我认为过度使用全局变量和静态关键字也会导致应用程序中实例的某些点的内存泄漏

在你的文章中有两个主要问题。

First, about static variables. Static variables are completelly unnecesary and it's use can be avoided easily. In OOP languajes in general, and in Java particularlly, function parameters are pased by reference, this is to say, if you pass an object to a funciont, you are passing a pointer to the object, so you dont need to define static variables since you can pass a pointer to the object to any scope that needs this information. Even if this implies that yo will fill your memory with pointers, this will not necesary represent a poor performance because actual memory pagging systems are optimized to handle with this, and they will maintain in memory the pages referenced by the pointers you passed to the new scope; usage of static variables may cause the system to load the memory page where they are stored when they need to be accessed (this will happen if the page has not been accesed in a long time). A good practice is to put all that static stuf together in some little "configuration clases", this will ensure the system puts it all in the same memory page.

Second, about static methods. Static methods are not so bad, but they can quickly reduce performance. For example, think about a method that compares two objects of a class and returns a value indicating which of the objects is bigger (tipical comparison method) this method can be static or not, but when invoking it the non static form will be more eficient since it will have to solve only two references (one for each object) face to the three references that will have to solve the static version of the same method (one for the class plus two, one for each object). But as I say, this is not so bad, if we take a look at the Math class, we can find a lot of math functions defined as static methods. This is really more eficient than putting all these methods in the class defining the numbers, because most of them are rarelly used and including all of them in the number class will cause the class to be very complex and consume a lot of resources unnecesarilly.

总之:在处理静态或非静态方法时，避免使用静态变量，并找到正确的性能平衡。

PS:抱歉我的英语不好。

静态变量有两个主要问题:

线程安全——静态资源根据定义不是线程安全的 代码隐式——你不知道一个静态变量何时被实例化，也不知道它是否会在另一个静态变量之前被实例化

从我的观点来看，静态变量应该只读取数据或变量创建的约定。

For example we have a ui of some project, and we have a list of countries, languages, user roles, etc. And we have class to organize this data. we absolutely sure that app will not work without this lists. so the first that we do on app init is checking this list for updates and getting this list from api (if needed). So we agree that this data is "always" present in app. It is practically read only data so we don't need to take care of it's state - thinking about this case we really don't want to have a lot of instances of those data - this case looks a perfect candidate to be static.

静态变量最重要的是会造成数据安全性问题(任何时间更改，任何人都可以更改，直接访问而不需要对象等)。

欲了解更多信息，请阅读这篇文章 谢谢。

邪恶是一个主观的术语。

你不能从创建和破坏的角度来控制静态。他们按照程序的要求生活装卸。

由于静态对象存在于一个空间中，所有希望使用它们的线程都必须通过您必须管理的访问控制。这意味着程序更加耦合，这种变化更难想象和管理(就像J Skeet说的那样)。这导致了隔离变更影响的问题，从而影响了如何管理测试。

这是我对他们的两个主要问题。