PEP 695: Rename typeparams to type_params in AST

[JelleZijlstra]

PEP-695 introduces a few new AST attributes that are currently called typeparams, as specified in https://peps.python.org/pep-0695/#ast-changes. However, @AlexWaygood rightly points out (#104642 (comment)) that type_params would be more readable and in line with most of the rest of the AST. Should we change it?

cc @erictraut for this PEP, @cdce8p who made some AST changes in this area before, @pablogsal @isidentical as AST experts.

Linked PRs

• gh-104656: Rename typeparams AST node to type_params #104657

[AlexWaygood]

Reposting my comment in #104642 (comment) here, for visibility:

Nearly all other attributes on AST nodes in Python use snake_case -- type_ignores, decorator_list, end_lineno, type_comment, end_col_offset, format_spec, is_async, kw_defaults, context_expr, optional_vars, kwd_attrs, kwd_patterns. And the attribute on TypeAliases, functions and classes at runtime is __type_params__, not __typeparams__.

There's a few examples of smushedtogethercase attributes on AST nodes -- finalybody, kwonlyargs, vararg, asname. But they're definitely in the minority.

I would greatly prefer it if this attribute was renamed to be type_params.

[JelleZijlstra]

The same goes for the typeparam type in the AST. Naming it type_param would be consistent with e.g. type_ignore.

I'm working on a draft PR.

[cdce8p]

From a users perspective either works fine. It would be more important to not change it unnecessarily after it has been released. I.e. I wouldn't recommend to go back and "fix" finalybody, kwonlyargs, etc.

[AlexWaygood]

@erictraut, I think we'd be especially interested in your thoughts on this as the author of the PEP, since typeparams is clearly specified by the PEP as the name for these nodes. Would you object to this change?

[erictraut]

I don't have a strong opinion one way or the other on the AST names. I tried to follow the patterns I could see when I implemented my prototype. If there are some other conventions or precedents that I missed, I have no objections to changing these names.

[gpshead]

I'm in favor of including the _ becauserunonwords_are_less_readable.

[JelleZijlstra]

Unless someone disagrees strongly I'll merge the change to type_params tonight, in time for the beta tomorrow.

[JelleZijlstra]

Wewilluseunderscores.