isnumber(::Number) undefined

[dalum]

Is there a reason isnumber(::Number) is left undefined? There are a billion other ways to check for it, I guess, but it's a bit odd that it has only been defined as a function on Char. I don't have a good use case for it, just one of surprise that the function tab-completed but didn't do what I expected. And also that I define my own isnumber with isnumber(::Number) = true. Would be nicer just to extend the one in Base with my own types, so other people wouldn't have to overwrite my isnumber for their personal number types that for some reason don't inherit from Number.

[yuyichao]

What do you want it to do? It's for checking if a character is a number so do you want it to be isnumber(::Number) = true or isnumber(a::Number) = isnumber(Char(a))? Better to leave this undefined I think.

[StefanKarpinski]

Yes, it does what's documented, but the name makes this quite confusing. It should at the very least have a better name or be scoped to a string-specific module. There has previously been talk of folding all character class checks into a single function with an enum argument for the class.

[dalum]

Sorry, yes, I would suggest isnumber(::Number) = true. I understand that it checks if a Char is in the number range, and there is nothing inherently wrong with the fact that it only does that, but it is a surprising behaviour when you just type isn<tab> and sees that "oh, there's a function called something that sounds like what I want". It's a bit like defining ismatrix to only check if a string parses into a Matrix.

[omus]

Maybe this should be renamed to isnumeric? I'm not sure isnumber(::Number) needs to be defined as you should be doing isa(x, Number) instead.

[yuyichao]

So this can be closed and any discussion about changing the name should go to #14347 where previous discussion is. It doesn't make sense to define isnumber(::Number).

[nalimilan]

So let's reopen #14347? :-)