On 15/09/2020 10:34 pm, Pablo Galindo Salgado wrote:
Thanks for the proposal Mark!
I wanted to make some comments regarding converting AST nodes to PyObjects internally. I see some challenges here:
- Not using an arena allocator for the nodes can introduce more
challenges than simplifications. The first is that deleting a deep tree currently is just freeing the arena block, while if the nodes were PyObjects it will involve recursive destruction. That could potentially segfault so we would need to use some custom trashcan mechanism of special deleters. All of this will certainly not simplify the code (at least the parser code) and will impact performance (although just in the parser/compiler phase).
Don't we need to do all of this in the _ast module, already? We already have an AST composed of Python objects.
- We would need to (potentially) reimplement the AST sequences into
proper owning-containers. That will involve changing a considerable amount of code and some slowdown due to having to use C-API calls.
I agree that the sequences would need additional code for them to be Python objects. They would be more expensive to create, but I don't the additional cost being prohibitive.
- The proposal seems to imply that the AST will be a fully public and
stable API. This has some danger as any internal optimization regarding AST can be braking changes to macro users. This will raise any problems that now linter and static analyst tools could have to all users (of macros), making of even more difficult to change it.
This is a good point.
I am going to change macros to take a testlist instead of an arglist. Historically, function arguments have been an unstable part of the API. Since the AST node for testlist is just a list of ast.expr, it should be stable. That way at least the top level of the AST will be stable.
Regards from sunny London, Pablo Galindo
On Tue, 15 Sep 2020, 20:28 Mark Shannon, <firstname.lastname@example.org mailto:email@example.com> wrote:
Hi all, I'd like to propose a new PEP for hygienic macros. I'm attempting to reduce the controversy and pain of adding new syntax to Python, ... by adding new syntax. That might sound silly, but I hope to persuade that it isn't. Adding a new feature to the language is always controversial. Before a new module or functionality is added to the standard library, it must usually prove itself as a package on PyPI. (Not always the same package, the value of dataclasses was demonstrated by "attrs".) However, this isn't currently possible with new language features. Any new language feature has pros and cons. It adds expressive power for some (usually experienced) users, but potentially confuses and inconveniences other users (especially beginners). However, it is very hard, if not impossible, to determine whether adding new syntax will be be beneficial or harmful before the syntax is added. Is there a way to add new syntax in a way that allows it be battle tested before adding to the released version of the language? Yes there is, Macros. I'm not talking about C-style macros which do textual substitution, but "hygienic" macros that rewrite the AST during compilation. Macros allow domain specific new syntactic features, without bloating the language or confusing newcomers. In general, I would except macros to be used within libraries, so that those libraries gain the power of their custom macros without making Python ever larger and more complex. However, there is no reason why they cannot be made more widely available. Should a macro become widely used and popular, then it can be considered for adoption into the language. The PEP: https://github.com/markshannon/peps/blob/hygenic-macros/pep-0700.rst Cheers, Mark. _______________________________________________ Python-Dev mailing list -- firstname.lastname@example.org <mailto:email@example.com> To unsubscribe send an email to firstname.lastname@example.org <mailto:email@example.com> https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://firstname.lastname@example.org/message/7BX4S6ROKPMTTJR64UPRJ4D66C23HMHI/ Code of Conduct: http://python.org/psf/codeofconduct/