Not all types of errors should count against RefreshWindow count

[andreas-bakriemer]

• I've read, understood, and done my best to follow the *CONTRIBUTING guidelines.

Problem

Whenever an access token refresh request/operation fails, it counts in the RefreshWindow. Using a RefreshWindow will eventually cause excessiveRefresh errors to be thrown. If the errors refreshing the access token are related to network problems (as it will come to pass on a mobile platform), the RefreshWindow can fill up rather quickly. I like the idea of excessiveRefresh errors, but only when it's related to actual errors concerning authentication - not timeouts and reachability.

Feature Request

It could be nice to be able to configure what types of errors count in the refresh window. A suitable default would obviously be all errors, so that not-configured RefreshWindows continue to operate as they do already.

[SamHastings1066]

Hi there,

I’ve begun resolving this issue and have developed a feature that allows users to configure which types of errors count towards the refresh window. The default behavior remains unchanged, counting all errors, while users can now provide a closure to specify their error handling logic.

I'm in the process of writing tests and updating the documentation to reflect this new feature. I plan to submit a draft PR soon to get feedback on the implementation.

Could this issue be assigned to me to avoid duplicate work?

Thank you!

[jshier]

Issue assignments aren't used like that, and there's no one else working on the issue, so feel to work as you want.