| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
Sorry, something went wrong.
There was a problem hiding this comment.
Don't merge this yet. I think we need more discussion about the desired behavior. And we need docs to match.
Sorry, something went wrong.
|
Can we include this in beta1? |
Sorry, something went wrong.
|
I have updated the PR to match the main branch and I am landing this as the discussion in bpo has been quiet for several years and everyone seems to be on the same page. @serhiy-storchaka, Дякую за чудову роботу! 😃 |
Sorry, something went wrong.
|
Wow! |
Sorry, something went wrong.
Summary: In cinder 3.8 we unintentionally allowed async comprehensions to nest inside non-async ones, due to comprehension inlining. When we ported comprehension inlining to 3.10, we closed this hole for better fidelity to upstream compiler behavior, but that means we have to fix the places in IGSRV that now break this rule. And this doesn't seem worth it considering the restriction is lifted in Python 3.11 anyway: python/cpython#6766 This diff just restores the 3.8 behavior of comprehension inlining allowing async comprehensions nested inside non-async ones. Reviewed By: itamaro Differential Revision: D39825885 fbshipit-source-id: bd2f27a
|
I noticed that no tests fail if I comment out this line: Line 5785 in 0898354 and I bisected it to this PR. (1) Does this make sense? |
Sorry, something went wrong.
I think it's fine because this PR added in symtable.c code that propagates the value of ste_coroutine from a comprehension to the enclosing scope (unless it's a generator, which the compile.c code excludes as well). So this is now happening already during symtable construction (as it should be), and we don't need the assignment in the compiler anymore. |
Sorry, something went wrong.
| Back | FazBrowse Home | New Git URL |
https://bugs.python.org/issue33346