]> git.ipfire.org Git - thirdparty/git.git/commit
worktree: don't die() in library function find_worktree()
authorEric Sunshine <sunshine@sunshineco.com>
Tue, 28 Aug 2018 21:20:18 +0000 (17:20 -0400)
committerJunio C Hamano <gitster@pobox.com>
Thu, 30 Aug 2018 16:28:02 +0000 (09:28 -0700)
commit4c5fa9e6c4091eea9d8df3466b18ddb2812cd934
treeedd8e082bbc5fe414466daad52cff4fc4c7a4126
parent2f743933341f276111103550fbf383a34dfcfd38
worktree: don't die() in library function find_worktree()

Callers don't expect library function find_worktree() to die(); they
expect it to return the named worktree if found, or NULL if not.
Although find_worktree() itself never invokes die(), it calls
real_pathdup() with 'die_on_error' incorrectly set to 'true', thus will
die() indirectly if the user-provided path is not to real_pathdup()'s
liking. This can be observed, for instance, with any git-worktree
command which searches for an existing worktree:

    $ git worktree unlock foo
    fatal: 'foo' is not a working tree
    $ git worktree unlock foo/bar
    fatal: Invalid path '.../foo': No such file or directory

The first error message is the expected one from "git worktree unlock"
not finding the specified worktree; the second is from find_worktree()
invoking real_pathdup() incorrectly and die()ing prematurely.

Aside from the inconsistent error message between the two cases, this
bug hasn't otherwise been a serious problem since existing callers all
die() anyhow when the worktree can't be found. However, that may not be
true of callers added in the future, so fix find_worktree() to avoid
die()ing.

Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
t/t2028-worktree-move.sh
worktree.c