Only Apple do. The Android Market isn't locked down in this way - and even if it were, you can install apps from "non-market sources" on any android phone, just by clicking the checkbox (no need to root it).
Well, you can always do an out-of-band payment method, or simply use the honour system:
"If you like my app, donate what you think it's worth by paypal, then check this box to hide the ad".
I can't be the only one who thinks that adverts on a smartphone are intolerable (perhaps with the exception of on the settings or about page). I'm a bit reluctant to use adblock on a free app, but that is the alternative: if an app shows annoying intrusive ads in the user's private, personal space, it's so viscerally annoying that the user will either adblock or uninstall.
The solution here is to ensure that the app store shows the funding model of the app to the user before installation.
Personally, I *really* object to adverts on my phone: it's my personal space, my privacy, and screen/bandwidth/battery are far too limited to waste. We should be able to filter the app funding model. For example, when given 50 different apps that do basically the same thing, I'd consider:
Indeed. If an HDD fails, you should physically destroy it. Your data privacy is usually worth more than the disk. Obviously if the disk is faulty, you can't wipe it. Also, I think it's crazy buying a refurbished disk anyway - the annoyance of disk failure (even with raid) is usually greater than the price of the drive.
I'm not sure a pardon is correct - because a pardon is an acknowledgement of guilt. As a question of legal terminology, it generally means "we, the state forgive you for what you did wrong". To that extent, an innocent man may actually reject a pardon, effectively "I refuse to accept forgiveness for something I didn't do, because it would require me to admit guilt".
It's doubly complex here: we aren't trying to "forgive" Turing; what we want to acknowledge is that the crime (according to the law of the day) that he committed shouldn't have been a crime at all.
Given that we can't change history, I'd suggest a posthumous knighthood would be a far better approach.
Indeed - Mandrake made stupid decision after stupid decision. For example, when they were doing really well at the dotcom boom, they wasted all their money on a failed diversification into e-learning.
Then they asked the community for support, which many of us gave, by selling club membership and DVDs. The stupid thing was this though: I had to pay $60 for a DVD I didn't want (after downloading the release ISOs weeks earlier), and I suspect Mandrake only got about $10 of that. I would have been happy to give them $20 for every release, if I knew that the money would go to more than just production and shipping of DVDs, and the included "commercial apps" which I also didn't want.
Another problem was lack of support of the released distro. For example, if you wanted to run the latest stable release (not cooker), but happened to purchase a printer with support in upstream CUPS, you couldn't always get it to work in the stable release. Bug fixes rarely got backported either, so the stable release that everyone was supposed to run always had bugs in it that were fixed (but only in the cooker release, which was frequently broken).
It's a shame: Mandrake did some really good stuff, including excellent documentation, a good set of KDE and Gnome defaults (including a unified theme), they supported i586 while most Linuxes still optimised for i386, had a really outstanding graphical installer (back in 2001 and before), and were deservedly at the top of the list for newbies, with tools that provided help, rather than dumbing down.
Mandrake also improved several defaults, for example in Debian/Ubuntu, the Webroot is "/var/www". In Mandrake, it's "/var/www/html". When serving a simple file, this means/var/www/myfile.html (Debian) vs./var/www/html/myfile.html (Mdk) - but it puts the webapps in a sensible place: Mdv use/var/www/mediawiki,/var/www/bugzilla etc, whereas Debian have to put it into/var/lib/ iirc. (On the other hand, Mandrake's Postgres configuration is weirdly in/var/lib/pgsql/data/postgresql.conf, whereas Debian put it in/etc/postgresql, where one would expect it. )
Hopefully Mageia can do something exciting; personally I've been running Mageia 1 for 8 months, and it's good, but not yet revolutionary.
It seems to me that the West really needs to force the issue of Chinese human rights and do it now. For our own self interest, if nothing else. Only by enforcing human rights on the Chinese government (and making it easier for workers to protest) can we make these factories less economically competitive. Otherwise their slave labour will outcompete our mostly decent working conditions, and it will be a race to the bottom.
