- From: Mark Watson <watsonm@netflix.com>
- Date: Tue, 12 Feb 2013 16:50:46 +0000
- To: Henri Sivonen <hsivonen@iki.fi>
- CC: Andrew Livingston <Andrew.Livingston@bbc.co.uk>, "<public-html-media@w3.org>" <public-html-media@w3.org>
On Feb 12, 2013, at 8:42 AM, Henri Sivonen wrote: > (Quotes reordered) > On Tue, Feb 12, 2013 at 6:23 PM, Andrew Livingston > <Andrew.Livingston@bbc.co.uk> wrote: >>> From: Mark Watson [mailto:watsonm@netflix.com] > ... >>> I can't speak for the BBC's requirements, but for our part the UA string is insufficient because it is not vouched for by the CDM or some other similarly secure component. > ... >> That's also our position - i.e. that the user agent string is unlikely to be sufficiently trustworthy. Sorry if this was unclear. > > Do you need more than the name of the operating system (as vouched for > by the CDM) to make the inferences you need to make? Or would > something less granular be sufficient (e.g. grouping operating systems > that run on at least desktops together and operating systems that run > at least on phones together and letting tablets fall into either > category depending on their lineage: Windows 8 tablets to the former > group and Android tablets to the latter group)? Or is it only a matter > of whether the system is considered to be a TV (set top box connected > to a TV) or not (e.g. Mac Mini that might be connected to a TV > anyway)? Today, for devices such as phones, TVs and Set Top Boxes, we know the exact make and model and possibly subcategories of that. We'd have to look at a specific proposal to analyze whether coarser granularity would be ok. …Mark > > -- > Henri Sivonen > hsivonen@iki.fi > http://hsivonen.iki.fi/ >
Received on Tuesday, 12 February 2013 16:51:18 UTC