Jean Klingler

> This looks great! In the future we want to allow the fallback to be any type but, as an optimization, this works great. :) OK, so the current approach...

Hi @Zurga , thank you for the suggestion. As you mentioned, this doesn't seem compatible with the performance goal of `Cmp`'s. Even as an option, checking options would be an...

Maybe we can update the documentation to include an example using https://hexdocs.pm/ex_cldr_collation/Cldr.Collation.Insensitive.html for instance, WDYT?

Hi, I was starting to look into this, quick question: Should we hard-code `:|>` here, or should it be a `:normalize_calls_on_pipe_operators` defaulting to `[:|>]`?

I just noticed after implementing this is something I already implemented [in the past](https://github.com/elixir-lang/elixir/pull/11845) and we ended up closing it because of the problem of `defmacro` 😅 ```diff -defmacro left...

> and I hope we will remind it next time, haha Haha, I hope so too, I can't believe I reimplemented almost the same thing twice without remembering 🤣 >...

I think now that we have the [`--migrate`](https://github.com/elixir-lang/elixir/pull/13846) flag we can re-open this one? Will start working on it if we agree.

> There is a downside here: every time we run --migrate in Elixir, it will make the code invalid, and we will have to manually fix it. This is already...

> Does migrate_unless currently make Elixir code itself invalid like pipe does? I guess the same Kernel.defmacro issue? Correct! ```elixir defmacro if(!condition, clauses) do ```

Although there is an argument that could be made that `--migrate` being best effort, and for unless we already have some Kernel-aware rewrite rules (guards, comparison operators...), so we could...