erl_lint: Warn on updating map/record literals

[jhogberg]

This PR adds a warning whenever a literal map or record is updated, catching errors like missing commas in lists of records:

-module(foo).
-export([list/0]).

-record(foo, {bar}).

list() ->
    [
        #foo{bar = foo} %% Missing comma!
        #foo{bar = bar}
    ].

https://erlangforums.com/t/maybe-a-bug-compiling-a-list-of-records-without-comma-between-them/3239

[github-actions]

CT Test Results

    2 files     93 suites   35m 18s ⏱️
2 016 tests 1 968 ✅ 48 💤 0 ❌
2 325 runs  2 275 ✅ 50 💤 0 ❌

Results for commit 239f7c1.

♻️ 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

[michalmuskala]

This is definitely a very welcome change.

How does this interact with macros? Some code I've seen is shaped like:

-define(FOO_DEFAULT, #foo{...}).

?FOO_DEFAULT#foo{custom = value}

I'm not sure this should be a concern, but just something to consider.

[jhogberg]

erl_lint runs after macro expansion, so yes, this PR would warn for that code. :-/

[michalmuskala]

As a reference, this can be bypassed in macros by wrapping the literal map that's meant to be updated immediately in a begin/end block, e.g:

-define(range_anno(Tok1, Tok2), begin #{
    location => map_get(location, ?anno(Tok1)),
    end_location => map_get(end_location, ?anno(Tok2))
} end).

And then ?range_anno(X, Y)#{...} does not issue warnings and should introduce no runtime overhead.

We hit this in erlfmt - WhatsApp/erlfmt#354