GH-132732: Use pure op machinery to optimize various instructions with _POP_TOP and _POP_TWO
#137577
Conversation
_POP_TOP and _POP_TWO relevant instructions_POP_TOP and _POP_TWO
Misc/NEWS.d/next/Core_and_Builtins/2025-08-09-04-07-05.gh-issue-132732.8BiIVJ.rst
Outdated
Show resolved
Hide resolved
|
When you're done making the requested changes, leave the comment: |
Python/optimizer_bytecodes.c
Outdated
| @@ -412,6 +413,7 @@ dummy_func(void) { | |||
| } | |||
|
|
|||
| op(_UNARY_INVERT, (value -- res)) { | |||
| REPLACE_OPCODE_IF_EVALUATES_PURE(value); | |||
There was a problem hiding this comment.
I think this isn't pure for bool. ~True raises a warning.
There was a problem hiding this comment.
Does it make sense to guard against booleans? Or are we saying that we can't do that/it's not worth it?
Also, can one of you explain why ~True isn't pure and how you discovered this/where the warning is raised?
There was a problem hiding this comment.
We could guard against booleans I think.
Anything that raises a warning isn't pure as that means the optimizer might cause a warning to be emitted during optimizing code, and we don't want that.
To test it out, try typing ~True into the current main's CPython repl. It's a recent deprecation so it wont show up on your system cpython probably, but a warning is emitted.
I found this out because I vaguely recalled reading about it somewhere https://discuss.python.org/t/bool-deprecation/62232
Uh oh!
There was an error while loading. Please reload this page.