Merge branch 'jc/merge-bases'

Optimise the "merge-base" computation a bit, and also update its
users that do not need the full merge-base information to call a
cheaper subset.

* jc/merge-bases:
  reduce_heads(): reimplement on top of remove_redundant()
  merge-base: "--is-ancestor A B"
  get_merge_bases_many(): walk from many tips in parallel
  in_merge_bases(): use paint_down_to_common()
  merge_bases_many(): split out the logic to paint history
  in_merge_bases(): omit unnecessary redundant common ancestor reduction
  http-push: use in_merge_bases() for fast-forward check
  receive-pack: use in_merge_bases() for fast-forward check
  in_merge_bases(): support only one "other" commit

11 files changed, 184 insertions, 112 deletions

diff --git a/Documentation/git-merge-base.txt b/Documentation/git-merge-base.txt
index b295bf8330..87842e33f8 100644
--- a/Documentation/git-merge-base.txt
+++ b/Documentation/git-merge-base.txt
@@ -11,6 +11,7 @@ SYNOPSIS
 [verse]
 'git merge-base' [-a|--all] <commit> <commit>...
 'git merge-base' [-a|--all] --octopus <commit>...
+'git merge-base' --is-ancestor <commit> <commit>
 'git merge-base' --independent <commit>...
 
 DESCRIPTION
@@ -50,6 +51,12 @@ from linkgit:git-show-branch[1] when used with the `--merge-base` option.
 	from any other.  This mimics the behavior of 'git show-branch
 	--independent'.
 
+--is-ancestor::
+	Check if the first <commit> is an ancestor of the second <commit>,
+	and exit with status 0 if true, or with status 1 if not.
+	Errors are signaled by a non-zero status that is not 1.
+
+
 OPTIONS
 -------
 -a::
@@ -110,6 +117,27 @@ both '1' and '2' are merge-bases of A and B.  Neither one is better than
 the other (both are 'best' merge bases).  When the `--all` option is not given,
 it is unspecified which best one is output.
 
+A common idiom to check "fast-forward-ness" between two commits A
+and B is (or at least used to be) to compute the merge base between
+A and B, and check if it is the same as A, in which case, A is an
+ancestor of B.  You will see this idiom used often in older scripts.
+
+	A=$(git rev-parse --verify A)
+	if test "$A" = "$(git merge-base A B)"
+	then
+		... A is an ancestor of B ...
+	fi
+
+In modern git, you can say this in a more direct way:
+
+	if git merge-base --is-ancestor A B
+	then
+		... A is an ancestor of B ...
+	fi
+
+instead.
+
+
 See also
 --------
 linkgit:git-rev-list[1],
diff --git a/builtin/branch.c b/builtin/branch.c
index e61b0ece21..5cb6d78f2c 100644
--- a/builtin/branch.c
+++ b/builtin/branch.c
@@ -130,7 +130,7 @@ static int branch_merged(int kind, const char *name,
 	if (!reference_rev)
 		reference_rev = head_rev;
 
-	merged = in_merge_bases(rev, &reference_rev, 1);
+	merged = in_merge_bases(rev, reference_rev);
 
 	/*
 	 * After the safety valve is fully redefined to "check with
@@ -140,7 +140,7 @@ static int branch_merged(int kind, const char *name,
 	 * a gentle reminder is in order.
 	 */
 	if ((head_rev != reference_rev) &&
-	    in_merge_bases(rev, &head_rev, 1) != merged) {
+	    in_merge_bases(rev, head_rev) != merged) {
 		if (merged)
 			warning(_("deleting branch '%s' that has been merged to\n"
 				"         '%s', but not yet merged to HEAD."),
diff --git a/builtin/fetch.c b/builtin/fetch.c
index 3f2ad7727f..97a849a9dc 100644
--- a/builtin/fetch.c
+++ b/builtin/fetch.c
@@ -323,7 +323,7 @@ static int update_local_ref(struct ref *ref,
 		return r;
 	}
 
-	if (in_merge_bases(current, &updated, 1)) {
+	if (in_merge_bases(current, updated)) {
 		char quickref[83];
 		int r;
 		strcpy(quickref, find_unique_abbrev(current->object.sha1, DEFAULT_ABBREV));
diff --git a/builtin/merge-base.c b/builtin/merge-base.c
index 2f223a6646..1bc7991048 100644
--- a/builtin/merge-base.c
+++ b/builtin/merge-base.c
@@ -26,6 +26,7 @@ static const char * const merge_base_usage[] = {
 	N_("git merge-base [-a|--all] <commit> <commit>..."),
 	N_("git merge-base [-a|--all] --octopus <commit>..."),
 	N_("git merge-base --independent <commit>..."),
+	N_("git merge-base --is-ancestor <commit> <commit>"),
 	NULL
 };
 
