Re: [PATCH 2/5] support fetching into a shallow repository

Hi,

On Mon, 13 Nov 2006, Junio C Hamano wrote:

Johannes Schindelin [off-list ref] writes:

quoted

diff --git a/shallow.c b/shallow.c
new file mode 100644
index 0000000..3cf2127
--- /dev/null
+++ b/shallow.c

...
+struct commit_list *get_shallow_commits(struct object_array *heads, int depth)
+{
+	int i = 0, cur_depth = 0;
+	struct commit_list *result = NULL;
+	struct object_array stack = {0, 0, NULL};
+	struct commit *commit = NULL;
+
+	while (commit || i < heads->nr || stack.nr) {
+		struct commit_list *p;
+		if (!commit) {
...
+		}
+		parse_commit(commit);
+		cur_depth++;
+		for (p = commit->parents, commit = NULL; p; p = p->next) {
+			if (!p->item->util) {
...
+			} else {
...
+			}
+			if (cur_depth < depth) {
+				if (p->next)
+					add_object_array(&p->item->object,
+							NULL, &stack);
+				else {
+					commit = p->item;
+					cur_depth = *(int *)commit->util;
+				}
+			} else
+				commit_list_insert(p->item, &result);
+		}
+	}
+
+	return result;
+}

I think the "commit = p->item" part is trying to do a tail
recursion optimization, but this is a bit too clever to my
liking (at first I mistook that the code forgot to re-point p at
its parents list and incrementing cur_depth).

I take it as a compliment ;-)

Seriously again, would you like me to add a comment, or rather do away 
with the tail recursion optimization? It is not a huge optimization 
anyway. Maybe a cleverer way would be to use an object_array instead of a 
commit_list?

Ciao,
Dscho