ok. You convinced me. Lets see if we can get it into 0.47... Maybe it is too late. I hope not.

ScislaC, can we commit this change to trunk?

Juca

On Mon, Aug 24, 2009 at 10:21 AM, Tavmjong Bah<tavmjong@...8...> wrote:

Hi,

I was wondering if there was some special reason you decreased the range for the Convolve Matrix Divisor from 1000 to 20 in commit 17056? If one constructs, for example, a 5x5 gaussian blur matrix using integer values in the kernel, one needs a divisor of 159. (I noticed because it broke this example in my book.) In general, I think it is better that ranges are a bit too large than too small as you never know what possible uses people will come up with or what values one might inherit from other drawing packages. What do you think about changing it back to 1000 or to 200?

Tav