You make a good point. I always thought it rather a shame that the excellent GPE (for iPaq) never went anywhere - GPE (the gnome-palmtop-environment) ran really well on the 266 MHz CPUs of its day, and contained regular linux + busybox + X + Gtk + some applications. It fitted into 16 MB of flash and 32MB of RAM. Sadly nobody ever created the phone-peripheral to make this into a smartphone, and we ended up evolving backwards - making phones gradually more smart, instead of fitting a voice-modem to a pre-existing portable computer. As a result, Android is 7 years late, and an atrocious resource-hog. Meanwhile, we had a diversion for QTopia etc (on, for example, the Zaurus). Qt was so much slower than Gtk for embedded devices (though it was prettier if one prized beauty over speed), and the resulting systems were unusable.
Part of the problem with Android (and iPhone) is that they run a Java GUI rather than X/Gtk (thereby making them incompatible with all the old, and fast apps); the other problem is that most apps aren't GPL. The consequence of this is that there is no central package manager (with dependency resolution and shared libraries). So every single app has to bundle its own icons, its own copy of the libraries, and run in its own sandbox. This makes them far more bloated. I do like Android, but we could get at least 10x better performance out of it if the environment were better engineered.
You can easily demonstrate this to yourself: take a look at MenuetOS, which fits an OS + GUI + browser + media-player + editor + source-code on a single floppy!
It is hugely more pleasant to work in a computer lab where the noise of fans isn't deafening. Actually, you can pretty much get total silence now, and I strongly recommend it. Specify computers with fanless coolers (usually this is $25 even for a high-powered i7 cooler), avoid rotating disks (use SSDs or etherboot), avoid case-fans, and use silent PSUs (these are usually equipped with fan for use when flat-out, but tend to run inaudibly; they cost a little more, but last much longer).
This isn't really a good system for computer control, because it has no feedback - it uses simple DC motors without potentiometers or servos. So there is no way for the computer to know the arm position.
Actually, an easier way to fix this (that will go away on reboot) is: 1. Student logs in as normal, opens the test app. 2. You SSH in and kill -9 the window-manager. 3. Result: your test app is running full-screen, and there is no way to exit.
To restart the WM, you would need a shell, or Alt-F2 [and the latter is usually a feature of the WM]. There is a neater way to make this all happen if you write a shell-script. Create a desktop startup script that does something like:
#!/bin/bash
killall -9 gnome-panel
killall -9 metacity
while : ; do java-test-app; done
Then shut down the machine afterwards using Alt-SysRQ-{R,S,E,I,U,B}
Fundamentally, you're trying for the impossible: you are trying to use the app to control the window manager. This is a bit like google trying to stop you closing a browser window!
BUT: If your test happens to be multiple-choice, you could consider making the app run full-screen maximised (windowless), and then unplugging the keyboard. That would work.
[My dept has some computer systems designed for tracking who is present in the buidling; they solve the lockdown problem with a special keyboard that has only alphanumeric keys - if you physically remove the Ctrl, Alt, Esc, Fx, etc keys you can reasonably make this work!]
An alternative would be to temporarily make the system run just a single X application. If you were to change the first line of the file "/etc/X11/Xsession" to be "exec your-java-app", then you'd get a single-window desktop that runs without a window manager.
Why is there no discussion of requiring the network operators to improve security?
After all, it's fairly easy for them to know who is placing a call, and to bill the right user: I can't spoof someone else's phone and make them get the bill for my call. So why can't the carriers lock voicemail to the device by default?
I don't want the hassle of using a PIN every time I dial my voicemail, but I am quite happy for my cellphone with my SIM to be the trusted token. Yes, if I'm unlucky enough to have my phone stolen, my voicemail security will be compromised, but otherwise, the voicemail should, in principle, be as secure as the degree to which I protect my phone. So, why is this not true in practice?
I bought one: it was a wonderful device (except for the touchscreen being resistive), but what killed it for us was that it had critical parts of the GUI being non-open. We would have deployed 200 of them, but we needed one minor bugfix: the ability to operate the camera during a voip phone call (which meant being able to disable the shutter sound so as to allow the sound-card not to block). Sadly, the camera library was crippled: it's very, very easy to use Hildon (basically GTK) to access the image capture with gstreamer, but we only if we didn't use the proprietary feature called "auto-focus"! So we couldn't fix it ourselves, and when we reported the bug, the Nokia team confirmed it, but didn't actually get round to fixing it (at least, not within many weeks).
