unpack-trees: only clear CE_UPDATE|CE_REMOVE when skip-worktree is always set

The purpose of this clearing is, as explained in comment, because
verify_*() may set those bits before apply_sparse_checkout() is
called. By that time, it's not clear whether an entry will stay in
checkout area or out. After $GIT_DIR/info/sparse-checkout is applied,
we know what entries will be in finally. It's time to clean unwanted
bits.

That works perfectly when checkout area remains unchanged. When
checkout area changes, apply_sparse_checkout() may set CE_UPDATE
or CE_WT_REMOVE to widen/narrow checkout area. Doing the clearing
after apply_sparse_checkout() may clear those widening/narrowing
bits unexpectedly.

So, only do that on entries that are not affected by checkout area
changes (i.e. skip-worktree bit does not change after
apply_sparse_checkout).

This code does not actually fix anything though, just
future-proof. The removed code and the narrow/widen code inside
apply_sparse_checkout are currently independent (narrow code never
sets CE_REMOVE, widen code sets CE_UPDATE, but ce_skip_worktree()
would be false).

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

1 files changed, 12 insertions, 0 deletions

diff --git a/t/t1011-read-tree-sparse-checkout.sh b/t/t1011-read-tree-sparse-checkout.sh
index 9189de8..81ab4c6 100755
--- a/t/t1011-read-tree-sparse-checkout.sh
+++ b/t/t1011-read-tree-sparse-checkout.sh
@@ -155,4 +155,16 @@ test_expect_success 'read-tree adds to worktree, dirty case' '
 	grep -q dirty sub/added
 '
 
+test_expect_success 'index removal and worktree narrowing at the same time' '
+	>empty &&
+	echo init.t >.git/info/sparse-checkout &&
+	echo sub/added >>.git/info/sparse-checkout &&
+	git checkout -f top &&
+	echo init.t >.git/info/sparse-checkout &&
+	git checkout removed &&
+	git ls-files sub/added >result &&
+	test ! -f sub/added &&
+	test_cmp empty result
+'
+
 test_done