or

The reason for the async API here is because anonymizeAppCenterUser internally relies on an async API. The synchronous API's for starting AppCenter are non-blocking methods that do a fire-and-forget call to StartAsync(string,bool).

Anonymous User Information

The component AppCenterSetup exposes a method called UseAnonymousUserIdAsync() which sets the UserId in AppCenter to the first 8 characters a GUID that is unique per app installation. This can be used as a support key for uniquely identifying application users for instrumentation and troubleshooting. The support key can be attached to all HTTP calls by using the DiagnosticDelegatingHandler

Error Reporting

The library exposes extension methods to the Exception class for conveniently reporting Exceptions to AppCenter

Example:

ITrackingCommand

This library provides 3 convenience implementations of ICommand that will report the action to AppCenter Analytics after successfully invoking the execute callback method

• TrackingCommand - This implementation accepts an Action as the Execute callback and a Func<bool> as the CanExecute callback
• TrackingCommand - This implementation accepts an Action<T> as the Execute callback and a Func<T, bool> as the CanExecute callback
• AsyncTrackingCommand - This implementation accepts a Func<Task> as the execute callback and a Func<bool> as the CanExecute callback. This also exposes a CompletionTask property that the consumer can await if desired. The Execute(object parameter) method here is a non-blocking call

Example:

Specifying the screenName argument in the constructor is optional and when this is not provided manually then it will use the declaring Type name from the method that instantiated the ITrackingCommand instance and convert it to a more analytics friendly event name using the ToTrackingEventName() extension method. In the example above, if the nameof(AboutViewModel).ToTrackingEventName() parameter is not provided then the owner declaring Type is AboutViewModel and the ScreenName will be set to "About"

Automatic Page Tracking

Automatic page tracking is enabled by replacing the base class of the ContentPage to classes to use TrackingContentPage class. By doing so the library will send page tracking information to AppCenter after leaving every page. Currently, the library will send the page Type, Title, and the duration spent on the screen. The library is rather opinionated on how to log information, and this will only change if I get a request to do so. Duration spent on screen is calculated using a Stopwatch that is started upon Page OnAppearing and is reported to Analytics upon OnDisappearing. The event name is based on the Type name of the Page and is split into multiple words based on pascal case rules and afterwards removes words like Page, View, Model, Async. For example: UserSettingsPage or UserSettingsView becomes User Settings

XAML Example:

Custom Trace Listener

This library includes a trace listener implementation that reports to AppCenter. The reason for this is to cater to those who have implemented error handling or reporting using Trace Listeners, these types of users can just swap out (or add on) the AppCenterTraceListener

This implementation implements the following methods:

• Write(object obj)
• Write(object obj, string category)
• WriteLine(object obj)
• WriteLine(object obj, string category)

If the object provided is an Exception then this is reported to AppCenter Crash Reporting. If the object provided is an instance of AnalyticsEvent then this is sent to AppCenter Analytics

The AnalyticsEvent exposes 2 properties:

• string EventName { get; } - self explanatory
• IDictionary<string,string> Properties { get; } - Additional properties to attach to the Analytics event

To set it up you simply add an instance of AppCenterTraceListener to your existing Trace listeners:

Here's an example of how to use System.Diagnostics.Trace to report errors

and here's an example of to use System.Diagnostics.Trace to send analytics data

Task Extensions

This library includes a few Task extension methods with AppCenter error reporting in mind. Possible exceptions that occur in the async operation are swallowed and reported to AppCenter. These extension methods will internally wrap the Task in a try/catch and await the Task using ConfigureAwait(false).

Here are usage some examples

• Fire and Forget on a Task (Note: Forget() returns void)

• Awaitable Task (also available for Task<T>)

HTTP Error Logging

The library provides a HttpMessageHandler implementation that logs non-successfuly HTTP results to AppCenter Analytics. This component will also attach HTTP headers describing the AppCenter SDK Version, Install ID, and a support key to all HTTP requests. The logged failed responses will contain the Endpoint URL (including the HTTP verb), Response status code, how the duration of the HTTP call. This will be logged under the event name HTTP Error

You will in most (if not all) cases would want to keep a singleton instance of the HttpClient. The DiagnosticDelegatingHandler is designed with unit testing in mind and accepts an IAnalytics and IAppCenterSetup interface, it also accepts an inner HttpMessageHandler if you wish to chain multiple delegating handlers.

Example:

In the example above we made an HTTP GET call to the RequestBin endpoint https://entbpr4b9bdpo.x.pipedream.net. This will result in the following we inspected in RequestBin

AppInsights Extensions

The library provides an extension method to IServiceCollection called AddAppCenterTelemetry() that should be called from the ConfigureServices(IServiceCollection services) in your ASP.NET Core Startup class. This enables logging the AppCenter diagnostic information described in the previous section to Application Insights

Here's an example taken from the Startup class in the sample web project

Once this is setup, AppCenter diagnostic information should now be searchable and visible in Application Insights. Here's a screenshot of search results for the x-supportkey header

and here's a screenshot of the details of a single request containing AppCenter diagnostic information logged in Application Insights

With this flow you can now correlate Crash Reports and Analytics data from AppCenter with the HTTP requests for your backend systems in Application Insights. In the systems that I have been involved with building we include the AppCenter diagnostic information from our API Gateway to all calls to our internal Microservices

For tips and tricks on software development, check out my blog

If you find this useful and feel a bit generous then feel free to buy me a coffee :)