I'm an android developer, and I *do* share his concerns. There are three aspects to the problem:
1. Unintended device differences.
I've had loads of emails from people saying that my app behaves incorrectly on their phone, whereas it works perfectly well on mine and many other phones. There are certain areas of development where the differences in behaviour on different devices can be pretty huge. In my case it is sleeping and waking the device, but there are others, like sound latency, graphics capabilities, and multitouch behaviour (*cough* stupid dual-touch *cough*). You really do need to test these apps on the actual phones in order to make sure they work (or wait for "I am shocked that this free, ad-free app doesn't work." emails).
2. Intended device differences.
There are a ton of different android phones. Suppose you want a layout to work nicely on all of them. Android has a pretty nice framework for selecting a layout file based on device differences, and these are only *some* of the things it can consider: orientation, whether the keyboard is open, screen size, screen aspect ratio and night mode. Multiply those and you have a lot of work. Ok presumably you wouldn't use all of them, but you could *easily* end up with 6 layouts for one screen. It doesn't help that Android's layout system is one of... no *the* least well behaved I've used. It frequently does stuff that makes no sense (search StackOverflow for examples).
3. Old versions of Android.
Yes it is a bigger problem than on iOS. 17% of users are still on Android 1.5 or 1.6. How many iOS users haven't upgraded their OS for a year? Actually I checked, and Apple stopped providing updates for the original iPhone 2.5 years after its release. It seems most Android phones don't even last a year before they are end-of-lined. This affects developers because it means you can't use the latest nice APIs without either using ugly reflection hacks (not possible with the NDK) or ignoring some users.
There's lots to like about Android, but don't pretend there aren't any flaws.
Anyone who has used android knows this is true. There are loads of apps that ask for permissions they clearly shouldn't need. Most often it is for internet access, your location, your phone ID (IMSI), and sometimes access to your contacts.
Obviously the crappy little 'content' apps like DailyHoroscope, backgrounds and ringtones are the main culprits.
My phone, my router, my PCs, my GPS, all have firmware I've compiled myself.
Who modded this insightful?
Do you even have the source code for your GPS firmware, the baseband in your phone, your PC's BIOS and so on? No. Even if you did, are you seriously saying that you've perfectly audited hundreds of thousands of lines of code?
Does DNSSEC allow storing SSL certificates in the DNS records? It would seem that this is an awesome way of getting free SSL certificates.
Also, I doubt anyone bothered with this, but does DNSSEC have any way of saying "this domain should only be contacted with SSL"? That would prevent SSL stripping MitM attacks.
It requires jailbreaking, and they said "Native applications are web applications. It's fully possible to do everything that you can do with native technology with web technologies". Which sounds pretty dubious, although it does apparently support NativeClient.
You're wrong. Even if you use common first and surnames, say 1000 of each, and say that 10 of the credit card digits are unknown that gives you 1000x1000x10^10 = 10^16 combinations.
For comparison, that's about the same as a 9-character random password containing lowercase, uppercase and digits. Good luck with that.
"lithium [...] now costs less than a dollar per kilogram" "researchers who worked on a seawater project in Japan for some 30 years concluded the technology was five times too expensive to commercialize" "a company called Simbol Mining [is] now “exploring the feasibility” of drawing lithium from geothermal sources, and the Times reports that some 60 mining firms are conducting feasibility studies"
Huh? You're saying Wii, which is accelerometer based, is similar to light gun game. But the technology used in the light gun games is more similar to vision-based people tracking or IMU tracking? That makes no sense.
Anyway, I agree that Wii was more revolutionary than evolutionary in the sense that it was the first console not to use game pads.
This ambient-light recording has to happen at a different time to when the laser is fired, so variable-light conditions or the lack of an incredibly steady camera, image object and reflective surface will make it basically impossible to render the image.
Yeah but you could record the ambient light 1 ms later. Light is pretty damn fast and ambient light conditions are essentially constant.
I absolutely love the concept. I just think that the nay-sayers whom Professor Raskar claims to be defeating were correct. It might not be theoretically impossible, but the practical limitations are so severe that I don't envisage them being "engineered" away - and if they are, such phenomenal engineering accomplishments would make this application appear trivial in comparison with the other things we could do.
I wouldn't be so sceptical. The main limitations are:
1. Miniaturisation. Obviously this is just an engineering problem. A damn hard one, sure. 2. Sampling rate of the light signal. This is the one that will really determine the image quality.
The door requirement is a pretty big limitation though. I think a fibre optic camera poked around the corner might be a bit easier!
The same is true on linux. Fairly often I accidentally allocate a huge (e.g. 16 GB) array in matlab, causing the computer to instantly go into "MUST SWAP EVERYTHING" mode. The entire desktop freezes and I can't kill matlab. Always ends in a hard reset.:-(
Slashdot Mirror
I'm an android developer, and I *do* share his concerns. There are three aspects to the problem:
1. Unintended device differences.
I've had loads of emails from people saying that my app behaves incorrectly on their phone, whereas it works perfectly well on mine and many other phones. There are certain areas of development where the differences in behaviour on different devices can be pretty huge. In my case it is sleeping and waking the device, but there are others, like sound latency, graphics capabilities, and multitouch behaviour (*cough* stupid dual-touch *cough*). You really do need to test these apps on the actual phones in order to make sure they work (or wait for "I am shocked that this free, ad-free app doesn't work." emails).
