I have a callback handler registered that listens to changes in the iOS Address Book. Due to some strange reason (for which a bug has been filed), this callback can sometimes be called more than once when the app returns from the background. I want my callback handler to run it's logic only once, even in cases the callback is called multiple times. This is how I register the callback:
ABAddressBookRegisterExternalChangeCallback(address_book, adressBookChanged, self);
This is how I structured my callback handler to take advantage of GCD to handle this. Unfortunately, it's not working, and GCD doesn't prevent the internal logic to be called twice...
void adressBookChanged(ABAddressBookRef ab, CFDictionaryRef info, void
*context)
{
NSLog(#"** IN addressBookChanged callback!");
ABAddressBookUnregisterExternalChangeCallback (ab, adressBookChanged, context);
__block BOOL fireOnce = FALSE;
dispatch_queue_t queue;
queue = dispatch_queue_create("com.myapp.abcallback", NULL);
dispatch_async(queue, ^{
if (fireOnce == FALSE) {
fireOnce = TRUE;
dispatch_queue_t queueInternal;
queueInternal = dispatch_queue_create("com.myapp.abcallbackInternal", NULL);
dispatch_async (queueInternal, ^{
NSLog(#"do internal logic");
});
dispatch_release(queueInternal);
}
});
dispatch_release(queue);
}
I'm pretty sure this code works for receiving multiple notifications, so are callbacks different? Do they spawn different threads automatically, making the fireOnce value to be FALSE each time? How should I write this code to prevent multiple callbacks from calling the internal logic more than once? I suppose I could use locks and/or synchronized blocks to achieve this, but GCD seemed like a cleaner way to achieve this.
The cause of multiple callbacks is due to the phone book iCloud background synchronization. Usually, if you have multiple devices logged in a same iCloud account, the synchronization will propagate to all devices, and echoed back to your testing device from where the change originated, thus, causes the callback to be invoked multiple times.
By the way, using a timer to constraint the duplicated invocations won't help resolve this issue completely, because you don't know when the next callback will be called depending on your network condition. You should instead program the logic to handle these duplicated invocations.
I ended up using NSTimers instead of GCD to prevent the duplicate callbacks from firing my critical method. Much simpler, and works quite well!
[self.changeTimer invalidate];
self.changeTimer = nil;
self.changeTimer = [NSTimer scheduledTimerWithTimeInterval:3.0
target:self
selector:#selector(handleAdressBookExternalCallbackBackground)
userInfo:nil
repeats:NO];
Whatever you are trying to use GCD for, you are negating any of its effects, since you are creating a queue for each time the callback is being called, and that queue is different from the others, so it always runs. You probably mean creating the queue outside of the callback and using it inside the callback (maybe a static global?).
Still, I don't understand how would that help you, since you would still be running each GCD block each time a callback fired. Unless your do internal logic part marks a record as has having been updated and you check this flag in your queued methods affecting the same record, you are still going to be running multiple times your code, GCD or not.
Not really a direct answer to you GCD question, but I think that every time a unique 'context' is provided when you register, this creates a new 'registration' such that you are called back for each 'context'. You may be able to avoid being called multiple times by providing the same 'context'.
I had similar problem. My solution was to save flag in NSUserDefaults, enable this flag after first addressbookChanged method, and disable it again, after my actions were done..
