Hacker News Reader

> The downside to this is that making a UI seems to even out your gained time--it's extremely messy and even complicated. Code that's valid in C# produces vague underwater bugs in Java code, which makes you keep hacking around until you find a working solution.

Interesting, I had the exact opposite experience, since the Java->C# mappings often reduce much of the boilerplate code that you have to write if you were using Java (e.g C# events, properties, etc..)

I also never came across any bugs in the mappings, but I can't say for sure that there aren't any.

Most of my problems often come from constructors. For example, using a custom parametered constructor in an Activity or Fragment seems impossible, so constructor injection is already out of the question. This is valid C#, but will not compile. Hence, I stated that using Xamarin feels like programming Java in C#, which makes me feel like I would actually be happier writing Java code.

Instead, for this example you could subclass Activity/Fragment and use reflection to find properties on the activity to inject services into.

This is only one of countless annoying Xamarin problems I have encountered in my few months of use, and this one was relatively easy to solve. Vague problems also ensue when forgetting to inherit a class from Java.Lang.Object which leads to vague exception messages like "Specified cast invalid".

Sure, sometimes parts can be made faster in C#, but more often than not I spend day(s) debugging extremely vague bugs related to non-documented mappings.

> Some of the most used API's are not documented at all, simply having a "To be added" description.

I rarely find myself reading the Xamarin documentation, and mostly browse documentation and StackOverflow as if I was coding a native iOS/Android app. This works because the Xamarin wrapper is pretty much identical to native APIs.

Agreed. I never read the Xamarin documentation, instead I'm looking at objc or Java and then impl as c# instead. The method names, signatures and idioms match almost one for one. When the API doesn't match it's because it makes sense to; Pickup a copy of the dotnet framework design guidelines and you'll see what I mean about the previous statement.

I've written a few apps in C++/COM with a UI written in Visual Basic. The VB app was 'button code', with the heavy lifting done in COM. I wonder if this approach would work for Xamarin - write the core logic in C# but leave the UI in Java.