Unicode: Supporting middle dot & real minus

[fbruetting]

Would be nice if you could equal · (U+b7) with the currently existing \cdot[TAB] (U+22c5), as on some keyboard layouts (like the german Neo2) you could directly produce that. You there also can produce the real minus glyph − (U+2212), which has the same width as the plus sign, but sadly isn’t supported in Julia yet, too.

[stevengj]

See #25157, which addresses U+00b7 equivalence.

I'm less sure of what the use-case is for normalizing − (currently disallowed in identifiers) to a hyphen, since the minus glyph is pretty hard to type by accident... See also #19464.

[fbruetting]

Ok, I see that the U+b7 is underway, nice!

With − being the real minus, I think it would be more than appropriate to support that. It’s not on every keyboard, that’s true. But you also get that when copying formulas, as some formula editors convert the quart em dash - to the real minus −, so that it doesn’t stick out negatively as much (because it’s much shorter than the plus sign).

[StefanKarpinski]

On the other hand, I can't see any reason not to treat a real minus sign as a minus sign.

[stevengj]

Since - is both a binary and a unary operator, this would add a bunch of cases into the parser... I just don't see the need here.

As I wrote in #19464, there are zillions of equivalences that we could potentially define, since Unicode has lots of easily confusable characters. But if we don't want to go full NFKC normalization, I think we should be conservative about adding custom normalizations: limit ourselves to confusable characters that are easy to type and hence show up frequently in practice. I don't think U+2212 qualifies here, since no one seems to have encountered it in actual code.

[fbruetting]

When you work with TeX/LaTeX/ConTeXt or simlar systems, you’ll encounter real minuses all day long. With Julia targeting more or less exactly the same area, I can’t imagine that it should be normal practice, having to substitute real minuses by hyphens for getting a rendered formula working in Julia by pasting.

[stevengj]

@fbruetting, I'm in a math department — of course I use typeset math with real minus glyphs. But I almost never type those glyphs, they are rendered for me. (The cases where I do type them are in illustrations, in software that lacks proper equation rendering, but I can't easily imagine copying and pasting from those illustrations into code.)

I'm skeptical of the use-case of copying and pasting a rendered formula into code. There aren't many occasions in which you can copy and paste a rendered formula into a text document, and in most of those cases you can just copy-and-paste from the LaTeX source (which uses hyphens) instead.

I tried googling this use case (since no popular language supports U+2212 for the minus operation as far as I know), and I can't find any case of anyone ever complaining about the inability to paste this into source code. Can you find any reported examples where this has arisen?

[fbruetting]

I can’t estimate the effort, but it would just be a nice feature. 🙂

sphinx-doc/sphinx#4136
https://groups.google.com/forum/#!topic/python-ideas/6neDDlxkaaM
https://stackoverflow.com/questions/3522745/using-minus-in-c-sharp
I just searched quickly, as I’m already late…

[stevengj]

@fbruetting, none of those links describe people using U+2212 by accident from copy-and-paste.

[mcabbott]

This bit me a few days ago, precisely by copying coefficients from a PDF while trying to make a plot. I agree that copying a formatted formula is never going to work well, but copying numbers seems like it ought to be straightforward.

The error was initially as follows, where − 5 was the copied coefficient (in reality with a few more decimals) which also seems odd for an illegal character:

julia> Ls = 1:10; 1 ./ (Ls .^ 2) .− 5 ./ (Ls .^ 3)
ERROR: syntax: space before "." not allowed in "(Ls .^ 2) ."

[stevengj]

Copying typeset numbers from a PDF does seem like a reasonable use-case.

[dpsanders]

I have had these issues copying and pasting from a PDF multiple times.

[fbruetting]

Please give the initial post a thumb-up emoji, to indicate the developers how much people would profit / are interested at a glance. 😉

[misyntropy]

I am excited enough about the real (unicode) operators sufficiently that I want to start using Julia for that reason (I am also academically from a scientific computation background)... Using − (U+2212) rather than - is one of the most visually pleasing changes I can make to arithmetic expressions in code. Please don't exclude that one operator from the list.

Treating it as a real minus sign, or allowing it as an operator with plus-like precedence, are both great options.