Why can't tracking references be used as class members? - c++-cli

The following code:
ref class A
{
private:
int% referenceToAnInt;
};
produces the following error in Visual Studio:
error C3160: 'int %' : a data member of a managed class cannot have this type
I thought tracking references were a managed thing - so why can't they be a member of a managed class?
Also: How would I store a reference to a value type in C++/CLI correctly?

The CLR doesn't allow storing tracking references as fields. Also, from the C++/CLI-Spec:
A program containing a tracking reference that has storage duration
other than automatic is ill-formed. (This precludes having a tracking
reference as a data member.)
I guess they wanted to avoid the problem where you keep a reference longer than the actual lifetime of the referenced object. An alternative would be to use a wrapper ref class to hold the value, or delegates for reading / writing.

Tracking references are “managed thing” which is why you can't use them as a member of a class. They are similar to C++ & references in that you can use them to refer to local variables in some other method. But local variables are stored in the short term store (a.k.a. the stack). That means that when the method that contains that local variable ends, the reference wouldn't be valid anymore, but the object could still exist.
This would not be memory-safe, which is why doing this is forbidden in .Net.

Related

What's the difference between an object and a data object?

The other day I noticed that I sometimes put data in front of objects and other times not:
object A
data object B
What's the difference between an object and a data object?
The fact that data is allowed on an object declaration is in fact a bug (KT-6486) which should be fixed.
data is an annotation which causes the compiler to generate equals, hashCode, toString, copy and componentN functions. It doesn't make much sense when applied to an object declaration for two reasons:
An object declaration cannot have a constructor, and all these functions work based on properties defined in the primary constructor.
There's only one instance of any object at runtime.
So no componentN functions would be generated, copy can't work, and the generated equals/hashCode/toString implementations will be equivalent to the default ones from Any which are based on identity.

Should static reference type variable be avoided? [duplicate]

This question already has answers here:
Are static local variables bad practice?
(2 answers)
Closed 8 years ago.
Consider the following functionally two code snippets in a single-threaded environment. Assuming there are no other methods in Foo I believe these are functionally identical.
Class Foo
Private _Bar As Bar
Public ReadOnly Property GetBar As Bar
Get
If IsNothing(_Bar) Then
_Bar = New Bar
End If
Return _Bar
End Get
End Property
End Class
And
Class Foo
Public ReadOnly Property GetBar2 As Bar
Get
Static _Bar As New Bar
Return _Bar
End Get
End Property
End Class
Today I was challenged on code following the 2nd method because "the New will be called each time". I already know that is false, but the primary objection was with regards to the use of Static. I found several references to Static variables indicating that they may be dangerous, but they were all talking about Java. However, I was not able to find any good explanations as to why.
How are these two methods different? Is the 2nd method dangerous? If so, why?
Static in VB.Net is not that same as static in Java, C#, C, or C++. VB.Net's analog to that construct is Shared. The documentation on the Static keyword is here:
http://msdn.microsoft.com/en-us/library/z2cty7t8.aspx
In particular, I'd like to point out this snippet:
Behavior
When you declare a static variable in a Shared procedure, only one copy of the static variable is available for the whole application. You call a Shared procedure by using the class name, not a variable that points to an instance of the class.
When you declare a static variable in a procedure that isn't Shared, only one copy of the variable is available for each instance of the class. You call a non-shared procedure by using a variable that points to a specific instance of the class.
It's likely the objection comes from believing that Static always behaves like the first paragraph, even in instance methods, when we can see here that it's clearly documented that this is not the case.
Instead, Static allows you to declare a variable whose lifetime-scope is that of the class instance, but whose access-scope is limited to a single method. It's a way to narrow the potential scope of a variable, and therefore is a good thing. Additionally, variables declared as Static are rewritten by the compiler to be protected via the Monitor class (at least for the Shared version), giving them a measure of thread-safety. In other words, a variable declared as Static is more likely to have any needed locking done verses a similar class-scoped variable.
In this particular case, though, I fail to see the point. You don't really gain anything beyond an auto-implemented property like this:
Public ReadOnly Property GetBar2 As New Bar()
This probably is confusing the VB.net concepts of Static and Shared because some languages use the keyword Static to mean what VB uses Shared for: a variable/field/property/method that is shared or common to all instances of a class.
But Static doesn't mean that in VB. Instead it means a routine-local variable that persists beyond the invocation of the routine (i.e., its lifetime is object-scoped rather than routine invocation-scoped).
REF: http://msdn.microsoft.com/en-us/library/z2cty7t8.aspx
So in VB, Static means "routine-scoped visibility, object-scoped lifetime".
Whereas Shared means "class-scoped visibilty, class/program-scoped lifetime".
I would avoid the second approach if for no other reason than the fact that C and C# have a static keyword whose meaning is totally different from that of the VB.NET Static keyword. I generally dislike language features which look like features of other languages but aren't. If it's necessary to use a language feature despite its unfortunate resemblance to the other language's feature, I'll use it, but the VB.NET static keyword doesn't really add much here. Effectively, it asks the compiler to make the variable Private field, give it an arbitrary name which differs from that of any other field, and replace all references to the variable's given name within the method with references to the invented name.
Conceptually, use of such "localized" fields may be regarded as dubious because while one may expect that a field will only need to be used within one method, that may turn out not to be true. I wouldn't worry too much about that issue in vb.net, however, because a Static variable may easily be turned into an ordinary private field if the need arises. If when that need does arise a field exists with the same name, one may easily rename the Static variable before moving it.

