Skip to Content.
Sympa Menu

cgal-discuss - Re: [cgal-discuss] Coding Conventions - CGAL methods VS custom implementations

Subject: CGAL users discussion list

List archive

Re: [cgal-discuss] Coding Conventions - CGAL methods VS custom implementations


Chronological Thread 
  • From: Marc Glisse <>
  • To:
  • Subject: Re: [cgal-discuss] Coding Conventions - CGAL methods VS custom implementations
  • Date: Fri, 15 Jan 2016 08:13:24 +0100 (CET)

On Thu, 14 Jan 2016, s_gandhi wrote:

I am confused about a coding convention in order to submit a code to CGAL.
I already read through the developer manual:
http://doc.cgal.org/latest/Manual/devman_code_format.html

Its is been observed, CGAL's functions work slower than custom
implementations.
For e.g. CGAL::left_turn() is significantly slower than my custom
implementation of LfTurn()
The cause is not that concealed. In order to reach CGAL's turn left(), many
wrappers have to be unrolled and therefore call stack increases, and hence
the additional time consumption.

Compiler optimizations (assuming you turn them on) are there precisely to remove all that overhead...

This holds true for other functions as well.
Another e.g.: In order to find mid point between three collinear points,
there is a possibility of using CGAL::collinear_are_ordered_along_line().
Now there are obviously multiple ways of achieving the same result. So if a
code is written, which achieves exact same task, and it gives a considerable
time improvement, should that custom implementation be used?

And more importantly - Will such a code will be accepted by "Editorial Board
of CGAL"?
I am new to this forum, and not aware if members of editorial board are
available, but I expect them to be, so they can put some light on this
issue.

--
Marc Glisse



Archive powered by MHonArc 2.6.18.

Top of Page