I have a silly question. Various questions on SO and tutorials on the web suggest saving a listener interface in your

AsyncTask

to still keep a reference and publish progress to your fragment/activity/service/anything. Question stands, shouldn't it be

WeakReference

? I'd assume that ART does not garbage collect running threads and if your application is suddenly ended, keeping a hard reference to your context will prevent ART from garbage collecting it.

dragas: if your application have suddenly ended, ART VM will be just killed with whole its heap. But if your UI is being killed due to screen rotation, you will get short-time memory leak, so yes, reference should be weak, and AsynkTask — static (if nested). Also, AsynkTasks are unaware of UI lifecycle, so it may be better to use Loaders or MVP.

or just don't use

AsyncTask

as there are much better alternatives

Nobody use

AsyncTask

those days because their limitations/problems

@abdellah wrong, I use `AsyncTaskLoader`s! :troll:

nice to meet you 🆘

@kirillrakhman I'd assume a better solution would be using Anko's

doAsync

?

doAsync is just a wrapper around

AsyncTask

and only mitigates the problem with leaks by keeping weak references. Using RxJava in combination with, say RxLifecycle will give you a clearer code structure and better leak protection