When we run a test helper function in test_submodule_switch_common(), we
sometimes specify a whole helper function as the $command. When we do
this, in some test cases, we just mark the whole function with
`test_must_fail`. However, it's possible that the helper function might
fail earlier or later than expected due to an introduced bug. If this
happens, then the test case will still report as passing but it should
really be marked as failing since it didn't actually display the
intended behaviour.
Instead of invoking $command as one monolithic helper function, break it
up into three parts:
1. $command which is always a git command.
2. $before which is a callback function that runs just prior to
$command.
3. $after which is a callback function that runs just after
$command.
If the command requires a filename argument, specify it as `\$arg` since
that variable will be set and the whole $command string will be eval'd.
Unfortunately, there is no way to get rid of the eval as some of the
commands that are passed (such as the `git pull` tests) require that no
additional arguments are passed so we must have some mechanism for the
caller to specify whether or not it wants the filename argument.
The $before and $after callback functions will be passed the filename as
the first arg. These callback functions are optional and, if missing,
will be replaced with `true`. Also, in the case where we have a
`test_must_fail` test, $after will not be executed, similar to how the
helper functions currently behave when the git command fails and exits
the &&-chain.
Finally, as an added bonus, `test_must_fail` will only run on $command
which is guaranteed to be a git command.
An alternate design was considered where $OVERWRITING_FAIL is set from
test_submodule_switch_common() and exposed to the helper function. This
approach was considered too difficult to understand due to the fact that
using a signalling magic environment variable might be too indirect.
Signed-off-by: Denton Liu <liu.denton@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since all invocations of test_submodule_forced_switch() are git
commands, automatically prepend "git" before invoking
test_submodule_switch_common().
Similarly, many invocations of test_submodule_switch() are also git
commands so automatically prepend "git" before invoking
test_submodule_switch_common() as well.
Finally, for invocations of test_submodule_switch() that invoke a custom
function, rename the old function to test_submodule_switch_func().
This is necessary because in a future commit, we will be adding some
logic that needs to distinguish between an invocation of a plain git
comamnd and an invocation of a test helper function.
Signed-off-by: Denton Liu <liu.denton@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
git stash push does not recursively stash submodules, but if
submodule.recurse is set, it may recursively reset --hard them. Having
only the destructive action recurse is likely to be surprising
behaviour, and unlikely to be desirable, so the easiest fix should be to
ensure that the call to git reset --hard never recurses into submodules.
This matches the behavior of check_changes_tracked_files, which ignores
submodules.
Signed-off-by: Jakob Jarmar <jakob@jarmar.se>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Test that the stash apply command updates the work tree as expected for
changes which don't result in conflicts. To make that work add a helper
function that uses read-tree to apply the changes of the target commit
to the work tree, then stashes these changes and at last applies that
stash.
Implement the KNOWN_FAILURE_STASH_DOES_IGNORE_SUBMODULE_CHANGES switch
and reuse two other already present switches to expect the known
failure that stash does ignore submodule changes.
Signed-off-by: Jens Lehmann <Jens.Lehmann@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
