Better handling for lots of parallel network requests and a bad network

[NerdyTherapist]

When the user is in a bad network environment (unreliable EDGE, for example) and/or sets off lots of requests by a lot of scrolling around (since requests take so long), the app handles it very poorly. A request timeout is now being logged in console and no longer crashes the app now, but the user is not being informed at all.

Optimal fix: Show a notification, that there are network problems, maybe even the localized error description.
Also: Is there a way to cancel pending network activity when a new request is being made? That would be quite nice for server load and app activity.

[mguentner]

In the Angular 1 Frontend, a lot of requests are saved by loading more than what is currently visible, assuming that most users will just pan around their current location.
Here's the code:

https://github.com/MateMonkey/frontend/blob/master/app/js/application.js#L148

and here the corresponding function that checks whether the current bounds (the view) are larger anywhere outside the previously fetched bounds rectangle:

https://github.com/MateMonkey/frontend/blob/04f775d2db4ecf89d6ab6ce1f820ea23f9484e97/app/js/map.js#L80

Using that you can pan / explore a bit until you need to load a whole new rectangle.
That is a good trade-off solution between a stitched-together multi-polygon (each request would append / add to that polygon), therefore managing a complex local state and requesting after each view change.