Just completed a summary of a small project, the following is the conclusion of partially taken, is willing to discuss with you. 1. Basic problems should be solved by the framework 2, the evaluation of the following points: "Therefore, the easiest and effective solution is to analyze the existing framework, refer to the C version to launch the Java version." Should be seen into two places: front The role is to use the existing framework for discovering potential problems, using ready-made solutions; negative role is easy to fall into all-in-one solutions; Object of mud. A bigger problem is that the offset of the work is possible, delaying the progress. 3, design pursuit is appropriate but perfect! 4. If you find the responsibility class preference "Shutdown Responsibility" (entrusted), wrap around the "Pocket", that is, it is over-design. 5, the smaller interface is not, the better, in the case of problem complexity, the more simple interface, the more complicated use. 6. The development of the interface should take care of the weak. 7. The understanding of things is gradually in-depth, but it is a contradiction when planning to make a shallow. 8. Clear important uncertain factors, planning can be more accurate. 9. When there is no problem, when the problem is difficult, the estimation workload is "buy lottery game". 10, if you pay attention to accumulation, multi-summary, multi-analysis, and workload estimates are possible to quantify, and it is even possible to introduce an accurate "workload calculation formula". But pay attention to the cost of paying. 11. It is definitely not simple to do planning, and it is necessary to do too much. 12, the frame can be gradually improved, so it is not necessary to have too much pressure. 13, habits, psychological expectations are also constraints on the interface. The interface can be simple, but the explanation is clear, otherwise the misunderstanding may make the interface user "very angry". 14. Don't conflict when you don't understand the foreman, you must carefully analyze. It is possible that you think is not right is the best solution under conditions. 15. Objective analysis is more, subjective mistakes are less. 16. Automatic generation code can not only reduce workload, but also help to find errors, more convenient for future maintenance. So a chance, do this.
