[RFC PATCH 0/3] Make fsck check other worktree HEADs

This patchset adds checking of other worktree HEADs to fsck.

The reason I've marked this RFC is that I'm worried my incidental reliance on "worktrees/$WORKTREE/HEAD" resolving as a ref (in patch 3) might raise some flags for others. In particular, in [1] Peff said that this refname resolves right now mostly by accident and will probably stop working in the future. However, I feel that since fsck checks the storage format as well as contents, it seems natural that a change of storage model would result in the fsck code changing and thus that I'm not locking in any particular ref format long term with these changes. But I want to flag this issue for discussion.

[1] https://public-inbox.org/git/20180207181706.GA4227@sigill.intra.peff.net/

Elijah Newren (3):
  fsck: Move fsck_head_link() to get_default_heads() to avoid some
    globals
  t1450-fsck: Add tests for HEAD of other worktrees
  fsck: Check HEAD of other worktrees as well

 builtin/fsck.c  | 73 ++++++++++++++++++++++++++++++++++++++++++++-------------
 t/t1450-fsck.sh | 27 +++++++++++++++++++++
 2 files changed, 84 insertions(+), 16 deletions(-)