Nicolas Berbiche

The infinite loop is caused by the recursive type returned by `jiffy:decode`. The workaround is `?assert_type(jiffy:decode(Thing), any())`.

One issue with checking refinement against the refined args is that `pick_value` is called against the refined type instead of the non refined type. For instance, the refined type of...

> > the refined type of a char is an integer range > > Can we turn it into a char range? Or is it normalized here? > > IIRC,...

> OK, good findings! Why did you close #411? I closed the PR after re-reading the comment before it multiple times. https://github.com/josefs/Gradualizer/blob/3b96c4568e29a51173f6427954120d9055543511/src/typechecker.erl#L3477-L3490 Also I had not found a reproducible issue...

I can reproduce the problem with your example. The issue can be avoided if the function `f` above (or `throw/1`) is called with an `assert_type` or `annotate_type` on the parameter:...

Hi, I'm hitting this issue in multiple functions in a large codebase. Most functions pattern match against a list of tuples: ``` erlang -spec should_not_error([{integer(), integer()}]) -> integer(). should_not_error([{A, 0}...

> @jesperes Could you try this branch - https://github.com/erszcz/Gradualizer/tree/start-app-in-rebar-plugin - to check if `noproc` goes away? I confirm that commit e0a1dbc286596be34cffd45807ffec656aec61b9 resolves the noproc issue.

Found an issue with where `pick_value` is called with `{type, _, any, []}`. I will fix this in a following commit.

> This is how I understand map types and a draft of an informal rule for map subtyping: https://github.com/josefs/Gradualizer/wiki/Map-types If I understand the conversation and the documentation above correctly, Gradualizer...

>Only way to fix it seems to be a restart of the whole system The following command will unmount the fuse filesystem. It resolved the issue in my case. ```sh...