Geometry Boost List: [geometry] 64 bit. Fix to a few 64 bit warnings

From: Johan Doré (jdh_at_[hidden])
Date: 2021-01-05 09:08:07

Dear All,

We were just compiling Boost Geometry Extension using VS 2017 64 bit and Warning Level 4.
Doing that showed a few warnings regarding size_t to int conversions which you might want to change.

It is in the file:
..\boost\geometry\extensions\algorithms\detail\overlay\dissolver.hpp
Line 436:
int n = boost::size(output_collection);
and Line 580
int size = 0, previous_size = 0;

We assume that both should be size_t.

Thanks a lot for all your hard work

Johan


Best regards,

Johan Doré
CTO, Vice President


[cid:logo_visiopharm_1line_rgb_blk_5b11a6fe-af82-480b-84be-c356b54bf0c6.png]


M: [phone]
P: [phone]
C: Hørsholm, Denmark
W: www.visiopharm.com<http://www.visiopharm.com/>






From: Geometry <geometry-bounces_at_[hidden]> On Behalf Of Chris Watts via Geometry
Sent: 31. december 2020 10:32
To: geometry_at_[hidden]
Cc: Chris Watts <Chris_at_[hidden]>
Subject: [geometry] Polygons appear to have serious rounding errors

Hi all,

I've been trying to debug an issue with geographic_point_circle wherein I can't seem to create a circle of radius 10 metres.

The easy-to-reproduce symptom is that the area of the circle varies greatly with only a small change in latitude or longitude.

See this StackOverflow post for details: https://stackoverflow.com/questions/65493910/boost-point-circle-coming-out-in-weird-shapes/

In order to produce a suitable circle of that scale, I need to use long double in my point type, but with an epsilon of ~1mm, a regular double ought to work fine. Is there something I'm missing about the implementation of point_circle or bg::area that would imply that this is expected behaviour?

Kind regards,
Chris