WMS-cache does not "tile" requests one same boundaries between session resulting in re-requests of already cached areas

[jjaf.de]

WMS does pre-load cached areas but since

cache.hasExactMatch()

is false since the viewport is slighty shifted although the complete area can be rendered the whole area is re-requested from WMS-server

Problem

Re-requests are resulting in having a huge cache of [session]-times the size and stress on network and WMS-server.

possible Solution

A solution could be to introduce "tiling" in maybe

downloadAreaToCache()

so that the same bounding-boxes for WMS-requests having shifted viewports are being calculated between sessions or have an algorithm to recognise total coverage from cache instead of

cache.hasExactMatch()

I would suggest the former is easier to implement maybe with help from the tiling-projection.

Workaround

Set viewport to exactly same coordnates and zoom before adding WMS-layer with cache e.g. by using "Jump To Position".

[bastiK]

When scale matches exactly, the cache should be used.

[stoecker]

We have WMS bookmarks, which solve this issue already. THought maybe a tiled approach and some sort of "native" request levels would be a good idea.

[jttt]

WmsLayer already has tiling, method getImageXIndex() convert coordinate to tile index. But scale needs to be exactly the same as before (that's what WMS bookmarks are for).

[bastiK]

Native request levels may decrease the image quality in some cases, but I guess for the average user, it would be an improvement.