thomas.bonfort at GMAIL.COM
Mon Oct 29 14:10:10 EDT 2007
On 10/29/07, Steve Lime <Steve.Lime at dnr.state.mn.us> wrote:
> I agree that both approaches can co-exist. There is renderer-based simplification
> already in place although it had an ugly bug with certain (rare) paths- hence Thomas'
those simplifications weren't applied when using agg (testing for
equality on integer rounded values with gd discarded some points, but
that never happened with floating point values with agg). Thus my
question if a similar strategy should be applied.
> On the second question, I had never worried about multipoint features because they
> were relatively rare, and even when encountered they had relatively few parts. You're
> talking layer-level generalization I take it. Get's a bit trickier though since you'd have to
> cache the point and the class being used. Wouldn't think it's worth it but I could be wrong.
actually, with your response on the bug report, my question is out of
date. I wasn't referring to multipoint features, just different point
features that happened to fall in the same pixel when rendering a
zoomed out view.
More information about the mapserver-dev