Description Ed Willink 2011-06-21 12:37:13 EDT

The current CST/AST synchronization in the editors relies on the ability to compute identical monikers from non-identical tree structures. This is hard and flaky.

Some from of synchronisation is needed since edits replace substantial CS sections.

Although the CS objects change, the CS URIs do not so a map of CS URI to AST element can be maintained and used to drive AST refreshes. This could render monikers almost obsolete eliminating their non-trivial creation algorithm and the need to define an extensibility mechanism for it.

Just need careful design of the CST model to provide good ID attributes so that the CS URIs are relatively stable for reorderings.