Regression in Mix.Tasks.Format.formatter_for_file introduced in 1.15.5

[lukaszsamson]

Elixir and Erlang/OTP versions

Erlang/OTP 25 [erts-13.2.2.2] [source] [64-bit] [smp:12:12] [ds:12:12:10] [async-threads:1] [jit]

Elixir 1.15.5 (compiled with Erlang/OTP 25)

Operating system

macOS

Current behavior

Elixir 1.15.5 introduced a regression in formatter plugin loading making it no longer able to format ~H sigils.

Repro:

1. mix phx.new formatter_regerssion
2. cd formatter_regerssion
3. mix deps.get
4. break sigil formatting in lib/formatter_regerssion_web/components/core_components.ex
5. execute this script in iex

Mix.start()
Code.compile_file("mix.exs")
Mix.Task.run("loadconfig")
{f, o} = Mix.Tasks.Format.formatter_for_file("lib/formatter_regerssion_web/components/core_components.ex")
text = File.read!("lib/formatter_regerssion_web/components/core_components.ex") 
text == f.(text)

on 1.15.5 it returns true
on 1.15.4 it compiles before and then returns false

Even if explicit Mix.Task.run("compile") is added on 1.15.5 it still produces a formatter not being able to format sigils.

Invoking Code.format_string! is also broken

IO.iodata_to_binary([Code.format_string!(text, o), ?\n]) == text
true

Most likely the regression was introduced in fixes for #12880 (9f977be or dc8cfcd)
Originally reported in elixir-lsp/elixir-ls#975

Expected behavior

Formatter should work when invoked via code. If plugin loading now requires some additional code this should be documented. Breaking changes should not be introduced in patch releases.

[josevalim]

I will investigate. Please do test the fix once it is out, since the previous commit was aiming to address another bug in Elixir LSP and the breaking change is non intentional.

[lukaszsamson]

I will investigate. Please do test the fix once it is out, since the previous commit was aiming to address another bug in Elixir LSP and the breaking change is non intentional.

The previous fix was not related to Elixir LS at all

[josevalim]

@lukaszsamson oh, I thought that was the case, apologies. Anyway, please give it a try a now and let me know.

[lukaszsamson]

Thanks @josevalim, It's working fine again on 6654f1b

[josevalim]

Excellent, i will ship a new version either this week or the beginning of next one.