- From: Diogo Resende <dresende@thinkdigital.pt>
- Date: Mon, 20 Dec 2010 15:43:43 +0000
Replied inline.. On Sat, 2010-12-18 at 21:56 +0000, Bjartur Thorlacius wrote: > On Fri, 17 Dec 2010 10:42:52 -0000, Diogo Resende > <dresende at thinkdigital.pt> wrote: > > > Let's think about an example, perhaps a bluetooth weather station. For > > the OS it's just another bluetooth device. What if a web app could have > > permission (granted by the user/browser) to scan for bluetooth devices > > and fetch weather information, save it on the weather history and > > process it in some way. > Then someone would create a web app that scans for specific models of > bluetooth devices, connects to them, fetches weather information and > processes it in some way. All is fine and dandy, as long as everybody is > using exactly the same model as the author (or one of the models supported > by the company maintaining the app), and needs to apply that process and > that process alone. Bluetooth was an example. We were discussing a serial connection in general. It could be BT or RSR232/USB or anything else. I was not thinking specifically about a web app that would do that, I was just thinking that talking to a serial device could be very handy. > Say someone creates a web app that applies a process to weather > information gathered from an hypothetical Weather-o-Meter connected via > Bluetooth. Say I'm doing a research on global warming and the Gulf stream, > and want to apply that process to weather information gathered from all > over the country. Even if all the weather stations in Iceland are using > Weather-o-Meter, I won't receive the information over Bluetooth. For this > to work weather information gathered on the weather stations must be > encoded in a standardized format, transferred to me over a network for me > to decode and apply said process to. I'm thusly required to persuade the > author to add an alternative input mechanism to the app, hack my Bluetooth > stack and create a virtual Bluetooth device or rewrite the app. > > > > - Does the OS need to know how to fetch this information? > Yes, the purpose of an OS is to abstract and multiplex hardware. > Does a web app need to know how to fetch this information? > > > - Is a browser plugin really a better idea? Which browser/version? Then > > how is the page going to fetch that? How secure is that? Can't another > > page do it? This reminds me the use of <embed> which I personally hate. > I agree that a browser plugin would the wrong approach, but I argue that a > web page would be as well. I can't imagine a scenario where I'm developing > software support for a Bluetooth weather station and I figure: "Heck, I > should put a web browser between the Bluetooth stack and the weather > station abstraction." Again, the app should not (but maybe could) listen for bluetooth-specific devices. If you develop web apps you wouldn't say that, it would be more like: "Heck, I have a wonderfull web app and now I have to do some kind of abstraction on several diferent OS (win,unix,..) so I can in some way export the bluetooth information to the browser." > > I hope this kind of example enlights some people to forget about > > storage, cameras, keyboards and all the stuff you can get in any general > > technology store. > > > Where do you draw the line? What's inherently different about hardware you > can find in general technology stores? General technology stores: cameras, lcds, mouses, keyboards, microphones. We're not talking about this kind of devices. > Why should there even be so > low-level webapps that they interface with hardware? HTML forms don't care > if their input comes from a pipe. No web technology should. I think I already made my point.
Received on Monday, 20 December 2010 07:43:43 UTC