Click Zoom + This runs reasonably fast, BUT the Features: 100. This is
clearly wrong. maxfeatures=50 and there are less displayed because of
the zoom in. So this looks like a bug in OL where on zoom in:
1) you should reuse the current feature set for the layer if maxfeatures
is NOT set
2) you should clear the current feature set if maxfeatures IS set and
re-query for new features.
3) ALWAYS clear the current features set and requery, because in a
multiuser environment the features on the server might have been updated
and you want to get a fresh copy.
4) Allow some combination of these behaviors to be set in the layer
definition similar to maxfeatures.
Click Zoom + again and it takes about 5 mins for the browser to process
the request. Features: 139
Reload the page to start over.
Click on Algers, DZ
Click Zoom -, This takes about 5 mins to return with NO poi displayed.
The mapfile has MAXSCALE 50000 set for the WFS LAYER
Step 5: after Step 4 returns, you can zoom out fast and zoom in until
you hit the WFS points and if you try to zoom out again you get the same
I hope this helps, let me know if you need more info.
Christopher Schmidt wrote:
> On Fri, Mar 30, 2007 at 05:09:00PM -0500, Stephen Woodbridge wrote:
>> Is there a work around to bug #541 like putting a square about the point
>> instead of a circle? How would I do that to give it a try?
> The workaround is to update to SVN HEAD. :) Schuyler and I were able to
> reproduce with the WFS example you had provided, and were able to track
> the bug down to bad behavior in Firefox when drawing circles of radius
> less than .0001. Opera and Safari were fine.
> All should now be fixed. Let us know if it isn't. (We'll probably be
> pushing for an RC soon: I tested with the imaptools URL you gave, but if
> you have some other failing case, please let us know as soon as
Re: [OpenLayers-Users] Work around for bug #541 ??
On Mon, Apr 02, 2007 at 04:31:30PM -0400, Stephen Woodbridge wrote:
> I updated to trunk today, did a build.py full and I'm using the
> OpenLayers.js - Thank you getting svn fixed up.
> I am still having serious performance problems.
> Step 1:
> http://imaptools.com/ld-test.html > Click on Algers, DZ link this displays just fine and is very fast notice
> the Features: 50, I have maxfeatures=50 set.
> Step 2:
> Click Zoom + This runs reasonably fast, BUT the Features: 100. This is
> clearly wrong. maxfeatures=50 and there are less displayed because of
> the zoom in.
Agreed. The WFS layer should be clearing out its feature list when the
map is redrawn (when more data is requested from the server). Up until
now, the renderer was being cleared, but the features were still on the
list. I've opened #622 to fix this.
> 3) ALWAYS clear the current features set and requery, because in a
> multiuser environment the features on the server might have been updated
> and you want to get a fresh copy.
^^ This is the one we'll go with for now.
> Step 3:
> Click Zoom + again and it takes about 5 mins for the browser to process
> the request. Features: 139
I'm not seeing this as being slow.
> Step 4:
> Reload the page to start over.
> Click on Algers, DZ
> Click Zoom -, This takes about 5 mins to return with NO poi displayed.
> The mapfile has MAXSCALE 50000 set for the WFS LAYER
Are you sure that this isn't your server? I saw several times while
testing yesterday that mapserver ended up just hanging out for 2-3
minutes and then never returning an image. If you zoom out from the
algers jump, no WFS is involved, just the Mapserver Untiled layer, and
it seems to me like your map is just slow :) Check the serverside on
this one, maybe?
> Step 5: after Step 4 returns, you can zoom out fast and zoom in until
> you hit the WFS points and if you try to zoom out again you get the same
Interesting. It's possible that your computer is slower than mine such
that things are just really slow for you, but I don't really understand
anything difficult that would be happening at that stage.
I'll work on getting the WFS to clear out its features, and we'll see if
that helps the rest of your performance issues, but, as I said, I'm not
entirely convinced, since I'm not seeing anything interesting that could
be causing this slow down, nor can I reproduce it in your application.