Rong,
Yes, there are known issues with the way labeling is occuring now. I've
sent quite a few message on this top to this and the geotools-devel
list.
The main problem is that the current labeling system is designed to
display road name labels, and hasnt been abstracted out to handle other
types of situations.
So, when you label something, you are actually grouping all the geometry
objects with the same label together. Each group is then evaluated and
and the best "representational" geometry is created. That one geometry
is then labeled.
So, if look at most road data, you'll see a set of "Broadway" line
segments - one for each block. You dont want to label ever segment
with "Broadway", so we network together all the lines with the
"Broadway" label and just label this "supra" line. It works extreamly
well for what its doing.
It also doesnt allow labels to overlap.
Unfortunately, for many other tasks, you dont want this to process to
occur. In your case, you have multiple objects with same label, so
they are being grouped together and only one label is being drawn.
A full "fixing" of this would involve:
1. Some type of geotools/geoserver-specific parameters in the SLD file
to control how the labeling occurs.
<VendorSpecificParameter name="...">...</VendorSpecificParameter>
* This shouldnt actually be too difficult to do, but it does mean
that these SLD file may not work on other WMS servers (which is one of
the main advanatages of using SLD).
* The PMC has already voted to accept this change, but no-one's
actually done it.
2. Change the render/labeler to actually properly process these
parameters. This could actually be HUGE amount of work since there are
some very fancy labeling algorthms available.
What needs to occur first is to figure out what people actually want in
their labeler(s) and then figure out how we're going to represent those
options (#1) and how to actually implement them (#2).
Now that I've got anti-aliasing on (my WMS pictures look good), I'm more
motivated to actually work on this. Do you want to work together on
this? Anyone else?
dave
----------------------------------------------------------
This mail sent through IMP: https://webmail.limegroup.com/