Hi and thanks Kyle, Fritz, Jens and everyone else.... On 2 במאי 2013, at 20:31, Kyle Sluder <k...@ksluder.com> wrote:
> On Wed, May 1, 2013, at 05:10 AM, Motti Shneor wrote: >> Hello. >> >> I have two NSObject subclasses - say MyA and MyB. >> >> - Each MyA instance. creates and owns a MyB instance. >> - MyB instances create an NSThread, and live their asynchronous life, >> communicating with a remote internet server, and sometimes with their >> owner (the MyA object) who lives in the main thread. > > Dumb question: can you not rely on Cocoa's built-in asynchronous > networking APIs? Unfortunately, MyB wraps in Obj-C a (rather weird) 15 year old C++ socket implementation which is nonstandard in many ways, and therefore cannot use the nicer and more modern asynchronous networking APIs. > >> What's the best way to get out of this? How can I simply synchronize the >> "release" calls, without blocking both threads? Maybe I should restrict >> myself to retaining/releasing in the main thread only? Is there some >> syntactic Obj-C sugar to ease such things? > > In such a circumstance, I would normally require MyA to use a factory > method to construct instances of MyB, and have that factory method add > the MyB instances to an object pool. When MyB operations are cancelled, > they perform their cleanup and message the thread they were created on > to remove themselves from the pool. Using dispatch queues should > simplify this quite a bit. > That doesn't change the problem much, because now the "object pool" will have the same issues as MyA. The problem is in the multithreaded allocation/deallocation domain. However, I think I found a good workaround the problem. If MyB calls [self retain] just BEFORE detaching its thread-function, and then calls [self release] AFTER the thread exits, and the thread-function returns, then all retain and release (hence dealloc) calls happen in the main-thread, and thus --- synchronized. I think I solved the problem. but now I'll exercise this code heavily, to see it is good. Last word --- I wonder why it isn't clearly written for every cocoa API (and especially for the very fundamental initialize/alloc/init/release/dealloc methods whether or not they're reentrant! In a multithreaded environment, that's a reasonable requirement from the documentation. Thanks again! Motti Shneor, Mac OS X Software Architect & Team Leader Spectrum Reflections Ltd. _______________________________________________ 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
