# operator not accepted about hlint

[ndmitchell]

You can get your code to pass HLint by using:

hlint Sample.hs -XNoUnboxedTuples

There are HSE issues with operators such as (#*:), see http://trac.haskell.org/haskell-src-exts/ticket/195 (however, it's their old bug tracker, so it probably needs reraising on their new one). However, in GHC, the (#) operator is not accepted with the UnboxedTuples extension, which HLint turns on by default. In general I turn on many extensions to accept more code, which has the cost of rejecting some normal programs. Is (#) the real example you were using, or did you reduce it first? I can't find that operator in the Lens library (but can find a zillion lens operators which HSE does invalidly reject).

In general, I think HSE are happy with bugs coming originally from HLint, although you might want to @ndmitchell me on them.

[mdorman]

That was actually what I was working with; I can't imagine how you weren't able to find it, lens being such a small package and all. :) http://hackage.haskell.org/package/lens-4.0.4/docs/Control-Lens-Review.html is where it's documented.

Anyway, it sounds like this particular instance probably isn't actually a bug in hlint or HSE, it's just a particular design choice for the default behavior?

[ndmitchell]

Yep, for (#) it's a design choice. However, given that a large and popular language uses # as an operator, and that HSE gets # prefixed operators wrong, I'm inclined to remove UnboxedTuples from the default list of extensions. My guess is that most code which actually uses that extension declares it in a LANGUAGE pragma, so will still work fine.