It's a real shame too: I could SSH into the phone, launch X-applications (on either $DISPLAY), install applications with a real package manager, and enjoy all the other Linux goodness.
It's still the case that digital camera sensors are far less capable than film: an $800 digital camera will be outperformed by a $100 second-hand film camera (eg Olympus OM-2). This is partly image quality, and partly sensitivity/noise (you can take a 2 year 'exposure' with film and see no noise, whereas most DSLRs can't exceed 30 seconds).
Part of the reason is that manufacturers of sensors tend to lie: a nominal "12 megapixel sensor" is really a "12 mega-subpixel-sensor", which has only 3M true pixels, and a major deconvolution problem too (see Bayer filter). So my advice would be to buy the best possible sensor you can get (perhaps the Canon 550D or Olympus XZ-1), and never mind the details of the rest.
Yes... but maybe we missed them. Neutrinos are really hard to detect, let alone identify the source direction. Given a non-directional, not-very-strong pulse, possibly widely distributed in time, an unknown amount of time before the supernova, which we weren't expecting, would it really be surprising to have missed it?
You can! Download an old copy of the Mandrake installation discs. 7.2 will get you KDE 1; 8.2 will get you KDE 2. Having used both in their time, KDE1 is very much a first try, while the final releases of KDE 2 were actually pretty good.
I don't think you are quite seeing the point. Apple have built a large amount of their success based on the hacker-culture/BSD/GCC/etc. In return, it seems fair to expect them to "play nice" with that culture from which they benefited.
In this example, I'm not asking that they should do lots of work to make iTunes run on Linux (although as a customer, I might want that too!); what I'm asking is that Apple should release the specs and be generally helpful to the libgpod folks. This would cost them almost nothing. Instead of which, they actively invest engineering effort in making the products harder to modify.
To summarise:
1. Imho, it's bad karma for Apple to be so uncooperative with the hacker-community. They do give back what they are obligated to, but in a very narrowly focused, rather than principled way.
2. As a geek and a potential apple customer, my purchases (and my time, development-effort, and recommendations) are not going to Apple because of some deliberate decisions on their part to make me not want to be a customer. This is all the more annoying because the openness-features that I want would be easy to implement, and would not negatively impact any of their existing customer-base.
It's true that Android has bugs, perhaps more than iOS does. But Android doesn't have mis-features like iOS. You can fix bugs yourself (or submit an issue to the tracker), but it's nearly impossible to remove the mis-features, which were put there deliberately to make your experience of the device more restrictive.
It's not really so much a failure to give back specific things (though honestly, a pretty GUI would seem a fair exchange for a stable kernel); it's the locking people out. The closed specifications are far worse than the closed-source GUI. I don't mind that Apple don't provide source for the pretty bits of OSX; what I object to is that they don't provide documentation on how to get an iPod to work with Linux, or that they don't provide instructions on how to root an iPad. They're also a really bad technical influence on the ecosystem, for instance by blocking Ogg from the iPod, or spoiling HTML5 video by getting out the patents.
I don't expect Apple to work for me for free; what I expect is that they don't add deliberate misfeatures to reduce the utility and hackability of a product I might buy, and that they don't work to undermine the open-source community.
There are 3 important differential-heights to get correct:
floor -> chair (so your legs are comfortable)
chair -> desk (so your arms are comfortable once sitting)
chair -> monitor (so your eyes are at a sensible level and you are not looking down too much).
Laptops are a disaster for this unless you use a laptop stand + external keyboard. Also, imho, most tables are too low, and you'll never get it right if you don't fix this first.
Also, get the monitor a comfortable distance from your eyes to avoid eye-strain; if necessary, then make the monitor larger, or put it on an angle-poise style arm.
Slashdot Mirror
Only Apple do. The Android Market isn't locked down in this way - and even if it were, you can install apps from "non-market sources" on any android phone, just by clicking the checkbox (no need to root it).
Well, you can always do an out-of-band payment method, or simply use the honour system:
"If you like my app, donate what you think it's worth by paypal, then check this box to hide the ad".
I can't be the only one who thinks that adverts on a smartphone are intolerable (perhaps with the exception of on the settings or about page).
