From: Patrick Steinhardt Date: Tue, 14 Apr 2026 07:24:02 GMT Subject: Re: [PATCH 07/12] t: prepare `test_when_finished ()`/`test_atexit()` for `set -e` Message-ID: In-Reply-To: On Mon, Apr 13, 2026 at 10:23:37AM -0700, Junio C Hamano wrote: > Patrick Steinhardt writes: > > > Both `test_when_finished ()` and `test_atexit ()` build up a chain of > > cleanup commands by prepending each new command to the existing cleanup > > string. To preserve the exit code of the test body across cleanup > > execution, we append the following logic: > > > > } && (exit "$eval_ret"); eval_ret=$?; ... > > > > The intent of this is to run the cleanup block and then unconditionally > > restore `eval_ret`. The original behaviour of this is is: > > > > +------------------+---------+------------------------------------+ > > |test body │ cleanup │ old behaviour │ > > +------------------+---------+------------------------------------+ > > │pass (eval_ret=0) | pass │ && taken -> (exit 0) -> eval_ret=0 | > > +------------------+---------+------------------------------------+ > > │pass (eval_ret=0) | fail │ && not taken -> eval_ret=$? | > > +------------------+---------+------------------------------------+ > > │fail (eval_ret=1) | pass │ && taken -> (exit 1) -> eval_ret=1 | > > +------------------+---------+------------------------------------+ > > │fail (eval_ret=1) | fail | && not taken -> eval_ret=$? | > > +------------------+---------+------------------------------------+ > > > > This logic will start to fail once we enable `set -e`. When `$eval_ret` > > is non-zero, the subshell we create will fail, and with `set -e` we'll > > thus bail out without evaluating the logic after the semicolon. > > > > Fix this issue by instead using `|| eval_ret=\$?; ...`. Besides being > > a bit simpler, it also retains the original behaviour: > > > > +------------------+---------+------------------------------------+ > > |test body │ cleanup │ old behaviour │ > > +------------------+---------+------------------------------------+ > > │pass (eval_ret=0) | pass │ || not taken -> eval_ret unchanged | > > +------------------+---------+------------------------------------+ > > │pass (eval_ret=0) | fail │ || taken -> eval_ret=$? | > > +------------------+---------+------------------------------------+ > > │fail (eval_ret=1) | pass │ || not taken -> eval_ret unchanged | > > +------------------+---------+------------------------------------+ > > │fail (eval_ret=1) | fail | || taken -> eval_ret=$? | > > +------------------+---------+------------------------------------+ > > > > Signed-off-by: Patrick Steinhardt > > --- > > t/test-lib-functions.sh | 4 ++-- > > 1 file changed, 2 insertions(+), 2 deletions(-) > > > > diff --git a/t/test-lib-functions.sh b/t/test-lib-functions.sh > > index f8bc77619b..0eac676109 100644 > > --- a/t/test-lib-functions.sh > > +++ b/t/test-lib-functions.sh > > @@ -1516,7 +1516,7 @@ test_when_finished () { > > test "${BASH_SUBSHELL-0}" = 0 || > > BUG "test_when_finished does nothing in a subshell" > > test_cleanup="{ $* > > - } && (exit \"\$eval_ret\"); eval_ret=\$?; $test_cleanup" > > + } || eval_ret=\$?; $test_cleanup" > > Hmph, it seems this "&& (exit $eval_ret)" pattern has been with us > forever since it was introduced at 3bf78867 (test-lib: Let tests > specify commands to be run at end of test, 2010-05-02), and survived > a slight modification to work around issues on FreeBSD done in > b6b0afdc (test-lib: some shells do not let $? propagate into an > eval, 2010-05-06). When a major part of test-lib.sh was split into > test-lib-functions.s at 12a29b1a (Move the user-facing test library > to test-lib-functions.sh, 2012-02-17), this part was copied intact. > > Does this interact well with the glitch b6b0afdc (test-lib: some > shells do not let $? propagate into an eval, 2010-05-06) tried to > work around, by the way? > > Thanks. Ah, that's a very good question. I'll spin up a FreeBSD VM to give this a test, thanks! Patrick