Re: [PATCH] Revive line type

On Sun, 2013-06-23 at 12:24 +0800, rui hua wrote:
> Regression tests are all succeed, but several problems have be found while
> ding some simple test. The updated document said that the points used in
> the output are not necessarily the points used on input. I understand that
> as long as they are marked on the same line. But when the input data
> represents a horizontal or vertical line, the output is not exactly the
> same line. It is another line parallel to it.

> postgres=# select line('1,3,2,3');
> line
> -----------------
> [(0,-3),(1,-3)]
> (1 row)

This was just using the wrong sign. Fixed.

> In addition, when a straight line coincides with coordinate axis, output
> appears -0, I do not know whether it is appropriate.

This is a matter of the floating-point operations. I don't think it's
on scope to worry about that. With the above fix, the cases you pointed
out go away anyway.

> Negative value appeared when use <-> to calculate the distance between two
> parallel lines.

Fixed with fabs().

New patch included. Still wondering whether to use a A,B,C-based output
format per Tom's comment.