APPLIES TO: SDK v4
Bots are complex apps, with a lot of different parts working together. Like any other complex app, this can lead to some interesting bugs or cause your bot to behave differently than expected.
Debugging, your bot can sometimes be a difficult task. Every developer has their own preferred way to accomplish that task; guidelines we present below are suggestions for you to use that apply to a large majority of bots.
After verifying your bot appears to work how you'd like it to, the next step is connecting it to a channel. To do this, you can deploy your bot to a staging server and create your own direct line client for your bot to connect to.
Creating your own client allows you to define the inner workings of the channel, as well as specifically test how your bot responds to certain activity exchanges. Once connected to your client, run your tests to set up your bot state and verify your features. If your bot utilizes a feature like speech, using these channels can offer a way to verify that functionality.
Use of both the Emulator and Web Chat via Azure portal here can provide further insight into how your bot performs while interacting with different channels.
Debugging your bot works similarly to other multi-threaded apps, with the ability to set breakpoints or use features like the immediate window.
Bots follow an event driven programming paradigm, which can be hard to rationalize if you're not familiar with it. The idea of your bot being stateless, multi-threaded, and dealing with async/await calls can result in unexpected bugs. While debugging your bot works similarly to other multi-threaded apps, we'll cover some suggestions, tools, and resources to help.
Understanding bot activities with the Emulator
Your bot deals with different types of activities besides the normal message activity. Using the Emulator will show you what those activities are, when they happen, and what information they contain. Understanding those activities will help you code your bot efficiently and allows you to verify the activities your bot is sending and receiving are what you expect.
Saving and retrieving user interactions with transcripts
Azure blob transcript storage provides a specialized resource where you can both store and retrieve transcripts containing interactions between your users and your bot.
Additionally, once user input interactions have been stored, you can use Azure's "storage explorer" to manually view data contained in transcripts stored within your blob transcript store. The following example opens "storage explorer" from settings for "mynewtestblobstorage." To open a saved user input select: Blob Container > ChannelId > TranscriptId > ConversationId
This opens the stored user conversation input in JSON format. User input is preserved together with the key "text:."
How middleware works
Middleware may not be intuitive when first attempting to use it, particularly regarding the continuation, or short-circuiting, of execution. Middleware can execute on the leading or trailing edge of a turn, with a call to the
next() delegate dictating when execution is passed to the bot logic.
If you are using multiple pieces of middleware the delegate may pass execution to a different piece of middleware if that is how your pipeline is oriented. Details on the bot middleware pipeline can help make that idea clearer.
next() delegate is not called, that's referred to as short circuit routing. This happens when the middleware satisfies the current activity and determines it's not necessary to pass execution on.
Understanding when, and why, middleware short-circuits helps indicate which piece of middleware should come first in your pipeline. Additionally, understanding what to expect is particularly important for built-in middleware provided by the SDK or other developers. Some find it helpful to try creating your own middleware first to experiment a bit before diving into the built-in middleware.
Keeping track of state is an important part of your bot, particularly for complex tasks. In general, best practice is to process activities as quickly as possible and let the processing complete so that state gets persisted. Activities can be sent to your bot at nearly the same time, and that can introduce very confusing bugs because of the asynchronous architecture.
Most importantly, make sure that state is persisting in a way that matches your expectations. Depending on where your persisted state lives, storage emulators for Cosmos DB and Azure Table storage can help you verify that state before using production storage.
The Cosmos DB storage class has been deprecated. Containers originally created with CosmosDbStorage had no partition key set, and were given the default partition key of _/partitionKey.
Containers created with Cosmos DB storage can be used with Cosmos DB partitioned storage. Read Partitioning in Azure Cosmos DB for more information.
Also note that, unlike the legacy Cosmos DB storage, the Cosmos DB partitioned storage does not automatically create a database within your Cosmos DB account. You need to create a new database manually, but skip manually creating a container since CosmosDbPartitionedStorage will create the container for you.
How to use activity handlers
Activity handlers can introduce another layer of complexity, particularly since each activity runs on an independent thread (or web workers, depending on your language). Depending on what your handlers are doing, this can cause issues where the current state is not what you expect.
Built-in state gets written at the end of a turn, however any activities generated by that turn are executing independently of the turn pipeline. Often this doesn't impact us, but if an activity handler changes state we need the state written to contain that change. In that case, the turn pipeline can wait on the activity to finish processing before completing to make sure it records the correct state for that turn.
The send activity method, and its handlers, pose a unique problem. Simply calling send activity from within the on send activities handler causes an infinite forking of threads. There are ways you can work around that problem, such as by appending additional messages to the outgoing information or writing out to another location like the console or a file to avoid crashing your bot.