Dirk Stöcker wrote:

> When downloading the whole world and rendering tiles based on that, these 
> tile-crossing problems of the name-suppression should vanish (or at least 
> be reduced a lot :-)

OK, that is true. however that problem mostly occurs on z0-11, right?
And we are already using the captionless layer for that, so it's merely
a matter of creating the caption layer for z0-11 that downloads the
whole world as a data source. This is mainly a setup thing on the client
side, I'd think (omitting the bbox for the lowzoom download and reuse
the downloaded data for all tiles).

> Making it dynamically should not be a problem. When properly cached, then 
> also CPU power should be no problem.

Caching uses disk space. Given taht our 931GB disk is 94% full, I'd
rather not anything at all right now. It seems we need to push for
multiple tah server instances to spread the disk (and network) load.

> Also I think instead of rendering captions at all, the captions should be 
> passes as position+direction+text information and rendered when 
> displaying. But that is probably a future project :-)

I'd agree. :-)

_______________________________________________
Tilesathome mailing list
[email protected]
http://lists.openstreetmap.org/listinfo/tilesathome

Reply via email to