MapServer 5.0 WMS, requiring styles...

Wed Oct 17 14:00:53 EDT 2007

It's really too bad this was not implemented initially.

..Tom

________________________________

From: UMN MapServer Developers List on behalf of Paul Spencer
Sent: Tue 16-Oct-07 10:17
To: MAPSERVER-DEV at LISTS.UMN.EDU
Subject: Re: [UMN_MAPSERVER-DEV] MapServer 5.0 WMS, requiring styles...

I find it amusing that the standard is about interoperability but 
implementing the standard breaks interoperability :)

All major web browsers implement support for various HTML standards 
in 'Standards Compliance Mode' and include a nasty thing called 
'Quirks Mode' which allows the browsers to handle all kinds of non-
compliant HTML markup.  Users of web browsers expect them to 'just 
work' when viewing content over the web.  I doubt a browser would get 
much adoption if some/most of the content on the web was not viewable 
because it wasn't compliant with some standard.

Users of ArcXxxx, GE, and any other WMS client are going to feel the 
same way.  If they connect to content, they want to be able to see 
it, not be told their client software isn't implementing the standard 
exactly and completely.  I know the analogy isn't quite right, but at 
the end of the day the same thing will happen.  Users will complain 
that they can't work with MapServer-based WMS servers and that will 
push people to use something other than MapServer to publish to WMS.

Paul

On 16-Oct-07, at 9:34 AM, Steve Lime wrote:

> Bart: In principle I agree with you.
>
> From a practical point point of view I think it's silly not to work 
> with the largest
> WMS clients out there without additional configuration on the users 
> part. I wasn't
> aware of the ArcMap issue but would advocate supporting that 
> variation too. It's
> not like we're talking about some minor client, I mean, it's Google 
> Earth and ArcGIS.
>
> Fine, we should let them know they are not compliant (good luck), 
> but making it
> hard for the users of those packages doesn't help. They are 
> unlikely to have
> read the WMS spec and will blame the service provider more often 
> than the
> client vendor. These are folks (on the ArcGIS side especially) that 
> we really want
> to have a positive experience with open source and open standards.
>
> I need to go back to the documents Tom references but at face value 
> I'd prefer to
> see MapServer work seamlessly with *major* clients if at all possible.
>
> Steve
>
>>>> Bart van den Eijnden <bartvde at GMAIL.COM> 10/14/07 3:22 PM >>>
> I tend to disagree. As a person having worked a lot with standards, 
> I think
> all vendors implementing a standard should comply with the contract 
> lay down
> in the standard.
>
> Remember ESRI's ArcMap up until 9.1 did not send the required 
> parameter
> QUERY_LAYERS in a GetFeatureInfo request. I did not hear anybody on 
> this
> list advocate that Mapserver should be made permissive and assume
> QUERY_LAYERS equals LAYERS. If you would go this path then the end 
> is near.
> Though it would have solved a long-standing irritating/frustrating 
> issue
> between Mapserver WMS and ESRI's ArcMap (and there are still a lot 
> of 9.0/9.1
> ArcMap's out there).
>
> Also, imagine falling back on the MAP file's width and height if not
> specified, I could write a simple WMS javascript client and forget 
> the width
> and height parameters in the url, the WMS client would write out <img
> src="..." width="650" height="450"> and the browser would totally 
> distort
> the image if the MAP file says something else then SIZE 650 450.
>
> It just shows which clients are not correctly implementing a spec, and
> frankly, the sooner the better. Go and bother the GE people and say 
> they are
> missing a required parameter.
>
> Personally I can not think of a reason why styles is required but 
> has a
> default value, but maybe the people who wrote the WMS spec had a 
> perfectly
> good reason for it. But if you do not agree it is a required 
> parameter, go
> to the OGC and debate why styles should not be a required 
> parameter ...
>
> Best regards,
> Bart
>
> On 10/14/07, Howard Butler <hobu.inc at gmail.com> wrote:
>>
>> On Oct 13, 2007, at 5:26 PM, Frank Warmerdam wrote:
>>
>>> Steve Lime wrote:
>>>> Hi all: Is anyone else running into problems with clients and
>>>> requiring the
>>>> styles parameter? I'm putzing with Google Earth which doesn't seem
>>>> to send
>>>> that parameter which makes using 5.0-based WMS servers a bit
>>>> useless. Or am
>>>> I missing something with?
>>>
>>> Steve,
>>>
>>> I haven't run into this myself, but I have heard problems on IRC.
>>>
>>> I personally think it is wrong headed for us to require the styles
>>> parameter.  If we want to have a "pendantic" mode that requires it
>>> that
>>> would be find, but I don't see why we should require it by default
>>> regardless of what the standard might say.
>>>
>>> I would suggest we be permissive in what we require.
>>>
>>> I think this is something that ought to be "fixed" in 5.0.1.
>>>
>>
>> I agree.  The benefit of enforcing this is negligible.  Same for
>> width and height too, frankly.  Do the CITE tests require these?
>>
>> I would like to go back to being more lax about these, especially
>> when we can get reasonable defaults from the map file for width/
>> height and when styles= infers absolutely nothing to anyone.
>>
>> Howard
>>

+-----------------------------------------------------------------+
|Paul Spencer                          pspencer at dmsolutions.ca    |
+-----------------------------------------------------------------+
|Chief Technology Officer                                         |
|DM Solutions Group Inc                http://www.dmsolutions.ca/ |
+-----------------------------------------------------------------+