Gervase Markham writes:

> Ideas such as this have been proposed in Bugzilla - please find the bugs
> and read them to see the current state of the debate.

I'm only interesting in what I actually see in the browser, as that's
what I have to use.

> However, I'd say that if we ever end up with a security UI 1/10 as
> complicated as Microsoft's, then we've failed in our duty to protect our
> users.

Actually, the complexity of MSIE's security options is one of the few
points in its favor.  The ability to precisely configure security is
extraordinarily important in any software that access the Net today.
Anything that does not allow extensive customization will doom users to
security breaches, which have very high visiblity.

> I suspect a lot of MS's security UI is necessary because they add
> features which are security holes, then can't remove them, so they 
> provide a way to turn them off - giving you a choice of letting your 
> site work and being exposed, or breaking it and being secure.

Security is always a choice between exposure and safety.  To be
completely secure you must disable everything; to be completely flexible
you must enable everything--but then you become vulnerable in terms of
security.  There isn't any way around this.  However, a good product
lets you decide exactly how much risk you want to take by allowing you
to precisely control exactly what the product will or will not do in its
interactions.  This is especially important for products like browsers
that have extremely high exposure to attack from the Net.

> You haven't yet established why these zones are necessary.

Yes, I have.  Zones allow you to control access with a high degree of
granularity at low administrative cost.  Controlling sites individually
is too much administrative overhead, and having all-or-nothing settings
for everything at once is too restrictive and insecure.

Most ACL schemes used in IT embody some sort of grouping mechanism to
ease administration without greatly sacrificing granularity.

> If Java is safe, it should be enablable everywhere with one switch. If
> it's not safe, it should be disabled until it is.

The world is not that simple.  Programming languages are not safe or
unsafe; only individual uses of those languages are safe or unsafe.
Just enabling or disabling Java for the whold world is so coarse that it
is useless.

All programming languages are useful; but all of them have security
issues.  There's no way around this.  But if one can individually
configure the ability to use a given language based on how much one
trusts the user, there is no problem.  If it's all-or-nothing, one must
either open one's machine to every conceivable attack, or completely
discard any utility offered by the programming language in question, and
neither option is particularly useful.

-- 
Anthony


_______________________________________________
Mozilla-security mailing list
Mozilla-security@mozilla.org
http://mail.mozilla.org/listinfo/mozilla-security

Reply via email to