w3ctag/private-mode

Specification for "private mode" browsing

private-mode's Issues

Accessing non-private data in private mode

Should private sessions be able to access cookies, etc. from "normal" browsing mode?

Impact on people who browse private by default

Some people currently browse in private mode by default. Some parts of this spec will make doing so more "in your face" / "chatty", which may cause them to stop using private by default.

Discuss.

HTTP request constraints

There are currently a number of constraints on the form of an HTTP request.

Is this level of precision for HTTP requests necessary? It doesn't completely address HTTP
fingerprinting, and could be a major implementation pain.

Needs implementer feedback.

HSTS / pins

... are state, and it might be better to persist them between sessions.

Taken up on 2 Nov

Discussed on call 2 Nov https://github.com/w3ctag/meetings/blob/gh-pages/2015/telcons/12-02-agenda.md - presentation API people wondering what's going on with it. Some possible evidence of demand?

"log information"

Browsers in private mode must not log information about the browsing session.

What does this mean, precisely?

JavaScript

TBD: history, fonts, introspection, geolocation, keypress, window, WebStorage, etc.

Sessions

The concept of a private browsing "session" and "terminating" it aren't well-defined.

ETags

Disallow completely, or constrain?

Password managers

How should they be handled? Integrated vs. extension doesn't seem to make a difference.

Extensions Private Mode

Some extensions (Flash, in particular) are extremely dangerous since they provide means to identify user (Flash cookie) AND at the same time user needs them to be enabled (to play video). Probably, we should talk about enabling private mode in plugins/extensions too?

References / sources

Tracking issue

Fingerprinting

• http://www.chromium.org/Home/chromium-security/client-identification-mechanisms
• https://www.torproject.org/projects/torbrowser/design/#fingerprinting-linkability
• https://www.browserleaks.com
• https://panopticlick.eff.org/index.php?action=log&js=yes

Side Channel Attacks

• http://research.microsoft.com/pubs/119060/WebAppSideChannel-final.pdf
• http://freehaven.net/anonbib/cache/LZCLCP_NDSS11.pdf

Cookie scoping

... needs discussion / refinement.

Trust and Intermediaries

... needs to be filled out.

This needs to cover trust issues around configuring proxies; they can be useful (e.g., for Tor) or harmful.

browsing data vs. configuration

Many extensions and browser settings are driven by the user; they are not "state" as set by the site.

Some of it falls into a grey area -- e.g., HSTS information (#10), password managers (#2).

How should this be handled? The spec needs to be a lot crisper here...