I am writing an application that periodically fetches data from a web server using ASI HTTP and then processes that data to display something relevant to the user on the UI. The data is retrieved from different requests on a single server. The data itself needs to be processed in a specific order. One of the blocks of data is much bigger than the other ones.
In order not to lock the UI while the data is being processed, I have tried to use the NSOperationQueue to run the data processing on different threads. This works fine about 90% of the times. However, in the remaining 10% of the time, the biggest block of data is being processed on the main thread, which cause the UI to block for 1-2 seconds. The application contains two MKMapViews in different tabs. When both MKMapViews tabs are loaded the percentage of time the biggest block of data is being processed on the main thread increases above 50% (which seems to point to the assumption that this happens when there is more concurrent activity).
Is there a way to prevent the NSOperationQueue to run code on the main thread?
I have tried to play with the NSOperationQueue –setMaxConcurrentOperationCount:, increasing and decreasing it but there was no real change on the issue.
This is the code that starts the periodic refresh:
- (void)refreshAll{
// Create Operations
ServerRefreshOperation * smallDataProcessor1Op = [[ServerRefreshOperation alloc] initWithDelegate:_smallDataProcessor1];
ServerRefreshOperation * smallDataProcessor2Op = [[ServerRefreshOperation alloc] initWithDelegate:_smallDataProcessor2];
ServerRefreshOperation * smallDataProcessor3Op = [[ServerRefreshOperation alloc] initWithDelegate:_smallDataProcessor3];
ServerRefreshOperation * smallDataProcessor4Op = [[ServerRefreshOperation alloc] initWithDelegate:_smallDataProcessor4];
ServerRefreshOperation * smallDataProcessor5Op = [[ServerRefreshOperation alloc] initWithDelegate:_smallDataProcessor5];
ServerRefreshOperation * hugeDataProcessorOp = [[ServerRefreshOperation alloc] initWithDelegate:_hugeDataProcessor];
// Create dependency graph (for response processing)
[HugeDataProcessorOp addDependency:smallDataProcessor4Op.operation];
[smallDataProcessor5Op addDependency:smallDataProcessor4Op.operation];
[smallDataProcessor4Op addDependency:smallDataProcessor3Op.operation];
[smallDataProcessor4Op addDependency:smallDataProcessor2Op.operation];
[smallDataProcessor4Op addDependency:smallDataProcessor1Op.operation];
// Start be sending all requests to server (startAsynchronous directly calls the ASIHTTPRequest startAsynchronous method)
[smallDataProcessor1Op startAsynchronous];
[smallDataProcessor2Op startAsynchronous];
[smallDataProcessor3Op startAsynchronous];
[smallDataProcessor4Op startAsynchronous];
[smallDataProcessor5Op startAsynchronous];
[hugeDataProcessorOp startAsynchronous];
}
This is the code that sets the ASI HTTP completion block that starts the data processing:
[_request setCompletionBlock:^{
[self.delegate setResponseString:_request.responseString];
[[MyModel queue] addOperation:operation]; // operation is a NSInvocationOperation that calls the delegate parse method
}];
I have added this block of code in all NSInvocationOperation Invoked method at the entry point:
if([NSThread isMainThread]){
NSLog(#"****************************Running <operation x> on Main thread");
}
The line is printed every time the UI freezes. This shows that the whole operation is run on the main thread. It is actually always the hugeDataProcessorOp that is run on the main thread. I assume that this is because it is the operation that always receives its answer last from the server.
After much investigation in my own code, I can confirm that this was a coding error.
There was an old call remaining that did not go through the NSInvocationOperation but was calling the selector that NSInvocationOperation should have called directly (therefore not using the concurrent NSOperationQueue.
This means that the NSOperationQueue DOES NOT use the main thread (except if it is the one retrieved by +mainQueue).
Override isConcurrent on your NSOperation and return YES. According to the documentation this will cause your NSOperation to be run asynchronously.
Related
folks! I'm implementing a shared cache in my app. The idea is to get the cached data from the web in the background and then update the cache and the UI with the newly retrieved data. The trick is of course to ensure thread-safety, since the main thread will be continuously using the cache. I don't want to modify the cache in any fashion while someone else might be using it.
It's my understanding that using #synchronized to lock access to a shared resource is not the most elegant approach in ObjectiveC due to it trapping to the kernel and thus being rather sluggish. I keep reading that using GCD instead is a great alternative (let's ignore its cousin NSOperation for now), and I'd like to figure out what a good pattern for my situation would be. Here's some sample code:
dispatch_queue_t queue = dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0);
// download the data in a background thread
dispatch_async(queue, ^{
CacheData *data = [Downloader getLatestData];
// use the downloaded data in the main thread
dispatch_sync(dispatch_get_main_queue(), ^{
[AppCache updateCache:data];
[[NSNotificationCenter defaultCenter] postNotificationName:#"CacheUpdated" object:nil];
});
});
Would this actually do what I think it does, and if so, is this the cleanest approach as of today of handling this kind of situation? There's a blog post that's quite close to what I'm talking about, but I wanted to double-check with you as well.