I'm a bit reluctant to use adblock on a free app, but that is the alternative: if an app shows annoying intrusive ads in the user's private, personal space, it's so viscerally annoying that the user will either adblock or uninstall.
Please allow your users to upgrade to a paid, ad-free version, or include it as a "pro" version in the app store.
The solution here is to ensure that the app store shows the funding model of the app to the user before installation.
Personally, I *really* object to adverts on my phone: it's my personal space, my privacy, and screen/bandwidth/battery are far too limited to waste.
We should be able to filter the app funding model. For example, when given 50 different apps that do basically the same thing, I'd consider:
[Best] ; F0SS (GPL/BSD etc) ; Free beer, closed source without ads. ; Paid ; Advert funded ; Demo ; Broken ; [Worst]
I filed a bug on this with google, but it was wontfixed.
What's really annoying is that some decent apps only exist in a free version with ads, and don't even have an option to get a paid, ad-free version.
Indeed. If an HDD fails, you should physically destroy it. Your data privacy is usually worth more than the disk.
Obviously if the disk is faulty, you can't wipe it. Also, I think it's crazy buying a refurbished disk anyway - the annoyance of disk failure (even with raid) is usually greater than the price of the drive.
I'm not sure a pardon is correct - because a pardon is an acknowledgement of guilt. As a question of legal terminology, it generally means "we, the state forgive you for what you did wrong". To that extent, an innocent man may actually reject a pardon, effectively "I refuse to accept forgiveness for something I didn't do, because it would require me to admit guilt".
It's doubly complex here: we aren't trying to "forgive" Turing; what we want to acknowledge is that the crime (according to the law of the day) that he committed shouldn't have been a crime at all.
Given that we can't change history, I'd suggest a posthumous knighthood would be a far better approach.
Indeed - Mandrake made stupid decision after stupid decision. For example, when they were doing really well at the dotcom boom, they wasted all their money on a failed diversification into e-learning.
Then they asked the community for support, which many of us gave, by selling club membership and DVDs. The stupid thing was this though: I had to pay $60 for a DVD I didn't want (after downloading the release ISOs weeks earlier), and I suspect Mandrake only got about $10 of that. I would have been happy to give them $20 for every release, if I knew that the money would go to more than just production and shipping of DVDs, and the included "commercial apps" which I also didn't want.
Another problem was lack of support of the released distro. For example, if you wanted to run the latest stable release (not cooker), but happened to purchase a printer with support in upstream CUPS, you couldn't always get it to work in the stable release. Bug fixes rarely got backported either, so the stable release that everyone was supposed to run always had bugs in it that were fixed (but only in the cooker release, which was frequently broken).
It's a shame: Mandrake did some really good stuff, including excellent documentation, a good set of KDE and Gnome defaults (including a unified theme), they supported i586 while most Linuxes still optimised for i386, had a really outstanding graphical installer (back in 2001 and before), and were deservedly at the top of the list for newbies, with tools that provided help, rather than dumbing down.
Mandrake also improved several defaults, for example in Debian/Ubuntu, the Webroot is "/var/www". In Mandrake, it's "/var/www/html". When serving a simple file, this means /var/www/myfile.html (Debian) vs. /var/www/html/myfile.html (Mdk) - but it puts the webapps in a sensible place: Mdv use /var/www/mediawiki, /var/www/bugzilla etc, whereas Debian have to put it into /var/lib/ iirc. (On the other hand, Mandrake's Postgres configuration is weirdly in /var/lib/pgsql/data/postgresql.conf, whereas Debian put it in /etc/postgresql, where one would expect it. )
Hopefully Mageia can do something exciting; personally I've been running Mageia 1 for 8 months, and it's good, but not yet revolutionary.
It seems to me that the West really needs to force the issue of Chinese human rights and do it now. For our own self interest, if nothing else. Only by enforcing human rights on the Chinese government (and making it easier for workers to protest) can we make these factories less economically competitive. Otherwise their slave labour will outcompete our mostly decent working conditions, and it will be a race to the bottom.
