Matthias Meulien <orontee@gmail.com> writes:

> Augusto Stoffel <arstoffel@gmail.com> writes:
>> (...)
>> This change implies that the "success" branch may run if the exit code
>> is nonzero (without knowing all the isort internals, it can't be
>> excluded that an exit code > 2 is used somewhere).  I suggest instead a
>> (pcase status ...) to construct the " (maybe... ?)" segment of the error
>> message.
>
> Thanks, I'll fix this.

>> We could also take this opportunity to distinguish between random
>> exceptions happening in the script (which likely leads to exit code 1)
>> and the ModuleNotFoundError case.
>
> Good point, I'll improve this too.

Here are two patches.

The first one revert part of the original patch where I introduced a
useless detailed check of isort exit status itself.

The second one relates to your remarks: It shift the custom exit status
of the Python script and make sure that an error is reported assoon as
exit code >0.