ABAP Objects - access message class of a class?

In the header information of an ABAP Objects class, I can enter a message class to use with the MESSAGE statement. This works like the MESSAGE-ID statement of a report or a function pool. Since I can't find the message class I entered in the header data anywhere in the generated sections, I assume that it's generated into the top-level CLASS-POOL statement somewhere.
For some libraries (for examples, the BAL application logging), it's necessary to specify the message class using a variable or a method parameter. Up to now, I've defined a constant that specified the message class and used that constant. I'm wondering if it's possible to access the message class specified in the header data in some other way so that I can get rid of that redundant variable.
Has anyone found a way to do so?
EDIT: The new way should be easier than the old one - I'm not crazy enough to add a CLASS-CONSTRUCTOR and perform some database access or SEO_* function calls just to get rid of that constant.
I think you need a CLASS-CONSTRUCTOR to set a class attribute with the message class.
The MESSAGE statement with INTO clause has the side effect of setting the SY- system variables. So you could put into your CLASS-CONSTRUCTOR something like:
DATA: lf_dummy TYPE string.
MESSAGE s999 INTO lf_dummy.
af_msgid = sy-msgid.
You could use the class builder API:
data the_class type ref to cl_oo_class.
create object the_class
exporting
clsname = `ZCL_SOMECLASS`.
data message_class type arbgb.
message_class = the_class->class-msg_id.
I haven't come across any syntax to do what you ask. For the reasons I outline below, I could believe that SAP never saw a need to include such functionality.
In my experience, the message class is an attribute of the message, not of the object that raises it, so it should be kept together with the type, number, and variables of the message. For example if my object is returning the number of an error, it should be returning the id (class) as well.
In this light I cannot see a reason why you would ever need to know the message-class assigned to an ABAP-OO class, you would only ever need to know the message-class of the messages returned by the ABAP-OO class.
The way I usually manage this is to raise my messages into a dummy field, and then use a subroutine to populate the contents of the sy-msg* fields into a BAPIRETURN structure. Then I return this BAPIRETURN structure to the caller. This way the type, id, number, and variables of the message are all kept together.

