of tiles in which that point lies is generated.
With this list, there are squid utilities to invalidate those objects in the cache.

Anyone interested in writing the code to generate the list of tiles from a single pair? 
It would replicate some of the tiling code in the clients. 

A complication: Tiles are generated by the Javascript slippy map, the Java editor (and maybe JSOM and other editors?).
The Javascript and Java tiles are slightly different. The Mercator projection is calculated slightly different, 
and there's a different level of precision being used. These need to be corrected, so that all clients are 
requesting the same tile set.  This is the complication of uses BBOX defined tiles rather than some other index.

Mikel

----- Original Message ----
From: Lars Aronsson <lars at aronsson.se>
To: openstreetmap at vr.ucl.ac.uk
Sent: Monday, January 23, 2006 2:03:35 AM
Subject: Re: [Openstreetmap] cache of editor applet background?

Frank Mohr wrote:

> now i'd like to draw those side roads, but the main trackpoints
> are still shown and it's hard to determine which points belongs
> where

Yes they are cached, and it's quite irritating.  Your situation 
appears both when you upload new tracks and when the cached image 
was generated during a time when the database refused to return 
any yellow dots.

You could either use a different zoom level (16 instead of 15) or 
you could wait two days (48 hours) for the cached image to expire.


-- 
  Lars Aronsson (lars at aronsson.se)
  Aronsson Datateknik - http://aronsson.se

_______________________________________________
Openstreetmap mailing list
Openstreetmap at vr.ucl.ac.uk
http://bat.vr.ucl.ac.uk/cgi-bin/mailman/listinfo/openstreetmap