Hi,
Why is “item” field of Item_tree of type List? I feel that it is causing a litle bit of overlap betwen Item_tree and item_table. Item_table has a rows property with type List, which is nice for representing multi column tables, but when it comes to a single column table, item_tree also seem to allow it, by using a List given that items property is List
Would not it be better to enforce semantics of a tree structure by making item property of Item_tree a single Cluster ? Then the meaning of the elements of the cluster would be enforced by the container class, item_tree, where each member of cluster (whether cluster or element) would mean a new row in a tree. This would also force a tree with a single parent node, but it would still allow expression of a tree.
Not a big deal, but I think it would be better to force the semantics of a type to make its use as clear as possible. Care to show me if I’m missing something?
Interesting thought! I think this comes down to a design choice, which
is whether we want to force single-rooted tree. By definition, tree
data structure has a single root. But when we model hierarchical
structure of clinical content, it's convenient to allow multiple nodes
at the top level of the tree without the need to create a root node
just to satisfy the notion of single-rooted tree. So it's a choice
between purity (according to computer science definition of a tree)
and convenience (which somewhat indirectly translated to usability of
the model and relevant tools).
. But when we model hierarchical
structure of clinical content, it’s convenient to allow multiple nodes
at the top level of the tree without the need to create a root node
just to satisfy the notion of single-rooted tree.
This point justifies the choice. Rule number one: do what the doctor says I asked to question from a purely technical point of view, knowing the answer should have a clinical argument somehow.
