Hi,
I'm wondering about transaction contents validation.
At the moment I know that insert elements in a transaction cannot be
validated.
However, I'm not really understanding the reason.
I'm wondering, wouldn't it be sufficient to use the schema builder to
build a full schema of all the feature types hosted by GeoServer and
pass that schema into the parser to get proper validation?
Or is it that doing so would result in quite an overhead and possibly
memory leaks? (the only point where I see the schema builder used
is the DescribeFeatureType WFS 1.1 response, which we know has a
small memory leak).
Cheers
Andrea
--
Andrea Aime
OpenGeo - http://opengeo.org
Expert service straight from the developers.
The problem is inherent in gml and wfs. The reason we don't validate is because a wfs does not force the client to specify the schema location of any application schemas. When a client says insert a "topp:states" it is up to the server to know what that means. If a client were to specify a schemaLocation attribute that contained the link to an application schema / DFT request then validating transactions could work.
We use pure sax validation, so building up an xsd object does not help us. However part of setting up the validator is registering schema locations for all the schemas, gml, ogc, wfs, etc... We could plug into that system to add schema locations for all application schema types. If we did so I think validation could work. Would have to try it out.
I think there is a jira open for this one. Be good to post this there.
-Justin
On 3/31/10 2:32 AM, Andrea Aime wrote:
Hi,
I'm wondering about transaction contents validation.
At the moment I know that insert elements in a transaction cannot be
validated.
However, I'm not really understanding the reason.
I'm wondering, wouldn't it be sufficient to use the schema builder to
build a full schema of all the feature types hosted by GeoServer and
pass that schema into the parser to get proper validation?
Or is it that doing so would result in quite an overhead and possibly
memory leaks? (the only point where I see the schema builder used
is the DescribeFeatureType WFS 1.1 response, which we know has a
small memory leak).
Cheers
Andrea
--
Justin Deoliveira
OpenGeo - http://opengeo.org
Enterprise support for open source geospatial.