- From: Mounir Lamouri <mounir@lamouri.fr>
- Date: Wed, 18 Dec 2013 23:43:56 +1100
- To: Marcos Caceres <marcos@marcosc.com>, public-web-mobile@w3.org
- Cc: Weiss Yoav <yoav@yoav.ws>
On Sat, Dec 7, 2013, at 13:19, Marcos Caceres wrote: > Hi, > I’ve been tracking the discussions that both the Web Perf and DAP WGs are > having around the network information APIs. However, those groups are > struggling to come up with use cases… hence, this sounds like a job for > our group to help them! > > I’ve started to gather real world usage from native platforms here: > https://github.com/w3c-webmob/netinfo/ I think gathering native application usage is not the best thing to do in that particular situation. What native applications currently do is basically asking the user whether they should do X while connected to a cellular connection or only do X while using wifi. Those are options that matches the current state of the technology, its pricing and its usage. This is evolving and cellular connections are becoming cheaper (you can have all you can eat data for 13 GBP in the UK) and faster (4G is being deployed in many countries) in some places. Also that kind of UI might end up being deprecated with different usage. For example, Africa is known to have a huge mobile potential with mobile being used more than landlines, internet being accessed a lot via mobile and even payments via mobile are incredibly more common than in Eurpoe. What if in a few years people would only access to the Internet via a cellular connection in Africa, a fast cheap and reliable connection and not even use broadband at home? If that happens, all those apps that requires opt-in to behave over cellular would be broken by default for those users. Trying to have APIs in order to reproduce those UI on the Web sounds short sighted. Native applications can be short sighted because there is a deprecation mechanism enforced by vendors. The Web can't and pushing new APIs that we know will help in the short term to solve very specific use cases do not sound like a great idea. This is the reason why the Network Information API was trying to provide the tools to be able to get the real information the applications were actually looking for. Another approach would be to have very high level APIs to access those information. I think Tobie proposed something like that on twitter. Anyhow, we should answer the real questions or give the tools to answer them, not fulfil short sighted use cases. -- Mounir
Received on Wednesday, 18 December 2013 12:44:21 UTC