void MyAddressBookExternalChangeCallback (ABAddressBookRef notifyAddressBook,CFDictionaryRef info,void *context)
{
NSLog(#"in MyAddressBook External Change Callback");
if([[[NSUserDefaults standardUserDefaults]objectForKey:#"addressBookChanged"] boolValue] == NO)
{
[[NSUserDefaults standardUserDefaults] setObject:#YES forKey:#"addressBookChanged"];
[[NSUserDefaults standardUserDefaults] synchronize];
//we save sync status to defaults to prevent duplicate call of this method
[[NSUserDefaults standardUserDefaults] setObject:[NSNumber numberWithBool:YES] forKey:#"addressBookSync"];
[[NSUserDefaults standardUserDefaults]synchronize];
[APICallWithCompletion:^(BOOL success, id object) {
[[NSUserDefaults standardUserDefaults] setObject:#NO forKey:#"addressBookChanged"];
[[NSUserDefaults standardUserDefaults] synchronize];
}];
}
}
While this may not be correct approach, it seems to be working for me, as my api call takes long enough to prevent duplicate call of this method... I guess you could replace it with
dispatch_after(dispatch_time(DISPATCH_TIME_NOW, 5 * NSEC_PER_SEC), dispatch_get_main_queue(), ^{
[[NSUserDefaults standardUserDefaults] setObject:#NO forKey:#"addressBookChanged"];
[[NSUserDefaults standardUserDefaults] synchronize];
});
I spent almost 2 days behind this issues. Even I was using timer but that was creating more issues. For eg. if you set timer for 5 sec and in that time duration if you go to contacts again and make some changes and come to the app it will end up ignoring that change because 5 secs are not over yet. so for that change you will have to kill the app and rerun the application.
I just did 2 steps and everything worked like magic
On
- (void)applicationDidEnterBackground:(UIApplication *)application
method I am registering to external changes
-(void) registerExternalChanges
{
dispatch_async(dispatch_get_main_queue(), ^{
ABAddressBookRef addressBookRef = [self takeAddressBookPermission];
ABAddressBookRegisterExternalChangeCallback(addressBookRef, addressBookChanged , (__bridge void *)(self));
});
}
And once you come to app after finish making changes in contacts database
UnRegisterExternalChanges
ABAddressBookUnregisterExternalChangeCallback(ntificationaddressbook, addressBookChanged,(context));
Thats it addressBookChanged method will only gets called once!!!
To execute a piece of code exactly once with the help of GDC, you can do:
static dispatch_once_t onceToken;
dispatch_once(&onceToken, ^
{
a piece of code
});
Related
Wiki Link for SDL_AddTimer
The wiki document for SDL_AddTimer claims that
"Note that it is possible to avoid the multithreading problems with SDL timers by giving to userevent.data1 the address of a function you want to be executed and to userevent.data2 its params, and then deal with it in the event loop."
How it is used to avoid multithreading problem?.
Can someone explain what is it i am unable to understand the statement ?
The first example assumes the working function, i.e. the function you want to execute ( my_function() ), resides in the my_callbackfunc() function.
SDL_AddTimer() specifies: Use this function to set up a callback function to be run on a separate thread after the specified number of milliseconds has elapsed.
This will introduce concurrency problems with my_function().
The solution (second example), assumes the event polling thread is the same thread that added the timer, and calls the function in that thread.
I've read that SDL documentation as well, and it makes a really bad suggestion for its "workaround". Specifically, it recommends casting a function address to a void pointer. This is not portable! Do not do it and please read
https://isocpp.org/wiki/faq/pointers-to-members#cant-cvt-fnptr-to-voidptr
If you feel that you must do this (or something like it) then I'd suggest wrapping the function pointer inside a struct/class.
struct Wrapper
{
void (*f)(void*);
Wrapper(void (*F)(void*)) { f = F; }
};
Create the wrapper when you want to push the custom event
SDL_Event event;
event.user.data1 = (void*) new Wrapper(your_function);
event.user.data2 = your_function_arg;
SDL_PushEvent(&user);
Then in your main loop, do the call, delete the wrapper
SDL_WaitEvent(&event);
if (event.type == SDL_USEREVENT)
{
Wrapper *p = ((Wrapper*) event.user.data1)
p->f(event.user.data2);
delete p;
}
(Pseudo-)Code
Here is a non-compilable code-sketch of the concepts I am having trouble with:
struct Data {};
struct A {};
struct B {};
struct C {};
/* and many many more...*/
template<typename T>
class Listener {
public:
Listener(MyObject* worker):worker(worker)
{ /* do some magic to register with RTI DDS */ };
public:
// This function is used ass a callback from RTI DDS, i.e. it will be
// called from other threads when new Data is available
void callBackFunction(Data d)
{
T t = extractFromData(d);
// Option 1: direct function call
// works somewhat, but shows "QObject::startTimer: timers cannot be started
// from another thread" at the console...
worker->doSomeWorkWithData(t); //
// Option 2: Use invokeMethod:
// seems to fail, as the macro expands including '"T"' and that type isn't
// registered with the QMetaType system...
