You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
First order of business is to remove insertable from the public interface.
add_child should then be reimplemented in terms of fast_insert + InsChild (fast_insert can do more checks, and using the same interface would allow spilling through fast_insert).
Then we get into the question of children vs child_alloc_id semantics. Currently these are made complex due to path dependence. We could simplify these by storing a sum type which is either a loc or an alloc_id, similar to how it was done on raw_node, but with an explicit discriminant.
flush_children will stay separate, because it is (most likely) a much smaller tree.
The text was updated successfully, but these errors were encountered:
g2p
added a commit
to g2p/wodan
that referenced
this issue
Apr 24, 2019
First order of business is to remove insertable from the public interface.
add_child should then be reimplemented in terms of fast_insert + InsChild (fast_insert can do more checks, and using the same interface would allow spilling through fast_insert).
Then we get into the question of children vs child_alloc_id semantics. Currently these are made complex due to path dependence. We could simplify these by storing a sum type which is either a loc or an alloc_id, similar to how it was done on raw_node, but with an explicit discriminant.
flush_children will stay separate, because it is (most likely) a much smaller tree.
The text was updated successfully, but these errors were encountered: