- From: Tab Atkins Jr. <jackalmage@gmail.com>
- Date: Thu, 11 Apr 2013 10:34:22 -0700
- To: Rick Byers <rbyers@chromium.org>
- Cc: Florian Rivoal <florian@rivoal.net>, Sylvain Galineau <sylvaing@microsoft.com>, "www-style@w3.org" <www-style@w3.org>, "Edward O'Connor" <eoconnor@apple.com>, Boris Smus <smus@google.com>, bokan <bokan@chromium.org>
On Thu, Apr 11, 2013 at 9:01 AM, Rick Byers <rbyers@chromium.org> wrote: > Sorry to revive this old thread. I've had some people request that we > provide a mechanism in Chromium to differentiate "coarse and fine" primary > pointer scenarios from "coarse only" ones (crbug.com/174553). I think the > argument really boils down to feeling that when there are multiple pointers, > the application should be in charge (or at least part of the decision) in > determining which to consider primary, rather than the UA. Of course they > can do this today by responding to events (i.e. when a touch is actually > seen, morph the UI or prompt to switch modes), but it's nicer to seed the > decision with a good guess. > > For example, for sites that have a functionally-reduced touchscreen version > and a more powerful mouse-based version - they'd prefer to give users to the > mouse version if it's at all possible for them to use it (and that's > probably what users would prefer too). Whereas for cases where the UI is > almost the same but just a little large by default when a touchscreen is in > use, it makes sense to default to the coarse mode. > > I think there's a legitimate argument here. I'm not sure how best to > address it though. I assume we don't want to allow a feature to take on > multiple values, right? We could replace the 'pointer' feature with a set > of 2-3 boolean features (fine-pointer, coarse-pointer, no-pointer), but that > would be ugly. There's nothing preventing a feature from taking on multiple values, actually. All the range-based ones like max-width can be thought of like this if you'd like. In general, each individual combination of query+value is independent, and can return true/false independently. I'd rather do that than break up the query into multiple new ones. I think that's more annoying to use, but more importantly, it's not upgradeable - if you didn't bring this point up before implementations were stable, and then *later* we decided we needed the ability to show when a pointer is both fine and course, we'd be stuck if "splitting them up" was the accepted way to handle things. ~TJ
Received on Thursday, 11 April 2013 17:35:09 UTC