Can we call a julia function %%?

I was trying to name a function as %% but I did not succeed. I got:

ERROR: LoadError: syntax: "%" is not a unary operator

Can we name a function in julia as %%?

julia> var"%%"() = 1
%% (generic function with 1 method)

julia> var"%%"()


Why do you want a function named like this?
If you want to define a new unary operator, you have to use one of the function names reserved for this purpose (e.g. ++).


because I have a function that is very close to mod (%), so I tried to rename it %%. After all, I used a text name. But I was wondering the reason why I could not use %% as a name.

No particular reason. Support would need to be added to the parser for %% as an infix operator. There might, however, be objections or complications. Worth filing an issue though if you’re interested in this being available in a future version.

1 Like

Given that %% has no standardized meaning anyway (unlike eg +, *), and we have a zillion free Unicode operators for custom purposes, I am not sure I understand the motivation for complicating things just for the sake of %%.

Yes, I know R has it (for mod), and a whole zoo of other operators sandwiched between %%s. But they have no Unicode. Also, once operators end up taking up several characters, and aptly named function may just be less cryptic.


You can call it %′ (tab-complete %\prime) or (%\hat) or (%\tilde), for example, or use a subscript or superscript after %.


Also, it might be good to reserve %% for mod since % means rem.


I am not sure mod deserves an infix operator. There are around 50 uses of it in Base, and 40 more in the standard libraries (as opposed to % as rem, which has a few hundred).

1 Like

If anything it’s rem that doesn’t deserve an infix operator. mod is by far the more useful of them whereas rem is a latent source of bugs whenever someone is careless with the sign of the first argument. (Yes, I know that hardware prefers rem, which is a historic mistake, and thus has its uses when performance is critical and you have full control of the signs.)

Out of which the large majority is x % Type, where rem and mod are identical. Of the rest I didn’t at a quick glance see any case where either argument could be negative, so mod would have worked as well as rem.