Optimizing ARC - weakness of [weak]

-
In Delphi, the [weak] attribute is used to break strong reference cycles between reference counted object instances. It can be used with the ARC compiler on both object and interface references, and with the classic compiler for interface references. Weak interface references were introduced in Delphi 10.1 Berlin.

Now, there is a teeny-weeny problem with the [weak] attribute. It denotes a zeroing weak reference that will be zeroed (niled) when the object it points to is no longer valid. In order to do that, the compiler has to track such objects at runtime and that introduces some overhead. If you are tracking many such references, that can introduce a significant performance loss.

Not all weak references that serve the purpose of breaking reference cycles need to be zeroed out. If you can guarantee that a [weak] reference will never outlive the object it points to, all that tracking and zeroing is useless. To avoid a performance penalty in such cases, you can use the [unsafe] attribute or a plain pointer.

While figuring out whether you can use [unsafe] instead of [weak] in ARC code is not always straightforward, there are some cases where you can be sure [unsafe] will work perfectly.
Since the classic compiler does not support the [weak] attribute for object references, any cross-platform code that needs to break strong reference cycles on the ARC side can safely use the [unsafe] attribute instead. Plain and simple.

Unfortunately, Delphi frameworks sometimes use [weak] attribute in places (cross-platform code) where they could use [unsafe], making such code less performant.

I have filed a general QP report for that issue https://quality.embarcadero.com/browse/RSP-20087

However, more specific bug reports that point to particular code that is experiencing a performance loss due to usage of the [weak] attribute might be resolved sooner. So if you know about such code, please file additional reports.

Comments

Post a Comment

Popular posts from this blog

Coming in Delphi 12: Disabled Floating-Point Exceptions

-
Delphi 12 is in the works. One of its small features that will have a huge impact is a change in the default handling of floating-point exceptions, which will now be masked on all platforms. Previously, floating-point exception handling was different across platforms, and most notably on the Windows platform: In VCL and console applications, floating-point exceptions were not masked and would raise an exception. On the other hand, FireMonkey applications had all floating-exceptions masked by default, regardless of which platform they were running on. Basically if you haven't explicitly changed the exception mask in your code to some value other than the default, and you had code that attempted to divide some number with zero, you would get an EZeroDivide exception in previous Delphi versions on the Windows VCL application. In Delphi 12, there is no exception, and the result of such a division will be +INF . So the following code will no longer raise an exception. If you were r
Read more

Beware of loops and tasks

-
Loops and anonymous methods don't go hand in hand. Any variable that changes for each iteration will cause problems if it is used within an anonymous method. I explained the mechanism behind variable capturing in the post Mysterious Case Of Wrong Value , along with the problem it causes with loops and its solution. Commonly, anonymous methods are used as callbacks or event handlers, where there are no loops in the context where variable capture happens. While there is a possibility to have an unexpected—wrong—value in such scenarios, it is rather easy to debug them and understand the order in which some code is called, and why the captured value is wrong in a given place. With the rise of multi-threading usage, and especially the Parallel Programming Library and tasks, writing anonymous methods inside a loop becomes more common, and it is much easier to write code where a captured variable will have the wrong value. The biggest problem with such code is that it will be harder to
Read more

Catch Me If You Can - Part II

-
It has been quite some time since I wrote the Catch Me If You Can post about the inability of LLVM-backed compilers to catch non-call hardware exceptions. In other words, hardware exceptions raised by code written in the same block as the try...except exception handler. Since then, more LLVM-backed compilers have been added to Delphi, and with those, more reasons to change common coding practices... but still, very few developers know about the issue. The original article covered only try...except exception handlers, and exploring what exactly happens with implicit and explicit try...finally handlers was left as an exercise to the readers and to some future, now long overdue, post. And Now: All Hell Breaks Loose When you read about non-call hardware exceptions not being caught by an immediate try...except block, the implications might not look too serious. After all, while plenty of code can raise exceptions, try...except handlers are not that frequently used in places, as mo
Read more