Elijah Newren <newren@xxxxxxxxx> writes: >> Because a new process instance of $PAUSE_SHELL is run, the options >> you add when inserting test_pause does not affect what happens in >> the tests after you exit the $PAUSE_SHELL [*], right? > > I think the warning was less about what happens after test_pause is > complete and the testcase resumes, and more intended to convey that if > the user tries to manually copy & paste snippets of code from the test > into $PAUSE_SHELL, then the use of these flags can cause the result of > those pasted commands to differ. But the difference of TERM and SHELL are much smaller issue when one cuts and pastes the lines from the test script. Even if you do not use -s or -t options, shell variables and helpers won't be available to the interactive session given by test_pause, and that is a much bigger difference between the "real" shell environment that is running the tests and the user's shell environment test_pause gives us. That is why I think the WARNING should be attached _directly_ to the description of test_pause and not to individual "this uses different X environment compared to the real shell environment running the tests" options. The updated $HOME does deserve a special mention, as .gitconfig from it would further affect the tests and hide the bug you are hunting for, and the CAUTION is a good thing to have, even if you do not plan to modify anything in the test_pause sight-seeing session.
