Would it be possible to avoid the side effect of delaying the PEP 765 syntax warnings to the code generation stage? Emitting those syntax warnings during AST construction is one of the key reasons I was comfortable suggesting turning them off globally in#139658 when static code analysis was in use.

Only if you are fine with double warnings in REPL or other places that useast.parse() +compile() instead of justcompile().

There are three kinds of warnings, by the emitter:

• Emitted in the lexer/tokenizer. They are always emitted when you parse the Python sources.
• Emitted in the AST optimizer. They may be emitted twice, inast.parse() andcompile(). Currently, this is only the PEP 765 warnings.
• Emitted in the code generator. They are only emitted if you usecompile().

So, you already see not all warnings if you only useast.parse(), without code generation. The fact that the PEP 765 warnings are emitted in the AST optimizer instead of the code generator is just an implementation detail. It could perhaps even be more convenient to emit them in the code generator.

The fact that the PEP 765 warnings are emitted in the AST optimizer instead of the code generator is just an implementation detail.

Actually I think this was a deliberate decision, we want this to happen during static analysis.

Can we add a kwarg to ast.parse to tell it whether to emit syntax warnings or not?

We can, but

• This is a new feature. It can perhaps be backported to 3.14.1 as a special exception.
• Every user ofast.parse() andcompile() will need to modify their code to pass the new option. And they will not get all warnings, they will only get the PEP 765 warning. Is it worth a new option?

It is safer to disable that warning by default inast.parse().

Every user of ast.parse() and compile() will need to modify their code to pass the new option. And they will not get all warnings, they will only get the PEP 765 warning.

The default should be to give the warning. Turn it off in cpython when we know that the code was just ast.parse'ed (rather than an ast given by the user - we do know that).

What is the name of the option you propose to add?

Could bewith_warnings or something like that.

No, because it should not control all warnings. Maybe_enable_pep_765_warnings?

This would complicated, and I need to revert#131993 because it is a blocker for my other changes.

See yet one option in#139719. It is based on the original solution by@tomasr8.

Why not all warnings?

The early PEP 765 warnings were indeed specified in the last paragraph ofhttps://peps.python.org/pep-0765/#specification

Edit: that said, I'm open to revisiting that decision and instead letting linters make the call on how to present syntax warnings in their ruleset. My suspicion is that they would end up turning off the parse-time warnings anyway, in which case we don't actually lose anything important by handling the PEP 765 warnings in the code generation step instead of the AST parsing step.

Why not all warnings?

There are three kinds of warnings.#139642 (comment)

If we silenceall warnings inast.parse() orcompile(), we will silence not only the PEP 765 warnings, but also warnings emitted in the tokenizer or in the code generator.

Anyway, adding an option for selective silencing requires complex changes will be a complex change comparable to#139652, but with less benefit. I would rather not backport it to 3.14.

So we have a choice for 3.14:

• Silence the PEP 765 warnings inast.parse(). This PR.
• Silence the PEP 765 warnings in PyREPL (and all other sites that useast.parse() +compile, including third-party code).gh-131927: Silence PEP 765 warnings in PyREPL #139719 and more.
• Live with double PEP 765 warnings.gh-139640: Fix swallowing syntax warnings in different modules (restore duplicated PEP 765 warnings) #139755
• Live with sporadic swallowing of syntax warnings (Syntax warnings can be swallowed in import #139640) and incorrect module for filtering (Improve/fix filtering SyntaxWarnings by module #135801). The worst option.

The early PEP 765 warnings were indeed specified in the last paragraph ofhttps://peps.python.org/pep-0765/#specification

This contrasts AST construction and execution of pre-compiled code. But there is a missed step -- the code generation from AST. You cannot normally execute AST.

Thanks@serhiy-storchaka for the PR 🌮🎉.. I'm working now to backport this PR to: 3.13, 3.14.
🐍🍒⛏🤖

Sorry,@serhiy-storchaka, I could not cleanly backport this to3.13 due to a conflict.
Please backport usingcherry_picker on command line.

cherry_picker ad0a3f733b23e7fc69aff13055c7fac8ab9dcd66 3.13

GH-140786 is a backport of this pull request to the3.14 branch.

Thank you for review.