MapServer / MapServer-import

3 stars 2 forks source link

[OWS-Server] tighter enumeration on SERVICE value #1503

Open tbonfort opened 12 years ago

tbonfort commented 12 years ago

Reporter: tomkralidis Date: 2005/10/21 - 20:56

Hi,

For OGC:WMS, OGC:WFS, OGC:WCS, there exists a service keyword.

The values it should accept for the OWSs we support are:

WMS
WFS
WCS

..currently, we accept

WMS
WFS
WCS
wms
wfs
wcs

The values must be upper case to be valid.
tbonfort commented 12 years ago

Author: assefa Date: 2005/10/24 - 19:46

I thing all/most of the comparing of parameters and values in OWS are done in 
non case sensitive manner.
 Should we review all the values to ensure that they are compliant to the 
specs ? 
tbonfort commented 12 years ago

Author: tomkralidis Date: 2005/10/24 - 20:13

Are we sure about this?  Wasn't MapServer run through OGC CITE for WMS and
WFS-basic?

In general, for OWS specs, keywords are case insensitive and keyword values are
case sensitive.  So:

request=getmap is invalid

ReQUEst=GetMap is valid

We should give a once over to all keyword/value pairs/params for supported OWS's.
tbonfort commented 12 years ago

Author: assefa Date: 2005/10/24 - 20:32

I have just tried getmap and getcapabilities (all lower case) and I have valid 
returns, so I guess there is a problem on all or part of the evaluation of the 
keyword values.
 I am not sure if case sensitivity is tested in OGC cite tests. I browsed 
quickly through them now and could not find one ( I may have missed it). 
tbonfort commented 12 years ago

Author: nsavard@mapgears.com Date: 2005/10/25 - 20:04

I checked all the OGC necessary tests to pass the compliance and none of them
verify the case sensitivity of the "request" parameter or the "service"
parameter.  All parameter values are always written the same way:

GetMap, GetCapabilities, capabilities, GetFeatureInfo.  The "6.4.1. Parameter
Ordering and Case" point of the specification is not verified anywhere.
tbonfort commented 12 years ago

Author: tomkralidis Date: 2005/10/25 - 20:39


Right.  So is this to say we obey the spec or the CITE tests?

I would go with the former.  The CITE tests are supposed to be tests which test
the implementation specification.

Perhaps someone can ask the WMS/WFS/WCS or CITE folks?
tbonfort commented 12 years ago

Author: assefa Date: 2005/10/25 - 22:19

OGC certification comes by passing the tests so who really cares about the 
specs :) 

Seriously I don't have any strong opinion on this. We could address them all 
(risk is to break existing apps). Or adress them partialy. Or status quo.  You 
could maybe bring this to the mapserver list. In any case I think here are the 
elements that we need to correct (I hope I did not miss some ):

  1)wms
    - seRvice=WMS
    - request = capabilities  or request = GetCapabilities
    - request = context  or request = GetContext (this one is not OGC standard)
    - request = GetMap
    - request = GetLegendGraphic
    - request = GetStyles
    - request = GetFeatureInfo
    - request = DescribeLayer

  2) wfs
     service = WFS
     request = GetCapabilities
     request = DescribeFeatureType
     request = GetFeature

  3)wcs
     service = WCS
     request = GetCapabilities
     request = DescribeCoverage
     request = GetCoverage
tbonfort commented 12 years ago

Author: dmorissette Date: 2005/10/25 - 23:47

Tom, MapServer being a bit more relaxed than what the spec requires for its
input parameters doesn't prevent or hurt interoperability in any way, well, if
anything that just allows semi-compliant clients to continue to exist. My vote
goes to leaving things as is (i.e. case-insensitive request parameter matching)
unless a good case can be made for changing them and risk breaking existing apps
based on semi-compliant clients.
tbonfort commented 12 years ago

Author: nsavard@mapgears.com Date: 2005/10/26 - 14:19

Normally the spec gets priorities over tests.  But I do agree with Daniel's
comment (#7).  We could not risk to break something in MapServer unless we have
a good reason to do so.
tbonfort commented 12 years ago

Author: tomkralidis Date: 2005/10/26 - 15:35

OK, I see your point.  We are still supporting the spec no less.