// QMetaObject::invokeMethod(worker,"doSomeGraphicsWork",Qt::AutoConnection,
// Q_ARG(T, t)
// );
// Option 3: use signals slots
// fails as I can't make Listener, a template class, a QObject...
// emit workNeedsToBeDone(t);
}
private:
MyObject* worker;
T extractFromData(Data d){ return T(d);};
};
class MyObject : public QObject {
Q_OBJECT
public Q_SLOTS:
void doSomeWorkWithData(A a); // This one affects some QGraphicsItems.
void doSomeWorkWithData(B b){};
void doSomeWorkWithData(C c){};
public:
MyObject():QObject(nullptr){};
void init()
{
// listeners are not created in the constructor, but they should have the
// same thread affinity as the MyObject instance that creates them...
// (which in this example--and in my actual code--would be the main GUI
// thread...)
new Listener<A>(this);
new Listener<B>(this);
new Listener<C>(this);
};
};
main()
{
QApplication app;
/* plenty of stuff to set up RTI DDS and other things... */
auto myObject = new MyObject();
/* stuff resulting in the need to separate "construction" and "initialization" */
myObject.init();
return app.exec();
};
Some more details from the actual code:
The Listener in the example is a RTI DataReaderListener, the callback
function is onDataAvailable()
What I would like to accomplish
I am trying to write a little distributed program that uses RTI's Connext DDS for communication and Qt5 for the GUI stuff--however, I don't believe those details do matter much as the problem, as far as I understood it, boils down to the following:
I have a QObject-derived object myObject whose thread affinity might or might not be with the main GUI thread (but for simplicity, let's assume that is the case.)
I want that object to react to event's which happen in another, non-Qt 3rd-party library (in my example code above represented by the functions doSomeWorkWithData().
What I understand so far as to why this is problematic
Disclaimer: As usual, there is always more than one new thing one learns when starting a new project. For me, the new things here are/were RTI's Connext and (apparently) my first time where I myself have to deal with threads.
From reading about threading in Qt (1,2,3,4, and 5 ) it seems to me that
QObjects in general are not thread safe, i.e. I have to be a little careful about things
Using the right way of "communicating" with QObjects should allow me to avoid having to deal with mutexes etc myself, i.e. somebody else (Qt?) can take care of serializing access for me.
As a result from that, I can't simply have (random) calls to MyClass::doSomeWorkWithData() but I need to serialize that. One, presumably easy, way to do so is to post an event to the event queue myObject lives in which--when time is available--will trigger the execution of the desired method, MyClass::doSomeWorkWithData() in my case.
What I have tried to make things work
I have confirmed that myObject, when instantiated similarly as in the sample code above, is affiliated with the main GUI thread, i.e. myObject.thread() == QApplication::instance()->thread().
With that given, I have tried three options so far:
Option 1: Directly calling the function
This approach is based upon the fact that
- myObject lives in the GUI thread
- All the created listeners are also affiliated with the GUI thread as they are
created by `myObject' and inherit its thread that way
This actually results in the fact that doSomeWorkWithData() is executed. However,
some of those functions manipulate QGraphicsItems and whenever that is the case I get
error messages reading: "QObject::startTimer: timers cannot be started from another
thread".
Option 2: Posting an event via QMetaObject::invokeMethod()
Trying to circumvent this problem by properly posting an event for myObject, I
tried to mark MyObject::doSomeWorkWithData() with Q_INVOKABLE, but I failed at invoking the
method as I need to pass arguments with Q_ARG. I properly registered and declared my custom types
represented by struct A, etc. in the example), but I failed at the fact the
Q_ARG expanded to include a literal of the type of the argument, which in the
templated case didn't work ("T" isn't a registered or declared type).
Trying to use conventional signals and slots
This approach essentially directly failed at the fact that the QMeta system doesn't
work with templates, i.e. it seems to me that there simply can't be any templated QObjects.
