Tal Leming

@belluzj This is fantastically helpful. Thank you so much for all the detail and examples! There are some great points raised in all of these comments. With regards to whitespace,...

> Another thought (re contents.plist) would be to come up with a glyph name to file name scheme that is reversible It's potentially more than contents.plist though. Any file that...

> That would also solve the other problems that @typesupply mentions about storing data outside GLIF files. How would it solve the cmap issue?

Plus: • It's weird that a glyph named A on layer "foo" can have a different unicode than a glyph with the same name on layer "bar".

> Note that my first argument is a bit of a slippery slope Yeah. So, doing my spec-writer-devil's-advocate job, I'll ask: Are these optimizations being added to solve tool limitations...

FWIW, I **strongly disagree** with the idea of making all fontinfo.plist fields mandatory. It would break many, many things. For example, if I'm making a font that will have CFF...

I agree 100% with this. I started a comprehensive UFO3 test suite a [long time ago](https://groups.google.com/forum/#!topic/ufo-spec/u1Zy-XFn71k) based on the [WOFF test suite](https://github.com/w3c/woff/tree/master/woff1/tests). It probably stalled because I was trying to...

This already works. I don't know why I thought that it didn't.

I just reviewed the code again and I don't see a way to make `fooInstance.bar = barInstance` add `fooInstance` to `barInstance` unless `fooInstance` is an orphan object. If it isn't,...

A lot of those old RoboFab methods were the result of me having a project that needed a few lines of code and I dumbly thought "I'll just put this...