Invalid task ids used for expiration handlers. #13

[garrettmoon]

This should fix two issues:

• Because taskID was being captured before being set, the expiration handler had an invalid taskID.
• If the expiration handler was called, and then PINCacheEndBackgroundTask was called, it would try to end a task that had already ended.

The second case is less of an issue because I don't see it causing any problems. I've used an atomic getter / setter for taskID which should make it far less likely for there to be a race condition where endBackgroundTask is called twice with the same taskID. However it's still possible. Since this just seems to generate warnings, this seems better than the overhead of introducing a lock of some sort.

[jparise]

If we're going to introduce any intermediate object to hold the task identifier, maybe just go all the way and wrap the full background task interaction (instead of using a macro):

PINBackgroundTask *task = [[PINBackgroundTask alloc] initWithExpirationHandler:^{ ... }];
// ...
[task end];

[jaredsinclair]

lgtm. It might be nice to add a convenience class method like:

PINBackgroundTask *task = [PINBackgroundTask start];

to keep the one-liner aesthetic of the original macro.

[jparise]

I'd prefer the PINBackgroundTask *task = [PINBackgroundTask start]; version, too.

[garrettmoon]

Sounds good, I'll update when I get access to my computer :)