如何让处理程序在重新帧中按顺序运行？

Question

从这两个处理程序开始，它们负责获取当前用户信息：

(re-frame/register-handler
  :got-user
  (fn [db [_ user]]
    (assoc db :user user)))

(re-frame/register-handler
  :get-user
  (fn [db [_]]
    (ajax/GET "/user" 
       {:handler #(re-frame/dispatch [:got-user %1])})
    db))

在一个页面中，我想显示朋友列表，但问题是它取决于首先获取的用户：

(re-frame/register-handler
  :get-friends
  (fn [db [_]]
    (when (nil? (:user db))
      (re-frame/dispatch [:get-user]))
    ; Here's the problem, as I need to way for get-users and got-users to run.
    (ajax/GET (str "/users/" (get-in db [:user :id])) 
       {:handler #(re-frame/dispatch [:got-friends %1])})
    db))

我应该如何构造这段代码？

Answer 1

First, the documentation specifically warns against creating handlers with side-effects (即从事件处理程序分派事件)。其次，不要尝试重用事件处理程序，而是将可重用的部分重构为helper函数。使用re-frame-http-fx，我会写下这样的代码：

(reg-event-fx
  :get-friends
  [trim-v]
  (fn
    [{db :db} [_]]
    {:http-xhrio  (if (:user db)
                  {:method          :get
                  :uri             (str "/users/" (get-in db [:user :id])) 
                  :on-success      [:got-friends]
                  :on-failure      [:get-friends-load-failed]}

                  {:method          :get
                  :uri             "/user" 
                  :on-success      [:get-friends-for-user]
                  :on-failure      [:get-friends-user-load-failed]})
     :db         db}
    ))

:get- friends -for-user将首先存储检索到的用户，并开始获取好友。

Answer 2

从我让get-friends调用自身的article about solving the CPU hog problem中获得了一些灵感，虽然用户是nil，但为了避免向服务器发送太多请求，它还添加了一个关于被请求用户的标志：

(re-frame/register-handler
  :get-friends
  (fn [db [_ user-requested]]
    (if-let [user (:user db)]
      (ajax/GET (str "/users/" (get-in db [:user :id]))
                {:handler #(re-frame/dispatch [:got-friends %1])})
      (do
        (when (not user-requested)
          (re-frame/dispatch [:get-user]))
        (re-frame/dispatch [:get-friends true])))
    db))

它可以工作，但在本地机器上，它会设法调用自己9次左右，直到got user设置了用户。我不确定这是否会是一个问题，如果存在其他解决方案，我也持开放态度。