-
Notifications
You must be signed in to change notification settings - Fork 0
component states
Unlike React, states in Respo is maintained manually for stablility during hot code swapping. At first, states is a HashMap inside the store:
defatom *store {}
:states $ {}
By design, if states is added, you would a tree:
{}
:states $ {}
:data $ {}
:todolist $ {}
:data $ {}
:input "|xyz..."
"task-1-id" $ {}
:data $ {}
:draft "|xxx..."
"task-2-id" $ {}
:data $ {}
:draft "|yyy..."
"task-2-id" $ {}
:data $ {}
:draft "|zzz.."
:data
is a special field for holding state of each component.
It has to be a tree since Virtual DOM is a tree.
You also notice that its structure is simpler than a DOM tree, it only contains states.
respo.core/>>
is the "picking branch" function. It also maintains a :cursor
field.
When you call (>> states :todolist)
, you get new states
variable for a child component:
{}
; "generated cursor, nil at top level"
:cursor $ [] :todolist
; "state at current level"
:data $ {}
:input "|xyz..."
; states for children
"task-1-id" $ {}
:data $ {}
:draft "|xxx..."
"task-2-id" $ {}
:data $ {}
:draft "|yyy..."
"task-2-id" $ {}
:data $ {}
:draft "|zzz.."
Then you call (>> states "task-1-id")
and you get new states
for child "task-1":
{}
; "generated cursor"
:cursor $ [] :todolist "|task-1-id"
; "state of task-1"
:data $ {}
:draft "|xxx..."
For state inside each component, it's nil
at first.
You want to have an initial state, use or
to provide one.
defcomp comp-task (states)
let
cursor (:cursor states)
state $ or (:data states) $ {}
:draft "|empty"
(div {}))
By accessing (:data states)
, you get nil
, so &{} :draft "|empty"
is used.
After there's data in states, you get data that was set.
Then you want to update component s:tate
defcomp comp-task (states)
let
cursor (:cursor states)
state $ or (:data states) $ {}
:draft "|empty"
div $ {}
:on-click $ fn (e dispatch!)
dispatch! cursor (assoc state :draft "|New state")
So (dispatch! cursor state)
sends the new state.
The last step to to update global states with respo.cursor/update-states
.
Internally (dispatch! cursor op-data)
will be transformed to (dispatch! :states ([] cursor op-data))
.
And then in updater
you add:
case-default op
; other actions
do store
; "where op-data is [cursor new-state]"
:states (update-states store op-data)
Let's wrap it. First we have empty states inside store:
{}
:states $ {}
And it is passed to (comp-todolist (>> states :todolist) data)
,
and then passed to (comp-task (>> states (:id task)) task)
.
In comp-todolist
, (:data states)
provides component state, (:cursor states)
provides its cursor.
Call (dispatch! cursor {:input "|New draft"})
and global store will become:
{}
:states $ {}
:todolist $ {}
:data $ {}
:input "|New draft"
In comp-task
of "task-1", you also get state
and cursor
, so call (dispatch! cursor {:draft "New text"})
you will get:
{}
:states $ {}
:todolist $ {}
:data $ {}
:input "|New draft"
"task-1-id" $ {}
:data $ {}
:draft "|New text"
And that's how Respo states is maintained.