ghilesz / geoml Goto Github PK

After make :

Error: missing source file "./gcode/utils.ml" for package "gcode"

Hello,
Since OCaml 4.00, there is the function hypot in the standard library. It is said to avoid undue overflow or underflow.
It could be used in Point.distance and Vector.magnitude. In my program, using hypot led me to different (and hopefully more precise) results.

Best,
Richard

title is self explanatory

Hi,

There is a mistake on Segment.contains. You should use Point.distance instead of Point.sq_distance.

Best regards,
Richard

Not sure if this is a pilot error...

After cloning the repo, compiling with make gives these error messages:

dune build src/geoml.cma
Entering directory '/Users/thomasbrupbacher/temp'
File "geoml/src/constraint.mli", line 7, characters 18-24:
7 | type t = private Line.t * comp
^^^^^^
Error: Unbound type constructor Line.t
File "geoml/src/polygon.mli", line 5, characters 17-24:
5 | type t = private Point.t list
^^^^^^^
Error: Unbound type constructor Point.t
File "geoml/src/curve.ml", line 4, characters 11-18:
4 | type t = Point.t * Point.t * Point.t
^^^^^^^
Error: Unbound type constructor Point.t
make: *** [build] Error 1

Being relatively new to Ocaml, I have no idea how to fix this.

Environment:
MacOS 10.14.x
Ocaml 4.08.1

Whoa, GeoML is amazing! I use it for all my ocaml geometry problems. But I'm so sad I can't directly get voronoi diagram :'( Have you planned VD for the future of GeoML?

drawing.ml : overload the drawing functions of graphics for points lines and geoml types

Using module Polygon :

open Polygon
let _ =
  let l = [(0.,0.);(0.,1.);(1.,1.);(1.,0.)] in
  let points = List.map (fun (x,y) -> Point.make x y) l in
  area points |> print_float 

returns -1

Hello,

I have discovered a bug several months ago on Circle.intersection. I fixed it in:
richardalligier@b3c59e3/src/circle.ml

I am not "git fluent" and the commit containing this fix also include others things so I cannot easily make a pull request so please feel free to make a copy/pasta... sorry...

Just a few additional suggestions:
_As the doc is automatically generated, I do not feel like it should be included in the git
_I feel like that modeling lines as (y=ax+b or x=c) is prone to numerical issues, so i have changed as (ax+by+c=0) (see richardalligier@b3c59e3/src/line.ml ). Likewise, the Line.intersection uses the package lacaml to solve the linear system. I think it will do a better job than me at dealing with numerical issues. However it introduces a dependence... If you are interested in all the things I suggested, you can merge the commit richardalligier@b3c59e3

Best,
Richard

self explanatory.