为什么继承类时构造函数需要super(X, self).__init__()

 pytorch构建模块时常常会看到

class ConvBlock(torch.nn.Module):
    def __init__(self, filters, s, f):
        super(ConvBlock, self).__init__()

为什么在构造新模块的时候必须super，不然会发生什么？

一句话：重写子类的__init__后不再自动继承父类的__init__，必须使用super把父类的拿过来

例子A：子类不重写__init__

class Person:
    def __init__(self, name, age):
        self.name = name
        self.age = age
        self.bodyTemperature = 36.5


class Female(Person):
    pass


femaleA = Female('Emma', 18)
print(femaleA.name, femaleA.age, femaleA.bodyTemperature)

这时候会执行父类的构造函数，返回正确的属性

例子B：子类重写__init__不super

class Person:
    def __init__(self, name, age):
        self.name = name
        self.age = age
        self.bodyTemperature = 36.5


class Female(Person):
    def __init__(self, name, age, height):
        self.height = height


femaleA = Female('Emma', 18, 175)
print(femaleA.name, femaleA.age, femaleA.bodyTemperature, femaleA.height)

会报错，因为Female类只需要一个height输入，没有继承父类的构造属性，即使把父类属性加入__init__()也没用

例子C：子类重写__init__并super

class Person:
    def __init__(self, name, age):
        self.name = name
        self.age = age
        self.bodyTemperature = 36.5


class Female(Person):
    def __init__(self, name, age, height):
        super(Female, self).__init__(name, age)
        self.height = height


femaleA = Female('Emma', 18, 175)
print(dir(femaleA))

一切正常