Re: [PATCH 1/8] update-index: add untracked cache notifications

Christian Couder <christian.couder@gmail.com> writes:

I think your "expectation" needs to be more explicitly spelled out.

"git -C /tmp --git-dir=/git/somewhere/else/.git" is a valid way to use that repository you have in somewhere else to track things under /tmp/ (as you are only passing GIT_DIR but not GIT_WORK_TREE, the cwd, i.e. /tmp, is the root level of the working tree), and for such a usage, the above command works as expected. Perhaps

    Attempting to flip the untracked-cache feature on for a random index
    file with

       cd /random/unrelated/place
       git --git-dir=/somewhere/else/.git update-index --untracked-cache

    would not work as you might expect.  Because flipping the
    feature on in the index also records the location of the
    corresponding working tree (/random/unrelated/place in the above
    example), when the index is subsequently used to keep track of
    files in the working tree in /somewhere/else, the feature is
    disabled.

may be an improvement.

The index already implicitly records where the working tree was and that is not limited to untracked-cache option. For example, if you have your repository and its working tree in /git/somewhere/else, which does not have a path X, then doing:

    cd /tmp && >tmp/X
    git --git-dir=/git/somewhere/else/.git update-index --add X

would store X taken from /tmp in the index, so subsequent use of the index "knows" about X that was taken outside /git/somewhere/else/ after the above command finishes and the subsequent use is made without the --git-dir parameter, e.g.

    cd /git/somewhere/else/ && git diff-index --cached HEAD'

would say that you added X, even though /git/somewhere/else/ may not have that X at all. And this is not limited to update-index, either. You can temporarily use --git-dir with "git add X" and the result would persist the same way in the index.

I think the moral of the story is that you are not expected to randomly use git-dir and git-work-tree to point at different places without knowing what you are doing, and we may need a way to help people understand what is going on when it is done by a mistake.

The patch to show result from xgetcwd() and get_git_work_tree() may be a step in the right direction, but if the user is not doing anything fancy, "Testing mtime in /long/path/to/the/directory" would be irritatingly verbose.

I wonder if it is easy to tell when the user is doing such an unnatural thing. Off the top of my head, when the working tree is anywhere other than one level above $GIT_DIR, the user is doing something unusual; I do not know if there are cases where the user is doing something unnatural if $GIT_WORK_TREE is one level above $GIT_DIR, though.