Missing some obvious ast optimizations

[wrongnull]

Feature or enhancement

Proposal:

suppose the following code

1 in [1, 2]

it produces such byte code

  0           0 RESUME                   0

  1           2 LOAD_CONST               0 (1)
              4 LOAD_CONST               1 ((1, 2))
              6 CONTAINS_OP              0
              8 RETURN_VALUE

however, interpreter can easily fold such an expression to a constant in the ast_opt stage. This only applies if both ast nodes are Constant_kind. But there is one obstacle. Types are not yet ready and have incomplete mro at the ast_opt stage. So we can't use functions like PyObject_RichCompare for this. We can compare pointers to PyObject instead, since we know that each object in a "constant expression" is internally a singleton (interned strings, integers from -5 to 255, etc.). Thus, for them, the == operator is semantically equal to the is operator. I'm going to make pull request for this

Has this already been discussed elsewhere?

No response given

Links to previous discussion of this feature:

No response

Linked PRs

• gh-123080: Additional ast optimizations #123081

[serhiy-storchaka]

The optimizer code is intentionally kept simple. Only necessary optimizations are performed, these that are common in user code. For example, '-'*80, 1/3 or 2**32-1.

This idea was already proposed and rejected earlier, for example in #70909. I think that this issue should be closed for the same reasons.

[wrongnull]

The optimizer code is intentionally kept simple. Only necessary optimizations are performed, these that are common in user code. For example, '-'*80, 1/3 or 2**32-1.

This idea was already proposed and rejected earlier, for example in #70909. I think that this issue should be closed for the same reasons.

IMO since there is no strict definition of necessity if interpreter CAN optimize something in a reasonable amount of time it SHOULD do so

[serhiy-storchaka]

It's not done that way. You should have good reasons to do something, not to not do it. Adding new optimizations increases maintenance burden and has a risk of introducing new bugs.