When it comes to mobile apps, performance is critical to user experience. Users expect apps to have smooth scrolling and meaningful animations free of stuttering or skipped frames, known as "jank." How to ensure that your app is free of jank on a wide variety of devices?

There are two options: first, manually test the app on different devices. While that approach might work for a smaller app, it becomes more cumbersome as an app grows in size. Alternatively, run an integration test that performs a specific task and records a performance timeline. Then, examine the results to determine whether a specific section of the app needs to be improved.

In this recipe, learn how to write a test that records a performance timeline while performing a specific task and saves a summary of the results to a local file.

This recipe uses the following steps:

1. Write a test that scrolls through a list of items.
2. Record the performance of the app.
3. Save the results to disk.
4. Run the test.
5. Review the results.

In this recipe, record the performance of an app as it scrolls through a list of items. To focus on performance profiling, this recipe builds on theScrolling recipe in widget tests.

Follow the instructions in that recipe to create an app and write a test to verify that everything works as expected.

Next, record the performance of the app as it scrolls through the list. Perform this task using thetraceAction() method provided by theIntegrationTestWidgetsFlutterBinding class.

This method runs the provided function and records aTimeline with detailed information about the performance of the app. This example provides a function that scrolls through the list of items, ensuring that a specific item is displayed. When the function completes, thetraceAction() creates a report dataMap that contains theTimeline.

Specify thereportKey when running more than onetraceAction. By default allTimelines are stored with the keytimeline, in this example thereportKey is changed toscrolling_timeline.

awaitbinding.traceAction(()async{// Scroll until the item to be found appears.awaittester.scrollUntilVisible(itemFinder,500.0,scrollable:listFinder,);},reportKey:'scrolling_timeline');

Now that you've captured a performance timeline, you need a way to review it. TheTimeline object provides detailed information about all of the events that took place, but it doesn't provide a convenient way to review the results.

Therefore, convert theTimeline into aTimelineSummary. TheTimelineSummary can perform two tasks that make it easier to review the results:

1. Writing a json document on disk that summarizes the data contained within theTimeline. This summary includes information about the number of skipped frames, slowest build times, and more.
2. Saving the completeTimeline as a json file on disk. This file can be opened with the Chrome browser's tracing tools found atchrome://tracing.

To capture the results, create a file namedperf_driver.dart in thetest_driver folder and add the following code:

import'package:flutter_driver/flutter_driver.dart'asdriver;import'package:integration_test/integration_test_driver.dart';​Future<void>main(){returnintegrationDriver(responseDataCallback:(data)async{if(data!=null){finaltimeline=driver.Timeline.fromJson(data['scrolling_timeline']asMap<String,dynamic>,);​// Convert the Timeline into a TimelineSummary that's easier to// read and understand.finalsummary=driver.TimelineSummary.summarize(timeline);​// Then, write the entire timeline to disk in a json format.// This file can be opened in the Chrome browser's tracing tools// found by navigating to chrome://tracing.// Optionally, save the summary to disk by setting includeSummary// to trueawaitsummary.writeTimelineToFile('scrolling_timeline',pretty:true,includeSummary:true,);}},);}

TheintegrationDriver function has aresponseDataCallback which you can customize. By default, it writes the results to theintegration_response_data.json file, but you can customize it to generate a summary like in this example.

After configuring the test to capture a performanceTimeline and save a summary of the results to disk, run the test with the following command:

flutter drive \  --driver=test_driver/perf_driver.dart \  --target=integration_test/scrolling_test.dart \  --profile

The--profile option means to compile the app for the "profile mode" rather than the "debug mode", so that the benchmark result is closer to what will be experienced by end users.

After the test completes successfully, thebuild directory at the root of the project contains two files:

1. scrolling_summary.timeline_summary.json contains the summary. Open the file with any text editor to review the information contained within. With a more advanced setup, you could save a summary every time the test runs and create a graph of the results.
2. scrolling_timeline.timeline.json contains the complete timeline data. Open the file using the Chrome browser's tracing tools found atchrome://tracing. The tracing tools provide a convenient interface for inspecting the timeline data to discover the source of a performance issue.

{"average_frame_build_time_millis":4.2592592592592595,"worst_frame_build_time_millis":21.0,"missed_frame_build_budget_count":2,"average_frame_rasterizer_time_millis":5.518518518518518,"worst_frame_rasterizer_time_millis":51.0,"missed_frame_rasterizer_budget_count":10,"frame_count":54,"frame_build_times":[6874,5019,3638],"frame_rasterizer_times":[51955,8468,3129]}

integration_test/scrolling_test.dart

import'package:flutter/material.dart';import'package:flutter_test/flutter_test.dart';import'package:integration_test/integration_test.dart';import'package:your_package/main.dart';​voidmain(){finalbinding=IntegrationTestWidgetsFlutterBinding.ensureInitialized();​testWidgets('Counter increments smoke test',(tester)async{// Build our app and trigger a frame.awaittester.pumpWidget(MyApp(items:List<String>.generate(10000,(i)=>'Item$i')),);​finallistFinder=find.byType(Scrollable);finalitemFinder=find.byKey(constValueKey('item_50_text'));​awaitbinding.traceAction(()async{// Scroll until the item to be found appears.awaittester.scrollUntilVisible(itemFinder,500.0,scrollable:listFinder,);},reportKey:'scrolling_timeline');});}

test_driver/perf_driver.dart

import'package:flutter_driver/flutter_driver.dart'asdriver;import'package:integration_test/integration_test_driver.dart';​Future<void>main(){returnintegrationDriver(responseDataCallback:(data)async{if(data!=null){finaltimeline=driver.Timeline.fromJson(data['scrolling_timeline']asMap<String,dynamic>,);​// Convert the Timeline into a TimelineSummary that's easier to// read and understand.finalsummary=driver.TimelineSummary.summarize(timeline);​// Then, write the entire timeline to disk in a json format.// This file can be opened in the Chrome browser's tracing tools// found by navigating to chrome://tracing.// Optionally, save the summary to disk by setting includeSummary// to trueawaitsummary.writeTimelineToFile('scrolling_timeline',pretty:true,includeSummary:true,);}},);}