add documentation for differences between kubo and iroh

iroh-resolver/src/balanced_tree.rs

@@ -138,7 +138,16 @@ fn create_unixfs_node_from_links(links: Vec<(Cid, u64)>) -> Result<UnixfsNode> {
         .into_iter()
         .map(|(cid, len)| dag_pb::PbLink {
             hash: Some(cid.to_bytes()),
+            /// ALL "stem" nodes have `name: None`.
+            /// In kubo, nodes that have links to `leaf` nodes have `name: Some("".to_string())`
             name: None,
+            /// tsize has no strict definition
+            /// Iroh's definiton of `tsize` is "the cumulative size of the encoded tree
+            /// pointed to by this link", so not just the size of the raw content, but including
+            /// all encoded dag nodes as well.
+            /// In the `go-merkledag` package, the `merkledag.proto` file, states that tsize
+            /// is the "cumulative size of the target object"
+            /// (https://github.com/ipfs/go-merkledag/blob/8335efd4765ed5a512baa7e522c3552d067cf966/pb/merkledag.proto#L29)
             tsize: Some(len as u64),
         })
         .collect();
@@ -174,7 +183,6 @@ impl TreeNode {
                 Ok((cid, bytes, len as u64))
             }
             TreeNode::Stem(links) => {
-                // keep track of `tsize`, aka the size of the encoded tree at the given link
                 let mut cumulative_len: u64 = links.iter().map(|(_, len)| len).sum();
                 let node = create_unixfs_node_from_links(links)?;
                 let (cid, bytes) = node.encode()?;