[GeoMoose-users] WFS identify is GeoMoose using MapServer

[Brent Fraser]

Hi Jim,

In my efforts to add WFS mapsources to our mapbook-test-servers.xml, I see that previously I had used

<map-source name="wfs-polygons"               type="wfs"     title="Parcel Polygons">
<url>http://localhost/cgi-bin/mapserv.exe</url>
<param name="map" value="C:/ms4w/apps/gm3-demo-data/demo/parcels/parcels.map"/>

which is not very "robust" for testing.  It's in our demo at demo.geomoose.org and will not work for most people experimenting with 
https://demo.geomoose.org/3.x/desktop/?mapbook=test

I'd like to change the mapbook-test-servers.xml  to use WFS at demo.geomoose.org  for parcel polygons, points and pipelines.  I guess I would rather not include a map param for each.  How much work do you think it would be to have the map files handled by the server for our demo?

Thanks!
Brent

----------------------------------------
From: "Jim Klassen" <klassen.js at gmail.com>
Sent: 1/3/23 10:51 AM
To: geomoose-users at lists.osgeo.org
Subject: Re: [GeoMoose-users] WFS identify is GeoMoose using MapServer

I would guess that it is non-compliant to send/ask for a SRS that isn't advertised in the GetCapabilities.  I think the bug is that MapServer is working at all.

Personally, I think type="wfs" should be strongly preferred and type="mapserver-wfs" is the hack.  It is a work around for `/cgi-bin/mapserv?map=/file/system/path&` being ugly.  "mapserver-wfs" really isn't designed to work across multiple mapserver sites either.  IMO MapServer shouldn't be exposing (full) filesystem paths to the web request at all and MapServer 7/8 have been making steps to avoid this more.  If it were up to me, I'd remove the "mapserver-*" types entirely.  I'd expect MS_MAPFILE_PATTERN is already set on the server correctly, otherwise it wouldn't work at all.

I'm still partial to adding an Apache handler for ".map" so mapfiles are "run" from within the webroot like a CGI or PHP file would normally be run, but instead of Apache passing it to PHP or Perl/Python/whatever for handling, the mapfile gets passed to MapServer (via environment variables) to run.  A side benefit is Apache sees the path to the mapfile and normal Apache authentication/authorization can be applied within the standard <Directory> or <Location> stanzas.  I don't see any reason to want MapServer to implement yet another authentication/authorization system when Apache is already there.  This works well at least from MapServer 5-7. I haven't tried it with 8 yet.  The catch is that MapServer doesn't auto detect the server URL correctly when doing this so "ows_onlineresource" needs to be set to the URL (as seen by end users) to the mapfile.  (I have a patch for MapServer that lets the auto-detection still work by looking at the SCRIPT_URI environment variable.)

On 1/2/23 16:44, Brent Fraser wrote:
Hmm,    I would say that GeoMoose should not include a MAP= in the request if there is no <file> in the mapsource definition.

From: "Dan Little" <theduckylittle at gmail.com>
Sent: 1/2/23 3:40 PM
To: bfraser at geoanalytic.com
Cc: Jeff McKenna <jmckenna at gatewaygeomatics.com>, GeoMoose Users <geomoose-users at lists.osgeo.org>
Subject: Re: [GeoMoose-users] WFS identify is GeoMoose using MapServer

I suspect MS_MAPFILE_PATTERN needs set. Ala: https://mapserver.org/mapfile/config.html
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/geomoose-users/attachments/20230104/115d2f5c/attachment.htm>