@@ -70,6 +71,20 @@ static int handle_octopus(int count, const char **args, int reduce, int show_all
 	return 0;
 }
 
+static int handle_is_ancestor(int argc, const char **argv)
+{
+	struct commit *one, *two;
+
+	if (argc != 2)
+		die("--is-ancestor takes exactly two commits");
+	one = get_commit_reference(argv[0]);
+	two = get_commit_reference(argv[1]);
+	if (in_merge_bases(one, two))
+		return 0;
+	else
+		return 1;
+}
+
 int cmd_merge_base(int argc, const char **argv, const char *prefix)
 {
 	struct commit **rev;
@@ -77,11 +92,14 @@ int cmd_merge_base(int argc, const char **argv, const char *prefix)
 	int show_all = 0;
 	int octopus = 0;
 	int reduce = 0;
+	int is_ancestor = 0;
 
 	struct option options[] = {
 		OPT_BOOLEAN('a', "all", &show_all, N_("output all common ancestors")),
 		OPT_BOOLEAN(0, "octopus", &octopus, N_("find ancestors for a single n-way merge")),
 		OPT_BOOLEAN(0, "independent", &reduce, N_("list revs not reachable from others")),
+		OPT_BOOLEAN(0, "is-ancestor", &is_ancestor,
+			    N_("is the first one ancestor of the other?")),
 		OPT_END()
 	};
 
@@ -89,6 +107,10 @@ int cmd_merge_base(int argc, const char **argv, const char *prefix)
 	argc = parse_options(argc, argv, prefix, options, merge_base_usage, 0);
 	if (!octopus && !reduce && argc < 2)
 		usage_with_options(merge_base_usage, options);
+	if (is_ancestor && (show_all | octopus | reduce))
+		die("--is-ancestor cannot be used with other options");
+	if (is_ancestor)
+		return handle_is_ancestor(argc, argv);
 	if (reduce && (show_all || octopus))
 		die("--independent cannot be used with other options");
 
diff --git a/builtin/receive-pack.c b/builtin/receive-pack.c
index 2cb854feb4..9145f1a595 100644
--- a/builtin/receive-pack.c
+++ b/builtin/receive-pack.c
@@ -480,7 +480,6 @@ static const char *update(struct command *cmd)
 	    !prefixcmp(name, "refs/heads/")) {
 		struct object *old_object, *new_object;
 		struct commit *old_commit, *new_commit;
-		struct commit_list *bases, *ent;
 
 		old_object = parse_object(old_sha1);
 		new_object = parse_object(new_sha1);
@@ -493,12 +492,7 @@ static const char *update(struct command *cmd)
 		}
 		old_commit = (struct commit *)old_object;
 		new_commit = (struct commit *)new_object;
-		bases = get_merge_bases(old_commit, new_commit, 1);
-		for (ent = bases; ent; ent = ent->next)
-			if (!hashcmp(old_sha1, ent->item->object.sha1))
-				break;
-		free_commit_list(bases);
-		if (!ent) {
+		if (!in_merge_bases(old_commit, new_commit)) {
 			rp_error("denying non-fast-forward %s"
 				 " (you should pull first)", name);
 			return "non-fast-forward";
diff --git a/commit.c b/commit.c
index 87268682f9..0ea441d13b 100644
--- a/commit.c
+++ b/commit.c
@@ -607,28 +607,12 @@ static struct commit *interesting(struct commit_list *list)
 	return NULL;
 }
 
-static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
+static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 {
 	struct commit_list *list = NULL;
 	struct commit_list *result = NULL;
 	int i;
 
-	for (i = 0; i < n; i++) {
-		if (one == twos[i])
-			/*
-			 * We do not mark this even with RESULT so we do not
-			 * have to clean it up.
-			 */
-			return commit_list_insert(one, &result);
-	}
-
-	if (parse_commit(one))
-		return NULL;
-	for (i = 0; i < n; i++) {
-		if (parse_commit(twos[i]))
-			return NULL;
-	}
-
 	one->object.flags |= PARENT1;
 	commit_list_insert_by_date(one, &list);
 	for (i = 0; i < n; i++) {
@@ -669,9 +653,34 @@ static struct commit_list *merge_bases_many(struct commit *one, int n, struct co
 		}
 	}
 
