Re: [Vala] Problem with programmatic generic object instantiation



So this is what I now have:

namespace ObjectFactory {

  [CCode (has_target = false)]
  public delegate Object Maker();

  Gee.HashMap<Type,Maker> construct_map;

  public  void add_maker(Type for_type, Maker maker) {
    if (construct_map == null) {
      construct_map = new Gee.HashMap<Type,Maker>();
    }
    construct_map[for_type] = maker;
  }

  public  Object? make(Type one_of_these) {
    if (construct_map != null && construct_map.has_key (one_of_these)) {
      Maker m = construct_map[one_of_these];
      return m ();
    }
    return null;
  }

}

and then in each instantiation:

public class IntArr : GArrSerializable<int> {

  class construct {
    ObjectFactory.add_maker(typeof (IntArr), () => { return new IntArr ();
});
  }

  public IntArr(int size = 0) {
    base (size);
  }
}

then I use ObjectFactory.make (type) in my deserialization procedure, and
it appears to work ok at last.

On Wed, Feb 17, 2016 at 10:26 AM, Andy Lees <andrewl oz gmail com> wrote:

Momentarily I thought that might work, but it does match the Json
serialization structure - the context of deserializing an array does not
provide the initialised array, the deserialization procedure has to do the
creation, and only has the type id to work with.  It looks like I'll have
to create an object factory of my own for this, as I don't want to get into
the internals of Json glib.

On Wed, Feb 17, 2016 at 10:09 AM, Daniel Espinosa <esodan gmail com>
wrote:

I run in the same issue for GXml and SerializableListArray<>, for example.

The way I resolve, in my case, I've defined a new Interface called
SerializableCollection, with a init() function. This function is called by
Serializable.deserialize() method in order to help implementators to
initialize its generic containers by calling MyObjectCollection.new()
providing any information it needs like the Generic Type it should work on.

In my other project, I've just initialize any public property as

MyProperty<int> prop { get; set; default = new MyProperty<int>(); }

or inside construct {}

The point is to call MyProperty.new() not Object.new() on generic classes.


2016-02-16 16:59 GMT-06:00 Andy Lees <andrewl oz gmail com>:

Yes, the problem is that for types derived from generics (and for the
generic constructor itself), the type information is not set in the
construct function, so very little can be done.

On Wed, Feb 17, 2016 at 9:53 AM, Daniel Espinosa <esodan gmail com>
wrote:

If you want to execute code on Object creation using Object.new(), you
should use

construct {
  // Your code here
}

When define code at MyObject.new(), you execute the code in this method
not Object.new(), then use the above should help.


2016-02-16 15:58 GMT-06:00 Andy Lees <andrewl oz gmail com>:

The problem with Object.new is that it doesn't call the object
constructor so far as I can tell, just calling g_object_new without any of
the Vala object construction code.

My Serializable is an interface.  I'm trying to make use of the Json
serialization framework, and for the most part it's working.  The problem
I'm having is with the construction of instances of array types - there
appears to be no factory framework for object creation in Vala, and so far
the only workaround I've come up with is very clunky.

Is there an object factory somewhere that someone could point me
towards?

On Wed, Feb 17, 2016 at 1:49 AM, Daniel Espinosa <esodan gmail com>
wrote:

Try use Object.new(typeof(IntArr))

Try to rename type() to any other, to avoid conflics with
Object.type()

You are trying to implement Gee.ArrayList<>, may you:

A) wrap Gee.ArrayList in your class

B) Derive directly from Gee.Arraylist then convert your base class
Serializable as an interface with all required methods implemented as
virtual, then you have two bases.

You may want to check at GXml.Serializable interface and its
GXml.SerializableArrayList implementation based on Gee.
El feb. 16, 2016 7:10 AM, "Andy Lees" <andrewl oz gmail com>
escribió:

I would have expected the generic type to be obtained from the type
id of the derived type (generic instantiation), which is obtained fresh
each execution, but it seems not.  Is the generic type a hidden property of
the class derived from the generic class, and if so, what would its name be
so that it can be set in the new call?

Alternatively, is there a way to call an object constructor given
the type id?

On Tue, Feb 16, 2016 at 11:16 PM, Daniel Espinosa <esodan gmail com>
wrote:

Take in account that Type system is dynamically created, then a
type number will change with each execution.

Use

YourCreatedObject is typeof (Object)

to check if the created object is of type you need, like a derived
object.

You can use

obj.get_type().is_a(typeof (Object))

And consider that your example is exactly the way to create an
object of Generic class, because you have set tje object type. Create a
generic instance with Object.new you should provide the object type.
El feb. 16, 2016 5:14 AM, "Andy Lees" <andrewl oz gmail com>
escribió:

Hi,

I have a generic array implementation like so:

public class GArrSerializable<T> : Serializable, Object  {
  public T[] data;
  public int length;
  public GArrSerializable(int size = 0) {
    data = new T[size];
  }

  public Type type () { return typeof (T); }

...etc

and a derived type:

public class IntArr : GArrSerializable<int> {

  public IntArr(int size = 0) {
    base (size);
  }
}

If I create an instance of the type, then instance.type () returns
24 as
expected.
If I create an instance like so:
var ia = Object  new (type) as IntArr;
var t = ia.type ();

Then t is 4 rather than 24.  As it is in the instance when it
tries to
deserialize.

Can you tell me why this is so, and how I can programmatically
create
objects derived from a parameterised type?

Thanks

Andrew Lees
_______________________________________________
vala-list mailing list
vala-list gnome org
https://mail.gnome.org/mailman/listinfo/vala-list






--
Trabajar, la mejor arma para tu superación
"de grano en grano, se hace la arena" (R) (en trámite, pero para los
cuates: LIBRE)





--
Trabajar, la mejor arma para tu superación
"de grano en grano, se hace la arena" (R) (en trámite, pero para los
cuates: LIBRE)





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]