It does sometimes make sense to have identifiers in non-English languages. If could be for a DSL, or it could be because the business logic concerns concepts which is not trivially translatable to english.

And this works right until a developer who doesn't speak the language will have to deal with that code, which will happen much sooner than you'd expect. Half of the fun is figuring out the meaning of variable names, the other half is figuring out how to type ß on a non-German keyboard.

Minor side note: I cannot for the life of me figure out why other operating systems won't borrow Apple's fantastic input scheme, in which this something as simple as a ß isn't a big deal. Need an emdash? Easy, option-shift-minus. Need an ñ? Easy, option-n-n. ß, ≠, ¡, ÷ or any number of other relatively common symbols? Easy. Most of them are even fairly mnemonic (e.g. ß is option-S, ¡ is option-1, ≠ is option-= and ÷ is option-/).

Lion made this even easier by stealing iOS's input scheme: Hold down a keyboard key for long enough and you can pick a special character out of a context pop out. îïíīįì šéē

I wish JS was better-suited to embedded DSLs. But without operator overloading, with required parens on function arguments, no (standard) method_missing, no atoms, and a limited set of characters for identifiers, it's pretty limited.

I agree that non ASCII identifiers are to be avoided, but they may be some exceptions, I have written code like this (in Go):

    type Polar struct {
        R, θ float
    }

First thing I think with that is : How the fuck can I make that symbol?

I'll load up google, then find unicode charts and locate this. So writing your variable name will take about 15 to 30 minutes. Thanks a lot

Not disagreeing with the sentiment, but if you’re using Windows, Linux, or OSX, you do have a searchable character map.

That fails when you don’t know the name of the character, in which case you can use something like Shapecatcher (http://shapecatcher.com/), but then how do you know whether it’s a Greek theta or a Coptic thethe or a Cherokee na or Cyrillic fita or Latin barred O?

Clearly our editing powers need to improve if we’re to feasibly use non–English-alphanumeric variable names.

If the variable name is Chinese, use an appropriate input method. Otherwise, maybe set up your keyboard with a compose key (and lowercase parens) http://canonical.org/~kragen/setting-up-keyboard.html and possibly install a bigger XCompose map so you can type Greek letters conveniently https://github.com/kragen/xcompose.

On Linux (and there's some similar utility for windows), there's a key sequence to enter arbitrary Unicode code points.

It's not per se entering the silly thing, it's figuring out which code it is.

Unless you have really, really, really good reasons, prefer radius and theta instead.

I tend to avoid that as inevitably when someone has to use the variable, it's a bugger to work out what to type unless you're a unicode whizz. I'd use "theta" instead.

Iunno, perhaps editors should just get better about supporting “special” characters. Programming language notation kinda sucks without proper symbols. Every time I type “->” or “<=” I have to sigh a little inside. Do we really need to be able to type code on 50-year-old typewriters?

Emacs is alright about it though. There are keybindings for entering Unicode characters by name or number, and for getting information about the character at the cursor. You can also use a TeX-like input mode where “\lambda” is automatically replaced with “λ”. But really, I’d rather we rethink input methods altogether.

Or you use compose or alt?