Re: [PATCH 1/7] t: move tests exercising the "files" backend

Patrick Steinhardt [off-list ref] writes:

We still have a bunch of tests scattered across our test suites that
exercise on-disk files of the "files" backend directly:

  - t1301 exercises permissions of reflog files when the config
    "core.sharedRepository" is set.

  - t1400 exercises whether empty directories in the ref store are
    handled correctly.

  - t3200 exercises what happens when there are symlinks in the ref
    store.

  - t3400 also exercises what happens when ".git/logs" is a symlink.

All of these are inherently low-level tests specific to the "files"
backend. Move them into "t0600-reffiles-backend.sh" to reflect this.

Makes sense, and they look like straight code movements.

These two are from t1400.

+test_expect_success 'empty directory removal' '
+	git branch d1/d2/r1 HEAD &&
+	git branch d1/r2 HEAD &&
+	test_path_is_file .git/refs/heads/d1/d2/r1 &&
+	test_path_is_file .git/logs/refs/heads/d1/d2/r1 &&
+	git branch -d d1/d2/r1 &&
+	test_must_fail git show-ref --verify -q refs/heads/d1/d2 &&
+	test_must_fail git show-ref --verify -q logs/refs/heads/d1/d2 &&
+	test_path_is_file .git/refs/heads/d1/r2 &&
+	test_path_is_file .git/logs/refs/heads/d1/r2
+'
+
+test_expect_success 'symref empty directory removal' '
+	git branch e1/e2/r1 HEAD &&
+	git branch e1/r2 HEAD &&
+	git checkout e1/e2/r1 &&
+	test_when_finished "git checkout main" &&
+	test_path_is_file .git/refs/heads/e1/e2/r1 &&
+	test_path_is_file .git/logs/refs/heads/e1/e2/r1 &&
+	git update-ref -d HEAD &&
+	test_must_fail git show-ref --verify -q refs/heads/e1/e2 &&
+	test_must_fail git show-ref --verify -q logs/refs/heads/e1/e2 &&
+	test_path_is_file .git/refs/heads/e1/r2 &&
+	test_path_is_file .git/logs/refs/heads/e1/r2 &&
+	test_path_is_file .git/logs/HEAD
+'

Luckily there aren't refs that were created at this point by earlier
tests that would contradict/block the creation of the new refs this
moved tests would want to create.

+test_expect_success 'directory not created deleting packed ref' '
+	git branch d1/d2/r1 HEAD &&
+	git pack-refs --all &&
+	test_path_is_missing .git/refs/heads/d1/d2 &&
+	git update-ref -d refs/heads/d1/d2/r1 &&
+	test_path_is_missing .git/refs/heads/d1/d2 &&
+	test_path_is_missing .git/refs/heads/d1
+'

And this is from the same t1400 but appears much later in the file.
Curiously, this tries to create d1/d2/r1 that an earlier test
already created, and this one passes because the branch gets removed
in the other test that also created the branch.  Tricky but not a
fault of this patch.

+test_expect_success SYMLINKS 'git branch -m u v should fail when the reflog for u is a symlink' '
+	git branch --create-reflog u &&
+	mv .git/logs/refs/heads/u real-u &&
+	ln -s real-u .git/logs/refs/heads/u &&
+	test_must_fail git branch -m u v
+'

This was migrated from t3200 and has no interaction with the new
context of t0600 as branch 'u' has never been used.  I notice that
this test is buggy since its inception at 16c2bfbb (rename_ref: use
lstat(2) when testing for symlink, 2006-11-29) in that we move the
log for branch 'u' up and then make the log for branch 'u' into a
dangling symlink, so it probably failed for a wrong reason even
before 16c2bfbb updated stat() to lstat().  Anyway, the current code
will see that logs/refs/heads/u is a symbolic link and fails,
regardless of what the link points at, so we are OK.  In any case,
not a fault of this patch.

+test_expect_success SYMLINKS 'git branch -m with symlinked .git/refs' '
+	test_when_finished "rm -rf subdir" &&
+	git init --bare subdir &&
+
+	rm -rfv subdir/refs subdir/objects subdir/packed-refs &&
+	ln -s ../.git/refs subdir/refs &&
+	ln -s ../.git/objects subdir/objects &&
+	ln -s ../.git/packed-refs subdir/packed-refs &&
+
+	git -C subdir rev-parse --absolute-git-dir >subdir.dir &&
+	git rev-parse --absolute-git-dir >our.dir &&
+	! test_cmp subdir.dir our.dir &&
+
+	git -C subdir log &&
+	git -C subdir branch rename-src &&
+	git rev-parse rename-src >expect &&
+	git -C subdir branch -m rename-src rename-dest &&
+	git rev-parse rename-dest >actual &&
+	test_cmp expect actual &&
+	git branch -D rename-dest
+'

OK.  As long as it cleans after itself, I won't read deeply into it
and stop at making sure this came verbatim from t3200.

+test_expect_success MINGW,SYMLINKS_WINDOWS 'rebase when .git/logs is a symlink' '
+	git checkout main &&
+	mv .git/logs actual_logs &&
+	cmd //c "mklink /D .git\logs ..\actual_logs" &&
+	git rebase -f HEAD^ &&
+	test -L .git/logs &&
+	rm .git/logs &&
+	mv actual_logs .git/logs
+'

As t0600 forces the default branch to be 'main', having this one
migrated from t3400 should be safe.

+test_expect_success POSIXPERM 'git reflog expire honors core.sharedRepository' '
+	umask 077 &&
+	git config core.sharedRepository group &&
+	git reflog expire --all &&
+	actual="$(ls -l .git/logs/refs/heads/main)" &&
+	case "$actual" in
+	-rw-rw-*)
+		: happy
+		;;
+	*)
+		echo Ooops, .git/logs/refs/heads/main is not 066x [$actual]
+		false
+		;;
+	esac
+'

And this is from t1301, another verbatim copy.

The rest of t1301 does not appear to depend on the reflog being
expired, so this move should be safe.

I've covered how these removals should be safe for the remaining
tests in this file already.  Good.

Likewise.

Looking good.

Thanks.