Is the absence of properties on Objects an omission by design or simply because no-one has gotten to it yet?

Similarly case statements as they appear in records are not supported. I would quite like to have the functionality to do some sort of union on a data structure that also supported (non-virtual) methods and properties.

I completely understand any reluctance to incorporate the record based case system into objects, it was not the best to begin with, but If the same behaviour could be achieved by another means, it would make coding some things a lot easier.

An example of the sort of thing where something like this would be useful would be in the Matrix unit, which uses objects for both vectors and matrices.

for example

        Tvector2_single=object
            data:Tvector2_single_data;
            constructor init_zero;
            constructor init_one;
            constructor init(a,b:single);
            function length:single;
            function squared_length:single;
        end;

If Length were a property with a setlength method then it would make it easy to do
MyVector.Length:=1;

instead of
MyVector:=MyVector/MyVector.length;

Similarly many users of vectors would like to have X and Y available as well as the data in array form.

Neil Graham wrote

Is the absence of properties on Objects an omission by design or simply because no-one has gotten to it yet?

They are supported in fpc and in objfpc mode.

Similarly case statements as they appear in records are not supported.

That's indeed not possible by design. You can always declare record fields which themselves consist of a variant record.

Jonas

They are? Wonderful! [tries]. Sweet it works :-)

It didn't look like it from

http://www.freepascal.org/docs-html/ref/refse24.html#x56-620005.1

That would do the trick it I can do

  property chicken : Integer read field.subfield;

[tries]. Yay! compiles!