On 9/4/2012 9:40 AM, ptheriault wrote:
> Now that we have more details of this API, I think we need to revisit the 
> permission model. I think the user should be asked before an app is allowed 
> to receive push notifications (inline with the transparency/no surpises 
> privacy principle). The user should also be able to see what push channels 
> are registered and turn them off. This could be achieved by requiring an 
> explicit permission installed, privileged and certified web apps.
> The only issue I see with this is that a user might not understand what  
> "Push Notification" actually is. 

I don't think they will understand.  How is this worse than using alarm API to 
poll a server?  We don't provide UX
anywhere to manage per-channel network access.  For consumption purposes we 
might want to have the ability to turn off
push on a per-device or per-app basis, but that's not a permission IMHO.
  Lucas.

> On Aug 8, 2012, at 5:45 PM, Lucas Adamski wrote:
>
>> Initial draft.  One of the trickier APIs to reason through as it really 
>> depends on the intended use cases.  Thoughts?
>>
>> ==Push Notifications API==
>>
>> References:
>> *https://wiki.mozilla.org/WebAPI/PushAPI
>> *https://bugzilla.mozilla.org/show_bug.cgi?id=747907
>> *https://groups.google.com/forum/?fromgroups#!topic/mozilla.dev.webapi/doBebGwUTNE
>>
>> Brief purpose of API: Asynchronous notification mechanism for apps with 
>> store and forward capabilities.
>>
>> General Use Cases: Provide an mechanism for websites to push small 
>> notifications to subscribed applications on the client, even when they 
>> aren't currently running.  
>> *IM messaging apps.  
>> *Website activity notifications (auctions, online price alerts, travel 
>> advisories and flight status, banking activity, etc).
>>
>> Inherent threats:
>> *Spoofing notifications could lead user to disclosing sensitive information
>> *Spoofing messages could trick an app into disclosing sensitive information 
>> (i.e. submit info to URL..) or otherwise take action on behalf of the 
>> attacker.
>> *Spoofing of notifications to system-critical applications could result in a 
>> variety of attacks, from information disclosure to device compromise.
>>
>> Threat severity: High, possibly Critical depending on usage
>>
>> == Regular web content (unauthenticated) ==
>> Use cases for unauthenticated code: Same
>>
>> Authorization model for normal content: None?
>>
>> Authorization model for installed content: Implicit
>>
>> Potential mitigations: Airplane mode?
>>
>> == Privileged (approved by app store) ==
>> Use cases for privileged code: Same
>>
>> Authorization model: Implicit
>>
>> Potential mitigations: Same
>>
>> == Certified (system-critical apps) ==
>> Use cases for certified code:  Do we use this API for any system-sensitive 
>> operations, like app updates, payments, etc?
>>
>> Authorization model: Implicit
>>
>> Potential mitigations: Same
>>
>> __NOTOC__
>>
>> _______________________________________________
>> dev-webapps mailing list
>> [email protected]
>> https://lists.mozilla.org/listinfo/dev-webapps

_______________________________________________
dev-webapps mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-webapps

Reply via email to