> On 11 Jun 2015, at 10:39 pm, Benjamin Francis <[email protected]> wrote:
> 
> On 11 June 2015 at 06:17, Paul Theriault <[email protected] 
> <mailto:[email protected]>> wrote:
> We discussed this in another thread, but Jonas convinced me that pinning will 
> not have any security side-effects, and I’ve updated the permission model to 
> reflect this [1].
> 
> To clarify, have you decided that granting extra storage, the ability to run 
> in the background and the ability to appear in the web activity picker will 
> have nothing to do with whether or not the site is pinned, so Gecko doesn't 
> need to know?

Thats correct. Though you are mixing things up a little here by throwing in 
Activities. That’s not a permission, that’s just something that is registered 
at install time. (In the same way that system message listeners are declared). 
Without an install step, we will need a new way for apps to register anything 
that was previously registered in the manifest that we want to continue to 
support.

> Does that mean that the user will always be prompted for these permissions 
> instead?

Yes, for now. But it doesn’t have to be, I would like to see improved UX so we 
don’t need to prompt up front for things that don’t really need user 
permission. For example, currently the only two manifest-declared web 
permissions are Alarms[1] & Storage. Neither of these would be terribly bad to 
grant to a website, so long as the user had a way to a) see that the permission 
was being used and b) had a way to respond if the website abuses the 
permission. I imagining something like a notification, when these APIs are 
used. In the case of alarms, we might have a persistent Alarm notification set, 
so the user is aware that there are pending alarms (and can remove them if they 
don’t want to be woken up!). I also imagine something an “about:permissions 
<about:permissions>” style UX that allows the user to block annoying sites, and 
set defaults (a user might want to decide that alarms should be denied by 
default, and only allowed by exception).

[1] I remember elsewhere Jonas suggested to merge Alarms into something like a 
“run in the background” permission. Imho we should keep alarms separate since 
its easy to explain and something the user can be expected to manage. This 
implications of “run in the background” aren’t very clear to me, especially 
when our app model already does’t suspend apps that are not foreground.


>  
> I assume “installation” will remain for add-ons and will remain as gate to 
> implicit permissions.
> 
> That's my understanding too, that add-ons will be "installed".
> 
> Ben 
> 

_______________________________________________
dev-b2g mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-b2g

Reply via email to