Mike Samuel wrote:
> I don't understand the details of how autocomplete can be exploited.  Do 
> you know of any documentation on this?

browser autocomplete is keyed on domain of page and name of form field.
if gadget 1 and gadget 2 are inlined in a page, or served from the same
iframe domain, then they'll be in the same autocomplete context, and
gadget 2 can read gadget 1's autocomplete values.

I think most autocomplete requires human intervention of some sort.
I don't really understand all the behavior yet, but most of that
seems harmless.

the cases that bother me:

* user might say "yes, remember name and password" for gadget 1 without 
realizing that the browser doesn't really know to associate the values 
with gadget 1 instead of gadget 2.  this feels like a real issue.

* filled-in values might be readable with a history attack.  this seems 
hard to make a real exploit, this might not be a real issue.

> What is the property we would want to enforce?

I think I'd like to enforce autocomplete=off for all cajoled forms and 
cajoled form inputs.  that sidesteps the problem of "what if this input 
changes type?"

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to 
http://groups.google.com/group/google-caja-discuss
To unsubscribe, email [EMAIL PROTECTED]
-~----------~----~----~----~------~----~------~--~---

Reply via email to