Radek Szymczyszyn

Hi, @hworld! Thanks for your interest and another example. Indeed, it seems to be another case of the same issue. We're aware that refinement is still somewhat limited and this...

Hi, @japhib! Thanks for the report. It's another one in a row about `gradualizer.pick_value/2` together with #169 and josefs/gradualizer#530. We'll take a look into it, as this function ought to...

Hey, Alan! This requires an explicit downcast, since not every `struct()` is a valid `t()`. You know this particular one is, but the typechecker doesn't. That's where `Gradient.TypeAnnotation` comes in...

Hmm, it seems we're not handling the code generated for the `with` statement properly. This needs some deeper research on our end.

Ok, here we have some findings: ```elixir with _value 2 end ``` compiles to ```erlang case _bar@1 of __value@1 -> 1; _ -> 2 end. ``` which shows what's already...

Thanks for the report, @l3nz! Indeed, it seems related to the issue you mention 👍 We'll look into it.

This seems to be a duplicate of #64.

Good ideas, thanks! We'll come up with something to address this.

Hi, @drselump14! Thanks for the report. It's a valid report about a problem in Gradualizer, the typechecking engine that Gradient uses. We'll look into it 👍

> On the other hand, how difficult would it be to have a parse transform changing the XML at compile time to the record representation? For `from_xml/1` it's trivial. For...