Sonoma Partners Microsoft CRM and Salesforce Blog

Compare Fields Across D365 Orgs to help Debug Solution Import Failures

Today's blog post was written by Matt Dearing, Principal Developer at Sonoma Partners.

When working with a solution it is common to make changes to customizations. Scenarios like an incorrect attribute data type that needs to be recreated are very common, especially early in development. Although it is easy to recreate attributes in a development environment, it can be challenging to import that solution into a target with a previous version of the solution. The import process may fail in the target organization with a generic error message. When this occurs, it is generally related to attributes being recreated. The following is a simple LINQPad script you can use to compare attributes in two different environments. It will print out any differences in casing of schema name or data type which are two of the more common reasons recreated attributes will cause a solution import to fail.

The script itself connects to two different CRM environments: a source where customization changes are made and a target for deploying the updated solution. The script scans the target to get all of the custom entities with a given prefix where the prefix matches that of the publisher for the solution. If the entity is new in source, there is no reason to compare it to target as it will not fail the import.

Next the script loops through the custom entities for the target environment capturing each entities attributes. Schema Name and Data Type are the two most important metadata properties to compare, so those are returned. If an attribute is deleted and recreated and the casing of the Schema Name does not match exactly, the solution import will fail. This is because CRM does a case sensitive comparison of attributes on solution import to determine if any attributes are new and should be created. It will see two attributes with differently cased schema names as two unique attributes, the failure will occur in the SQL database as two columns cannot have the same name. Unfortunately this won't show up in the solution import UI as a helpful error. The second more common failure is that the data type has changed. Maybe the attribute was an integer/whole number and should have been of type money, or it was a picklist and should have been a string. In that case the solution import will also fail when Schema Names match but data type differs.

Next the source environment is queried for its attributes to do the comparison. If the entity no longer exists in the source environment, an exception will be thrown and caught and printed to the screen. This would mean the entire entity no longer exists in the source environment, so there is no comparison that can be made.

Finally the comparisons are done by Schema Name and Type and any discrepancies are printed out to the screen. If the source attribute no longer exists, there is no need to compare.

Knowing the discrepancies in attributes between environments can help tremendously when attempting to figure out why a solution import is failing. Here is the full source code:

Let us know if you need any help or have any questions by commenting below.

Topics: Microsoft Dynamics 365

QnA Maker: An Easy Way to Explore Chat Bots

Today's blog post was written by Bryson Engelen and Kevin Yamashita, Sales Engineers at Sonoma Partners.

Today we'd like to show you a nifty tool from Microsoft that we've been using to add some pizzazz to the end of our service demonstrations. There are already a number of nifty tie-ins between Dynamics 365 and Azure, including product recommendations, documentation recommendations, connected field service, and sentiment analysis (via Flow). We find that it's compelling to show the breadth of the Microsoft stack as much as we can, even if that means leaving the safe space of Dynamics 365 sometimes. One of the limitations of showing ancillary Azure functionality is that these services can be clunky to maintain. Well if you're lazy like us, you'll love QnA Maker.

With this service, you can easily provision and train a chat bot to answer FAQ-type queries.

With this tool, you can very quickly load a list of question/answer pairs; we recommend pulling realistic examples from a prospect's own website. QnA Maker actually includes functionality to automatically scrub FAQ pages to generate this content, but in our own experiences our mileage varied depending on the site itself. Regardless, it's quick and easy to show off a client's own knowledge content in a compelling way.

Chatbot1a
Click on the image to expand.


If you want to embed your bot in a Dynamics portal or to expose your bot via Skype for Business, you'll need to actually put the effort in to provision a service within Azure. However, QnA Maker does expose a web interface to interact with your bot directly from the web. Our prospects often find that this is compelling enough and they can easily visualize how this service could be embedded and surfaced in other ways.

Chatbot2
Click on the image to expand.


This web interface has additional elements exposed to help train your bot, and talking about this underpinning functionality during the demonstration is actually useful for prospects so that they can better understand how this technology actually operates.

Chatbot3
Click on the image to expand.