You make a good point. I always thought it rather a shame that the excellent GPE (for iPaq) never went anywhere - GPE (the gnome-palmtop-environment) ran really well on the 266 MHz CPUs of its day, and contained regular linux + busybox + X + Gtk + some applications. It fitted into 16 MB of flash and 32MB of RAM. Sadly nobody ever created the phone-peripheral to make this into a smartphone, and we ended up evolving backwards - making phones gradually more smart, instead of fitting a voice-modem to a pre-existing portable computer. As a result, Android is 7 years late, and an atrocious resource-hog. Meanwhile, we had a diversion for QTopia etc (on, for example, the Zaurus). Qt was so much slower than Gtk for embedded devices (though it was prettier if one prized beauty over speed), and the resulting systems were unusable.
Part of the problem with Android (and iPhone) is that they run a Java GUI rather than X/Gtk (thereby making them incompatible with all the old, and fast apps); the other problem is that most apps aren't GPL. The consequence of this is that there is no central package manager (with dependency resolution and shared libraries). So every single app has to bundle its own icons, its own copy of the libraries, and run in its own sandbox. This makes them far more bloated. I do like Android, but we could get at least 10x better performance out of it if the environment were better engineered.
You can easily demonstrate this to yourself: take a look at MenuetOS, which fits an OS + GUI + browser + media-player + editor + source-code on a single floppy!
It is hugely more pleasant to work in a computer lab where the noise of fans isn't deafening. Actually, you can pretty much get total silence now, and I strongly recommend it. Specify computers with fanless coolers (usually this is $25 even for a high-powered i7 cooler), avoid rotating disks (use SSDs or etherboot), avoid case-fans, and use silent PSUs (these are usually equipped with fan for use when flat-out, but tend to run inaudibly; they cost a little more, but last much longer).
This isn't really a good system for computer control, because it has no feedback - it uses simple DC motors without potentiometers or servos. So there is no way for the computer to know the arm position.
You do realise there is a "dog" command? It's written as "an improved version of cat".
Actually, an easier way to fix this (that will go away on reboot) is:
1. Student logs in as normal, opens the test app.
2. You SSH in and kill -9 the window-manager.
3. Result: your test app is running full-screen, and there is no way to exit.
To restart the WM, you would need a shell, or Alt-F2 [and the latter is usually a feature of the WM]. There is a neater way to make this all happen if you write a shell-script. Create a desktop startup script that does something like:
#!/bin/bash
killall -9 gnome-panel
killall -9 metacity
while : ; do java-test-app; done
Then shut down the machine afterwards using Alt-SysRQ-{R,S,E,I,U,B}
Fundamentally, you're trying for the impossible: you are trying to use the app to control the window manager.
This is a bit like google trying to stop you closing a browser window!
BUT: If your test happens to be multiple-choice, you could consider making the app run full-screen maximised (windowless), and then unplugging the keyboard. That would work.
[My dept has some computer systems designed for tracking who is present in the buidling; they solve the lockdown problem with a special keyboard that has only alphanumeric keys - if you physically remove the Ctrl, Alt, Esc, Fx, etc keys you can reasonably make this work!]
An alternative would be to temporarily make the system run just a single X application. If you were to change the first line of the file "/etc/X11/Xsession" to be "exec your-java-app", then you'd get a single-window desktop that runs without a window manager.
Why is there no discussion of requiring the network operators to improve security?
After all, it's fairly easy for them to know who is placing a call, and to bill the right user: I can't spoof someone else's phone and make them get the bill for my call. So why can't the carriers lock voicemail to the device by default?
I don't want the hassle of using a PIN every time I dial my voicemail, but I am quite happy for my cellphone with my SIM to be the trusted token. Yes, if I'm unlucky enough to have my phone stolen, my voicemail security will be compromised, but otherwise, the voicemail should, in principle, be as secure as the degree to which I protect my phone. So, why is this not true in practice?
I bought one: it was a wonderful device (except for the touchscreen being resistive), but what killed it for us was that it had critical parts of the GUI being non-open.
We would have deployed 200 of them, but we needed one minor bugfix: the ability to operate the camera during a voip phone call (which meant being able to disable the shutter sound so as to allow the sound-card not to block). Sadly, the camera library was crippled: it's very, very easy to use Hildon (basically GTK) to access the image capture with gstreamer, but we only if we didn't use the proprietary feature called "auto-focus"! So we couldn't fix it ourselves, and when we reported the bug, the Nokia team confirmed it, but didn't actually get round to fixing it (at least, not within many weeks).
