The current OP technology is particularly important in the direction of the data carrier in the field of ORM. One is the way to use DataSet represented by Microsoft .NET technology, this data carrier has a congenital advantage in data consistency in data consistency in its flexible inclusive and convenient interface. The other is a fully object-oriented technology represented by Hibernate, CMP, etc., with standardized objects as a data carrier, which has a unique advantage in the field of purely-faced objects with its powerful and absolute standard flexible design support.
Nothing data carrier container, advantages: 1. Due to its data, the interface defined, the interface defined, the unified data carrier, easy to unify many convenient operations and functions 3. Simplify programmers need to understand the programming API, only need to remember an infinitely possible data operation API4, standardized output formats ensure a complete consolidation of the level.
Disadvantages: 1. Specification also gives a certain limit to the design of the object, sacrificing flexibility. 2, no data type brings the function of losing strong type values, increasing the running error. (Typed Dataset is a feasible way) 3, the determined output data carrier, from the design to increase the coupling of this data carrier between the layers, thereby there will be a variety of functions, not very compass
Separation of the separation of object function particleization attention. (This approach can be considered to be solved with Decorator mode) 4, although the flat data carrier structure design is advantageous for the upper layer, it is inevitably not suitable for the relationship and the description of the collection object, will weaken the underlying function. But complex ways to support the collection objects have the innate vulnerability in data performance. (This contradiction can be achieved by expanding the flat structure, that is, the relationship between the flat structure is supported, but the complex relationship caused by the secondary meaning of the flat structure must be solved. Another idea is through the collection The object is adapted, which makes it characterized by the working mode of the flat structure, but will cause a large amount of resource consumption or complexity of operation)
Standard object-oriented mode. Advantages: 1, more object-oriented, suitable for object-oriented programmers. 2. The data object exists in a class, adds programmable interface and flexibility. 3, and existing design tools can be better integrated.
Disadvantages: 1. This is actually not completely unregulated. Just look more object-oriented, still have design constraints. 2. Increase the amount of maintenance, and the objects that are passed in the way. 3. The structure of this object mode is not necessarily suitable for the needs of the performance layer. 4, the existence of multiple classes, adding the API of the data operation.
From my personal, it may be the relationship between the past, and it is more preferred, that is, there is no data type container, and in the direction of hope that the software can industrialize
I feel that this approach has the following advantages. 1. Avoid excessively alienated structures, unified structures are destined to have an alienated structure that procedures cannot use. 2. Good scalability of the versatility of data, many details can be unified, more importantly, this way has laid the foundation for future graphics. 3, more critical is: This design method of OP has certain versatility, which will not be excluded from the language that cannot be reflected by the kernel. 4, this design should be said that it is not a good independent technology (because he defines the output data format, not any object), but there is a well design of a local reuse and industrialized running operation.
