iOS: Why do I need to use sleep(1) after NSURLConnection sendAsynchronousRequest? - ios

Question

I want to display an image on the screen which I take from the internet. I have used
NSURLConnection to create an asynchronous call to take the data and, in the response block, I called the code to assign it to an UIImage object.
My question is why do I need to call sleep(1) after the block execution? If i'm not calling it, then my image is not drawn on the screen. Is it another, more elegant way to achive this?
-(void)loadImage:(NSString *)url
{
NSURL *imageURL = [NSURL URLWithString:url];
NSOperationQueue *queue = [[NSOperationQueue alloc]init];
NSURLRequest *imageRequest = [NSURLRequest requestWithURL:imageURL cachePolicy:NSURLRequestReloadIgnoringLocalAndRemoteCacheData timeoutInterval:5.0f];
[NSURLConnection sendAsynchronousRequest:imageRequest queue:queue completionHandler:^(NSURLResponse *response, NSData *data, NSError *connectionError) {
if(!connectionError) {
if(data) {
//there goes the main thingy
self.myView.wallpaperImage = [UIImage imageWithData:data];
[self.myView setNeedsDisplay];
} else {
NSLog(#"No data found at url:%#",url);
}
} else {
NSLog(#"Could not connect to %#",url);
}
}];
sleep(1);
}

Answer 1

This:
self.myView.wallpaperImage = [UIImage imageWithData:data];
[self.myView setNeedsDisplay];
Is happening on the thread managed by the NSOperationQueue passed to sendAsynchronousRequest. Those methods need to be called from the main thread.
Your sleep may be causing the main thread's runloop to iterate, after which those calls appear to have worked.
To fix this, and to avoid a whole bunch of other problems your current approach will have, do this:
[NSURLConnection sendAsynchronousRequest:imageRequest queue:queue completionHandler:^(NSURLResponse *response, NSData *data, NSError *connectionError) {
if([data length] > 0) {
//there goes the main thingy
[[NSOperationQueue mainQueue] addOperationWithBlock:^{
self.myView.wallpaperImage = [UIImage imageWithData:data];
[self.myView setNeedsDisplay];
}];
} else {
// Perform your error handling here.
}
}];
This will use [NSOperationQueue mainQueue] to perform those UIKit calls from the main queue - not libdispatch. libdispatch is a low level interface, it is a recommended best practice to always prefer the higher level interface - in this case, NSOperationQueue. UIKit is only safe when called from the main thread (or queue).
It also changes your error handling behavior to follow the best practices for the platform - check the result of your call (in this case, data) and THEN process any error returned.
Your code is actually a good example of why blocks retain captured objects (in this case self). If there was no retain cycle here, ARC could destroy queue as soon as it goes out of scope, and the block would never execute. Instead, because of the retain cycle, the queue stays around until the block has executed.