Production issues with GML validation

[carlospzurita]

Since some days ago, we have been experiencing some problems in the INSPIRE validator. This problems are related to an issue with the GML validation. There is a bug with the srsName and srsDeclaration detection from the gml:posList tag. By default the dimension is set to 2, and this provokes an error in the geometry parsing that throws an uncontrolled exception, and it ends in a memory leak.

We are working on a hotfix for this issue, but for now, it would be advisable to include the attributes srsName and srsDimension not only on the wrapping Geometry tag, but also on the gml:posList tag. This prevents the memory issues from happening and the validation can be performed.

We have an open discussion for this issue on the bugfix repository etf-validator/etf-webapp#214 , which you can use as a technical reference.

Dear @DeordD , we have found references to the state Baden-Württemberg (BW) in some of the datasets that were related to this issue. Do you have any information on the user that is submitting the datasets, and maybe propose the workaround of the srsDimension?

[ghost]

Dear @carlospzurita, @MarcoMinghini

thanks a lot for the information.

Dear @DeordD , we have found references to the state Baden-Württemberg (BW) in some of the datasets that were related to this issue. Do you have any information on the user that is submitting the datasets, and maybe propose the workaround of the srsDimension?

This is already communicated to the user which submitted the dataset. GML dataset won't be validated until this problem is fixed.

Kind regards

[carlospzurita]

We have uploaded a workaround for this issue on staging . During validation, we take the srsDimension declared on the parent node of the GML entity, and we include it on the gml:posList if it exists. This way we prevent the exception in parsing the geometry. This is a temporary solution, and needs to be further analyzed, but at least you can submit your GML datasets to staging, check the results and share some feedback about the workaround.

[carlospzurita]

Dear @DeordD
Did you or you're data providers had the opportunity to validate the GML with the problematic CRS, an check if the solution works fine?

[ghost]

Dear @carlospzurita ,

data provider has tested the same GML dataset on staging instance against following CCs:

Following problem

this provokes an error in the geometry parsing that throws an uncontrolled exception, and it ends in a memory leak.

didn't occur. GML Validation was fully performed.

Thanks a lot for the fix.

[danielnavarrogeo]

Thank you @DeordD for the feedback.

Regards