Open tbonfort opened 12 years ago
Author: dmorissette Date: 2004/11/22 - 22:39
I believe we have code in mapwmslayer.c to add the trailing "&" or "?" that
could easily be extracted to a function and shared between the WMS and WFS code.
Author: tylermitchell@shaw.ca Date: 2004/11/24 - 06:10
I'm setting up WFS server and think I'm hitting a related bug, using
4.4b3/current CVS.
When trying to get capabilities of my WFS server, I'm getting:
----
XML parsing error
fatal parsing error: letter is expected in line 19, column 126 (points to the
& at the end)
<OnlineResource>http://spatialguru.com/cgi-bin/mapserv440b3?map=/srv/www/vhosts/spatialguru/maps/apps/global/globalwfs.map&</OnlineResource>
----
My online resource in the map file is set to:
<snip>...?map=/srv/www/vhosts/spatialguru/maps/apps/global/globalwfs.map"
I've tried adding & or & or & with no change.
Is this related?
Author: dmorissette Date: 2004/11/24 - 16:49
> I've tried adding & or & or & with no change.
Are you sure that "&" doesn't fix the issue? I can't see any other reason
for the XML parser to complain. Can you please try with "&" again?
Author: tylermitchell@shaw.ca Date: 2004/11/25 - 05:38
Yep, same kind of response. Note that it appears to ignore the & and tries
to replace it with &.
XML parsing error
fatal parsing error: letter is expected in line 19, column 131
<snip>...mapserv440b3?map=/srv/www/<snip>/globalwfs.map&&</OnlineResource>
The error points to col 131, the "<" symbol.
Author: tylermitchell@shaw.ca Date: 2004/11/25 - 05:45
Before you look further let me make sure I've got the version right here, I
have some doubts...
Author: tylermitchell@shaw.ca Date: 2004/11/25 - 06:03
Still confirmed with fresh CVS sources. Problem with my parser?
Author: dmorissette Date: 2004/11/25 - 06:18
not a parder problem: there is indeed a problem with the trailing '&' in the XML
output that you quoted. What I don't understand is that the behavior you
describe would be complete ly the opposite of what Tom initially reported.
In Tom's case he was complaining that MapServer was *not* appending a trailing
"&", and in your case you do get one but it's just not encoded properly.
I think I know how to fix this, but I'm wondering if you are both talking about
the same thing. Um... I think I understand now:
- I think Tom may have been talking about the use of wfs_onlineresource in WFS
client layers
- I think you are referring to problems with WFS server
If that's the case then I know how both can be fixed, and I think we need this
fixed in 4.4
Author: dmorissette Date: 2004/11/25 - 06:24
Tyler, I have verified that the WFS server calls msOWSGetOnlineResource(), so it
does the right thing already with respect to appending the trailing "&" and
encoding the string. Can you please attach a mapfile that reproduces your problem?
Author: dmorissette Date: 2004/11/25 - 07:28
Okay, it turns out that there were two different problems and they are both fixed:
1- The original problem that was reported in this bug was that we didn't check
fr the presence of "?" or "&" at the end of the layer->connection when building
the WFS client request. This is fixed now by calling the new
msOWSTerminateOnlineResource() function (code extracted from
msOWSGetOnlineResource().
2- Tyler's problem, was in WFS server mode. If wfs_service_onlineresource was
not specified then we were using the value of wfs_onlineresource unencoded. I
have fixed the server code to use the encoded value.
Author: assefa Date: 2004/11/26 - 17:20
Just to note that I have tested the initial bug (trailing ? or & missing) for
wfs client layers and all seems to work well.
Author: dmorissette Date: 2004/11/26 - 17:34
Thanks.
Reporter: assefa Date: 2004/11/22 - 22:10