What I would like help with
After spending about a week on attempting to fix this, reading up on threads (and uncovering some other issues in my code), I would really like to get this done right.
As such, I would really appreciate if :
somebody could show me a generic way of how a QObject's member function can be called via a callback function from another 3rd-party library (or anything else for that matter) from a different, non QThread-controlled, thread.
somebody could explain to me why Option 1 works if I simply don't create a GUI, i.e. do all the same work, just without a QGraphcisScene visualizing it (and the project's app being a QCoreApplication instead of a QApplication and all the graphics related work #defineed out).
Any, and I mean absolutely any, straw I could grasp on is truly appreciated.
Update
Based on the accepted answer I altered my code to deal with callbacks from other threads: I introduced a thread check at the beginning of my void doSomeWorkWithData() functions:
void doSomeWorkWithData(A a)
{
if( QThread::currentThread() != this->thread() )
{
QMetaObject::invokeMethod( this,"doSomeWorkWithData"
,Qt::QueuedConnection
,Q_ARG(A, a) );
return;
}
/* The actual work this function does would be below here... */
};
Some related thoughts:
I was contemplating to introduce a QMutexLocker before the if statement, but decided against it: the only part of the function that is potentially used in parallel (anything above the return; in the if statement) is--as far as I understand--thread safe.
Setting the connection type manually to Qt::QueuedConnection: technically, if I understand the documentation correctly, Qt should do the right thing and the default, Qt::AutoConnection, should end up becoming a Qt::QueuedConnection. But since would always be the case when that statement is reached, I decided to put explicitly in there to remind myself about why this is there.
putting the queuing code directly in the function and not hiding it in an interim function: I could have opted to put the call to invokeMethod in another interim function, say queueDoSomeWorkWithData()', which would be called by the callback in the listener and then usesinvokeMethodwith anQt::AutoConnection' on doSomeWorkWithData(). I decided against this as there seems no way for me to auto-code this interim function via templates (templates and the Meta system was part of the original problem), so "the user" of my code (i.e. the person who implements doSomeWorkWithData(XYZ xyz)) would have to hand type the interim function as well (as that is how the templated type names are correctly resolved). Including the check in the actual function seems to me to safe typing an extra function header, keeps the MyClass interface a little cleaner, and better reminds readers of doSomeWorkWithData() that there might be a threading issue lurking in the dark.
It is ok to call a public function on a subclass of QObject from another thread if you know for certain that the individual function will perform only thread-safe actions.
One nice thing about Qt is that it will handle foreign threads just as well as it handles QThreads. So, one option is to create a threadSafeDoSomeWorkWithData function for each doSomeWorkWithData that does nothing but QMetaMethod::invoke the non-threadsafe one.
public:
void threadSafeDoSomeWorkWithData(A a) {
QMetaMethod::invoke("doSomeWorkWithData", Q_ARG(A,a));
}
Q_INVOKABLE void doSomeWorkWithData(A a);
Alternatively, Sergey Tachenov suggests an interesting way of doing more or less the same thing in his answer here. He combines the two functions I suggested into one.
void Obj2::ping() {
if (QThread::currentThread() != this->thread()) {
// not sure how efficient it is
QMetaObject::invoke(this, "ping", Qt::QueuedConnection);
return;
}
// thread unsafe code goes here
}
As to why you see normal behaviour when not creating a GUI? Perhaps you're not doing anything else that is unsafe, aside from manipulating GUI objects. Or, perhaps they're the only place in which your thread-safety problems are obvious.
Starting from the standard Xcode Document-based Application w/ CoreData template on 10.7, I'm experiencing some frustrating behavior. I'm sure it's something simple that I'm overlooking.
Let's say in my NSPersistentDocument subclass, I have something like this, hooked up to a button in the window:
- (IBAction)doStuff:(id)sender
{
NSEntityDescription* ent = [[self.managedObjectModel entitiesByName] valueForKey: #"MyEntity"];
NSManagedObject* foo = [[[NSManagedObject alloc] initWithEntity: ent insertIntoManagedObjectContext: self.managedObjectContext] autorelease];
[self.managedObjectContext save: NULL];
}
If I create a new document and click that button, I'll get the following error: This NSPersistentStoreCoordinator has no persistent stores. It cannot perform a save operation. I get this. We haven't saved yet, there are no persistent stores. Makes sense.
