问题
Any equivalent as the decorator @property
in python?
That is to have one field automatically updated (on access) due to an update of another field, instead of recomputing it before access.
UPDATE 2014-02-06
By defining "one" field as an activeBindingFunction
of "another" field turns on the automatic updates (See @jdharrison's answer). However, is there anyway to check whether such updates are lazy evaluation? If not, how can we achieve this?
(A later edit: an "informal" print function inside the activeBindingFunction
is evident that the evaluation is lazy. Any way to check this issue in a more formal manner?)
For better illustration of my question, below is a piece of python code producing the desired lazy behavior via decorator @property
- bear with me to put it in an R-tagged question.
class Test(object):
def __init__(self):
self._num1=None
self._num2=None
@property
def num1(self):
if (self._num1 is None):
self._num1=1
return(self.num1)
else:
return(self._num1)
@num1.setter
def num1(self, value):
self._num1 = value
@property
def num2(self):
self._num2=self._num1*2
return(self._num2)
The interactive session
In [2]: obj=Test()
In [3]: obj.__dict__
Out[3]: {'_num1': None, '_num2': None}
In [4]: obj.num1
Out[4]: 1
In [5]: obj.__dict__
Out[5]: {'_num1': 1, '_num2': None}
In [6]: obj.num2
Out[6]: 2
In [7]: obj.__dict__
Out[7]: {'_num1': 1, '_num2': 2}
In [8]: obj.num1 = 5
In [9]: obj.__dict__
Out[9]: {'_num1': 5, '_num2': 2}
In [10]: obj.num2
Out[10]: 10
In [11]: obj.__dict__
Out[11]: {'_num1': 5, '_num2': 10}
回答1:
Not sure if its what you are after
test <- setRefClass("test",
fields = list(
num1 = "numeric"
, num2 = function(){
print("hello")
2*.self$num1
}
),
methods = list(
initialize = function(num1 = 1){
num1 <<- num1
},
show = function(){
print(c(num1 = num1, num2 = num2))
}
)
)
> tt <- test()
> tt
[1] "hello;"
num1 num2
1 2
> tt$num1<-4
> tt
[1] "hello;"
num1 num2
4 8
来源:https://stackoverflow.com/questions/21587410/auto-update-of-a-field-data-member-in-r-reference-class