GeezerVison

_BUILD_MAP returns a dict not a map() object.

@Fidget-Spinner I think I still have some work for this PR

A few days ago, In #132289, @brandtbucher setup optimization path for _GUARD_NOS_DICT/_GUARD_NOS_LIST, But after using sym_new_type instead of sym_new_not_null, the _GUARD_NOS_DICT/_GUARD_NOS_LIST is not existed in the final generate result. So the test failed here https://github.com/python/cpython/actions/runs/14419426452/job/40440141201?pr=132434#step:4:1788

I'm not sure this is what we want to get(I guess it's, so maybe we just need to patch the test?)

@Zheaoli, that's actually a good thing! The additional information that you've added to the optimizer for _BUILD_MAP and _BUILD_LIST means that the JIT is able to remove those guard instructions. You should just update the tests to check for 0 instead of 1, since these guard instructions shouldn't be present at all anymore.

(I don't think writing a test for the _BUILD_SLICE change is going to be very helpful, given how this instruction is used right now. So don't worry about that one.)

that's actually a good thing! The additional information that you've added to the optimizer for _BUILD_MAP and _BUILD_LIST means that the JIT is able to remove those guard instructions. You should just update the tests to check for 0 instead of 1, since these guard instructions shouldn't be present at all anymore.

Yes, I guess the _GUARD_NOS_DICT is not needed here(just for the test) when we setup extra information for ops when I see the failed test. You confirmed my guess. Updating the PR

@Zheaoli please don't force push, just normal push so we can see the commit diff. We will squash and merge at the end as one commit. Thanks!