Casting Early bound object type into Late bound object type

I've got a piece of code in a project (MyProject) that contains early bound object from a referenced assembly (We'll call it CommonAssembly):
Dim myObject As CommonAssembly.MyEarlyBoundType
now I have another assembly that is dynamically loaded because it is not present in all projects:
Dim myLateBoundObject As Object = AppDomain.CurrentDomain.CreateInstanceAndUnwrap("Utils", "Utils.MyLateBoundType")
MyLateBoundType derives from CommonAssembly.MyEarlyBoundType, and I want to cast myObject to myLateBoundObject and then programmatically invoke the additional member methods
via reflection.
I'd have thought that by extracting the type of myLateBoundOject and casting myObject into myLateBoundObject, that'd work, but both CType() and DirectCast() methods won't accept the extracted type citing a "Keyword does not name a type" error:
myLateBoundObject = DirectCast(myObject, GetType(myLateBoundObject))
I'm not entirely sure why a dynamically loaded type cannot be used against the DirectCast (pretty sure it's not type saftey checked?) method, since if Utils.MyLateBoundType was referenced in "MyPrjoect", I could execute:
myLateBoundObject = DirectCast(myObject, Utils.MyLateBoundType)
without any problems - but this is not a dynamic solution.
Any suggestions?
Cheers,
Yum.
Yeah, that's not going to work like that. Any attempt you'd make in your code to cast to the dynamically loaded type will make your program have a non-dynamic dependency on the assembly.
You should use an interface type. Declare that type, with all the properties and methods you'd want to have available in your main program, in a separate assembly. Both your main program and your plug-in will have a dependency on it. The dynamic type should inherit it to provide the implementation. You can now cast the return value of CreateInstance to that interface type.

ABAP create object

Below is a code snippet that is creating object.
Form userexit_save_document_prepare.
data: /bks/exitmanager type ref to /bks/exit_manager.
create object /bks/exitmanager
exporting main_prog = 'SAPMV45A'
exit_form = 'USEREXIT_SAVE_DOCUMENT_PREPARE'.
include /bks/exitman.
ENDFORM.
I got this from the documentation
For performance reasons, the parameters "main_prog" and "exit_form" should be filled, in the case of userexits, which are performed very often like "user_field_modification" in "SAPMV45A" which is called for every single screen-field.
1) What happened exactly behind when create object /bks/exitmanager is called? memory allocated for the object etc?
2) Why for performance reasons the exporting parameters of create object needs to be filled?
I'm not 100% sure, but here is my best guess:
an object called /bks/exitmanager is constructed (which is an oject of the class /bks/exit_manager or more specific a reference/"pointer" to an object of this class) .. the required memory allocated etc., but also the "constructor" code is called (probably sets some instance variables as passed to the call).
If you're explicitly passing these parameters, they don't have to be "calculated" at run-time (e.g. by looking at the call stack). This should save some time, especially if it would have to be done quite often (as described in the documentation).
It would help to see what /bks/exit_manager actually is, and a brief explanation of what you are trying to accomplish.
Expanding on what IronGoofy wrote:
data: /bks/exitmanager type ref to /bks/exit_manager
This creates a reference pointer in the ABAP memory of your program, much like a field symbol. Also it must be already delared. If it is in the include, you need to move the include.
create object /bks/exitmanager
exporting main_prog = 'SAPMV45A'
exit_form = 'USEREXIT_SAVE_DOCUMENT_PREPARE'.
This creates an object instance based on the declared class, and assigns it to the referance pointer. It does this by calling the constructor method first.
Only by examing /bks/exit_manager will you find out exactly what you need to export.
It's impossible to tell what's going on and why the parameters should be passed without taking a look at the constructor of /BKS/EXIT_MANAGER. It's a common pattern though to keep a buffer of settings (think of a static hashed table with the key being the parameters and the value holding whatever is complicated and time-consuming to fetch). In this case, I would have expected a protected constructor that cannot be accessed directly, but only using a static factory method that uses a hashed table to keep the references of the exit handler itself - optimally using weak references...