Jiří Beneš

Personally, I use something like: ```sh #! /usr/bin/env bash -e # ... add node to PATH ... exec long/full/path/to/bin/java -jar another/long/path/to/effekt.jar "$@" ``` for my version of `effekt.sh` both on...

Fascinatingly, the error doesn't go away if I annotate the return type as Int: ``` def foo(): Int = { val x = Nil[Int]() head(x) } ``` So this is...

As @marvinborner noted, this issue is fixed by handling the `Exception` effect. The following code doesn't report the error anymore: ```scala def foo(): Int = { val x = Nil[Int]()...

The issue is also fixed by annotating the `Exception` effect: ```scala def foo(): Int / Exception[EmptyList] = { // ^^^^^^^^^^^^^^^^^^^^ val x = Nil[Int]() head(x) } ```

Hi @marvinborner, looks great! Could we add some pos/neg tests for multiline strings? I'm especially worried about things like newline and indentation handling (so whitespace handling in general).

If you get some time, could you check if this also fixes https://github.com/effekt-lang/effekt/issues/353?

Quick question, any idea on what's the overhead of the timers? It would be nice to verify that there's no unexpected performance loss -- for example, by running the benchmarks...

I think this is already solved by #495, right? cc @dvdvgt

Counter proposal: we could also forbid `return` _everywhere_ (or at least in the "early return" position) and instead force the users to use `boundary`/`break` for early returns (like in Scala...

Agreed, parametrising `Control` sounds good [1. changing `Control` to `Control[R]`; 2. changing the signature of `break`; 3. adding some general `boundary` handler], the biggest problem I can foresee is the...