IDE0055 triggered on no space after spread operator

[jez9999]

I get the following warning because there's no space after my spread operator:

I don't even see what formatting option would apply to this, and in any case I don't want a space after my spread operators. It shouldn't give this warning.

[CyrusNajmabadi]

As with most of formatting, there is no option to control things. The formatting for spreads simply is this way. If not desired, it can be suppressed in these locations, or the analyzer can be entirely disabled.

Requests to support arbitrary formatting rules falls under a general formatting extensibility system which we are considering, but have no timeline on.

[jez9999]

@CyrusNajmabadi Is there a Github issue for said new formatting system?

[sharwell]

I agree that the space after .. looks incorrect.

[CyrusNajmabadi]

Space there was chosen by ldm, and corresponds to the same space we chose for the corresponding slice pattern form.

This is not an operator, and does not 'bind tightly' (which is when we think there should be no space.

As an example: !b ? c : d here we want !b as that binds tightly to b and not the whole expression.

But that's not the case with the spread element. .. a ? b : c is .. (a ? b : c) not (..a) ? b : c. The space is intentionally there to not make it feel like it's operating directly on the things it touches.

Put another way, it's not a unary expression, so it gets a space

[sharwell]

The space is intentionally there to not make it feel like it's operating directly on the things it touches.

Can you show sample code defining a, b, and c such that both of the following compile?

1. .. (a ? b : c)
2. (..a) ? b : c

The post implicitly claims that users may be confused about the binding precedence, but I'm having a hard time coming up with a case where that could happen.

It feels like we penalized all actual cases to account for a case that is either rare or doesn't exist, or maybe I'm just not seeing it yet.

[CyrusNajmabadi]

The point isn't that both compile, it's that the formatting implies one over the other.

[CyrusNajmabadi]

It feels like we penalized all actual cases

There's no penalty. We made an informed choice to align with slice patterns so that construction and deconstruction would pair up and format uniformly. And we used our general view here on precedence to dictate for both how we felt the formatting should be.