项目管理的一些思考----致原boss的一封信

XXXXX:<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" />

How about you and our staff recently? How about xxxx? Send my best wishes to them.

I do not take much work this period, so I have much time to review our old work and have conclude some suggestion to your future work. I will explain them in three opinions: management, code and database. I will say what I feel and they may affront you. If they do, I apologize for your forgiveness.

?

First ----- Management.

In this opinion, I think you should have more experiences than me so all followed is my own idea, just propose for advices.

When we got a project or make our decision to set up a project, the first job is to make sure we have clear known the project and can propose the project or product vision, which containing the goal and the development of the project. Then call a meeting on project manager to analysis the executive possibility. (Though xxxx is a very young project manager, I think I should define her as a project keeper or recorder, so to find a good project manager or to promote one team leader is necessary.) Then the project manager should be assign to take an investigation on the new plan to make sure your decision. The investigation should contain many areas such as the project status, the attitude of customers and counterwork, how to gain the profit. Through this investigation, you will find whether the vision is executable. Then you can assign it to one team. Make sure everyone in that team can understand the vision and let he or she express his or her own ideas or advices. The first job is done. Then second is to conclude the need of the project containing feathers, problems, customers’ proposal and the basic need on usefulness, the basic need on system environment, congregation (choosing server, database and other function models), the plan on maintenance and training on customer. All these should be done carefully and all after work should follow it. Then the third job is to make the priority of the creation on function models. In our old deploring pattern, we do all function models in the same time. They have triggered many problems during the congregation process. So I think this job should be done before us deploring one project. The fourth job is to write all kinds documents: the assignment on everybody in the developing team, the plan on the function models, the schedule, test plan, User manual and others. Though they are too rough, all developer should keep them and revise them in following work. Move on the implementing period, we have got many documents, they we should keep them in mind and transform to the codes. When some one has got a better idea or some one has changed the design pattern, we should change the documents. The old should be named by name plus time, for example, advices2004-8-1, and following the name rule. During the implements, we can publish our news on the new project. All we provide is the UI designed by the art designer and the describing on our vision proposal. The implements method I will mention in the next part. Then the next will be test period. All what we done before are wrong. We have a wrong cognition that the period takes a short time. No. It will take about half time of this project. So if we make the publish date, we should calculate this period into it. The test contains function model test and the whole test. The developer and the tester should test the function model. If it is ok then put it into the whole test, make sure it will not interfere other used function model. (Arrange them by the priority plan; do not add all function models together to test). How to judge the project have the least bug and do not have the logical mistake, which are not allowed in real project? After all model tests and the congregation test, there should be fewer bugs, but we should take others to use our project to find out our bugs. This called beta edition. So we should publish some beta edition to our customer our others such like our old customers or unknown customers. They all will provide some information to us by collection the bug report, which should be transferred from them by Internet. Then we can publish our real product.

?

Second-----Code

I have written many codes. Though they work well, but now I say they should be recoded. All that we programmed are too easy to collapse. Once our customer ask to change one mistaken name, what we written will be rewrite, the old is worthless. What’s wrong with them, it is that we have many duplicated code which make our code hard to alter, I think you should know what I said. Take an instance. Once a database table has been altered, we should regenerate our bean and then alter the base code and the pages we written. If the function model is isolate, maybe the work is easy to complete. Once the name Is used in many place. How could we alter the changes and change the framework, for example, once we change one item in Employee table.?

The second is the generation of bean. I think it is a problem existing before I leave. xxxxx have provide a new way to generate a bean which do not include Resultset. I have read several books. All of them do not give a precise definition on bean. I think it is a good way to include ResultSet in a bean, but using this method we must change the way we using SQL to use database. You can think about it. The connection between database and the code, the insertion, update method all should be altered. So I think you should using some method to judge the benefit of each method. And wish you can tell me your answer.

The third is the supporting on different server. For J2EE server there is a powerful mechanism to find the bugs. So when you using JRUN you will not find some problem which will happen on Tomcat. What’s the problem during this? We do not know well o
4000
n different server. For this reason , the code we write should be in a general way and could not be different. but for connection different database and using different server, we should provide different Server.xml and web.xml which are under WEB-INF.

?

Third-----database

Now we pay our attention on SQLSEVER for its cheap price. But you know the oracle have published his new product 10G, which means a new generation, is coming-----he computing grid era. For its coming we should take a research on it and make some changes. I think in the recent future, this tech will be using everywhere, so we should take an eye on it. Another is to support different database and improve the speed for searching something. You know we can insert some SQL code in java to create table and column, so this method could solve the problem I proposed above. So we can dynamically alter our table’s strut. There’s another good method to solve the above problem, using XML file to record the frame of the table and using the alias as the name. So we using the alias in our code, all will be easy. Using the code implemented in different database will improve speed apparently. So this is a good method to improve our performance. Maybe xxxx will do this job to help others know Database Well

?

At the end, I wish our company could get a better position in our city. To achieve this goal depends on many factors. Wish you may make it!

?

Best wishes

Louis wang

?

2004-8-1? 2:04 a.m.

?