CQRS: how do you retrieve information about an executed command?

Question

In most command interfaces I've seen, there is typically an "Execute" method which takes takes a command input and either returns void or some generic structure indicating if the command executed successfully or not (we are using the latter). Now, I've never thought of this before, but we suddenly got the need to know some more details about the result of the command than what you can expose generically.

Consider the following example:

you have a team and you are creating a screen where you can add members to your team. The members of the team are shown in a grid below the "add new member"-stuff. Now, when you press "add new member" you want to run some jquery/roundohuse/whatever and add the new member to the list of team members. No problems so far, but: you also want to include some identification data in a hidden field for each member and this id-data comes from the server.

So the problem is: how can I get that id-data from the server? The "AddNewTeamMember" command which I am pushing through the "ExecuteCommand"-method does not give me anything useful back, and if I add a new query method to the service saying something like: "GetLastAddedTeamMember" then I might just get the last entry added by someone else (at least if this is data which is very aggressively added by different users). In some situations you have a natural unique identifier generated on the client side which we can use, but for team members we did not.

Answer 1

Given that you have no choice but to update an on-page widget when another command completes, I see two choices for you:

1. Shoot off the command, display something locally that indicates it is submitted, and then wait until you get a notification from the server that the team member list has changed. Update the widget to reflect that.

2. Add a correlation ID to your command when you submit it, and add the team member provisionally locally to the list. When you get a confirmation from the server that a team member update happened because of your correlation ID, update your local data.

I would suggest the first approach, where the "provisional indicator" could be throwing a marked version of the normal indication into place; then, when you finally get an update you should have the data you need.

Given you went with CQRS to solve this problem I assume you have frequent updates to the content of those widgets happening in the background already, so have presumably solved the "background update" problem.

If not, I suggest you either ditch CQRS as a bad - over-complicated - solution in your problem space, or solve the background update problem first.

Answer 2

If you want to add an existing team member, you should query the read side of your application for this data. If you need to add a new team member, you have to consider if it's necessary to show the user in the grid below at once. Can you wait until the team member is in place on the read side? You can also query a service on the server side to get an unique ID (it can return a GUID). Then you add the team member to the grid, and of course, send the command to the server. But, if it's possible, try to design the application in a way that you don't have to show the team member at once. It's also possible to give the user a message saying something like this: "Team member added, waiting for response from server.". Then use AJAX to query the read side for new team members. When it appears on the read side, show it in the grid. You might have to deal with team members added by other users, but does it matter? CQRS gives you a great way to collaborate with other users, so maybe you should take advantage of that. As I see it; CQRS forces you to think different, and that may not be a bad thing.