bpo-29622: make AST constructor accepts less than enough number of positional arguments#249
bpo-29622: make AST constructor accepts less than enough number of positional arguments#249methane merged 3 commits into
Conversation
Sorry, something went wrong.
methane
changed the title
|
I implemented minimum number of arguments. |
Sorry, something went wrong.
|
Could I ask that we don't make arguments required, even if you need them to construct a valid AST? I have a pair of little side projects (astsearch and astcheck) which rely on the ability to instantiate incomplete AST objects. I reuse these as templates, so you can check for things like 'for loop with an else clause' without specifying any of the required pieces of a for loop. Obviously this isn't what the AST classes are intended for, but it works pretty nicely. And yes, I am being this guy: |
Sorry, something went wrong.
methane
changed the title
methane
changed the title
|
Thanks ! |
Sorry, something went wrong.
…n API Stackless does not support custom frame evaluation functions defined by PEP 523. If an extension module sets a custom frame evaluation function, Stackless now terminates to prevent undefined behavior. (cherry picked from commit d57b317)
…n API Stackless does not support custom frame evaluation functions defined by PEP 523. If an extension module sets a custom frame evaluation function, Stackless now terminates to prevent undefined behavior.
…athsep Honor '/'-separated names in ResourceContainer.joinpath.
Reviewers
+1 more reviewer
5 participants
Currently, AST constructor accepts
a. empty arguments
b. positional arguments, only when it's length is exactly same to number of fields
c. keyword arguments. No check for missing required fields.
Only (b) is strict. And it require argument even if matching field is optional.
This pull request removes the strict check.
Missing required field can be detected when compiling AST though.