Now let's say I split this out into two actions, hooked up to different buttons, like so:
- (IBAction)doStuff:(id)sender
{
NSEntityDescription* ent = [[self.managedObjectModel entitiesByName] valueForKey: #"MyEntity"];
NSManagedObject* foo = [[[NSManagedObject alloc] initWithEntity: ent insertIntoManagedObjectContext: self.managedObjectContext] autorelease];
}
- (IBAction)doOtherStuff:(id)sender
{
[self.managedObjectContext save: NULL];
}
If I create a new document and press the first button, then at some indeterminate time after pressing that button (dirtying the document), autosave will come along and autosave the document, which will create a store in the temp location. If I then press the second button, there are no complaints (since there's now a store.)
I need my document to be able to do managedObjectContext saves from the get-go. I'm kicking off some stuff on a background thread, and I need the background context's save operation (and notification) in order to merge the changes made by the background thread into the main thread's managedObjectContext.
I thought about trying to force an autosave, but the autosave process appears completely asynchronous, so I'd have to jump through hoops to disable any UI interaction that could cause a managedObjectContext save until the first autosave operation completes.
I also considered creating an in-memory store to bridge the gap between creating a new document and the first autosave, but then it's not clear to me how I would migrate that in memory store to the disk store and remove the memory store synchronously with the first autosave operation.
Anyone have any ideas on how I might handle this?
So I fooled around with this for a while, including trying #Aderstedt's suggestion. That approach didn't work, because faking a notification appears to merely tell the receiving context "hey, check with the persistent stores, I've updated them!", when in reality, I haven't, because there are none. I eventually found an approach that worked. Unfortunately it relies on Lion-only features, so I'm still looking for a way to do this that doesn't require Lion.
Background
I wanted to work with the NSPersistentDocument approach. Although I've not found this documented explicitly anywhere, I found several forum posts, and experienced a bunch of empirical evidence that you cannot call -[NSManagedObjectContext save:] on a context that belongs to an NSPersistentDocument. As mentioned in the question, if you call that before the document has ever been saved, it'll have no stores, so the save will fail. Even after the store exists, by saving the context directly (and not via the document save API) you are effectively changing the on-disk representation behind the NSPersistentDocument's back, and you will get the document popping sheets that say:
File has been modified by another application
In short, the NSPersistentDocument expects to control the saving action of the associated NSManagedObjectContext itself.
Also worth mentioning up front: the goal here was to make sure that the context used by the UI would trigger no (or at least minimal) I/O in order to remain responsive. The pattern I eventually settled on was to have 3 contexts. One context owned by the NSPersistentDocument, which would be responsible for doing file I/O in conjunction with the document. A second, effectively read-only, context for binding the UI to. (I realize that many folks want UI that mutates the model, so this may be less thrilling for them, but it wasn't a requirement for me.) And a third context for use on a background thread that asynchronously loads data from a web service, and hopefully pushes it into the other contexts so that it can be both saved on disk and presented in the UI without potentially blocking the UI on network I/O.
Lion-only Solution
The new parent/child NSManagedObjectContext feature in Lion's CoreData implementation is perfect for this. I replaced the NSPersistentDocument's NSManagedObjectContext with an new MOC of concurrency type NSPrivateQueueConcurrencyType. This will be the "root" context. Then I made the UI context with NSMainQueueConcurrencyType concurrency, and made it a child of the root context. Lastly I made the network-loading context a NSPrivateQueueConcurrencyType context which is a child of the UI context. The way this works is that we kick off a network load operation in the background, it updates the network context. When it's done, it saves the context. With parent/child relationships, saving a child context pushes the changes up into the parent context (the UI context) but does not save the parent context to the store. In my case, I also listen for the NSManagedObjectContextDidSaveNotification notification from the network context and then tell it's parent to save as well (which will push the changes from the UI context into the root/disk context, but will not save it to disk.)
