Re: [W3C WebMob] Data Info API

(NetInfo Editor's + Mozilla hat on... )

On October 14, 2014 at 9:46:13 AM, Daniel Appelquist (appelquist@gmail.com) wrote:
> Comments below:
> I would definitely echo your comments on the usefulness of the Network Info API. I’m not  
> entirely sure what the perceived down-side is.

Please see the Discussion section:
http://w3c-webmob.github.io/netinfo-usecases/#discussion

>  The browser people seem to think the Netinfo  
> API is “not needed” but I agree it’s just exposing info that the device has anyway.

It's a case of "just because you can, don't mean you should". In trusted contexts, this info might not be a big deal... on the Web, it might be. Also, there are a lot of foot guns (e.g., trying to do responsive images with this info, which would actually be harmful for performance ... this is not such a huge issue, now that <picture> is coming) - and even after many years, very few use cases have surfaced: strongest are analytics and dropping some ads from 2G connections. 

Having said that, we have, or are working to, surfaced some bits (connection type, and working on max theoretical bandwidth). 

Spec is still actively being worked on by Mozilla and Google outside DAP... we want to roll it back into W3C at some point:
http://w3c.github.io/netinfo/

> However, I still think a “data plan” info (network) API could also be useful, especially  
> if this info could be surfaced to the (web) application itself, which would know more  
> about the context. For example, a streaming video service could evaluate how much of  
> the user’s data plan would be impacted if they streamed a specific video and warn the user  
> using service-specific meaningful language “you are going to use up 30% of your allowance”  
> or “you’re going to go over your data allowance if you stream this - you might want to switch  
> to Wifi”. 

Yes, this is the idea.  

> So I see this as enabling clearer messaging to the end user which should hopefully  
> be accessible to a wider audience - not only the tech-savvy.
>  
> But agreed, knowing the current bearer (and roaming status) would be essential in order  
> for this to make any sense.
>  
> By the way, this is all privacy-sensitive and finger-print-enabling information so  
> it plays into the whole permissions question as well, see the latest TAG discussion on  
> this topic here https://github.com/w3ctag/meetings/blob/gh-pages/2014/sept29-oct1/09-30-f2f-minutes.md#permissions  
> with proposal from Alex Russell https://gist.github.com/slightlyoff/43cd8c2f64a0719358fe.  

Agree. Initially, we would like to get this surfaced at the OS/System level (i.e., to Android/FxOS/etc.) and then we can work out safe ways to actually expose it to the Web Platform.