On Mon, Oct 07, 2002 at 08:21:46PM +0200, Benja Fallenstein wrote:
But then, in the fisheye case, if I draw a rectangle, why's it
*distorted* in the end? I mean, it's just taking four points; they may
not yield a rectangle when connected, but the lines should be straight,
not distorted? -- Yet, if this is the case, the fisheye doesn't really
fulfill its purpose, does it?
No, see, the point is that the fisheye is then transforming the *parent*
coordinate system. I.e. instead of having everything scaled according to
the root coordinate system, we use the fisheye.
But basically, the point is that we've had these semantics in gl earlier
(the scale defines the size of things, esp. boxes), and it was so
difficult to work with. I simply don't want to continue working in this
system.
The reason it was difficult was that we didn't have all the parts.
The really important part that was missing was a call like
cs3 = coordsys_at(cs1, cs2, 0.3, 0.3)
which would create a coordsys which is a translation inside cs1, but whose
center
is always at the point (0.3, 0.3) of cs2.
This is what I've been trying to explain.