At the end of this chain of events all the contexts are consistent, and we still haven't forced a real save of the underlying root context, and so we haven't run afoul of the NSPersistentDocument in its role of managing the on-disk representation.
One catch is that if you want to prevent saves of child contexts from generating undo (i.e. this was a network loading operation, there's nothing to undo) you have to disableUndoRegistration on each parent context as you propagate the changes up the chain.
Pre-Lion Efforts
I would really have liked to find a pre-Lion-compatible solution for this issue. I tried a few things before giving up. I first tried associating an in-memory store with the PSC on document init, so that I could do NSManagedObjectContext saves before the document saved, then migrate the in-memory store on the first save. That part worked great. But once an on-disk store existed, this approach was bogus because after it's saved to disk we have the same problem where any saving of MOCs connected to PSCs owned by an NSPersistentDocument have to be done by the document.
I also tried hacking up a mechanism to move changes from one context to another using the NSManagedObjectContextObjectsDidChangeNotification payload. Although I was able to get this to work (for some nominal definition of 'work'), I saw big problems looming on the horizon with this approach. Specifically, it's easy to migrate those changes once but what if it changes again before a save operation? Then I'd be stuck maintaining a long lived mapping of OIDs in the source context to OIDs in the destination context(s). This got ugly really fast. If anyone's interested, here's what I came up with:
#interface NSManagedObjectContext (MergeChangesFromObjectsDidChangeNotification)
- (void)mergeChangesFromObjectsDidChangeNotification: (NSNotification*)notification;
#end
#implementation NSManagedObjectContext (MergeChangesFromObjectsDidChangeNotification)
- (void)mergeChangesFromObjectsDidChangeNotification: (NSNotification*)notification
{
if (![NSManagedObjectContextObjectsDidChangeNotification isEqual: notification.name])
return;
if (notification.object == self)
return;
NSManagedObjectContext* sourceContext = (NSManagedObjectContext*)notification.object;
NSAssert(self.persistentStoreCoordinator == sourceContext.persistentStoreCoordinator, #"Can't merge changes between MOCs with different persistent store coordinators.");
[sourceContext lock];
// Create object in the local context to correspond to inserted objects...
NSMutableDictionary* foreignOIDsToLocalOIDs = [NSMutableDictionary dictionary];
for (NSManagedObject* foreignMO in [[notification userInfo] objectForKey: NSInsertedObjectsKey])
{
NSManagedObjectID* foreignOID = foreignMO.objectID;
NSManagedObject* localMO = [[[NSManagedObject alloc] initWithEntity: foreignMO.entity insertIntoManagedObjectContext: self] autorelease];
[foreignOIDsToLocalOIDs setObject: localMO.objectID forKey: foreignOID];
}
// Bring over all the attributes and relationships...
NSMutableSet* insertedOrUpdated = [NSMutableSet set];
[insertedOrUpdated unionSet: [[notification userInfo] objectForKey: NSInsertedObjectsKey]];
[insertedOrUpdated unionSet: [[notification userInfo] objectForKey: NSUpdatedObjectsKey]];
for (NSManagedObject* foreignMO in insertedOrUpdated)
{
NSManagedObjectID* foreignOID = foreignMO.objectID;
NSManagedObjectID* localOID = [foreignOIDsToLocalOIDs objectForKey: foreignOID];
localOID = localOID ? localOID : foreignOID;
NSManagedObject* localMO = [self objectWithID: localOID];
// Do the attributes.
[localMO setValuesForKeysWithDictionary: [foreignMO dictionaryWithValuesForKeys: [[foreignMO.entity attributesByName] allKeys]]];
// Do the relationships.
