The concern I have about this change is that ! is a shell builtin operator and this could cause confusion/result in bugs if someone tries to fun the command on the command line in certain shells where ! has special meanings in different contexts (we can't always assume atf-sh is running /bin/sh, unfortunately..).

How about atf_check -s '!0' false instead? If things were done that way the negation operator could be extended to -e and -o without any sort of ambiguity/confusion.

In D58099#1333191, @ngie wrote:

The concern I have about this change is that ! is a shell builtin operator and this could cause confusion/result in bugs if someone tries to fun the command on the command line in certain shells where ! has special meanings in different contexts (we can't always assume atf-sh is running /bin/sh, unfortunately..).

Can you provide an example?

How about atf_check -s '!0' false instead? If things were done that way the negation operator could be extended to -e and -o without any sort of ambiguity/confusion.

No, that completely negates the point of the patch.

ping

This still feels ambiguous in a non-desirable way to me, but I can see the value in making this possible so I want to give this the space to get more feedback instead of me serving as an unnecessary impediment.

Please send out a message with a brief summary of the proposed change/asking for feedback to hackers@ and testing@.
Please add tests for the new feature:
- Positive case: '! false' in m_argv[0]: negation behavior should be applied.
- Negative case: 'true !' in m_argv[1]: negation behavior should not be applied.
I would leave existing tests that could be simplified with the new feature as-is until enough time has been given to get feedback on the new feature.
I would set the MFC timer to a higher value like 2 weeks or a month.