On Mar 27, 2012, at 2:23 PM, Rick Mann wrote: > > On Mar 27, 2012, at 14:16 , Conrad Shultz wrote: > >> As David already pointed out you should probably be using file URLs. > > I made that change, and it behaves the same way. Even so, I don't see why > that particular difference would allow it to work on the sim and not on the > device. > >> There are a couple other issues: >> >> 1) You shouldn't handle errors by testing (err != nil). This might work >> (since you initialized err to nil) assuming that removeItemAtURL: isn't >> fiddling with err anyway, but it's not guaranteed. The preferred >> approach is to test the return value of removeItemAtURL: (it returns >> BOOL for a reason!) and examine err iff the method returns NO. > > Okay, I made that change, but it seems blatantly incorrect to create an error > return if it is also going to return true.
Despite the impoliteness, the documentation used to explicitly warn that this was possible. The contract is "return value indicates success; if successful, *error is meaningless." Nowadays with ARC, you can't trash *error without breaking ARC's rules. > >> 2) Less likely, but still reasonably plausible since you are working >> with cache files (which I imagine have other clearance mechanisms), is >> that you are encountering a race condition. It's possible that another >> thread removed the file in the intervening time between your >> fileExistsAtPath: and removeItemAtURL: calls. This is a major reason >> why it's usually not useful to to ask "can I delete a file?" but rather >> to try "delete the file!" and handle any errors appropriately. > > A race is possible, although I think unlikely (the call to the method that > does the deletion is serialized on the main queue, although I suppose it's > possible two operations are getting queued for the same file). In any case, I > wasn't checking for it first, and then deleting it. I was checking for it > only as a sanity check once I started seeing the problem. I was just deleting > it and handling the error. The more unlikely you think a race condition is, the more likely you are going to hit it at a very bad time. Don't play around with race conditions. This is a fairly canonical example of a filesystem race. --Kyle Sluder _______________________________________________ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
