[mapserver-users] WMS 1.3.0 questions

Yves Moisan yves.moisan at boreal-is.com
Wed Nov 10 16:50:37 EST 2010


> 3- That may not be obvious, but MapServer honors the VERSION that you
> set in a GetMap request...
Hi Daniel,

I would say it's confusing to see VERSION=1.2222 going through ...

> it's just that it works with ranges of
> versions for each version-specific difference in the spec instead of
> discrete version numbers in an attempt to be forgiving of clients that
> support only intermediate versions that MapServer may not be explicitly
> built for. More specifically:
>
> (0.0.0<  VERSION<  1.0.7) is treated as 1.0.0
>
> (1.0.7<= VERSION<  1.1.0) is treated as 1.0.7
>
> (1.1.0<= VERSION<  1.1.1) is treated as 1.1.0
>
> (1.1.1<= VERSION<  1.3.0) is treated as 1.1.1
>
> (VERSION>= 1.3.0) is treated as 1.3.0

A word from a coder is worth a thousand [GetMap] images ;-).  That's 
really clear.  But misleading for the client/map integrator IMO.
> 4- After seeing this thread, I would tend to lean towards producing an
> exception if version is not one of the explicitly supported values to
> prevent any confusion (i.e. anything not in the following list would
> produce an exception: 1.0.0, 1.0.6, 1.0.7, 1.1.0, 1.1.1 and 1.3.0)
>
I think that's a very sensible solution.  Especially since the VERSION 
parameter is required.  Trying to push a GetMap request without a 
VERSION url parameter already generates an error : "msWMSDispatch(): WMS 
server error. Incomplete WMS request: VERSION parameter missing".  I 
think failing to find an explicit VERSION number in a set of allowed 
values should trigger a similar exception,  something along the lines of 
"msWMSDispatch(): WMS server error. Offending (or misformed or 
something) WMS request: VERSION parameter not in {1.0.0, 1.0.6, 1.0.7, 
1.1.0, 1.1.1, 1.3.0}".

That behaviour would at least comfort me as a user that the VERSION 
number I'm passing in is the one used, short of having a response header 
confirming that (which appears to be an impossible task anyways).

Thanx,

Yves



More information about the mapserver-users mailing list