Simon Cozens

Looks good. We will probably have to fix the diff function but I think go for it.

I'm also getting the same problem here. This is before it tries to post anything: ``` $ PYTHONPATH=~/others-repos/gftools/Lib gftools qa -f *.ttf --diffenator INFO:__main__:No regression checks possible since there are...

Actually this is probably because it can't find the previous fonts, d'oh. Adding `-fb ~/others-repos/fonts/ofl/notosans/*.ttf` works...

If you add `-a`, yes?

Building all Rubik targets with gftools-builder: 1m 52s. Building all Rubik targets with gftools-builder-ninja: 23s.

There is still some problem here with instance UFO filenames. Sometimes fontmake puts them into `instance_ufo/` regardless of what's in the designspace file and sometimes it goes with the path...

Just an update on this: I've used it to build hundreds of Noto fonts, and for that it works well. I haven't tracked down the issue with instance UFOs ending...

This is clever but doesn't work. If you decompile the cmap table (as we did to find the existing Unicode mappings), the mappings are stored internally in the font object...

Yeah, basically. I've put it into a custom script in my font's repo, but I figured if this is something that fontbakery has a check for, it should also be...

Let’s keep this out of the next release until the tag issues are fixed. Who’da thought that OpenType having its own bespoke almost-but-not-quite-BCP47 language tag system would have caused such...