-	/* Clean up the result to remove stale ones */
 	free_commit_list(list);
-	list = result; result = NULL;
+	return result;
+}
+
+static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
+{
+	struct commit_list *list = NULL;
+	struct commit_list *result = NULL;
+	int i;
+
+	for (i = 0; i < n; i++) {
+		if (one == twos[i])
+			/*
+			 * We do not mark this even with RESULT so we do not
+			 * have to clean it up.
+			 */
+			return commit_list_insert(one, &result);
+	}
+
+	if (parse_commit(one))
+		return NULL;
+	for (i = 0; i < n; i++) {
+		if (parse_commit(twos[i]))
+			return NULL;
+	}
+
+	list = paint_down_to_common(one, n, twos);
+
 	while (list) {
 		struct commit_list *next = list->next;
 		if (!(list->item->object.flags & STALE))
@@ -709,6 +718,60 @@ struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 	return ret;
 }
 
+static int remove_redundant(struct commit **array, int cnt)
+{
+	/*
+	 * Some commit in the array may be an ancestor of
+	 * another commit.  Move such commit to the end of
+	 * the array, and return the number of commits that
+	 * are independent from each other.
+	 */
+	struct commit **work;
+	unsigned char *redundant;
+	int *filled_index;
+	int i, j, filled;
+
+	work = xcalloc(cnt, sizeof(*work));
+	redundant = xcalloc(cnt, 1);
+	filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
+
+	for (i = 0; i < cnt; i++) {
+		struct commit_list *common;
+
+		if (redundant[i])
+			continue;
+		for (j = filled = 0; j < cnt; j++) {
+			if (i == j || redundant[j])
+				continue;
+			filled_index[filled] = j;
+			work[filled++] = array[j];
+		}
+		common = paint_down_to_common(array[i], filled, work);
+		if (array[i]->object.flags & PARENT2)
+			redundant[i] = 1;
+		for (j = 0; j < filled; j++)
+			if (work[j]->object.flags & PARENT1)
+				redundant[filled_index[j]] = 1;
+		clear_commit_marks(array[i], all_flags);
+		for (j = 0; j < filled; j++)
+			clear_commit_marks(work[j], all_flags);
+		free_commit_list(common);
+	}
+
+	/* Now collect the result */
+	memcpy(work, array, sizeof(*array) * cnt);
+	for (i = filled = 0; i < cnt; i++)
+		if (!redundant[i])
+			array[filled++] = work[i];
+	for (j = filled, i = 0; i < cnt; i++)
+		if (redundant[i])
+			array[j++] = work[i];
+	free(work);
+	free(redundant);
+	free(filled_index);
+	return filled;
+}
+
 struct commit_list *get_merge_bases_many(struct commit *one,
 					 int n,
 					 struct commit **twos,
@@ -717,7 +780,7 @@ struct commit_list *get_merge_bases_many(struct commit *one,
 	struct commit_list *list;
 	struct commit **rslt;
 	struct commit_list *result;
-	int cnt, i, j;
+	int cnt, i;
 
 	result = merge_bases_many(one, n, twos);
 	for (i = 0; i < n; i++) {
@@ -748,28 +811,11 @@ struct commit_list *get_merge_bases_many(struct commit *one,
 	clear_commit_marks(one, all_flags);
 	for (i = 0; i < n; i++)
 		clear_commit_marks(twos[i], all_flags);
-	for (i = 0; i < cnt - 1; i++) {
-		for (j = i+1; j < cnt; j++) {
-			if (!rslt[i] || !rslt[j])
-				continue;
-			result = merge_bases_many(rslt[i], 1, &rslt[j]);
-			clear_commit_marks(rslt[i], all_flags);
-			clear_commit_marks(rslt[j], all_flags);
-			for (list = result; list; list = list->next) {
-				if (rslt[i] == list->item)
-					rslt[i] = NULL;
-				if (rslt[j] == list->item)
-					rslt[j] = NULL;
-			}
-		}
-	}
 
