I've never understood this rationale. Pretty much any feature can be abused by "less experienced" developers. I think you should never leave out useful features just because they could be abused--trying to protect developers from their own incompetence is never going to be completely successful and is rather arrogant at that.

I could see leaving out or modifying features that lead to a lot of mistakes, like manual memory management, but this is just a flag useful for debugging--you never have to use it and it does not affect your code at all if you don't use it.

Also, this feature is more like replacing unsafe functions with undefined rather than making everything dynamically typed. For example, it will always error if you run a poorly typed function, regardless of what argument you pass in. And, of course, if you're worried about your co-workers using this flag, you can just recompile without it and fix all the errors.

So you're a perl programmer then?

Amusingly, I used to be :).

Also, Perl's philosophy is actually not that far from Haskell's in some ways: Haskell embraces TMTOWTDI and tends to favor more concise code.

How? Your continuous integration system won't have this flag on, so their non-typechecked code will never reach production. What they do while developing is not much of a concern; this is simply a faster way to put "--" in front of a lot of lines of code.

This is not about making Haskell dynamically typed. It's about not type-checking code that never runs.

I completely concur. This seems to not be "haskelly" at all.

Reminds me of zedshaw's post last week.