Why does python not have a mechanism for data hiding?

[Patrick Mullen]

Here is a quick, ugly way you can easily hide data, although persistent
clients can still bypass it.  x.__dict__ is ugly, though not any uglier than
x.__Ugly__Internal__Variable__Do__Not__Set

Another option would be to use another structure, such as a dictionary or
list, and store variables in there proxying to it from get/set functions (or
__getattr__ __setattr__).  Or just don't document this.

class HideTest(object):
    def __init__(self):
        self.__dict__["beans"] = 5
    def __setattr__(self,key,val):
        if key == "beans":
            raise Exception("Cannot set this attribute")
        return super(HideTest,self).__setattr__(key,val)

b = HideTest()
print "b.beans is",b.beans
try:
    b.beans = 5
except:
    print "Cannot set"


Also the point of underscore rules have never been to keep clients from
accessing variables.  Read about them in PEP8:
http://www.python.org/dev/peps/pep-0008/

They are primarily to handle specific name conflict situations, and
secondarily to inform users when they might be mucking about with something
at a lower level.  Using a keyword could work I suppose for the module-level
convention, as well as for the name mangling.  Implementation wise it seems
a bit tricky to use keywords for this.  For the purposes of informing users
however, keywords versus underscore is a terrible idea.  dir() as is makes
it clear, with underscores, which variables are safe to use and which are
potentially dangerous.  Rewriting dir() to print extra information about
names seems incredibly problematic as well.  I really like the idea of
having variables that may be unsafe to use but I can still use them if I
want to, and the underscore rules are the most clean way I can see for this
purpose (not to mention all of the history).  I agree that they do look ugly
of course.  I also tend not to use them in my code as well.

This has been brought up so many times before, and turned down again and
again.  Working around it as best you can is your best bet, especially since
a major change like this could never happen until well after python 3.0.
(Who knows how long it will take before 3.0 is fully adopted enough to move
on from that)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/python-list/attachments/20080525/55982510/attachment-0001.html>