You'll certainly want to show this whenever Live Assist or Dynamics Portals are in play, as chat bots could represent an initial line of defense for portal users before being routed to a live chat agent. So we invite you to give this tool a try if you haven't already. It's quick to set up, easy to demonstrate, and offers another way to hook people on the power of Azure. Clearly we love showing Azure as it relates to Dynamics. If you have any interesting Azure-related tips or tricks to share, we'd love to hear from you to keep the conversation going.

Topics: Microsoft Dynamics 365

Hey Cortana, Say Hello to CRM

Today's blog post was written by Angel Shishkov, Principal Developer at Sonoma Partners.

Cortana integration for CRM has been available for preview since the 2016 Update. Here are some commands you can ask it to do. This feature was geared towards mobile phones, but it also works on a Windows PC or laptop.

As the Cortana integration becomes more developed and users explore its possibilities, we expect to see some requests for custom Cortana integrations with CRM as well.

In this post, I will give a brief, high-level introduction to the steps necessary to set up a custom integration between Cortana and CRM and the possibilities it offers us. In this example, there will be no user authentication, and we will hardcode our CRM connection string. I will assume you have an Azure subscription, Visual Studio 2015+, and some knowledge of C#.

Overview

Here are the steps we will go through:

  1. Environment Setup
  2. Implementation
  3. Register the Bot
  4. Test the Bot
  5. Deploy the Bot to Azure
  6. Talk to Cortana

Environment Setup

First, let’s set up our development environment. Visual Studio is free, and Azure has a free trial.

  1. You should have access to Visual Studio IDE, as well as an Azure subscription to deploy the bot.

  2. Download and install the Azure SDK for Visual Studio here.

  3. Download and install Cortana Skill template here.

Implementation

We will implement a Cortana bot in C# .NET and connect to CRM through the CRM SDK. Our bot will be very simple – it will only respond to a couple of phrases and will only perform one query into CRM. For the scenario, I have used an entity from our internal Sonoma CRM system. We use CRM to track our work in records called Items. The items are assigned to people and have Due Dates. The bot we are going to build will respond to two questions regarding Items in CRM: “How many items are due this week?” and “How many items are due next week?”

1. Create VS project with the “Bot Application – Cortana Skill” template we installed earlier. I called mine, HelloCRM.

Angelshish1

2. Restore NuGet packages and run a build to make sure everything is good. The template creates a sample bot with a controller and dialog.

3. Add CRM SDK NuGet packages: Microsoft.CrmSdk.CoreAssemblies and Microsoft.CrmSdk.XrmTooling.CoreAssembly. These are necessary so we can connect our bot to CRM.

Angelshish2

 4. Open RootDialog.cs and replace it with the code below. The code does the following:

  1. RootDialog is an implementation of a dialog between Cortana and our bot.

  2. The StartAsync method is called one when a new dialog starts.

  3. The MessageReceivedAsync method is called each time a message is received from Cortana. That is, whenever the user speaks a phrase.

  4. The spoken message passes in with the result parameter, as a text string.

  5. First, we create a connection to CRM. We are using the Xrm.Tooling.Connector library, which makes it as easy as passing a connection string to a constructor. You will need to modify this connection string to point to your instance of CRM, with your credentials.

  6. Next, we extract the spoken message into a text string.

  7. Then, we use Regex to match the spoken message to a question our bot can handle. Note that the text string is a representation of what Cortana thinks the user said, so we have built in some fuzzy matching of like-sounding terms. For example, the words “week” and “weak” sound the same, so we’re matching on both.

  8. We check for two questions; “how many items this week?” and “how many items next week?” They will both trigger the same query in CRM, but the filter will be different based on which week we are looking at.

  9. Within each match, we first call the CountCRMItemsDue method. This does a simple FetchXML and returns a count of the results. Then, we send back a response using the SayAsync method. This method takes a parameter for the text to display, the text to speak and whether to listen for further questions, or end the conversation. We are passing the AcceptingInput hint, so Cortana will expect further questions in the conversations, after speaking the answer.

