Open
Description
class S(tuple[*tuple[()]]): ...mypy 1.15 stacktrace (mypy-play):
Traceback (most recent call last):
File "<frozen runpy>", line 198, in _run_module_as_main
File "<frozen runpy>", line 88, in _run_code
File "/layers/google.python.pip/pip/lib/python3.12/site-packages/mypy/__main__.py", line 37, in <module>
console_entry()
File "/layers/google.python.pip/pip/lib/python3.12/site-packages/mypy/__main__.py", line 15, in console_entry
main()
File "mypy/main.py", line 119, in main
File "mypy/main.py", line 203, in run_build
File "mypy/build.py", line 191, in build
File "mypy/build.py", line 267, in _build
File "mypy/build.py", line 2937, in dispatch
File "mypy/build.py", line 3335, in process_graph
File "mypy/build.py", line 3430, in process_stale_scc
File "mypy/semanal_main.py", line 97, in semantic_analysis_for_scc
File "mypy/semanal.py", line 7621, in apply_semantic_analyzer_patches
File "mypy/semanal.py", line 2587, in <lambda>
File "mypy/semanal_shared.py", line 305, in calculate_tuple_fallback
NotImplementedErrorand on current master (mypy-play)
Traceback (most recent call last):
File "<frozen runpy>", line 198, in _run_module_as_main
File "<frozen runpy>", line 88, in _run_code
File "mypy/__main__.py", line 37, in <module>
console_entry()
File "mypy/__main__.py", line 15, in console_entry
main()
File "mypy/main.py", line 127, in main
res, messages, blockers = run_build(sources, options, fscache, t0, stdout, stderr)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "mypy/main.py", line 211, in run_build
res = build.build(sources, options, None, flush_errors, fscache, stdout, stderr)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "mypy/build.py", line 191, in build
result = _build(
^^^^^^^
File "mypy/build.py", line 267, in _build
graph = dispatch(sources, manager, stdout)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "mypy/build.py", line 2939, in dispatch
process_graph(graph, manager)
File "mypy/build.py", line 3337, in process_graph
process_stale_scc(graph, scc, manager)
File "mypy/build.py", line 3432, in process_stale_scc
mypy.semanal_main.semantic_analysis_for_scc(graph, scc, manager.errors)
File "mypy/semanal_main.py", line 99, in semantic_analysis_for_scc
apply_semantic_analyzer_patches(patches)
File "mypy/semanal.py", line 7740, in apply_semantic_analyzer_patches
patch_func()
File "mypy/semanal.py", line 2630, in <lambda>
self.schedule_patch(PRIORITY_FALLBACKS, lambda: calculate_tuple_fallback(base))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "mypy/semanal_shared.py", line 305, in calculate_tuple_fallback
raise NotImplementedErroranother way to trigger this crash is through a typo in the variadic type parameter name:
class MyTuple[*Ts](tuple[*_Ts]): ... # oh no a typothis produces an identical stacktrace (mypy-play)
some observations:
- it doesn't matter if
typing.Unpackis used instead of* - I checked all versions from 1.9.0-master, and all crash in the same way
class S(tuple[()]): ...does not cause a crash
(althoughS()is revealed astuple[(), fallback=__main__.S], which is kinda sus)- in case you didn't notice; the first example is valid at runtime, the second isn't
- my ear is a bit itchy
And it's probably not what you want to hear, but it might be relevant info, so I'll carefully whisper it: Pyright does not have this problem (playground 1, playground 2), and the same goes for ty (playground 1, playground 2)
ref: #19100 (comment)
Metadata
Metadata
Assignees
Projects
Milestone
Relationships
Participants
Activity
[-]`class S(tuple[*tuple[()]]): ...` crashes mypy[/-][+]`class S(tuple[*tuple[()]])` crashes mypy[/+]tuplesubtype not assignable totuple[int, ...]#19110sterliakov commentedon May 19, 2025
Ough, note this isn't specific to empty tuples, the following crashes just as hard:
sterliakov commentedon May 19, 2025
Quick braindump: we normalize such trivial tuples in
semanal_typeargs, butTypeArgumentsAnalyzerruns after we have calculated all fallbacks. Fallback calculation logic assumes normalized inputs, which is most probably wrong. I'm not sure whether what I did is the best approach, but at least it doesn't crash anymore - let's continue this discussion in the linked PR.