I'm thinking that as long as I only ever access shared the shared resource on the same thread/queue (main in my case) and only ever update UI on main, then I will effectively achieve thread-safety. Is that correct?
Thanks!
Yes.
Other considerations aside, instead of shunting read/write work onto the main thread consider using a private dispatch queue.
dispatch_queue_t readwritequeue;
readwritequeue = dispatch_queue_create("com.myApp.cacheAccessQueue", NULL);
Then update your AppCache class:
- (void)updateCache:(id)data {
dispatch_sync(readwritequeue, ^{ ... code to set data ... });
}
- (id)fetchData:... {
__block id data = nil;
dispatch_sync(readwritequeue, ^{ data = ... code to fetch data ...});
return data;
}
Then update your original code:
dispatch_queue_t queue = dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0);
// download the data in a background thread
dispatch_async(queue, ^{
CacheData *data = [Downloader getLatestData];
**[AppCache updateCache:data];**
// use the downloaded data in the main thread
dispatch_async(dispatch_get_main_queue(), ^{
[[NSNotificationCenter defaultCenter] postNotificationName:#"CacheUpdated" object:nil];
});
});
If you ask 100 developers here was is the most elegant way to do this, you will get at least 100 different answers (maybe more!)
What I do, and what is working well for me, is to have a singleton class doing my image management. I use Core Data, and save thumbnails directly in the store, but use the file system and a URL to it in Core Data for "large" files. Core Data is setup to use the new block based interface so it can do all its work on a private thread managed by itself.
Possible image URLS get registered with a tag on the main thread. Other classes can ask for the image for that tag. If the image is not there, nil is returned, but this class sets a fetchingFlag, uses a concurrent NSOperation coupled to a NSURLConnection to fetch the image, when it gets it it messages the singleton on its thread with the received image data, and the method getting that message uses '[moc performBlock:...]' (no wait) to process it.
When images are finally added to the repository, the moc dispatches a notification on the main queue with the received image tag. Classes that wanted the image can listen for this, and when they get it (on the main thread) they can then ask the moc for the image again, which is obviously there.
This code works
[[MyManager sharedManager] makeRequestAndParsingfor:someParameters
success:^(NSDictionary * dictionary){
// Sucessful response
NSLog(#"Success!!");
}
failure:^(NSError* error){
// Error response
NSLog(#"Failure!");
}];
But this whenever I run the same in the background it never enters in success or failure block.
dispatch_async(dispatch_get_global_queue( DISPATCH_QUEUE_PRIORITY_LOW, 0), ^{
[[MyManager sharedManager] makeRequestAndParsingfor:someParameters
success:^(NSDictionary * dictionary){
// Sucessful response
NSLog(#"Success!!");
}
failure:^(NSError* error){
// Error response
NSLog(#"Failure!");
}];
}];
Can anybody explain me what happens? The method makeRequestAndParsingfor: makes an asynchronous request (with blocks again) and then parses the result. My debugger shows that it never gets in its own success/failure in the second case. In the first case it works like a charm. Any ideas?
Does your method 'makeRequestAndParsingfor' use block_copy() of the args, and store the return in a strong variable (to get the blocks in the heap)? Also add asserts() so you can verify you get both blocks in 'makeRequestAndParsingfor', and even retest before calling one or the other. [In the past use of block_copy() was necessary but now not 100% sure.]
Note that in the second case, where 'makeRequestAndParsingfor' runs in a concurrent queue that multiple of these can call the blocks concurrently - not sure what your success/failure block does, but it better be thread safe, or you should run the block on the main queue in 'makeRequestAndParsingfor' (which I do in my similarly constructed app).
Application Specific Information:
com.my-app failed to launch in time
Elapsed total CPU time (seconds): 20.090 (user 20.090, system 0.000), 100% CPU
Elapsed application CPU time (seconds): 17.598, 87% CPU
I've made a modification to my app and as a result I now run a function from applicationDidFinishLaunching which will do some database processing.
I'm basically creating some new records and updating some existing ones.
For one of my existing beta testers / real customers, this is taking 20 seconds to complete.
Although in this case this is a one off, users could experience this situation if they haven't used the app for a while.
Normally the process wouldn't take long at all, as there would only be a few transactions to process.
I'm unsure how to proceed, any suggestions ?
I suggest you to do your db processing in the background. Maybe you could disable the interface or display a waiting indicator while you are updating the db in the background thread. Then, once finished you could enable the interface or hide the indicator.
