Discussion #2682

Predefined (datasets) or not Predefined that is the question

Added by James Passmore over 4 years ago.

Status:New
Priority:Normal
Assignee:-

Description

I have removed this discussion (but not the table) from the draft WCS TG where I had it as an editor note and am adding it to the wiki for wider discussion (or not)

 

1

2

3

4

5

6

Get Download Service Metadata

Get

Spatial

Data Set

Describe Spatial

Data Set

Link Download Service

Get

Spatial Object

Describe Spatial Object Type

Direct access

download services

MUST

SUPPORT

MUST

SUPPORT

MUST

SUPPORT

MUST

SUPPORT

MUST

SUPPORT

MUST

SUPPORT

Other download services

MUST

SUPPORT

MUST

SUPPORT

MUST

SUPPORT

MUST

SUPPORT

 

CAN’T

SUPPORT

 

This is by definition, if a service can support these operations it becomes a Direct Access DLS

Table 4: Summary of operations that must be supported by download services

 

Current(WFS+ATOM) TG refer (in the WCS draft TG it would be section 4) to pre-defined data (download services) verses direct-access download services at some length. So using the above table as reference the current (WFS+ATOM) TG tells us that operations 1 to 4 are pre-defined data (download services) and operations 5 & 6 are direct access download services, and this doesn’t make sense ~ the table shows us (and the regulation tells us) that operations 1 to 4 are equally applicable to direct access download services, so they can’t have different definitions.

So we might say that operations 1 to 4 apply to predefined data sets whether they they are acted upon by a direct access download service or not, but does this help us?

If we think about this a bit further we can see that when we do a GET SPATIAL DATA SET [2] or a GET SPATIAL OBJECT [5] operation (when we actually get the data) the INS NS tells us that for operations that result in the retrieval of a data set, that we must always be supplying the spatial data set identifier.

So if we know the spatial data set identifier in the request doesn’t this imply that the data set is ALWAYS predefined?

I think it does.

All we are doing in a search query then is to apply/request some representation of that predefined dataset.  If we look instead at INS NS Annex V, that deals with transformational services and operations therein, we can see that there is no requirement to specify an existing spatial data set identifier, the result of a transformation is  a new data set, that is, the resultant data set is not predefined.

A better way of thinking about this might be that in operations 1 to 4 the query request is constrained to some degree that is, it is the query that is predefined; and depending on implementations the result of such a predefined query might be a set of data that is always the same, for example an atom feed result will return the same data in two successive requests, but in a direct access service the actual data might be different, for example if a WCS is providing data from a database that is constantly being updated.

I have therefore removed the discussion of pre-defined data (download services) verses direct-access download services in this TG just because it makes no sense.

Luckily neither the INSPIRE directive, INS ISSDS, or INS NS actually mention ‘predefined’ when talking about download services, so we don’t have to mention it in this TG.

So technically it shouldn't be a problem.  I understand though that it means that this TG is in conflict with the current understanding of how things work for Atom and WFS services, so I'm thinking that I/we should put together some additional prose to explain the differences.

 

Also available in: Atom PDF