-
Notifications
You must be signed in to change notification settings - Fork 5
Algorithms
Whenever a client makes any of the API calls described on the Clients page, the LDE initiates the four consecutive processing steps shown in the bottom row of the figure on the Design Overview page. Naturally, the LDE begins by modifying the IT in obedience to the specific API method called. It also calls markDirty()
in the modified node(s), which cascades upward, meaning that all ancestors are also marked dirty for interpretation.
Some time shortly thereafter, the LDE runs the Modification and Interpretation phases, the latter of which processes the entire IT starting at the root and recurring downward, in
The first run of the Interpretation phase creates the OT; each subsequent run replaces the previous OT with a new one. (Although the entire tree is replaced with a new one, it will often be that the new tree was created from large re-used pieces of the old tree, and only a few changes.) It might seem natural to call this process "reinterpretation," but we will not make things confusing by introducing two terms, so will just use the term "interpretation" consistently.
Upon completion of the Interpretation phase, the Scoping phase begins, and traverses the OT in
Each of these four processing steps--Modification, Interpretation, Scoping, and Validation--are covered in detail below, each in its own section. Because each is lengthy, we do not include them all on this page, but rather break them into their own pages:
Not seeing typeset math? GitHub doesn't render MathJax; try a browser plugin.
Main sections:
More to come!