In order to adapt to the spirit of the times, this paper has conducted a less typical attempt. I suggest that it is slightly away from the transaction manager we have already familiar with. Similarly, I hope that you can guess what kind of special framework I am going to analyze. Because almost as long as there is a place where there is a microprocessor exists, there is a framework. The J2EE application server can be said to be a framework, similarly, a message delivery system or even an operating system can be referred to as a framework. At the same time, it can easily draw this argument: the framework is the only component of the successful implementation of this sacred target in the software. Although truly reusable business logic instances are still difficult to determine, there are many frames (especially when you accept the definition of my framework, that is, the framework includes application servers, such as commercial products that are widely used, such as BEA. The code of WebLogic Server is more reusable. The framework is an entity present in each layer, and each frame is responsible for some general problems existing in its architectural layers, providing a higher abstraction layer for customers located at the high-level. For example, the application server is responsible for techniques such as threads and socket management, so that the programmers using components can be more concentrated in their or her attempted business functions. Although the application server provides higher abstraction, it is known to know that each application server has considerable customers, which is not just running projects; this causes some EJBs and servlets created from the underlying. Business functions. In almost all cases, the application server customers have written a framework yourself, and then their application developers write code that exists in the scheduled frame. Typically, these frameworks perform tasks such as reading data from a centralized enterprise repository to complete the configuration of static reference data and business rules. In most cases, the framework also limits the ability to provide many architectural decisions to their users from the degree of freedom -J2EE provided to the application developer. Is my logic in the servlet or in EJB? EJB is a session bean or an entity bean? The session bean is status or stateless ... There are still many choices such as this. These options provide a powerful toolset that is widely existing in server-side, logical issues, but 80% of the application server customers try to solve the problem is not widely presented. Many applications use very common standard usage modes. These modes drive the common set of J2EE usage modes, or at least they should be. By providing a powerful tool to an application developer, you have forced them to correctly make decisions such as how to best utilize their tools. What is the business value of the experienced J2EE architect team? Why do you need to have a person who can make a decision on these difficulties? And how will you keep them? If they spend a lot of time to solve the problem from the technical perspective, they will be able to find another job - ("" With a wisdom mind, I am parking here ") or will spend a lot Time to make a daydream and write a new framework that may not be required for you or cannot be profitable from it. Even if the status you have an expert team and they follow discipline, and there is no adventure activity in architecture, you have many talented individuals, they do their own things, pursue their own goals, will Inevitably lead to diversity of code style and variation in practice; this will result in rising maintenance costs when the next generation of programmers attempt to maintain the value of J2EE business applications in the production of J2EE business. In production, rules and consistency is your friends! Therefore, it is necessary to provide a framework (not just to make Master!) Provide consistency to developers to make "ordinary personnel" can achieve high production efficiency rather than just super heroes. However, most application server customers will find potential dangers in their predetermined framework.
When they recruit new team members, these individuals do not immediately have output - they will be hired for a few weeks to solve the problem of the scheduled framework they will work. Inevitably, due to lack of or no existing documents, no matter how much J2EE experience will take a lot of time to ask for teachers in the corner, how they should do. This will not only waste them for many times, the master in the corner also spend his or her half of the time to answer the foundation, rather than teach students to condense their high level code. This problem becomes worse when you pack some development work to other companies. Another organization, or even another mainland developers will not ask the masters to obtain significant benefits to the masters in the corner - because the master (he or her) is not in the corner. Frames bring us consistency and have multi-tape level teams, and its cost is the reasons why we use J2EE in our beginning. So, in turn, when we need a framework to improve production efficiency, we have defined any of the needs sets that are required to have a highly useful framework, ie, it must be widely adopted and have a sufficient document description to make developers transferable maximize. It must be easy to use to minimize developers understand the needs of the server-side calculation of low-level details. This means that it must be tooltime. Entering the next-generation runtime framework of the honeycomb frame allows today's BEA WebLogic platform application to be more powerful and consists of three basic components: page streams, controls, and JSR 181 Web Services. All three components use code annotations to drive the code deployed from the runtime view - the runtime behavior declaration set will limit most of the code-level developers interactively limit the implementation of business needs. The code annotation on the controller allows the page crem system to handle the detail portion of the support; the category level annotation allows things related to business targets to be easily handled as web services; at the same time, the annotation makes the developers can pass in the control The reusable function block is declared. The best news is: Although the comments and the relevant code collected together, and can also operate it through the trusted editor, through the idea of the honeycomb (first is WebLogic Workshop, but there are other tool suppliers) (Don't say Apache Polinate project) is also there, to make sure it is not the only one), so that it is suitable for intelligent operations; this makes developers to easily implement custom controls with honeycomb development models and custom methods. Interaction - For example, look at today's Database Controls - Create a control in the IDE, you need to find the data source from the running server, and even what tables in the mode are needed. Such intelligence, environmentally sensitive tips will bring fast learning curves to developers, unlike wizard-based code generator methods, beginners will never face hundreds of code lines and other products, these codes Or the product is not his or her own written, it is difficult to understand, but now you must maintain (still remember 4GL COBOL generator tool ?!). Of course, this development method today has benefited BEA's existing customers, just as in the study of Gartner and Crossvale or others; but the commander Paranoia is whispering on their head whispering "private" The word and also complained to the locked supplier. So for these people, the hive represents the possibility of benefiting from the BEA's innovation but is not subject to the life cycle of the BEA (although that is why any reason in the foregoing people!).
