[Tilecache] Read Only Cache and Coverage...

[Christopher Schmidt]

On Tue, Mar 27, 2007 at 09:20:20AM +0930, Andrew Hughes wrote:
> Hi All,
> 
> I'll try and be very concise here, it would be nice to have the live 
> (external) WMS-C service serve only cached tiles, and an internal 
> process to control what populates the cache. (This is for disk cache).
> 
> 
> Here's why:
> 
> Limiting the cache to a simple "extent"/"bbox" means that when we get 
> new coverage we can't increase the extent without modification to our OL 
> clients. We want world-wide..... even if we don't have EU maps... yet!
> 
> Further more, having an internal process to populate the cache means the 
> cache coverage can be quite sophisticated. For example, we might cache 
> different resolutions for different area's and also complex polygons 
> e.t.a. or maybe draw on polygons from our spatial database to execute 
> the seed script.
> 
> Next, most "extents"... (ie a BBox)... is a very inefficient use of 
> resources as the cache has a good proportion of useless tiles (ie sea or 
> desert).
> 
> Performing "live caching" from client requests could be based on another 
> approach... Where the tile must me certain conditions (ie coverage or 
> resolution), this is normally a slow process that will degrade 'live' 
> service. It is also most likely that the conditions that need to be met 
> are defined and dependent on external spatial data. This is most likely 
> not TC's responsibility.

Schuyler and I have no problem with the idea of seperate caches for each
layer. If we were to go that route, then the implementation of a
non-writing cache would be simple (just return false from the write()),
and you could set up two different layer names -- one for internal use,
one for external use.

Does that solve your problem? I'm not sure if I'm understanding it. 

Regards,
-- 
Christopher Schmidt
MetaCarta