Will Robertson

Should this work? ``` \documentclass{article} \usepackage{unicode-math} \setmainfont{[texgyrepagella-regular]} \begin{document} hello \emph{hello} \end{document} ``` It currently does but only by coincidence (without the square brackets it doesn't work, but with a literal...

See recent test "mergeshapes.lvt". All sorts of problems with ``` \renewcommand\itdefault{ii} \renewcommand\scdefault{ss} ```

Local vs global: \fontspec_set_family:Nnn etc

3

Should functions like `\fontspec_set_family:Nnn` be always local, always global, or have `gset` variants?

See: https://github.com/lualatex/luaotfload/issues/424#issuecomment-428616588

Add API for babel's check for languages/scripts

1

## Description Babel currently has some code: \{fontspec-opentype}{Script/\bbl@cs{sname@#1}}% {\\\newfontscript{\bbl@cs{sname@#1}}{\bbl@cs{sotf@#1}}}% \{fontspec-opentype}{Language/\bbl@cs{lname@#1}}% {\\\newfontlanguage{\bbl@cs{lname@#1}}{\bbl@cs{lotf@#1}}}}} This will break if I ever change fontspec's keyval processing. But I could add something like `\fontspec_if_script_option_defined:nTF` or similar.

`Extension = .xyz` needed unnecessarily

1

## Description It looks like I incorrectly strip off extensions when I shouldn't be when `Extension = .xyz` is NOT included. ## Check/indicate - [x] Relevant for XeTeX - [x]...

The current system is rather baroque!

The fontspec code is very bad at using `\cs_new:Nn` when it should be using `\cs_new_protected:Nn`. Also check `set` vs `new` usage.

Support XeTeX's Graphite renderer

1

You can currently write things like (I think!) ``` \newfontfamily{\LanglToneContourFontFamily}[Renderer=Graphite, RawFeature={Special=Hide tone contour staves}]{Charis SIL} ``` but that's not very friendly.

Should have a test per OpenType tag, plus more, basically.