It's a real shame too: I could SSH into the phone, launch X-applications (on either $DISPLAY), install applications with a real package manager, and enjoy all the other Linux goodness.
It's still the case that digital camera sensors are far less capable than film: an $800 digital camera will be outperformed by a $100 second-hand film camera (eg Olympus OM-2). This is partly image quality, and partly sensitivity/noise (you can take a 2 year 'exposure' with film and see no noise, whereas most DSLRs can't exceed 30 seconds).
Part of the reason is that manufacturers of sensors tend to lie: a nominal "12 megapixel sensor" is really a "12 mega-subpixel-sensor", which has only 3M true pixels, and a major deconvolution problem too (see Bayer filter). So my advice would be to buy the best possible sensor you can get (perhaps the Canon 550D or Olympus XZ-1), and never mind the details of the rest.
Also, take a look at dpreview.com
I think you are confused with "O, Tannenbaum" - note the double "n", which is the song translated as "O, Christmas tree".
Yes... but maybe we missed them. Neutrinos are really hard to detect, let alone identify the source direction. Given a non-directional, not-very-strong pulse, possibly widely distributed in time, an unknown amount of time before the supernova, which we weren't expecting, would it really be surprising to have missed it?
You can! Download an old copy of the Mandrake installation discs. 7.2 will get you KDE 1; 8.2 will get you KDE 2. Having used both in their time, KDE1 is very much a first try, while the final releases of KDE 2 were actually pretty good.
I don't think you are quite seeing the point. Apple have built a large amount of their success based on the hacker-culture/BSD/GCC/etc. In return, it seems fair to expect them to "play nice" with that culture from which they benefited.
In this example, I'm not asking that they should do lots of work to make iTunes run on Linux (although as a customer, I might want that too!); what I'm asking is that Apple should release the specs and be generally helpful to the libgpod folks. This would cost them almost nothing. Instead of which, they actively invest engineering effort in making the products harder to modify.
To summarise:
1. Imho, it's bad karma for Apple to be so uncooperative with the hacker-community. They do give back what they are obligated to, but in a very narrowly focused, rather than principled way.
2. As a geek and a potential apple customer, my purchases (and my time, development-effort, and recommendations) are not going to Apple because of some deliberate decisions on their part to make me not want to be a customer. This is all the more annoying because the openness-features that I want would be easy to implement, and would not negatively impact any of their existing customer-base.
Try CyanogenMod first :-)
It's true that Android has bugs, perhaps more than iOS does. But Android doesn't have mis-features like iOS. You can fix bugs yourself (or submit an issue to the tracker), but it's nearly impossible to remove the mis-features, which were put there deliberately to make your experience of the device more restrictive.
It's not really so much a failure to give back specific things (though honestly, a pretty GUI would seem a fair exchange for a stable kernel); it's the locking people out. The closed specifications are far worse than the closed-source GUI. I don't mind that Apple don't provide source for the pretty bits of OSX; what I object to is that they don't provide documentation on how to get an iPod to work with Linux, or that they don't provide instructions on how to root an iPad. They're also a really bad technical influence on the ecosystem, for instance by blocking Ogg from the iPod, or spoiling HTML5 video by getting out the patents.
I don't expect Apple to work for me for free; what I expect is that they don't add deliberate misfeatures to reduce the utility and hackability of a product I might buy, and that they don't work to undermine the open-source community.
There are 3 important differential-heights to get correct:
floor -> chair (so your legs are comfortable)
chair -> desk (so your arms are comfortable once sitting)
chair -> monitor (so your eyes are at a sensible level and you are not looking down too much).
Laptops are a disaster for this unless you use a laptop stand + external keyboard. Also, imho, most tables are too low, and you'll never get it right if you don't fix this first.
Also, get the monitor a comfortable distance from your eyes to avoid eye-strain; if necessary, then make the monitor larger, or put it on an angle-poise style arm.
Use the gel-wrist-rests under the elbows though... that can help if you have a hard desk and don't wear a sweater.