DataType Enum Design about libplctag.net HOT 8 CLOSED

@jkoplo - I'd be keen to get your view on this.

from libplctag.net.

from libplctag.net.

What about something like this:

The consumer provides the preferred C# they want to use, and they provide the method by which the tag is encoded/decoded.

public class Tag<T, U> where T : IDataType<U>, new()
{

    T dataType = new T();

    Tag tag;

    public void Read(TimeSpan timeout)
    {
        tag.Read(timeout);
        Value = dataType.Decode(tag);
    }

    public void Write(TimeSpan timeout)
    {
        dataType.Encode(tag, Value);
        tag.Write(timeout);
    }

    public U Value { get; set; }
}

public interface IDataType<T>
{
    int ElementSize { get; }
    int CipCode { get; }
    void Encode(Tag tag, T t);
    T Decode(Tag tag);
}

Our library could implement the DataType for the basic types but if they wanted to extend this (PID tags, other UDTs, etc) they could just implement IDataType.

This is probably better than the abstract classes I did last year for my own projects in my opinion. The main thing missing is array indexers.

from libplctag.net.

That's really nice and just about what I had in mind. I particularly like the interface. You kinda did the work already, but I'm hoping I'll have time to integrate this tonight.

from libplctag.net.

Cool, nice - I haven't compiled/tried it but seems roughly right. Do you know if there is a way to make classes be arrays in a generic way:
i.e. a single class can have 0, 1, 2 or 3 dimensions?

from libplctag.net.

Nice!

Perhaps you could call something like your DataType.Decode(), but it would take two arguments: the tag and an offset. The offset would be the byte offset to that element of the array in the tag's data. After Read you would calculate the element offsets, and then call Decode() on one element at a time?

In C++ this is painful. Easier to have a class that overloads the array operator and pretends to be an array than to use the normal notation for array declarations. I think that there is some template magic to make this a bit more palatable.

Minor nit: in your example code, if the timeout is too short or zero, Read() will return before there is any data and perhaps throw an exception.

from libplctag.net.

@kyle-github I think we're thinking along the same lines: maybe we should only support 0D and 1D access because that is what libplctag does. I know that AB have 2d and 3d arrays, but not sure about other systems.
If users want to inherit this class and provide 2d or 3d indexers and different constructors (with array sizes) they could do so. I don't know how this would work in C++but in C# it is fairly straightforward to provide 2D indexers.

So the base type would provide access via a Value property, and 1d access via a 1D indexer.

What do you think?

public class Tag<T, U> where T : IDataType<U>, new()
{

    // .....

    public U Value { get; set; } // check that element_count == 1 and throw error if not
    public U this[int index] { get; set; } // Of course do checks on index bounds
}

Hmmmm... it doesn't feel right, maybe only the indexer? hmmmm...

from libplctag.net.

This is long done

from libplctag.net.