NSDictionary* rByName = foreignMO.entity.relationshipsByName;
for (NSString* key in [rByName allKeys])
{
NSRelationshipDescription* desc = [rByName objectForKey: key];
if (!desc.isToMany)
{
NSManagedObject* relatedForeignMO = [foreignMO valueForKey: key];
NSManagedObjectID* relatedForeignOID = relatedForeignMO.objectID;
NSManagedObjectID* relatedLocalOID = [foreignOIDsToLocalOIDs objectForKey: relatedForeignOID];
relatedLocalOID = relatedLocalOID ? relatedLocalOID : relatedForeignOID;
NSManagedObject* localRelatedMO = [self objectWithID: relatedLocalOID];
[localMO setValue: localRelatedMO forKey: key];
}
else
{
id collection = [foreignMO valueForKey: key];
id newCollection = [NSMutableSet set];
if ([collection isKindOfClass: [NSOrderedSet class]])
{
newCollection = [NSOrderedSet orderedSet];
}
for (NSManagedObject* relatedForeignMO in collection)
{
NSManagedObjectID* relatedForeignOID = relatedForeignMO.objectID;
NSManagedObjectID* relatedLocalOID = [foreignOIDsToLocalOIDs objectForKey: relatedForeignOID];
relatedLocalOID = relatedLocalOID ? relatedLocalOID : relatedForeignOID;
NSManagedObject* localRelatedMO = [self objectWithID: relatedLocalOID];
[newCollection addObject: localRelatedMO];
}
[localMO setValue: newCollection forKey: key];
}
}
}
// And delete any objects which pre-existed in my context.
for (NSManagedObject* foreignMO in [[notification userInfo] objectForKey: NSDeletedObjectsKey])
{
NSManagedObjectID* foreignOID = foreignMO.objectID;
NSManagedObject* localMO = [self existingObjectWithID: foreignOID error: NULL];
if (localMO)
{
[self deleteObject: localMO];
}
}
[sourceContext unlock];
}
#end
Conclusion
Between the improvements in concurrency management and this parent/child feature, I quickly lost interest pursuing a pre-Lion solution. I'm beginning to gather that the pre-Lion solution would effectively be "Don't use NSPersistentDocument." As best I can tell, all these pain points go away if I drop that requirement. Without that, you can save contexts and migrate stores whenever you want, but naturally you would have to do all that work yourself.
The problem here is that the default NSManagedObjectContext that is created by the NSPersistentDocument is of concurrency type NSConfinementConcurrencyType. CoreData does not allow to create a child context of a context with that type.
As a workaround, this is working for me. The NSManagedObjectContext is created by your NSPersistentDocument, so you can override that method:
- (NSManagedObjectContext *)managedObjectContext {
if (!_context) {
NSManagedObjectContext *_default = [super managedObjectContext];
_context = [[NSManagedObjectContext alloc] initWithConcurrencyType:NSMainQueueConcurrencyType];
_context.persistentStoreCoordinator = _default.persistentStoreCoordinator;
}
return _context;
}
I had no idea where to get the persistentStoreCoordinator from, so I called the super implementation and get it there from. With this context you should be able to create child context from that you can use in background operations.
If you don't have a store then you can't save. It seems that you want to save the document in order to merge the changes made on a background thread; well, you can merge those changes manually. When the background thread completes, tell the main thread which objects have been updated / inserted and then make the same changes on the main thread.
If the changes are near arbitrary and thus tedious to duplicate, you can even construct your own NSManagedObjectContextDidSaveNotification on the background thread and then merge it using -[NSManagedObjectContext mergeChangesFromContextDidSaveNotification:] on the main thread.
I found a very nicely written-up solution along the lines of ipmcc's solution (also running the NSPersistentDocument's MOC in a background thread, and making it the parent context of the main thread's MOC) here.
It includes the full BSD-licensed code for a document app based on that (basically a Mac OS version of iOS's UIManagedDocument).
I have a Silverlight application that uses WCF services and also uses the Wintellect Power Threading library to ensure logic executes fully before the application continues. This is achieved by calling back to the application using delegates so it can continue after the service call has completely finished.
I wish to achieve the same thing in another part of my application but without the use of callbacks e.g. call method that uses WCF service to say load an object from the database, wait for this to return and then return the Id of the object from the original method called.
The only way I could see to do this was to carry out the call to the WCF service in a helper library which loads the object on a different thread and the original method would keep checking the helper library (using static variables) to wait for it to complete and then return it.