There are different ways to create background thread.
Create a thread manually using NSThread class
Using NSOperation and NSOperationQueue classes
Using Grand Central Dispatch (GCD)
Hope it helps.
Edit
Here simple code for your goal (following #JeremyP suggestion).
First, create a NSOperation subclass
// .h
#interface YourOperation : NSOperation
{
}
//.m
#implementation YourOperation
// override main, note that init is executed in the same thread where you alloc-init this instance
- (void)main
{
// sorround the thread with a pool
NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
// do your stuff here..
// you could send a notification when you have finished to import your db,
// the notification is sent in a background thread,
// so in the place where you listen it, if you need to update the interface,
// you need to do it in the main thread (e.g. performSelectorOnMainThread)
[[NSNotificationCenter defaultCenter] postNotificationName:kImportComplete object:self];
[pool drain];
pool = nil;
}
Then, in your application delegate for example call [self import]; that could be defined as follow:
if (!(self.operationQueue)) {
NSOperationQueue* q = [[NSOperationQueue alloc] init];
[q setMaxConcurrentOperationCount:1];
self.operationQueue = q;
[q release];
YourOperation *op = [[YourOperation alloc] init];
[self.operationQueue addOperation:op];
[op release], op = nil;
}
I would (and have in the past in a few apps) perform the DB update on a background thread and show the user a 'please wait' screen while the updates complete.
I want to handle freezing my program, when it load an xml from bad address.
I tryed it with using #try and #catch, but it doesn't work.
Can I use some alternative handling?
#try{
NSString *test=[[NSString alloc] initWithContentsOfURL:[NSURL URLWithString:[NSString stringWithFormat:#"http://%#:%#",addressLabel.text,portLabel.text]] encoding:NSUTF8StringEncoding error: nil];
}
#catch (NSException *ex) {
NSLog(#"Bad IP address");
return;
}
Run your XML Parser in NSThread and use notification for errors.
initWithContentsOfURL is a synchronous call. The control will return back from the function only on complete. Try this function is a worker thread so that your main thread will not be blocked.
If you use use NSThread then you have to dive into the memory management unless you are working in XCode 4.2 and using ARC.
So there are two ways for fetching the XML from the server.
1) Use NSURLConnection to get the xml as a NSData object and when you finish loading the data you can simply use that data to initialize an NSString Object. NSURLConnection sends asynchronous call to the server so it will not freeze your view.
2) You can use NSIncovationOperation and NSQueue to fetch your XML and it will also not effect your main thread. like
-(void)myMethod{
NSString *test=[[NSString alloc] initWithContentsOfURL:[NSURL URLWithString:[NSString stringWithFormat:#"http://%#:%#",addressLabel.text,portLabel.text]] encoding:NSUTF8StringEncoding error: nil];
[self performSelectorOnMainThread:#selector(handleString:) withObject:test];
}
You can use NSInvocationOperation object as follow
NSInvocationOperation *opr = [[NSInvocationOperation alloc] initWithTarget:self selector:#selector(myMethod) object:nil];
NSOperationQueue *queue= [NSOperationQueue new];
[queue addOperation:opr];
When the perform selector will be call then you can pass that XML string object to the handleString: method. like
-(void)handleString:(NSString*)xmlString{
// Do something with string
}
I hope that it clarifies a little bit of your confusion. All this was to give you an idea how can you achieve your goal without freezing your interface i.e main thread.
regards,
Arslan
You need to launch all long time operations on a second thread to avoid blocking the main thread. Use [self performSelector:#selector(yourXmlDownloadMethod)].
I am trying to create class that will handle multiple downloads at same time (I need to download a lot of small files) and I have problems with "disappearing" connections.
I have function addDonwload that adds url to list of urls to download, and checks if there is free download slot available. If there is one it starts download immediately. When one of downloads finishes, I pick first url form list and start new download.
