Move -epmd_module and -erl_epmd_port into kernel parameters

[jonatanklosko]

This introduces epmd_module and erl_epmd_listen_port parameters for the kernel application. I updated the docs to deprecate -epmd_module and -erl_epmd_port, similarly to -connect_all in 1d0fe6f.

The kernel application already has a lot of parameters related to distribution, so I believe semantically this makes sense.

For context, in Livebook we started using a custom epmd_module. In Escript and Elixir releases, there are ways to set the -epmd_module flag, however there is no way to do that in development, other than setting ERL_AFLAGS env var. We already start distribution manually using net_kernel:start/2, so the idea is that now we could do application:set_env(kernel, epmd_module, Module) right before that.

[CLAassistant]

All committers have signed the CLA.

[github-actions]

CT Test Results

    5 files    206 suites   1h 58m 18s ⏱️
3 078 tests 2 791 ✅ 287 💤 0 ❌
4 002 runs  3 657 ✅ 345 💤 0 ❌

Results for commit 1714f2f.

♻️ This comment has been updated with latest results.

To speed up review, make sure that you have read Contributing to Erlang/OTP and that all checks pass.

See the TESTING and DEVELOPMENT HowTo guides for details about how to run test locally.

Artifacts

• Complete CT logs (Download Logs)
• HTML Documentation (Download HTML Docs)
• Windows Installer

// Erlang/OTP Github Action Bot

[jonatanklosko]

Note the fake nonode, nohost, those arguments are ignored by erl_epmd:listen_port_please/2, and this way we use the public API. Alternatively we could expose erl_epmd:erl_epmd_listen_port/0 and make it -doc false.

[josevalim]

We could also expose it as net_kernel:epmd_listen_port, to mirror net_kernel:epmd_module.

[jonatanklosko]

epmd_module is a broader config, while erl_epmd_listen_port is specific to erl_epmd, as per the other comment.

[jonatanklosko]

A follow up question, if this change is accepted, would it be possible to have it in the next maintenance release? If yes, I will change the target branch, and then I can submit a separate PR altering the -erl_epmd_port 0 behaviour (if you agree that it's desired).

[jhogberg]

Thanks for the PR! I think we'll want to discuss this a bit more internally before merging, and most of us are on vacation right now. We'll revisit this later in the summer/early autumn. :-)

[jonatanklosko]

@jhogberg sounds good, thanks for the heads up! :)

[kikofernandez]

The comments are from the OTP team (not just me, so credit to them for this review)

Please, address the issues mentioned and feel free to rebase on top of maint.

[kikofernandez]

If there is nothing left to fix, I will merge this next Monday (some of us are in a conference this week and I do not want to rush the merging before having a really final good look at it :) )

[jonatanklosko]

@kikofernandez sounds good, thanks :)

[jonatanklosko]

@kikofernandez kindly ping :)

[kikofernandez]

We have a code freeze for 27.1. I will merge as soon as I am given the OK.
This change was not breaking backwards compatibility, so we can merge in maint and release with OTP 27.2

[kikofernandez]

Merging tomorrow if no internal tests break

[jonatanklosko]

@kikofernandez thank you :)