[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH] Fixes to AffineTransform
From: |
Olga Rodimina |
Subject: |
Re: [PATCH] Fixes to AffineTransform |
Date: |
04 Feb 2004 11:19:40 -0500 |
On Wed, 2004-02-04 at 00:33, graydon hoare wrote:
> Olga Rodimina wrote:
>
> > Here is the patch that fixes few errors in the AffineTransform class.
> > It corrects a small error in shear transformation and fixes
> > createInverse() to return correct inverse matrix.
>
> hi olga,
>
> this stuff looks good, thanks a lot! it's surprising this just sat there
> broken for so long. I've checked it against a few algebra texts I have
> kicking around and some geometry libraries (including cairo) and they
> seem to mostly agree with your calculation; though just to be difficult
> they all use different notation and half of them do it upside down :)
>
Hi,
I fixed the suggested changes. Attached is the revised patch that I'll
be committing.
>
> it might be nice to write up a mauve testsuite which generates a bunch
> of pseudorandom vectors and runs them through a bunch of pseudorandom
> scaled, sheared and translated transformations (and their inverses) and
> checks that the vectors pop out the same as they went in, within some
> epsilon to account for floating point error. anything to make this stuff
> less fidgety and prone to eyes-glossing-over would probably be good.
> it's too easy to make a minor arithmetic mistake which completely breaks it.
>
Sure. I'll do that.
Olga.
patch
Description: Text document