Is this the best way to achieve this functionality? If so here are details of my implementation which is not working correctly.
public class MyHelper
{
private static Thread _thread;
private static User _loadedObject;
public static GetUser()
{
return _loadedObject;
}
public static void LoadObject(int userId)
{
_loadedObject = null;
ParameterizedThreadStart ts = new ParameterizedThreadStart(DoWork);
_thread = new Thread(ts);
_thread.Start(userId);
}
private static void DoWork(object parameter)
{
var ae = new AsyncEnumerator();
ae.BeginExecute(DoWorkWorker(ae, Convert.ToInt32(parameter)), ae.EndExecute);
}
private static IEnumerator<Int32> DoWorkWorker(AsyncEnumerator ae, int userId)
{
// Create a service using a helper method
var service = ServiceHelper.GetService<IUserServiceAsync>();
service.BeginGetUserById(userId, ae.End(), null);
yield return 1;
_loadedObject = service.EndGetUserById(ae.DequeueAsyncResult());
_thread.Abort();
}
}
My method then is:
public int GetUser(int userId)
{
MyHelper.LoadObject(userId);
User user = MyHelper.GetUser();
while (user == null)
{
Thread.Sleep(1000);
user = MyHelper.GetUser();
}
return user.Id;
}
The call to the get the user is executed on a different thread in the helper method but never returns. Perhaps this is due to the yield and the calling method sleeping. I have checked the call to get the user is on a different thread so I think everything should be kept separate,
The whole construct you are using does not match current best practices of Silverlight. In Silverlight your data access methods (via WebServices of course) are executed asynchronously. You should not design around that, but adapt your design accordingly.
However calling services sequentially (which is different than synchonously) can be valid in some scenarios. In this blog post I have shown how to achieve this by subscribing the Completed event of the remote call and block the UI in the meantime, with which the workflow looks and feels like normal async calls.
I believe calls to the server from Silverlight apps use events that fire on the UI thread; I think that's part of the Silverlight host environment in the browser and can't be worked around. So trying to call back to the server from another thread is never going to end well. If you are waiting in program code in the UI thread, your never going to get the call result events from your WCF calls.
You can simulate a synchronous call from a non-UI thread with a callback on the UI thread, but that is probably not what you want. It's better to bite the bullet and make your program logic work with the async calls Silverlight gives you.
If you code against the Interface created for your service reference you can call the Begin and End methods 'synchronously' for each one of your service calls, we then pass in an Action<T> to execute after the End methods has completed. Take note that you have to do this from a dispatcher. This is very close to making a synchronous call as the code to run after the call is still written where the call is made, and it executes after the service call is completed. It does however involve creating wrapper methods but we also worked around that by hiding our wrappers and generating them automatically. Seems like a lot of work but isn't, and ends up being more elegant than all the event handlers etc. Let me know if you need more info on this pattern
My app must run on iOS 3.2 and the methods such as -addOperationWithBlock: only work in > 4.0.
But NSOperationQueue was available since iOS 2.0 so I'd like to give it a try the "old way". Does anyone know of a handy tutorial that shows the basics of how to use NSOperationQueue without blocks?
it's pretty straightforward with Invocation operations. These are operations that allow you to send a message to a particular object with some object parameter (optional).
So given this method that you want to invoke:
- (void)doSomething {
NSLog (#"Did it!");
}
You can do something like this to make it happen:
// Get or create some queue
NSOperationQueue *someQueue = [NSOperationQueue mainQueue];
// create an invocation operation
NSInvocationOperation *invocationOp = [[NSInvocationOperation alloc] initWithTarget:self
selector:#selector(doSomething)
object:nil];
[someQueue addOperation:invocationOp]; // Add the operation to the queue
[invocationOp release];
Hope that helps.
#Firoze Lafeer gave an example with NSInvocation operation, but you can also use your own NSOperation subclass.
The official documentation shows with example every type of operation you can use. Even with blocks available, it's sometime preferred to use NSOperation subclass for bigger tasks.
Found a pretty good tutorial here. It also goes beyond the topic and gives info on why it's not always a good idea to fetch data on the main thread.