EchoBot | A tiny bot to test Tox audio and video | Software As A Service library

client.send_message is an old, outdated method, it has been replaced with Messageable.send

The EchoBot is a basic bot sample. You will need to add more code while following the article for restarting conversations.

The sample code in this article begins with the structure of a multi-turn bot.

If you want to follow along this article, I suggest using this sample.

https://github.com/microsoft/BotBuilder-Samples/blob/main/samples/csharp_dotnetcore/05.multi-turn-prompt/Bots/DialogBot.cs

This is a simple typo. You've written "context-type" instead of "content-type." The YouTube tutorial you linked to tells you to write a line like this:

You're declaring captchaResponse inside onMessage's function scope, meaning it's not available on the outside.

Declare it before the class, then access the outer captchaResponse with the global keyword to override it from inside the function.

Alright, I got it to work.

Basically I duplicated all the code for the conversationReferences, because those get stored in the Bot and are used in NotifyController.

Some code, in case somebody else is trying to do the same:

Create another Singleton.

When Entity Framework throws random, unintelligible exceptions from deep within its internal classes, it's usually a threading issue. A DbContext is not thread-safe. You can't add entities from multiple threads.

From comments:

Looks like the I forgot to add an await calling the AddPatientAsync method. Maybe that's why it was multithreaded

If you don't await the call to your AddPatientAsync() and call it in a loop, multiple threads are accessing the same _context. You can't do that; await the call.

All the code that generates the bearer tokens for user-to-bot messages in the Azure Bot Service is proprietary, so there's not really a publicly-facing explanation for how to generate such a token. The token that you're currently generating is for bot-to-user messages, which I guess is what your question is about anyway. So all your bot has to do to send messages is generate a token just like your channel is doing.

You actually don't need to generate a token manually since the Bot Framework SDK will do that for you. You can create a connector client like I explained in the answer to your other question, and you just pass in your app ID and password as credentials: Changing the service url for a direct line activity that is sent to our bot

Of course, all this authentication is optional anyway if you're building your own API. If you don't care about authenticating requests sent to your bot then you don't need to do that, and if you don't care about authenticating requests sent to your channel then you don't need to do that either. Alternatively, you can come up with your own means of authentication rather than using the builtin AAD app registration credentials.

In the Bot Framework, a turn is the time between the bot receiving an activity in an HTTP request and the bot responding to that request. Note that responding to an HTTP request is different from replying to an activity. The bot can reply by sending new activities to ABS in their own HTTP requests, and it can do this many times in a turn. The HTTP response is not another activity, it's just a status code (like 200 OK) that signifies the end of the turn.

There's not really a concept of a "dialog turn." There are "steps" in waterfall dialogs and adaptive dialogs, though they don't correlate to turns since a step can span multiple turns and a turn can span multiple steps. There is an "End Dialog Turn" action in adaptive dialogs which I think is what you're talking about, but it just ends the turn. The word "Dialog" may be superfluous/misleading there.

There also isn't really a concept of "pausing" a conversation. A conversation is understood to be a series of turns and activities exchanged between a bot and one or more users. Your bot always needs to know what to do about every request that reaches its endpoint, so it's up to you to define what pausing a conversation means.

I'm guessing you want the bot to respond differently or not respond at all while the conversation is paused. You will need some sort of bot state for the bot to know that it's paused for a given user or conversation, and dialogs use bot state so a dialog would do. Whatever you do to indicate to the bot that the conversation is paused, you can just undo it to unpause it. Just ending the turn won't work because that doesn't add anything to state and the next turn will start as soon as the user sends a message.

As shown in the docs for connecting Rasa to Microsoft Bot Framework you need to include the following in your credentials.yml file: