[MAPSERVER-USERS] WMS server with non-EPSG SRS?

[Neil Best]

Frank Warmerdam-2 wrote:
> 
> The normal practice if you want a non-EPSG coordinate system to be usable
> for WMS serving with MapServer is to just extend the epsg init file with
> the definition.  Is this unacceptable for some reason?
> 

I have been reluctant to do this because:

1)  I was concerned about tinkering with core software files in a production
system, without any hard evidence that there was any chance for an
end-to-end solution (see #3).  At the moment I don't have a segregated test
bed where I can test the effect of such a change on my Mapserver application
and guarantee no interruption.  Is it still true then that Mapserver won't
even respect "init=esri:xxxxx"?  

2)  I would be extending a popular, mature standard in an ad hoc, maverick
fashion (which is central to some business models, but not ours!), mainly
because of the risk of future complications as the standard evolves and the
maintenance overhead of propagating my amendments and remembering to reapply
them any time that software is upgraded or new systems are provisioned.

3)  I have no reason to expect that the client (ArcGIS Server in this case)
will understand what it is that is being advertised, assuming that an
analogous look-up table is contained somewhere within its WMS client
implementation.  Is there any reason to think that Arc's WMS client relies
on the proj support files that are included with the GDAL code that ships
from ESRI?  Do I really want to get into the scenario of violating my
customer's end user license or voiding their warantee by editing files
installed by ESRI software?

On the other hand, it is useful to know that such an edit is considered a
"normal practice" by someone I can trust to make such a judgment.  That
said, is there anyone else who considers the standard unnecessarily
constraining in this respect or am I shouting into the abyss?

Given the situation the best path forward that I can see is to serve the
data in some geeric EPSG SRS and have the client re-reproject it and accept
the performance hit in favor of taking the path of least resistance, but I
don't actually know if the client can do that -- I wouldn't expect QGIS to
be able to, for example, unless it is now reprojecting rasters on the fly
where it has not in the recent past.  We need to test this possibility on
the Arc side -- maybe someone has experience with this . . .

-- 
View this message in context: http://www.nabble.com/WMS-server-with-non-EPSG-SRS--tp17342340p17343514.html
Sent from the Mapserver - User mailing list archive at Nabble.com.