Bytecode for empty f-string != bytecode for empty normal string

[thatbirdguythatuknownot]

Bug report

The bytecode for an empty f-string does not match the bytecode for an empty normal string.

import dis
dis.dis('""')
dis.dis('f""')

Output:

0           0 RESUME                   0

1           2 LOAD_CONST               0 ('')
            4 RETURN_VALUE
0           0 RESUME                   0

1           2 BUILD_STRING             0
            4 RETURN_VALUE

Environment

• CPython versions tested on:
  3.11, 3.12.0a2

Linked PRs

• gh-99606: consistent empty f-string bytecode #99608
• gh-99606: Make code generated for an empty f-string identical to that of a normal empty string #112407

[pochmann]

Why do you call that a bug? And why would you write f""?

[thatbirdguythatuknownot]

Why do you call that a bug? And why would you write f""?

Aren't they the same thing so shouldn't they have the same bytecode? It's pretty inconsistent with f"string contents" without the replacement fields and the other prefix combos with empty strings (they all use LOAD_CONST). Also sometimes I can forget to remove or subconsciously add f.

[ericvsmith]

The intention was that an f-string without any substitutions would be equivalent to a normal string. These examples are 3.12+. It's true for non-empty strings, it should probably be the same for empty strings. Not that I think it's a big deal.

>>> dis.dis("f'x'")
  0           0 RESUME                   0

  1           2 LOAD_CONST               0 ('x')
              4 RETURN_VALUE
>>> dis.dis("'x'")
  0           0 RESUME                   0

  1           2 LOAD_CONST               0 ('x')
              4 RETURN_VALUE
>>> dis.dis("f''")
  0           0 RESUME                   0

  1           2 BUILD_STRING             0
              4 RETURN_VALUE
>>> dis.dis("''")
  0           0 RESUME                   0

  1           2 LOAD_CONST               0 ('')
              4 RETURN_VALUE