-	/* Surviving ones in rslt[] are the independent results */
+	cnt = remove_redundant(rslt, cnt);
 	result = NULL;
-	for (i = 0; i < cnt; i++) {
-		if (rslt[i])
-			commit_list_insert_by_date(rslt[i], &result);
-	}
+	for (i = 0; i < cnt; i++)
+		commit_list_insert_by_date(rslt[i], &result);
 	free(rslt);
 	return result;
 }
@@ -780,6 +826,9 @@ struct commit_list *get_merge_bases(struct commit *one, struct commit *two,
 	return get_merge_bases_many(one, 1, &two, cleanup);
 }
 
+/*
+ * Is "commit" a decendant of one of the elements on the "with_commit" list?
+ */
 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
 {
 	if (!with_commit)
@@ -789,28 +838,28 @@ int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
 
 		other = with_commit->item;
 		with_commit = with_commit->next;
-		if (in_merge_bases(other, &commit, 1))
+		if (in_merge_bases(other, commit))
 			return 1;
 	}
 	return 0;
 }
 
-int in_merge_bases(struct commit *commit, struct commit **reference, int num)
+/*
+ * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
+ */
+int in_merge_bases(struct commit *commit, struct commit *reference)
 {
-	struct commit_list *bases, *b;
+	struct commit_list *bases;
 	int ret = 0;
 
-	if (num == 1)
-		bases = get_merge_bases(commit, *reference, 1);
-	else
-		die("not yet");
-	for (b = bases; b; b = b->next) {
-		if (!hashcmp(commit->object.sha1, b->item->object.sha1)) {
-			ret = 1;
-			break;
-		}
-	}
+	if (parse_commit(commit) || parse_commit(reference))
+		return ret;
 
+	bases = paint_down_to_common(commit, 1, &reference);
+	if (commit->object.flags & PARENT2)
+		ret = 1;
+	clear_commit_marks(commit, all_flags);
+	clear_commit_marks(reference, all_flags);
 	free_commit_list(bases);
 	return ret;
 }
@@ -819,51 +868,31 @@ struct commit_list *reduce_heads(struct commit_list *heads)
 {
 	struct commit_list *p;
 	struct commit_list *result = NULL, **tail = &result;
-	struct commit **other;
-	size_t num_head, num_other;
+	struct commit **array;
+	int num_head, i;
 
 	if (!heads)
 		return NULL;
 
-	/* Avoid unnecessary reallocations */
-	for (p = heads, num_head = 0; p; p = p->next)
-		num_head++;
-	other = xcalloc(sizeof(*other), num_head);
-
-	/* For each commit, see if it can be reached by others */
-	for (p = heads; p; p = p->next) {
-		struct commit_list *q, *base;
-
-		/* Do we already have this in the result? */
-		for (q = result; q; q = q->next)
-			if (p->item == q->item)
-				break;
-		if (q)
+	/* Uniquify */
+	for (p = heads; p; p = p->next)
+		p->item->object.flags &= ~STALE;
+	for (p = heads, num_head = 0; p; p = p->next) {
+		if (p->item->object.flags & STALE)
 			continue;
-
-		num_other = 0;
-		for (q = heads; q; q = q->next) {
-			if (p->item == q->item)
-				continue;
-			other[num_other++] = q->item;
+		p->item->object.flags |= STALE;
+		num_head++;
+	}
+	array = xcalloc(sizeof(*array), num_head);
+	for (p = heads, i = 0; p; p = p->next) {
+		if (p->item->object.flags & STALE) {
+			array[i++] = p->item;
+			p->item->object.flags &= ~STALE;
 		}
-		if (num_other)
-			base = get_merge_bases_many(p->item, num_other, other, 1);
-		else
-			base = NULL;
-		/*
-		 * If p->item does not have anything common with other
-		 * commits, there won't be any merge base.  If it is
-		 * reachable from some of the others, p->item will be
-		 * the merge base.  If its history is connected with
-		 * others, but p->item is not reachable by others, we
-		 * will get something other than p->item back.
-		 */
-		if (!base || (base->item != p->item))
-			tail = &(commit_list_insert(p->item, tail)->next);
-		free_commit_list(base);
 	}
-	free(other);
+	num_head = remove_redundant(array, num_head);
+	for (i = 0; i < num_head; i++)
+		tail = &commit_list_insert(array[i], tail)->next;
 	return result;
 }
 
diff --git a/commit.h b/commit.h
index d617fa3f28..6edce87673 100644
--- a/commit.h
+++ b/commit.h
@@ -171,7 +171,7 @@ extern struct commit_list *get_shallow_commits(struct object_array *heads,
 		int depth, int shallow_flag, int not_shallow_flag);
 
