remove glyphsLib.affine, use fontTools.misc.transform instead #854
Conversation
Fixes #257 The glyphsLib.affine module has different copyright/license and is not actually needed because we already depend on fontTools which has an equivalent module for affine transforms. This removes that and replaces it with fontTools Transform.
this is to fix regression tests spuriously flagging things like:
<contour>
- <point x=50 y=784 type=line/>
- <point x=950 y=784 type=line/>
- <point x=950 y=826 type=line/>
- <point x=50 y=826 type=line/>
+ <point x=50.0 y=784.0 type=line/>
+ <point x=950.0 y=784.0 type=line/>
+ <point x=950.0 y=826.0 type=line/>
+ <point x=50.0 y=826.0 type=line/>
</contour>
https://github.com/googlefonts/glyphsLib/actions/runs/4025928471/jobs/6919775408
|
ugh.. I really don't understand why the regression tests keep failing with spurious diffs where integers are turned to floats e.g. 575 => 575.0: - <point x="575" y="524" type="line"/>
+ <point x="575.0" y="524.0" type="line"/>Any help appreciated. I give up for today... |
|
oh shoot, I misundestood the unified diff output. Confusingly enough, the minus-prefixed lines in the regression tests diffs are the newly generated files, whereas the plus-prefixed lines are the regression tests' reference files. I was expecting the opposite, i.e. reference or expected files on the left ( |
|
thanks guys |
Fixes #257
The glyphsLib.affine module has different copyright/license and is not actually needed because we already depend on fontTools which has an equivalent module for affine transforms. This removes that and replaces it with fontTools Transform.
The existing tests still pass. Let me know if you want me to add additional tests.