Re: filled rectangles differ ?

[Soeren Sandmann <sandmann daimi au dk>]

Havoc Pennington <hp redhat com> writes:

> Yes, GDK just uses the X11 primitives. X draws filled primitives 1
> pixel smaller. If you dig through the X docs you can probably find 
> some sort of rationale, I don't know offhand.

The rationale I invented for myself is like this: Think of the
coordinates as denoting the infinitely thin line _between_
pixels. When such a coordinate has to be converted to a pixel, it gets
converted to the pixel to the right/below of the thin line, but the
position it denotes is really _between_ two pixels.

If you think of this way, the filled rectangles make sense: they fill
the area between the two passed in thin lines.

Unfilled rectangles can be though of as four lines with width 1, so
each coordinate has to converted to a pixel.

It is often useful to think of C opinters and C++ iterators this way
too. The iterators in the TextView and the TreeView can also be
thought of like this.

Well, that works for me, YMMV.

Søren