Micha <[EMAIL PROTECTED]> writes:

> Yes, there was a thread on this list where jidanni points to a possible
> solution for caching https (IIRR). Now i'm asking myself the same
> question....but i can't find this message. And I remember there's an
> archive somehwere, but i lost the link. Could somebody help me with this
> archive URL or just repeating what this idea was about ?

The e-mail archive is
http://www.mail-archive.com/wwwoffle-users%40gedanken.demon.co.uk/

> I've some thoughts about the problem.
> 
> wwwoffle can't cache or modify pages received via https:// 
> since this traffic is tunneled (that is, encrypted) between the web
> server and the browser-- is that right ?

That is correct, the link is encrypted with a one-time encryption key
that is sent to you with a certificate.  The browser will complain
about accepting data unless it comes with a certificate with the
correct server's name that has been signed by a Certificate Authority
(CA).  When you install a browser you will install a number of
certificates from these certificate authorities.

There is a chain of trust when you visit an https server.  You trust
the browser, the browser trusts the certificate authority and the
certificate authority trusts the server.  The browser will then load
the page without complaint.  If the server has a certificate that is
not signed by one of the certificate authorities that your browser
trusts then the browser will complain.

If you install Apache with SSL enabled then you can have your own
https server.  You will need a certificate for it and it is likely
that you will create your own and sign it yourself.  When you access
the server your browser will complain, but it will let you do it.  You
can stop the browser from complaining by telling it to trust your
certificate as a certificate authority.


> But, from a browser, i can save https pages on disk (with content
> directory). So there *is* a way for a user to get the unencrypted
> content, besides through the eyes.

Yes, there is nothing magical about the link.  The chain of trust only
goes one way, the browser is not trusted by the server (in certain
cases you can have a personal certificate that is used by the server
to make sure that you really are you).  Any program can fetch https
data from a server (e.g. 'curl' was one of the first command line
https clients).


> Is there at least theoretically a way for a proxy (or any user process)
> to get the 'https' page contents from the browser, without modifying the
> browser code ?

It is all a question of trust.  If you trust your proxy enough then
the proxy can make an https connection to the server, decrypt the
data, store it and re-encrypt it to send to you.  The certificate the
proxy uses to send the data to you would be signed by its own
certificate and not the original certificate authority's certificate.

This is the model that Dan (jidanni) was talking about in his original
e-mail on the subject.

-- 
Andrew.
----------------------------------------------------------------------
Andrew M. Bishop                             [EMAIL PROTECTED]
                                      http://www.gedanken.demon.co.uk/

WWWOFFLE users page:
        http://www.gedanken.demon.co.uk/wwwoffle/version-2.8/user.html

Reply via email to