[Tilecache] GetCapabilities document is not valid

[Holger Jaekel]

> -----Original Message-----
> From: tilecache-bounces at openlayers.org 
> [mailto:tilecache-bounces at openlayers.org] On Behalf Of 
> Christopher Schmidt
> Sent: Tuesday, March 04, 2008 3:19 PM
> To: tilecache at openlayers.org
> Subject: Re: [Tilecache] GetCapabilities document is not valid
> 
> On Tue, Mar 04, 2008 at 10:01:33AM +0100, Holger Jaekel wrote:
> > <Layer queryable="0" opaque="0" cascaded="1">
> >   <Name>linktothepast</Name>
> >   <Title>linktothepast</Title>
> >   <SRS>EPSG:4326</SRS>
> >   <BoundingBox SRS="EPSG:4326" minx="0.000000" miny="0.000000"
> >                         maxx="4096.000000" maxy="4096.000000" />
> >   <Layer queryable="0" opaque="0" cascaded="1">
> >     <Name>boston-freemap-remote-mapserver</Name>
> >     <Title>boston-freemap-remote-mapserver</Title>
> >     <SRS>EPSG:4326</SRS>
> >     <BoundingBox SRS="EPSG:4326" minx="-180.000000" 
> miny="-90.000000"
> >                           maxx="180.000000" maxy="90.000000" />
> >   </Layer>
> 
> This isn't correct: that reads (to me) as if the boston freemap is a
> sublayer of the zelda link to the past layer. But that's not true:
> they're both children of the TileCache parent.

You are right, boston-freemap-remote-mapserver should not be defined as
a sublayer of linktothepast. This would be a technically valid document,
but it's not correct.

> 
> So, there's two possibilities here:
>  1. WMS doesn't need the <Layer> container. This is possible: 
> I was told
>     it did, and I confirmed that other WMS servers used it.
>  2. TileCache needs to create a bogus layer with a name, title, srs,
>      boundingbox? That doesn't make sense to me. Which I 
> think means the
>      answer is 1. Which means I should have been more diligent in
>      pushing back on stvn.
> 

I re-read the section 7.1.4 (GetCapabilities Response) of the WMS spec v
1.1.1 and I suggest possibility three:

3. WMS needs a <Layer> container which defines at least a <Title>. Table
7 of the spec defines which attributes of the enclosing <Layer> are
inherited by the enclosed layers. Enclosing layers does not define
sublayers, but inheritance of attributes between layers.

So maybe you create a enclosing layer with the name "TileCacheLayers"?
This would make the document valid and it is conforming the
specification. An even better solution would be to add a [map] section
to tilecache.cfg where you can define the title (and maybe other
attributes).

> > The attribute "SRS" of "BoundingBox" must be written in 
> upper case. The
> > values for maxx and maxy look strange for me, but this may be a
> > configuration problem.
> 
> 'Problem' is a relative term. the Zelda layer powers
> http://tilecache.org/demos/linktothepast.html , which is not quite
> Geographic.

I did not know that. But "srs" should be changed to "SRS".

Regards,
  Holger

---
Holger Jaekel
phone: +49 (89) 121528-75	mailto:holger.jaekel at gaf.de	
fax:   +49 (89) 121528-79	http://www.gaf.de	
GAF AG     Arnulfstr. 197     80634 Muenchen     Germany

Vorstand: Dr. Peter Volk, Aufsichtsratsvorsitzender: Marcello Maranesi

Amtsgericht Muenchen HRB 140 509, Firmensitz: Muenchen