(x:y for y in arr) should express an array of objects, no?

[dbrans]

Why doesn't

x:y for y in arr

behave like

x y for y in arr

or like

x = y for y in arr

Thanks.

[michaelficarra]

It looks like {x: (y for y in arr)} to me. As per my normal recommendation, if you're not positive it's absolutely unambiguous, use explicit braces: {x: y} for y in arr.

[TrevorBurnham]

Michael's right; if you can write x: foo bar, then it's perfectly consistent to be able to write x: y for y in arr. Of course, the spacing (x:y vs. x: y) makes all the difference to human eyes, but we really shouldn't allow compilation precedence to turn on that.

[jashkenas]

Explicit braces are great, but I think that dbrans has a point here. If postfix forms close off implicit ), then they should probably also close off implicit } in the same fashion.

[TrevorBurnham]

Hmm, good point; it would be nice for x:y for y in arr and x y for y in arr to parse similarly.

[michaelficarra]

Yeah, @jashkenas made a good point there. x: y for y in arr is visually ambiguous, but I guess the technically correct parse should be consistent with the other cases. That is, as long as there isn't a strong use case for the current parse (or, similarly, no use case for the suggested one).

[dbrans]

The compiled form of

f x:y if z

also took me by surprise as well. I was expecting it to behave like

f x y if z

Should this be discussed in a separate issue? Thanks.

[TrevorBurnham]

I think it's an example of the same issue, dbrans. And yes, I'd say it feels like a bug that

f x:y if z

parses as

f {x: y if z}

rather than

f {x:y} if z

[satyr]

→ #1871