Unicode characters and semantics

[Larivact]

• - should become minus sign (U+2212) −
• ... should become horizontal ellipsis (U+2026) …
• := should become colon equals (U+2254) ≔
• varepsilon should become greek small letter epsilon (U+03B5) ε to be consistent with TeX

Also shouldn't NN, QQ, RR, ZZ, Omega, Phi, Pi and Theta be identifiers instead of operators?

[drlippman]

I can't speak for certainty, but I'm guessing that those were not originally mapped to Unicode characters because they may not have been present in commonly available fonts at the time the script was written. We'd have to experiment to see if remapping them to Unicode would cause any issues with MathJax. I know that the minus mapping could be problematic, as ideally negation would be distinct from subtraction (minus), but it isn't currently.

I'm not sure why U+025B was selected over U+03B5 for varepsilon, or why RR and the others were tagged with an mo; perhaps @jipsen can shed some light on that? I know Pi and Theta both have meanings as operators in certain contexts, so I can at least imagine why those are mo.

[saivan]

There is also the possibility of adding other symbols like

• √ for sqrt
• ∂ for partial
• ∫ for int

They could just be used as alternatives for the full ascii versions of the same words, but they would make the source more readable, eg:

(∂f)/(∂x) is probably nicer in text form than (partial f) / (partial x)

Especially where many keyboards have keyboard shortcuts to type these symbols these days, it might be good to include them (eg on a mac you type option + d to get the partial symbol)