Boost logo

Geometry :

Subject: [ggl] Snapping
From: Stephen Leary (sleary)
Date: 2009-12-01 05:17:18

I think the issue is confused... Geometric tolerance is used in geometry
engines because it helps give details about the character of the data. I.e.
minimum line lengths etc. The factor of ten is something that I have found
in my experienc and is down to the character of the data not numerical
precision. This not something ever enforced by algorithms it is a choice by
the coder (but is probably worth keeping in the documentation)

Snapping tolerance is as you say a buffer round a shape which snaps all
parts of a second shape within or on that buffer down onto the nearst
points. How we do that should be to get the best performace.


On Mon, Nov 30, 2009 at 10:59 PM, Mateusz Loskot <mateusz_at_[hidden]> wrote:

> Stephen Leary wrote:
> > 3. Yes and No.
> So, it's an acceptable deviation of something else or
> number of properties.
> > A geometric tolerance is used when for example two vertices are less
> > than the geometric tolerance a comparison operator will return true.
> Yes, it's been clear since the beginning, but...
> > This is not always the numerical precision so much as the limit of
> > what we care about in real life. i.e. less than 1mm on a countrywide
> > map might be the geometric tolerance. Snapping would be 1cm in this
> > case.
> It seems to be confirming my assumption that geometric tolerance will be
> based on actual size and coordinate of geometry.
> ...bug it is not clear to me how the snapping would work:
> - for non-GIS geometries (no CRS, no projection)
> - no precision model
> - the only information we have is: type of coordinate (float, integral),
> dimension, size of geometry (minimal bounding rectangle)
> What would be geometric tolerance in that case?
> A fraction of a size of geometry (or MBR)?
> I imagine, that for vertex-to-vertex snapping this kind of issues could
> be solved relatively easy by generating buffers of vertices using
> snapping tolerance as buffer distance and finding if buffers intersect.
> IOW, point equlity is more fragile than intersection of larger
> geometries. However, that's the simplest case of snpaping :-P
> > 6. I'm happy to once I can include everyones comments.
> Sure. Though, my suggestion was that Wiki is an alive and growing
> animal, no need to wait for all details first :-)
> Best regards,
> --
> Mateusz Loskot,
> Charter Member of OSGeo,
> _______________________________________________
> ggl mailing list
> ggl_at_[hidden]

Stephen Leary
-------------- next part --------------
An HTML attachment was scrubbed...

Geometry list run by mateusz at