fix: reporters receive copy of message

[Smixi]

Type of Changes

	Type
✓	🐛 Bug fix

Description

MultiReporter was sending message by reference, which then could be changed by reporter. The colorized reporter was one of them, leading to changing any other reporters. This was broken since 2.14.0 and affected mostly people using colorized with a custom reporter.

Closes #7214

[coveralls]

Pull Request Test Coverage Report for Build 3260364257

• 2 of 2 (100.0%) changed or added relevant lines in 1 file are covered.
• 26 unchanged lines in 2 files lost coverage.
• Overall coverage increased (+0.0008%) to 95.354%

---

Files with Coverage Reduction	New Missed Lines	%
pylint/checkers/refactoring/refactoring_checker.py	6	98.2%
pylint/checkers/utils.py	20	95.23%

Totals
Change from base Build 3245605436:	0.0008%
Covered Lines:	17138
Relevant Lines:	17973

---

💛 - Coveralls

[clavedeluna]

should this be Closes?

[Smixi]

I wasn't sure of semantics of this, but yes ! Let me fix that.

[Pierre-Sassoulas]

Thank you for the investigation. I wonder what are the performance implication and also if the side effect of being able to modify other reporter's message wasn't seen as a feature by some users. We'd need to do it as a breaking change if that's the case.

[Smixi]

I guess copy works well with dataclass instances, Message contains only litterals.
I don't know if anyone is using multiple formatter to change data, but I think each formatter should handle themselve the message they receive. For instance, colorized was breaking any logic a reporter would have for a Message, like handling a path, category, module ...

[github-actions]

🤖 According to the primer, this change has no effect on the checked open source code. 🤖🎉

This comment was generated for commit ece4392

[DanielNoord]

We could also change the handle_message in the colour reporter to create a copy. That way reporters that depend on being able to modify for others aren't broken?

[Smixi]

@

We could also change the handle_message in the colour reporter to create a copy. That way reporters that depend on being able to modify for others aren't broken?

Do you have an usecase example ? Maybe that would be another option ? One issue is that we would need solve is to specify ordering. From 2.13.9 to 2.14.0, coloriezd/text reporter is now always the first reporter. If two reporters are necessary to achieve a specific output, maybe it should be only one reporter with options ?

[DanielNoord]

Yeah you're right. If anybody complains about this and has a usecase we can always update to allow it again.