A small collection of WTF code snippets sorted by language.
I love that the contribute is just a mailto link. I want to see more of this & less “join the Discord chatroom & create a Microsoft GitHub account today”
Python item 1
Mutable default arguments don’t get re-initialized with each function call.
got an audible wtf from me… Been using python for years, totally unaware.
Mutating function arguments is pretty wtf to begin with.
true, and i can’t think of a legitimate case where it would have tripped me up. but if someone, a novice perhaps, wrote
def some_func(foo, bar=[1, 2, 3]): bar.reverse() # for whatever reason print(bar) some_func('hello') # output [3,2,1] some_func('hello') # output [1,2,3]i think they would be within their rights to be surprised that calling this function twice has different results. that’s what i was surprised by; it feels like bar would be re initialised each time with a scope of the function but apparenty not
For Haskell:
-
I’d say this is definitely a wtf. Tuples should have never been given Foldable instances and I’d immediately reject any code that uses it in code review.
-
I actually didn’t know, so TIL. Not surprising since common wisdom is that a lot of the type class instances for Doubles are pretty weird/don’t make a lot of sense. Just like in any language, floating point needs special care.
-
This is jjust expected syntax and not really a wtf at all. It’s very common for languages to require whitespace around operators.
-
Agreed as wtf, the
NegativeLiteralsshould be on by default. Just would be a pretty significant breaking change, unfortunately -
Not a wtf… What would you expect to happen? That operation is not well-defined in any language
-
Perl not included in list.
Conclusion: In Perl, everything works as expected
I’ll share with you this gem from someone who tried to cause a syntax error on purpose, but the script ran just fine: https://stackoverflow.com/questions/11695110/why-is-this-program-valid-i-was-trying-to-create-a-syntax-error
Correction: In Perl, everything works as can be expected.
The WTF in the C# example seems to be that people don’t understand anonymous functions and closures?
Go had the same behavior until recently. Closures captures the variable from the for loop and it was a reference to the value.
They changed it because it’s “common” in Go to loop over something and run a goroutine that uses the variable defined in the loop. Workaround was to either shadow the variable with itself before the loop, or to pass the value as an argument.
It’s been a long time since I wrote c# so idk if the same is expected from the avg dev, but in Go it’s really not explicit that the variable will be a reference instead of a plain valueiis still a value type, that never changes. Which highlights another issue I have with the explanation as provided. Using the word “reference” in a confusing way. Anonymous methods capture their enclosing scope, soisimply remains in-scope for all calls to those functions, and all those functions share the same enclosing scope. It never changes from being a value type.
Some of the examples seem to be more “unintuitive for newbies”, but there are still some good ones in there
Yeah. I didn’t understand what they meant by the wtf there. Seemed to me someone wondered if the Action would have a localised version of i (making this stay lowercase on a phone was harder than it should be) or if it used the same i. So made a simple test for it.
Not really sure it’s a wtf unless they expected a different result.
I think the explanation they provide is a bit lacking as well. Defining an anonymous function doesn’t “create a reference” to any variables it uses, it captures the scope in which it was defined and retains existing references.
Not only funny, but also learned something today: you can use NegativeLiterals to not have to write (-123) in Haskell
Enjoyed this :D
Are we supposed to be able to scroll through the examples? Also, it’s odd to see expected behavior in there.
Make sure, you’re not blocking fontawesome.com in NoScript or whatever content blocker you might be using. All the UI buttons depend on that to be available…
I’m raw-dogging Chrome and Android and there are no navigation controls, just the dropdown and reveal.
The C# example works exactly as you’d expect… Actually most of them do
The compiler optimization example (IIRC #6) was unexpected, when just looking at the code
C++ is just fried hahahhahah
Hey, don’t you curse on my bread and butter! ;-)