2. Intended device differences.
There are a ton of different android phones. Suppose you want a layout to work nicely on all of them. Android has a pretty nice framework for selecting a layout file based on device differences, and these are only *some* of the things it can consider: orientation, whether the keyboard is open, screen size, screen aspect ratio and night mode. Multiply those and you have a lot of work. Ok presumably you wouldn't use all of them, but you could *easily* end up with 6 layouts for one screen. It doesn't help that Android's layout system is one of... no *the* least well behaved I've used. It frequently does stuff that makes no sense (search StackOverflow for examples).
3. Old versions of Android.
Yes it is a bigger problem than on iOS. 17% of users are still on Android 1.5 or 1.6. How many iOS users haven't upgraded their OS for a year? Actually I checked, and Apple stopped providing updates for the original iPhone 2.5 years after its release. It seems most Android phones don't even last a year before they are end-of-lined. This affects developers because it means you can't use the latest nice APIs without either using ugly reflection hacks (not possible with the NDK) or ignoring some users.
There's lots to like about Android, but don't pretend there aren't any flaws.
Or Urban Dictionary. That was blocked by my mobile provider, but it's occasionally a useful site.
Anyone who has used android knows this is true. There are loads of apps that ask for permissions they clearly shouldn't need. Most often it is for internet access, your location, your phone ID (IMSI), and sometimes access to your contacts.
Obviously the crappy little 'content' apps like DailyHoroscope, backgrounds and ringtones are the main culprits.
It's well worth reading the comments under the Guardian article!
My phone, my router, my PCs, my GPS, all have firmware I've compiled myself.
Who modded this insightful?
Do you even have the source code for your GPS firmware, the baseband in your phone, your PC's BIOS and so on? No. Even if you did, are you seriously saying that you've perfectly audited hundreds of thousands of lines of code?
Where's the "-1 this is really stupid" option?
Does DNSSEC allow storing SSL certificates in the DNS records? It would seem that this is an awesome way of getting free SSL certificates.
Also, I doubt anyone bothered with this, but does DNSSEC have any way of saying "this domain should only be contacted with SSL"? That would prevent SSL stripping MitM attacks.
I'm right handed but often use a mouse with my left (my dad is left-handed). Swapping the buttons would just be really really confusing!
It requires jailbreaking, and they said "Native applications are web applications. It's fully possible to do everything that you can do with native technology with web technologies". Which sounds pretty dubious, although it does apparently support NativeClient.
Well, no. They cost the same as bigger tablets because small electronics are expensive. Same reason laptops are more expensive than desktops.
The carrier 'subsidy' trickery just lets them sell more.
2 is in every android phone, and 3 is in nearly all of them.
Seriously. Why does anyone use them?
Yeah but now it's concave!
I'm pretty sure the journalist is just confused (I know - surprising right?), and it is actually using Foxit's PDF library:
http://www.foxitsoftware.com/pdf/sdk/dll/
This also means it won't end up in Chromium.
It doens't load web bugs until you tell it to.
That's a standard feature on every mail client I've used for at least the past 5 years.
I wonder if they can translate US patents into English...
You're wrong. Even if you use common first and surnames, say 1000 of each, and say that 10 of the credit card digits are unknown that gives you
1000x1000x10^10 = 10^16 combinations.
For comparison, that's about the same as a 9-character random password containing lowercase, uppercase and digits. Good luck with that.
You can mine it indefinitely from seawater for about $70 per kg.
Huh apparently this is true. Sort of. It seems to be still in the "way too expensive and theoretical" stage.
http://gigaom.com/cleantech/will-seawater-stave-off-a-lithium-squeeze/
Quotes:
"lithium [...] now costs less than a dollar per kilogram"
"researchers who worked on a seawater project in Japan for some 30 years concluded the technology was five times too expensive to commercialize"
"a company called Simbol Mining [is] now “exploring the feasibility” of drawing lithium from geothermal sources, and the Times reports that some 60 mining firms are conducting feasibility studies"
You can get in trouble 'just for saying something' in the US too. You have libel, copyright, and discrimination laws that limit your speech a bit.
Huh? You're saying Wii, which is accelerometer based, is similar to light gun game. But the technology used in the light gun games is more similar to vision-based people tracking or IMU tracking? That makes no sense.
Anyway, I agree that Wii was more revolutionary than evolutionary in the sense that it was the first console not to use game pads.
The Wii was a direct outgrowth of similar devices in arcades.
What arcade game is it similar to?
This ambient-light recording has to happen at a different time to when the laser is fired, so variable-light conditions or the lack of an incredibly steady camera, image object and reflective surface will make it basically impossible to render the image.
Yeah but you could record the ambient light 1 ms later. Light is pretty damn fast and ambient light conditions are essentially constant.
I absolutely love the concept. I just think that the nay-sayers whom Professor Raskar claims to be defeating were correct. It might not be theoretically impossible, but the practical limitations are so severe that I don't envisage them being "engineered" away - and if they are, such phenomenal engineering accomplishments would make this application appear trivial in comparison with the other things we could do.
I wouldn't be so sceptical. The main limitations are:
1. Miniaturisation. Obviously this is just an engineering problem. A damn hard one, sure.
2. Sampling rate of the light signal. This is the one that will really determine the image quality.
The door requirement is a pretty big limitation though. I think a fibre optic camera poked around the corner might be a bit easier!
So if you only use GUI apps, it has no effect at all?
Well it's not completely undocumented. There is a lot of documentation, but there is also a hell of a lot of missing documentation.
http://www.google.com/search?q=android+%22completely+undocumented%22
Android is documented
Ha ha good one.
The same is true on linux. Fairly often I accidentally allocate a huge (e.g. 16 GB) array in matlab, causing the computer to instantly go into "MUST SWAP EVERYTHING" mode. The entire desktop freezes and I can't kill matlab. Always ends in a hard reset. :-(