As an opportunity to get my latest project so damn fast, I looked into key-based cache expiration, also known as the Russion Doll approach to caching.
There is something I don’t understand.
“You deal with dependency structures by tying the model objects together on updates.”
So if you change a todo that belongs to a todolist that belongs to a project, you update the updated_at timestamp on every part of the chain, which will automatically then update the cache keys based on these objects
Okay, so the hierarchy is project → todolist → todo. Then a single todo item, probably keeps track of its creator, right?
In Rails it could be declared like this:
class Project < ActiveRecord::Base
end
class Todolist < ActiveRecord::Base
belongs_to :project, touch: true
end
class Todo < ActiveRecord::Base
belongs_to :todolist, touch: true
belongs_to :creator
end
in app/views/project/show.html.erb
<% cache project do %>
All my todo lists:
<%= render project.todolists %>
<% end %>
in app/views/todolists/_todolist.html.erb
<% cache todolist do %>
<%= todolist.name %>:
<%= render todolist.todos %>
<% end %>
in app/views/todo/_todo.html.erb
<% cache todo do %>
<%= todo.name %>(by <%= todo.creator.name %>)
<% end %>
This will not trigger todo.touch!,
which in turn does not trigger todo.todolist.touch!
and does not trigger todo.todolist.project.touch!
todo.creator.update_attributes(name: ‘John Doe’)
What do I need to change in order to make this process trivial to implement caching schemes and trust that I am never going to serve stale data?
