1

Initial problem:
I'm using the typical RelayCommand implementation in my MVVM Application. I realized that the CanExecute of my RelayCommand is not always called even if meaningful properties of my ViewModel have changed.

I read that we can manually call CommandManager.InvalidateRequerySuggested to raise a RequerySuggested event. This will eventually make the Command Source call the CanExecute method.

RequerySuggested is automatically raised by the CommandManager when some UI events occurs. I feel like it would be useful if it would also be raised on PropertyChanged.

My current solution:
I modified the property changed method of my ViewModelBase class like this:

protected virtual void OnPropertyChanged(string propertyName)
{
    PropertyChanged(this, new PropertyChangedEventArgs(propertyName));
    Application.Current.Dispatcher?.Invoke(CommandManager.InvalidateRequerySuggested);
}

Concerns:
I have not found any ViewModelBase online implemented this way and I feel like this should be a red flag.

RelayCommand is highly recommended in MVVM for it's simplicity so I find weird that I end up having to manually call the CommandManager. Why not preferring DelegateCommand then?

I know that this causes the CanExecute to be called more often, but it is already spammed every time a UI event happens anyway.

Spirielle
  • 23
  • 1
  • 3
  • Usually implementations of ICommand have some method of raising the CanExecuteChanged event. Have you tried that? IMO much cleaner and more efficient than CommandManager.InvalidateRequerySuggested – Dave M Sep 07 '19 at 00:37
  • Look at this [thread](https://stackoverflow.com/questions/22285866/why-relaycommand) how implementation is done – Pavel Anikhouski Sep 07 '19 at 07:45
  • To trigger the `RequerySuggested` from the `PropertyChanged` event invocator of your base view model is a bad idea. It will trigger the `RequerySuggested` event when _any_ property has changed. No matter if the property is part of a command logic or not. This creates a huge overhead given that the `ViewModelBase` is used throughout the whole application and extended by a many classes. Usually the `CanExecute` is called on focus changes or keyboard and mouse events. This is sufficient most of the time. For special situations you can explicitly call `CommandManager.InvalidateRequerySuggested` – BionicCode Sep 07 '19 at 16:44
  • @DaveM You mean having a public `RaiseCanExecuteChanged`? The one from Josh Smith does not. I agree that it could be cleaner. @BionicCode That's where I'm confused I guess. I feel like having to _sometimes_ manually call the command manager defeats the purpose of `RelayCommand`. I'll try thinking of a better long term solution for when the CanExecute could change without any user interaction. – Spirielle Sep 09 '19 at 14:39

1 Answers1

0

"typical" relaycommand?

If you're using MVVMLight's relaycommand then the source is:

https://github.com/lbugnion/mvvmlight/blob/master/GalaSoft.MvvmLight/GalaSoft.MvvmLight%20(PCL)/Command/RelayCommand.cs

You can call CanExecutechanged on the relaycomand.

When you add the using statement to any viewmodel, make sure you use the CommandWPF version.

InvalidateRequerySuggested is called when there is a user input event. You would therefore be raising it twice when a user inputs anything and that changes a property you raise propertychanged on. Meaning your approach is a bad idea.

It's rare to have some data comes from a model to a viewmodel on a property by property basis. Much more usual that an event is raised and the viewmodel then composes data from a viewmodel into properties. You could probably just raise canexecutechanged the once after you get a new record or a change event or however your data changes happen when they're not user input driven.

Andy
  • 11,864
  • 2
  • 17
  • 20
  • Thank you. I used Josh Smith's RelayCommand which does not contain a way of raising the CanExecuteChanged manually. I'll modify my version and just raise the event when I need it. – Spirielle Sep 09 '19 at 15:00