5. Rebuild again to make sure everything is good.

Register the Bot

We haven’t deployed the bot to Azure yet, but we can already register it on our Microsoft account. This is necessary if we want to test It locally first. We are going to register a new bot and connect it to the Cortana channel, so it can receive messages from Cortana.

1. Log in to https://dev.botframework.com with your Microsoft account.

2. Go to My Bots and click Register.

3. Type in your bot display name, handle and description.

4. Click on Generate App ID and password and generate both. Copy and save both, we will need them later!

Angelshish3

5. Agree to the terms at the bottom and click Register.

6. You should now be on the Hello CRM bot page, in the Channels tab. Under “Add a channel”, click the Cortana image.

Angelshish4

7. Keep the defaults and click Save

Test the Bot

We can test our bot locally, before we deploy it to Azure. To do this, we need to complete the bot registration above and install the Bot Framework Emulator.

1. Install the Bot Framework Emulator here.

2. In Visual Studio, open the Web.config file for your bot and near the top, fill in the App ID and password we copied earlier.

Angelshish5

3. Run the bot in Visual Studio with the green Run button. It should be set to open in your default browser.

Angelshish6

4. Take note of the URL and port number in the address bar of the browser that popped up. Now, start the Bot Framework Emulator.

5. At the top, click the blue bar, update the URL and port to match what is in your browser and enter the App ID and password you configured in your Web.config.

Angelshish7

6. If everything is good, on the bottom right the Log should show a POST 200 message.

Angelshish8

7. Now you can use the textbox on the bottom to send test messages to your bot, or click the microphone icon to speak to it. It will reply by text message, as well as speech (although it is not Cortana speaking).

Deploy the Bot to Azure

Now that we’ve tested the bot, it is ready to deploy to the cloud. We will be deploying it as an Azure App Service.

  1. Make sure the App ID and password have been added to your Web.config file.

  2. If you have a WebApp already set up in Azure, you can publish directly into it. Otherwise you can create a new WebApp as part of the publish process:

    1. In Visual Studio, right-click the project and select Publish.

    2. Select “Microsoft Azure App Service” as the publish target.

    3. On the App Service page, you can either select an existing WebApp to deploy into, or click New to create a new one.

    4. On the Connection page, keep the defaults, but copy the Destination URL; we will need it later.

      Angelshish9
    5. Click Next through the rest of the dialog and click Publish at the end. Visual Studio will deploy the bot to the WebApp location and open a browser window to it.

  3. Log in to https://dev.botframework.com with your Microsoft account.

  4. Go to My Bots, open the “Hello CRM” bot and click Settings on the top right.

  5. In the Configuration section, fill in the “Messaging endpoint” with the URL of your bot copied earlier, with /api/messages appended to the end.

    Angelshish10

  6. Click Save Changes at the bottom. Now the bot is deployed to the Azure cloud, instead of the local machine, and its registration has been updated with the URL, so Cortana knows where to find it.

  7. By default, the bot is only registered on your Microsoft account, and will only be available to Cortana running on your account. It is possible to deploy it to groups of users, or to make it fully public. On the bot page under My Bots, you can click on the “Manage in Cortana dashboard" link next to the Cortana channel to explore these options.

Talk to Cortana

Now we’re all set to have Cortana use our bot. To invoke custom bots, you need to use the phrase “Hey Cortana, ask <botname> <question>”.

  1. You can activate Cortana by saying “Hey Cortana” (if you have that setting turned on), or clicking the Cortana circle icon in the taskbar.

  2. In our case, we can say “Hey Cortana, ask Hello CRM how many items are due next week?”

  3. The first time Cortana connects to the custom bot, it will ask your permission to send your request to the bot.

    Angelshish11
  4. Click Yes, and you can now converse with the bot.

    Angelshish12


  5. Yay, no items!

I am excited to see the interesting use cases our customers will come up with for custom Cortana integrations. While controlling a system with your voice might seem awkward at first, it will become easier and more natural as voice assistants become more popular in homes.

I hope this was useful, thanks for reading!

Topics: Microsoft Dynamics 365