Implement TranslateScale * Affine (both directions) #287
Conversation
As part of testing this, I also added `float_cmp` to dev dependencies and implemented it for Point/Affine. I propose that we start to standardize on this method for checking approximate equality. Finally, I've added some MathML to the docs to make it so people can understand the code. In the TranslateScale * Affine (and rev) case, one probably doesn't need more docs, but having them proves the concept.
| /// ``` | ||
| /// | ||
| /// [matrix-norm#entry-wise]: https://en.wikipedia.org/wiki/Matrix_norm#%22Entry-wise%22_matrix_norms | ||
| pub fn norm_l_1_1(self) -> f64 { |
There was a problem hiding this comment.
I added this function for error checking, before deciding that max(element-wise error) was a better way. I've left it in to stimulate conversation about matrix norms, but fully expect to need to remove it before merging.
There was a problem hiding this comment.
If we did keep it I almost certainly think we should remove the 1, otherwise |a - a| != 0.
|
(from @DJMcNab on Zulip) In terms of the first point, I do agree that precomputing the MathML seems like the best approach |
As part of testing this, I also added
float_cmpto dev dependencies and implemented it for Point/Affine. I propose that we start to standardize on this method for checking approximate equality.Finally, I've added some MathML to the docs to make it so people can understand the code. In the TranslateScale * Affine (and rev) case, one probably doesn't need more docs, but having them proves the concept.