Christopher Schmidt wrote:
It would prevent it, because the problem only presents itself at tile
edges -- and the data from MapServer no longer *has* any visible tile
edges in a meta-tiled TileCache instance. To see similar behavior to
TileCache metatiling, set the 'gutter' option on your WMS layer to '15'
or something similar (see http://openlayers.org/dev/examples/gutter.html)
and I bet you'd see the problem disappear.
In your example, I can still see the problem even with the gutter turned
on. It is almost gone, but there is a still a small defect in the
rendering at the edge. It is masked by the anti-aliasing and thickness
of your lines. In my case, with no anti-aliasing and single-pixel lines,
it is far more visible ... or rather, completely invisible when it
shouldn't be.
My understanding of how lines are rasterized still says to me that it is
impossible to get pixel-perfect seemlessness unless the renderer on
either side of the tile boundary is rendering the exact same line. If
you make the "gutter" big enough, it will help to make this the case,
but if your lines are too long the gutter solution still won't work.
If you're not convinced, try the gutter option. If that doesn't fix it,
you'd be right -- but I bet it will.
Hmm, I tried it, I'm using mapserver 4.8.4, and it didn't help at all. I
even tried making it really big, to the point that I really did expect
it to help, and it still didn't so I'm not convinced either way yet. You
can have a look at the problem here, with the gutter set to 100:
http://draum.refractions.net/gisi/test.html
Notice the missing border between the Metchosin and Malahat polygons, if
you zoom in far enough it will draw it.
Looking forward to your feedback.
Chris
