INSPIRE Thematic Clusters

Separate WFS endpoint for each dataset in TN - one dataset per GetCapabilities response - how to do it?


Separate WFS endpoint for each dataset in TN - one dataset per GetCapabilities response - how to do it?

Started by Tomas KLIMENT - KLIMETO Replies (4)

Dear TN implementers,

I would like to ask for a help in providing INSPIRE direct download service for TN data using OGC WFS. In Slovakia we currently have the following "single" endpoint:

which operatesOn 20 featureTypes as tn-a:AerodromeArea, tn-c:CablewayLink, tn_ra:RailwayArea, tn-ro:RoadArea, etc. each linking to a single INSPIRE dataset metadata record.

The question is whether this is not against the requirement 52 of having one dataset per GetCapabilities response.

If yes, how should data provider do it correctly, provide a separate endpoint for each featureType, separate endpoint for each application schema (tn-ro, tn-ra, tn-a).

Thank you in advance for sharing your experiences in implementing direct download service based on WFS for TN data.

And also, is there any definition of a INSPIRE dataset and its relation to WFS featureType. Could not find definition of INSPIRE dataset anywhere (incl. INSPIRE registry - glossary, TG for download service, etc...)



    • Public

    By Peter PARSLOW

    Please note this is not an authoritative answer!

    The definition of "dataset" in INSPIRE is taken from ISO 19115, and is at You can see this in the Directive, Chapter I General Provisions, Article 3, bullet 3 (English:

    In my opinion, that makes requirement 52 of the Download Services TG self fullfilling, in that 'all the data returned from this endpoint' is a way of identifying all that data as a data set!

    In many discussions here and at INSPIRE conferences, it is clear that an INSPIRE dataset can certainly contain a variety of feature types. I think the most common approach is to consider it somewhat akin to an 'application schema' in the INSPIRE specifications, so you would have one dataset each for Air, Rail, Road. But I don't think that's a requirement.

    Speaking on behalf of Ordnance Survey, GB, we have made some INSPIRE datasets that correspond to individual application schemas, but in others we have included feature types from more than one application schema - e.g. our vector Elevation dataset also includes LandWaterBoundary features, because our "elevation" customers are used to getting at least some indication of the tide line with their contours. Noone has challenged the validity of that, but I haven't actually tried it with the recent validator.

    I hope that helps.

    • Public

    By Iurie MAXIM


    You may see this thread that covers this topic: 

    A dataset that contains more aplication schemas and that is implementing short URLs, is this one:

    It should be understood that a dataset is not linked to an individual application schema. If you are familiar with ESRI, one dataset = one ESRI Feature Dataset that contains multiple feature classes (application schemas).

    Best regards,

    Iurie Maxim

    • Public

    By Katharina SCHLEIDT

    So, to follow up on Peter's self fulfilling requirement 52, worst case, we reverse engineer our metadata to suit the blocks of data we provide under one end point?

    For institutions that can argue that all their data is one dataset, things get very simple :)

    Question is what to do if an institution does has multiple datasets (blocks of data too divergent to claim as one set) that also span INSPIRE Themes? Has anybody experimented with providing the same INSPIRE Theme in multiple versions? (does this work with the GeoServer fix for unique endpoints?)

    • Public

    By Iurie MAXIM

    Hi Kathi,

    Should try with Geoserver 2.13 that implemented isolated workspaces. We did not tried it yet. The 2.13 stable version was just released on 19 March, so one month ago (beta was released on 28 February).