f() in []

[satyr]

$ coffee -bpe 'f() in []'
var _ref;
;

$ coffee -bpe 'f() in [x]'
var _ref;
(_ref = f()) === x;

[michaelficarra]

The second one is just an optimization issue, right? An unnecessary _ref.

[thejh]

Pull request #1242 contains a fix.

[thejh]

closed by 0497c07: instead of nothing, compile to false

"if a in []"

[satyr]

$ coffee -bpe 'f() in []'
var _ref;

false;

$ coffee -bpe 'f() not in []'
var _ref;

false;

• f() should still be evaluated.
• not in result is incorrect.

[xixixao]

Any reason for this optimization? I'll make a PR to remove otherwise. Seems nonsensical.

[vendethiel]

Whatever's the reason for this optimisation (avoid a call to the helper seems obvious), it clearly introduces bug (coco removed this optimisation totally).
I tried to take a stab at fixing this, but the tests are failing in several (weird, ofc) ways for me, and git bisect gives me segfault in rm and other thing. Well, windows...

[xixixao]

Let's just remove it, why would you ever want to write x in []? The only optimization that would make sense would be to check for length in the helper and I assume that wouldn't speed it up much. +1 to remove it.

[jashkenas]

. +1 to remove it.

Go for it. That is quite silly indeed.

[michaelficarra]

Don't remove the optimisation for a in [b] -> a === b, a in [b, c] -> a === b || a === c, etc, though.