I use NSURLConnection for downloading, here is some code
- (bool) TryDownload:(downloadInfo*)info
{
int index;
#synchronized(_asyncConnection)
{
index = [_asyncConnection indexOfObject:nullObject];
if(index != NSNotFound)
{
NSLog(#"downloading %# at index %i", info.url, index);
activeInfo[index] = info;
NSURLRequest *request = [NSURLRequest requestWithURL:info.url cachePolicy:NSURLRequestUseProtocolCachePolicy timeoutInterval:15];
[_asyncConnection replaceObjectAtIndex:index withObject:[[NSURLConnection alloc] initWithRequest:request delegate:self startImmediately:TRUE]];
//[[_asyncConnection objectAtIndex:i] scheduleInRunLoop:[NSRunLoop currentRunLoop] forMode:NSDefaultRunLoopMode];
return true;
}
}
return false;
}
- (void)connectionDidFinishLoading:(NSURLConnection*)connection
{
[self performSelectorOnMainThread:#selector(DownloadFinished:) withObject:connection waitUntilDone:false];
}
- (void)DownloadFinished:(id)connection
{
NSInteger index = NSNotFound;
#synchronized(_asyncConnection)
{
index = [_asyncConnection indexOfObject:(NSURLConnection*)connection];
}
[(id)activeInfo[index].delegate performSelectorInBackground:#selector(backgroundDownloadSucceededWithData:) withObject:_data[index]];
[_data[index] release];
[activeInfo[index].delegate release];
#synchronized(_asyncConnection)
{
[[_asyncConnection objectAtIndex:index] release];
[_asyncConnection replaceObjectAtIndex:index withObject:nullObject];
}
#synchronized(downloadQueue)
{
[downloadQueue removeObject:activeInfo[index]];
[self NextDownload];
}
}
- (void)NextDownload
{
NSLog(#"files remaining: %i", downloadQueue.count);
if(downloadQueue.count > 0)
{
if([self TryDownload:[downloadQueue objectAtIndex:0]])
{
[downloadQueue removeObjectAtIndex:0];
}
}
}
_asyncConnection is my array of download slots (NSURLConnections)
downloadQueue is list of urls to download
What happens is, at the beginning everything works ok, but after few downloads my connections start to disappear. Download starts but connection:didReceiveResponse: never gets called. There is one thing in output console that I don't understand I that might help a bit. Normaly there is something like
2010-01-24 21:44:17.504 appName[3057:207]
before my NSLog messages. I guess that number in square brackets is some kind of app:thread id? everything works ok while there is same number, but after some time, "NSLog(#"downloading %# at index %i", info.url, index);" messages starts having different that second number. And when that happens, I stop receiving any callbacks for that urlconnection.
This has been driving me nuts as I have strict deadlines and I can't find problem. I don't have many experiences with iphone dev and multithreaded apps. I have been trying different approaches so my code is kinda messy, but I hope you will see what I am trying to do here :)
btw is anyone of you know about existing class/lib I could use that would be helpful as well. I want parallel downloads with ability o dynamically add new files to download (so initializing downloader at the beginning with all urls is not helpful for me)
You've got a bunch of serious memory issues, and thread synchronization issues in this code.
Rather than go into them all, I'll ask the following question: You are doing this on a background thread of some kind? Why? IIRC NSURLConnection already does it's downloads on a background thread and calls your delegate on the thread that the NSURLConnection was created upon (e.g., your main thread ideally).
Suggest you step back, re-read NSURLConnection documentation and then remove your background threading code and all the complexity you've injected into this unnecessarily.
Further Suggestion: Instead of trying to maintain parallel positioning in two arrays (and some sketchy code in the above relating to that), make one array and have an object that contains both the NSURLConnection AND the object representing the result. Then you can just release the connection instance var when the connection is done. And the parent object (and thus the data) when you are done with the data.
I recommend that you take a look at this:
http://allseeing-i.com/ASIHTTPRequest/
It's a pretty sophisticated set of classes with liberal licensing terms (free too).
It may provide a lot of the functionality that you are wanting.
This snippet can be the source of the bug, you release the object pointed to by the activeInfo[index].delegate pointer right after issuing async method call on that object.
[(id)activeInfo[index].delegate performSelectorInBackground:#selector(backgroundDownloadSucceededWithData:) withObject:_data[index]];
[_data[index] release];
[activeInfo[index].delegate release];
Do you use connection:didFailWithError: ? There may be a timeout that prevents the successful download completion.
Try to get rid of the #synchronized blocks and see what happens.
The string inside the square brackets seems to be thread identifier as you guessed. So maybe you get locked in the #synchronized. Actually, I don't see a reason for switching thread - all the problematic code should run in the main thread (performSelectorOnMainThread)...
Anyhow, there is no need to use both the #synchronized and the performSelectorOnMainThread.
BTW, I didn't see the NSURLConnection *connection = [[NSURLConnection alloc] initWithRequest:request delegate:self]; line. Where do you initiate the connection?
As for the parallel downloads - I think that you can download more than one file in a time with the same code that you use here. Just create a separate connection for each download.
Consider just keeping a download queue along with a count of active connections, popping items off the top of the queue when downloads complete and a slot becomes free. You can then fire off NSURLConnection objects asynchronously and process events on the main thread.
If you find that your parallel approach prohibits doing all of the processing on the main thread, consider having intermediary manager objects between your main thread download code and NSURLConnection. Using that approach, you'd instantiate your manager and get it to use NSURLConnection synchronously on a background thread. That manager then completely deals with the downloading and passes the result back to its main thread delegate using a performSelectorOnMainThread:withObject: call. Each download is then just a case of creating a new manager object when you've a slot free and setting it going.