[Geoserver-devel] [jira] Created: (GEOS-354) GetBounds optimization

GeoServer GeoServer Developer
1

GetBounds optimization
----------------------

         Key: GEOS-354
         URL: http://jira.codehaus.org/browse/GEOS-354
     Project: GeoServer
        Type: Improvement
    Reporter: dblasby
Assigned to: dblasby
     Fix For: 1.3.1

Add support (for "SPEED" or some new strategy) to calculate the FeatureCollection's boundedBy element without re-querying.

Basically, stick a "look at the end of the XML" reference in the GML output for the boundingBox. Once all the features have been written (and the bounds calculated), stick the actual answer in.

Most people dont really need the boundedBy element, and it requires querying the dataset twice to calculate it.

--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
   http://jira.codehaus.org/secure/Administrators.jspa
-
For more information on JIRA, see:
   http://www.atlassian.com/software/jira

2

hrmmm....

what I really think is that datastores should be fast in calculating q
query's bbox, and indeed all bbox related oeprations should be optimized
at the datastore or native level.
I'll put the example of ArcSDE.
ArcSDE (for MSSQL) stores the actual geometries in a related table's
binary field. this related table also contains minx/y maxx/y fields for
fast calculations.
So when you request a query bbox, the cost is the cost of executing a
sql query like "SELECT MIN(minx), MIN(miny), MAX(maxx), MAX(maxy) FROM
<geometry_table> WHERE <where clause>"

Shapefiles should use spatial indexing.

I guess postgis does something similar with its gist indexing?

why need we to traverse the result to aquire the request bbox? can't the
bbox be queried explicitly in postgis?

just questions of an ignorant.