Visitor pattern
Study case: employee structure
- start with a "flat" data structure
- transform it into an in-memory object tree
- add the ability to print the structure into HTML
- introduce visitor pattern
- define traversing order
- refactor "printing" using the visitor pattern
Data model
- Manager
- name, profile picture
- role
- list of subordinaries
- Team lead
- name, profile picture
- list of team members
- Engineer
- name, profile picture
- technology
Bueno
- add external behavior without affecting the underlying data structure
- remove auxiliary behavior from the data structures
No bueno
- when a new "type" of data structure is added (or removed), all visitors have to be updated
- [if needed] visitor has to internally track the state of traversal
- exposing richer public interface
Object tree
- 12 different data types
- primitive
- text, numeric, date, select
- composite
- root, struct, array, array item
- dynamic
- computed
- reference
Visitors
- populate data structure
- from the incoming client request
- from the database
- finding nodes in a tree
- export(JSON)
- computing dynamic fields
- resolving references
- persisting the tree (3 variations)