Boost logo

Boost :

Subject: Re: [boost] [gsoc19] Seeking a mentor for Boost.Geometry project related to triangulation and random sampling
From: Tinko Bartels (tinkobartels_at_[hidden])
Date: 2019-03-19 11:19:51


Hi Vissarion,

sorry for the double post, I have only received your e-mail after posting my
response to Adam.

> Hi Tinko, Adam,
>
> thanks for the interest in the project!
>
> In general I find your proposal good but a bit overambitious.

I understand, and since this is the second advice in that direction I will
try to limit the proposal.

> I am mainly covered by Adam's response but I have a few comments.
>
> 1. Apart from the two projects (triangulation and random generators)
> your project includes one more, namely "robust geometric predicates"
> that could be a solely GSOC project. This is not described in the
> proposal but what I am understanding is something like this
> http://www.cs.cmu.edu/~quake/robust.html (there are many solutions
> for
> robust predicates I just named one) to be done for both "orientation"
> (aka side) and "in-circle" predicate and for all 3 coordinate
> systems.

This (what Shewchuck described as Adaptive A in his short paper "Robust
Adaptive Floating-Point Geometric Predicates") was indeed the solution that
I had in mind.

> What I propose is to either focus on one topic (e.g. Delaunay
> triangulations) and go further in coordinate system support etc. or
> do
> Delaunay and random generators only for cartesian. In any case you
> can
> include more topics as extras. In other words there will not be a
> problem if you do more that you propose.

I understand. Both (Delaunay triangulation for more then cartesian
coordinates and Delaunay + random generators for cartesian) are interesting.
I have a slight preference towards doing Delaunay triangulation and random
generators for cartesian coordinates and add extras if the work is done
early in the GSoC. In that case, I would specify in my proposal that I
propose these things for cartesian coordinates. This is just a slight
preference, though, and if the maintainers would deem Delaunay for cartesian
and other coordinates more useful, I would do that and rewrite my proposal
accordingly.

Thank you for your feedback as well and kind regards,
Tinko Bartels

--
Sent from: http://boost.2283326.n4.nabble.com/Boost-Dev-f2600599.html

Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk