compiler: fix nondeterministic behaviour

[delitrem]

Fixes nondeterministic behaviour related to new doc system. See #8579.

[CLAassistant]

All committers have signed the CLA.

[github-actions]

CT Test Results

    2 files    324 suites   10m 16s ⏱️
  817 tests   815 ✅ 2 💤 0 ❌
5 420 runs  5 418 ✅ 2 💤 0 ❌

Results for commit ed02c1a.

♻️ 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

[garazdawi]

Hello! Thanks for the fix! Would you mind adding a testcase to lib/compiler/test/beam_doc_SUITE.erl that tests that it works?

[garazdawi]

Also, would you mind targeting maint instead of master so that we can get a fix out before next year :)

[delitrem]

Hello and thank you very much for the feedback. I will write a testcase.

[delitrem]

Still I have no great idea how to reproduce 'bad behaviour'. I've used the following code:

    {ok, _, Reference} = compile:file("ssh", [binary]),

    false =
        lists:all(fun(_) ->
                     {ok, _, Testing} = compile:file("ssh", [binary]),
                     Testing =:= Reference
                  end,
                  lists:seq(1, 100)),

Meanwhile, it can be reproduced easily when using erlc from shell with the same files:

## the following files are from lib/ssh/src/
$ ls
ssh.erl  ssh.hrl  ssh_connect.hrl
$ for a in $(seq 1 100); do $ERL_TOP/bin/erlc ssh.erl; md5=$(md5sum ssh.beam | awk '{print $1}'); mv ssh.beam ssh.beam.${md5}; done
$ ls
ssh.beam.0c2b4dbfa37b12c2478782670083c40b  ssh.beam.6e58cd3ec2f0ba1de51959191be6a128  ssh.erl  ssh.hrl	ssh_connect.hrl

Any ideas, advices, etc.?

[garazdawi]

The behaviour relies on two different VMs being started when compiling so that the atom indexes end up being different for the metadata. You can use the ?CT_PEER macro to start a new node in your testcase.

[delitrem]

You can use the ?CT_PEER macro to start a new node in your testcase.

Thank you very much for you hint! It helped.

I will test a little bit more and then commit a test case into my branch.

[delitrem]

Pushed changes to the compile.erl and test suite.

[garazdawi]

I rebased your commit on maint-27, so that it will be part of the 27.0.1 release and added the primary bootstrap.

[delitrem]

Thank you very much for your assistance!