A Cartesian coordinate system is generally assumed i.e there exists an x-y coordinate system so there is an inherent ordering property here.
Regarding atan2, this makes interesting reading: http://en.wikipedia.org/wiki/Atan2
All I am asking is the documentation for atan2 conform with the correct definition. You are actually using atan2(y,x) in postgresql.
The inverse tangent is defined as arctan(y/x). Hence atan2 should be atan2(y,x) to be consistent with this definition. This conforms with C++, C usage.
Andrew
-----Original Message-----
From: Bruno Wolff III [mailto:bruno@wolff.to]
Sent: Wednesday, 5 September 2007 12:34
To: a.maclean@cas.edu.au
Cc: pgsql-general@postgresql.org
Subject: Re: [GENERAL] Documentation fix regarding atan2
On Wed, Sep 05, 2007 at 10:37:18 +1000,
Andrew Maclean <andrew.amaclean@gmail.com> wrote:
> In Table 9.4 of the documentation atan2 is described as follows:
> atan2(*x*, *y*) inverse tangent of *x*/*y*
>
> I am sure it should read as:
> atan2(*y*, x) inverse tangent of y/x
Aren't those two statements sayiong the same thing?
You've just switched the names 'x' and 'y' and not changed their relationships.
>
>
> You can easily test this:
> If y = 2, x = 1, then degrees(atan(y/x)) =63.4 but if we proceed according
> to the documentation; degrees(atan2(x,y))=degrees(atan2(1,2))=25.6 which is
> not the same as degrees(atan(y/x)).
In this example you switched things around part way thorugh. atan2(1,2)
is the atan of (1/2), not atan(2/1) as used at the beginning of the example.
> So it must be degrees(atan2(y,x))=degrees(atan2(2,1))=63.4.