gh-104683: Rework Argument Clinic error handling #107551
Conversation
- Introduce ClinicError and ClinicWarning - fail() raises ClinicError - warn() uses warnings.warn(msg, ClinicWarning) - the CLI runs main(), catches ClinicError, formats the error message prints to stderr and exits with an error - adapt the test suite to work with ClinicError
|
Ready for a new round of review! Finally, all (or most) line number are now correct. If I introduce an error in Lib/test/clinic.test.c, I now get an accurate line number. Yay! I'm not sure why some of the tests are off by one, though. Perhaps it has to do with FakeClinic. |
Tools/clinic/clinic.py
Outdated
| try: | ||
| self.state(line) | ||
| except ClinicError as exc: | ||
| raise ClinicError(str(exc), filename=exc.filename, lineno=line_number) |
There was a problem hiding this comment.
IMO, this is a workaround! We should ensure that line numbers are correctly updated in the state machine. Let's create an issue about this.
There was a problem hiding this comment.
I'd like to keep the workaround for now; it really helps to have accurate line numbers when debugging.
Co-authored-by: Alex Waygood <[email protected]>
|
I had an idea for how to deduplicate some code between |
|
Great, thank you; this PR is in a lot better shape now 😄 |
Introduce ClinicError and ClinicWarning:
In main(), catch ClinicError, format the error message,
print to stderr, and exit with an error. Adapt the test
suite accordingly.
Tools/clinic/#104683