Hi,
so I'm about to leave for one week vacation and I thought
I should make the point of the geoSearch (vector super-overlay,
regionation, call it as you like) subsystem.
The main point, which was, making it return data faster, has
been obtained thanks to a well designed algorithm (from David
and Arne) and some extra kicks of mine: using the native-sorting
strategy, which btw works only against databases for the moment,
we got the first feature returned back in roughly 1s instead
of 1h (3600x speedup, woot!).
The chats and the crucible reviews helped in setting up things
straight (btw, still two crucible reviews going, I guess I'll
see the results when I come back).
There are still some things that need doing. For a few I could
open jira issues:
http://jira.codehaus.org/browse/GEOS-2047
http://jira.codehaus.org/browse/GEOS-2048
http://jira.codehaus.org/browse/GEOS-2049
(neither of them is that high priority btw).
One thing that I don't know exactly how to manage is data
outside of the lat/lon standard range, the current algorithm
won't simply work with it (think a map that centers
around the antimeridian, and that for that reason uses
lat > 180).
Then there are a couple of issues in GWC, maybe I should have
opened tickets, but anyways:
* it does not seem possible to specify the regionation
policy and the other params in calls to GWC... it would
be useful to be able to (what about allowing the user
to append params to the request that get reflected
directly on the server request?)
* sometimes GE just reports back "invalid request" errors,
but I cannot see any error log from neither GWC nor
GeoServer (I set up the logging so that GWC messages
show up from INFO level upwards).
Cheers
Andrea