At 05:12 13-05-2012, Andreas Petersson wrote:
>If someone is writing a document specifying extensions to this
>document, and they also want it as an RFC I don't think it is
>that strange to demand that they are notifying IANA?
Requirement language tend to be overused in some documents.
>Maybe it however would be better to write something similar to what is
>written in httpbis-p2/3.1?
>"Extensions are registered using the procedures described in
That sounds better.
>What information you forward is up to you as the proxy owner. If that
>information is "leakage" in matter of privacy, depends on the situation.
>Well, I would see no huge difference between using a proxy that adds
>this information and using no proxy at all.
>The header field "for" is essentially for non-internal nodes, if you
>by "internal nodes" mean nodes belonging to the same organization.
>I believe it is hard to write something specific about privacy, but it
>is worth mentioning.
It is generally difficult to come up with text as there isn't much in
terms of previous work to borrow from.
Web proxies are used for various reasons, e.g. piracy, anonymity,
etc. Given that "hostile government" is being considered in threat
scenarios in some IETF WG, it would be good to have some text about
that from a privacy angle.
The X-Forwarded-For: header is used in existing implementations to
pass on the client IP address. This draft defines "Forwarded: for="
as a replacement. There is already some discussion about information
leak (Section 8.2) but it only takes into account use cases where the
nodes belong to the same organization, e.g. there is a proxy in front of a NAT.
I suggest removing Section 6.3. If you don't want to disclose
information, don't add the "Forwarded: for" header field. Adding
random identifiers ends up providing information about:
(i) a proxy is in use (most people in this WG could easily figure
that out anyway)
(ii) another means to track down users even though the IP address
is not being passed
As a starting point, here's some suggested text for Section 8.2:
In recent years, there has been growing concerns about privacy. There is a
tradeoff between ensuring privacy for users versus disclosing information
which is useful for debugging. The Forwarded HTTP header field, by design,
exposes information which affects the privacy of users. This header field
should not be used if the proxy is being operated as a privacy service.
If you use a privacy considerations section, the text may have to be
expanded. I'll send text if you decide to do that.