Re: [PATCH 3/4] check-non-portable-shell: improve `VAR=val shell-func` detection

From: Junio C Hamano <gitster@pobox.com>
Date: Jul 22, 2024, 21:57 UTC
Message-ID: <xmqq34o19jj1.fsf@gitster.g>
In-Reply-To: <CAO_smVhd_fWkC1=9r_ASCEPoM_rRap3DAWq--nq+6dQ8M8qzjQ@mail.gmail.com>

Kyle Lippincott <spectral@google.com> writes:

> Having just looked at the POSIX standard (I don't think I'm allowed to
> copy from this document), the POSIX standard (POSIX.1-2024, at least)
> explicitly leaves it unspecified whether the variable assignments
> remain in effect after function execution.

True.

https://pubs.opengroup.org/onlinepubs/9799919799/utilities/V3_chap02.html#tag_19_09_01_02

also says that it is unspecified if the variable gets exported, and older version of dash that comes on Ubuntu 20.04 chooses *not* to export, which was the test breakage that triggered this whole discussion. The thread can be seen here:

https://lore.kernel.org/git/xmqqbk2p9lwi.fsf_-_@gitster.g/

> Thanks for indulging my curiosity; should we include a statement in
> the linter along the lines of `# POSIX.1-2024 explicitly does not
> specify if variable assignment persists after executing a shell
> function; some shells, such as ksh, have these variables remain.`?

Giving a review (either positive or negative is fine, as long as it is constructive) on the update to CodingGuidelines

  https://lore.kernel.org/git/xmqqbk2p9lwi.fsf_-_@gitster.g/

may be a good place to start.

Thanks.