Re: [Jts-topo-suite-user] Computing with fixed precision model

[Martin D. <mtn...@gm...>]

Yes, I'm aware of this.

Using fixed precision tends to exacerbate robustness errors, due to the
nature of the JTS noding algorithm.  A better approach using snap-rounding
would help with this. As you point out, the ability to compute in a
fixed-precision is essential if the destination store for the data requires
fixed precision itself.

As for buffer performance, I'm not sure why this is the case, but it may
also be due to robustness errors showing up more freqeuently, and thus
triggering the heuristics for handling them - which are slow.

If you have the WKT for the buffer test case I'd be interested in trying it
out.

On Sun, Jan 27, 2013 at 4:24 PM, Michaël Michaud <mic...@fr...>wrote:

>  Hi Martin,
>
> I encounter much more problems performing computations with the fixed
> precision model than with the floating point one.
> Is this something you're aware of ?
>
>