 int is_descendant_of(struct commit *, struct commit_list *);
-int in_merge_bases(struct commit *, struct commit **, int);
+int in_merge_bases(struct commit *, struct commit *);
 
 extern int interactive_add(int argc, const char **argv, const char *prefix, int patch);
 extern int run_add_interactive(const char *revision, const char *patch_mode,
diff --git a/contrib/examples/builtin-fetch--tool.c b/contrib/examples/builtin-fetch--tool.c
index 0d54aa7061..8bc8c7533a 100644
--- a/contrib/examples/builtin-fetch--tool.c
+++ b/contrib/examples/builtin-fetch--tool.c
@@ -96,7 +96,7 @@ static int update_local_ref(const char *name,
 	strcpy(oldh, find_unique_abbrev(current->object.sha1, DEFAULT_ABBREV));
 	strcpy(newh, find_unique_abbrev(sha1_new, DEFAULT_ABBREV));
 
-	if (in_merge_bases(current, &updated, 1)) {
+	if (in_merge_bases(current, updated)) {
 		fprintf(stderr, "* %s: fast-forward to %s\n",
 			name, note);
 		fprintf(stderr, "  old..new: %s..%s\n", oldh, newh);
diff --git a/fast-import.c b/fast-import.c
index eed97c8fa9..c2a814ec66 100644
--- a/fast-import.c
+++ b/fast-import.c
@@ -1691,7 +1691,7 @@ static int update_branch(struct branch *b)
 			return error("Branch %s is missing commits.", b->name);
 		}
 
-		if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
+		if (!in_merge_bases(old_cmit, new_cmit)) {
 			unlock_ref(lock);
 			warning("Not updating %s"
 				" (new tip %s does not contain %s)",
diff --git a/http-push.c b/http-push.c
index a832ca77a3..8701c1215d 100644
--- a/http-push.c
+++ b/http-push.c
@@ -1610,9 +1610,8 @@ static int verify_merge_base(unsigned char *head_sha1, struct ref *remote)
 {
 	struct commit *head = lookup_commit_or_die(head_sha1, "HEAD");
 	struct commit *branch = lookup_commit_or_die(remote->old_sha1, remote->name);
-	struct commit_list *merge_bases = get_merge_bases(head, branch, 1);
 
-	return (merge_bases && !merge_bases->next && merge_bases->item == branch);
+	return in_merge_bases(branch, head);
 }
 
 static int delete_remote_branch(const char *pattern, int force)
diff --git a/submodule.c b/submodule.c
index 19dc6a6c0d..d133796c9c 100644
--- a/submodule.c
+++ b/submodule.c
@@ -788,7 +788,7 @@ static int find_first_merges(struct object_array *result, const char *path,
 		die("revision walk setup failed");
 	while ((commit = get_revision(&revs)) != NULL) {
 		struct object *o = &(commit->object);
-		if (in_merge_bases(b, &commit, 1))
+		if (in_merge_bases(b, commit))
 			add_object_array(o, NULL, &merges);
 	}
 	reset_revision_walk();
@@ -803,7 +803,7 @@ static int find_first_merges(struct object_array *result, const char *path,
 		contains_another = 0;
 		for (j = 0; j < merges.nr; j++) {
 			struct commit *m2 = (struct commit *) merges.objects[j].item;
-			if (i != j && in_merge_bases(m2, &m1, 1)) {
+			if (i != j && in_merge_bases(m2, m1)) {
 				contains_another = 1;
 				break;
 			}
@@ -865,18 +865,18 @@ int merge_submodule(unsigned char result[20], const char *path,
 	}
 
 	/* check whether both changes are forward */
-	if (!in_merge_bases(commit_base, &commit_a, 1) ||
-	    !in_merge_bases(commit_base, &commit_b, 1)) {
+	if (!in_merge_bases(commit_base, commit_a) ||
+	    !in_merge_bases(commit_base, commit_b)) {
 		MERGE_WARNING(path, "commits don't follow merge-base");
 		return 0;
 	}
 
 	/* Case #1: a is contained in b or vice versa */
-	if (in_merge_bases(commit_a, &commit_b, 1)) {
+	if (in_merge_bases(commit_a, commit_b)) {
 		hashcpy(result, b);
 		return 1;
 	}
-	if (in_merge_bases(commit_b, &commit_a, 1)) {
+	if (in_merge_bases(commit_b, commit_a)) {
 		hashcpy(result, a);
 		return 1;
 	}