I’m writing my first WinRT app for Windows 8 in C#/XAML managed code and I obviously have to handle the UI in the different sizes and orientations that could occur (FullScreenLandscape, FullScreenPortrait, Filled and Snapped).
Most people seem to suggest handling the UI changes through the Visual State Manager in XAML, however me being more comfortable with writing the code-behind rather than XAML, I thought I would simply handle the Current_SizeChanged event with a switch statement for each of the states.
I tried both ways and both seem to work for me (though the VSM was decidedly more work – at least for me).
Can someone tell me why I should use the VSM over code, or what benefits I would receive?
It’s a good question, and certainly one that entered my head when starting WPF & WinRT development. Having coded C# projects for years, it seemed a strange concept to start building UI logic in the Xaml (which is usually more lines than C#, but also more declarative).
In actual fact (if you don’t mind me saying) I think we can abstract your question out to ‘Is there any benefit to writing UI code in Xaml over C#/VB.net?’.
Let me give you an example, in work I’m on a project team with several developers, and several graphic designers. The designers are pretty awesome at laying out that Xaml, and creating a consistent feel for the application (something I can’t say I would be that good at) – but will have little idea when it comes to writing web-services and Data Access layers – which is our job as developers. And that’s how it should be right?
Well whenever you start writing a lot of View/UI related logic in C#.net, this can lead to all sorts of problems. The designers all of a sudden can’t focus on the Xaml at hand, and must get up to speed with OO programming. In our project this isn’t that much of a concern, as the designers are actually pretty competent developers (must have been all that UI code I forced them to understand last year :)) But what I think it boils down to is a ‘Separation of Concerns’ at a job description level. If we take the paradigms associated to WPF I think things like Databinding ‘guides’ developers down a road where they create separated, testable UI and Business layers – the same way that the nature of Xaml allows for a very declarative approach to writing UI where the View Logic is created and maintained in a very readable manner without getting too much into the nature of OO programming.
So, coming back to your question – No, I don’t think there any immediate benefits, and if you come from a mainly .Net background to write a lot of these things is Xaml can be little tricky. However, if you are in a team of people – or if you find best-practices a very important factor in development, then writing View related code such as Visual State manager configuration in Xaml is the way to go.
One last point – you’ll notice I used the terminology ‘View related logic’ a lot. This is because it’s widly viewed that writing UI related code in code-behind is acceptable (if not best-practice) but sometimes due to the nature of the WPF or WinRT framework you are dragged down this route for some functionality. However, if you are writing business logic in the UI files codebehind this is viewed a particular no-no. This breaks the ‘seperation of concerns’ and can make testing very difficult. If you are following the MVVM pattern (as many WPF or WinRT projects do) then this is what the ViewModel is responsible for.