mirror of
https://github.com/rails/rails.git
synced 2022-11-09 12:12:34 -05:00
0547b1646d
* Add Server Timing middleware What is server timing? It's a specification that defines how the server can communicate the user-agent performance metrics about the request it is responding to. Here's the official specification: https://www.w3.org/TR/server-timing/#dfn-server-timing-header-field How does it work? This introduces a new `ServerTiming` middleware only on `development` by default, this is done using the `config.server_timing` setting. It works by subscribing to all `ActiveSupport::Notifications` and adding their duration to the `Server-Timing` header. Why is this useful? It makes looking at performance metrics in development much easier, especially when using Chrome dev tools which includes the metrics in the Network Inspector. Here's an example: ![](https://d3vv6lp55qjaqc.cloudfront.net/items/371h2y3B3a0U470j040u/Image%202019-05-15%20at%205.40.37%20PM.png?) Paired on this with @guilleiguaran
33 lines
809 B
Ruby
33 lines
809 B
Ruby
# frozen_string_literal: true
|
|
|
|
require "active_support/notifications"
|
|
|
|
module ActionDispatch
|
|
class ServerTiming
|
|
SERVER_TIMING_HEADER = "Server-Timing"
|
|
|
|
def initialize(app)
|
|
@app = app
|
|
end
|
|
|
|
def call(env)
|
|
events = []
|
|
subscriber = ActiveSupport::Notifications.subscribe(/.*/) do |*args|
|
|
events << ActiveSupport::Notifications::Event.new(*args)
|
|
end
|
|
|
|
status, headers, body = begin
|
|
@app.call(env)
|
|
ensure
|
|
ActiveSupport::Notifications.unsubscribe(subscriber)
|
|
end
|
|
|
|
header_info = events.group_by(&:name).map do |event_name, events_collection|
|
|
"#{event_name};dur=#{events_collection.sum(&:duration)}"
|
|
end
|
|
headers[SERVER_TIMING_HEADER] = header_info.join(", ")
|
|
|
|
[ status, headers, body ]
|
|
end
|
|
end
|
|
end
|