Re: lynx-dev Javascript

[Bela Lubkin]

Lalo Martins wrote:

Bela>>   - administrators of public access sites will want to be able to
Bela>>     prevent anonymous users from using it, which means integration with
Bela>>     the "-anonymous" and "-restrictions" flags and probably a
Bela>>     system-wide override in lynx.cfg;

Lalo> Uh, why? Unless it can access local files and do other things
Lalo> that can be restricted, and in this case it should just use the
Lalo> default restrictions (for example, if lynx isn't allowed to
Lalo> access local local files, neither should javascripts).

Well, I said "will want", not "will need".  Out of paranoia, if nothing
else.  There might not be any technical reason, but they'll still want
to be able to disable it until someone "proves" that it's safe.  I have
no actual opinion of my own as to *whether* it's safe.

But your own statement casts some doubts.  "if lynx isn't allowed to
access local files, neither should javascripts" -- OK, but that's a
matter of implementation.  Anonymous account administrators need to be
confident of the *implementation* before they'll want to provide the
facility to their users.

Similarly, you might have seen people argue that Java is safe -- *if*
the VM implementation can be trusted.  Followed by quite a number of
examples where the Java VM implementations can not, in fact, be trusted,
so Java is *not* safe.  Any argument that rests on implementation
quality needs to be proven, not just claimed.

(You might argue: administrators should build a separate binary for
their anon users, which omits libjs entirely.  Maybe so.  I still
predict demand for this.)

>Bela<