[Requests] comment #1 on candidate OGC WMTS Simple Profile (OGC 13-082)

Keith Pomakis pomakis at cubewerx.com
Tue Jun 17 11:59:21 EDT 2014


PART A
------

1. Evaluator:
         Keith Pomakis <pomakis at cubewerx.com>
         CubeWerx Inc.
         815 boulevard de la Carrière, bureau 202
         Gatineau, Québec, Canada
         J8Y 6T4
         Tel: 819-771-8303, ext. 202

2. Submission: candidate OGC WMTS Simple Profile (OGC 13-082)


PART B
------

1. Requirement: General

2. Implementation Specification Section number: 6.1

3. Criticality: Editorial

4. Comments/justifications for changes:

Section 6.1 states:

     In a general WMTS the only way of knowing the extent of a tile
     expressed in CRS coordinates is to extract the TileMatrixSet
     and the TileMatrix identifier from the request, retrieve the
     ServiceMetadata document and read the TileMatrix definition (xpath:
     /Capabilities/Contents/TileMatrixSet[ows:Identifier={TileMatrixSet}]/
     TileMatrix[ows:Identifier={TileMatrix}]) in order to know
     the TopLeftCorner, ScaleDenominator, TileWidth and TileHeight
     values. Then, by extracting the TileRow and TileCol from the request
     it is possible to apply the math explained in section H.2 of the
     WMTS core standard (OGC 07-057r7) and finally, get the bounding
     box of the tile which will enable an integrated client to place
     the tile in the right position on the screen.

This is exaggerating the complexity of the calculations required.
It implies that the client application must parse its own tile-request
URLs and backtrace the information to the matching TileMatrixSet and
TileMatrix in the ServiceMetadata document to figure out the bounding
boxes of the tiles.  But this is certainly not the way WMTS clients would
be implemented.  A general WMTS client starts by reading the server's
ServiceMetadata document and locating the appropriate TileMatrixSet
and TileMatrix to use for the layer to be rendered.  The appropriate
tile-request URLs are prepared with this knowledge, so the client
application already knows (with the simple calculation presented in
Annex H.2) exactly what the bounding box is for each requested tile,
and can also easily calculate the pixel location it should be placed
at on the screen.

-- 
Keith Pomakis <pomakis at cubewerx.com>
Senior Software Developer, CubeWerx Inc.


More information about the Requests mailing list