revision.c: make default history consider bottom commits

Previously, the default history treated bottom commits the same as any
other UNINTERESTING commit, which could force it down side branches.

Consider the following history:

   *A--*B---D--*F         * marks !TREESAME parent paths
     \     /*
      `-C-'

When requesting "B..F", B is UNINTERESTING but TREESAME to D. C is
!UNINTERESTING.

So default following would go from D into the irrelevant side branch C
to A, rather than to B.  Note also that if there had been an extra
!UNINTERESTING commit B1 between B and D, it wouldn't have gone down C.

Change the default following to test relevant_commit() instead of
!UNINTERESTING, so it can proceed straight from D to B, thus finishing
the traversal of that path.

Signed-off-by: Kevin Bracey <kevin@bracey.fi>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

revision.c

@@ -684,7 +684,7 @@ static void try_to_simplify_commit(struct rev_info *revs, struct commit *commit)
 			    sha1_to_hex(p->object.sha1));
 		switch (rev_compare_tree(revs, p, commit)) {
 		case REV_TREE_SAME:
-			if (!revs->simplify_history || (p->object.flags & UNINTERESTING)) {
+			if (!revs->simplify_history || !relevant_commit(p)) {
 				/* Even if a merge with an uninteresting
 				 * side branch brought the entire change
 				 * we are interested in, we do not want

t/t6111-rev-list-treesame.sh

@@ -146,8 +146,8 @@ check_result '(LH)M (E)H (J)L (I)J (E)I' E..M --ancestry-path --simplify-merges
 # to G.
 check_result 'M L K J I H' G..M
 check_result 'M H L K J I' G..M --topo-order
-check_outcome failure 'M L H' G..M -- file # includes J I
-check_outcome failure '(LH)M (G)L (G)H' G..M --parents -- file # includes J I
+check_result 'M L H' G..M -- file
+check_result '(LH)M (G)L (G)H' G..M --parents -- file
 check_result 'M L J I H' G..M --full-history -- file
 check_result 'M L K J I H' G..M --full-history --parents -- file
 check_result 'M H L J I' G..M --simplify-merges -- file
@@ -161,8 +161,8 @@ check_result 'M H L J I' G..M --ancestry-path --simplify-merges -- file
 # But --full-history shouldn't drop D on its own - without simplification,
 # we can't decide if the merge from INTERESTING commit C was sensible.
 check_result 'F D C' B..F
-check_outcome failure 'F' B..F -- file # includes D
-check_outcome failure '(B)F' B..F --parents -- file # includes D
+check_result 'F' B..F -- file
+check_result '(B)F' B..F --parents -- file
 check_result 'F D' B..F --full-history -- file
 check_result '(D)F (BA)D' B..F --full-history --parents -- file
 check_result '(B)F' B..F --simplify-merges -- file
@@ -174,8 +174,8 @@ check_result 'F D' B..F --first-parent
 check_result 'F' B..F --first-parent -- file
 
 # E...F should be equivalent to E F ^B, and be able to drop D as above.
-check_outcome failure 'F' E F ^B -- file # includes D
-check_outcome failure 'F' E...F -- file # includes D
+check_result 'F' E F ^B -- file # includes D
+check_result 'F' E...F -- file # includes D
 
 # Any sort of full history of C..F should show D, as it's the connection to C,
 # and it differs from it.