ReactiveCommand.Execute() not delievered on outputScheduler

[csmager]

Do you want to request a feature or report a bug?

Potential bug. A difference in behaviour in 7.x vs 6.x.

What is the current behavior?

The results from ReactiveCommand.Execute() are not delivered on the outputScheduler, only results from subscribing to the command itself.

This differs from the observable returned by ReactiveComand.ExecuteAsync() in v6.x.

What is the expected behavior?

I'm unclear whether this change was intentional or not, but thought I'd raise as it bit us when migrating some of our commands.

[bradphelan]

I just found this as well. The documentation is clear on this but it seems a bit strange. The 99% case is

var command = ReactiveCommand.Create(()=>DoStuff(), schedular = TaskPoolSchedular.Instance)

where I want DoStuff to run on the task pool. Rather that generate some value and then subscribe to the value for the body of the command like below

var command = ReactiveCommand.Create(()=> Unit.Default, outputSchedular = TaskPoolSchedular.Instance);
command
    .Subscribe(x=>DoStuff());

[csmager]

@bradphelan I'm not sure that's related. I'd agree that's made very clear in the big 'Warning' box in the documentation.

Here I'm talking about the difference between these two subscriptions:

command.Subscribe(x => ...);
command.Execute().Subscribe(x => ...);

Take this example:

Debug.WriteLine("Main Thread [{0}]", Thread.CurrentThread.ManagedThreadId);

var command = ReactiveCommand.Create(() => Unit.Default, 
    outputScheduler: RxApp.MainThreadScheduler);

command.Subscribe(_ => 
    Debug.WriteLine("command [{0}]", Thread.CurrentThread.ManagedThreadId)
);

await Task.Run(() =>
    command.Execute().Subscribe(_ =>
        Debug.WriteLine("command.Execute() [{0}]", Thread.CurrentThread.ManagedThreadId))
);

I get this as output:

Main Thread [1]
command.Execute() [3]
command [1]

Where I would expect (or would have seen in v6.x):

Main Thread [1]
command [1]
command.Execute() [1]

[kentcb]

Thanks for the report @csmager. This was indeed something I missed in the rewrite. Curiously, I had created a test intended to check this case, but I neglected to fix the test code up (copy/paste error). See #1325 for the PR.

[csmager]

@kentcb thanks for that. Pleased it wasn't a deliberate change!