LWN: Comments on "Tasting the Ice Cream Sandwich"

Tasting the Ice Cream Sandwich

hummassa — Sun, 01 Jul 2012 03:38:19 +0000

Interesting; my personal experience has been exactly the opposite. I have upgraded my kubuntu system (which uses network manager exclusively, all configured via GUI, and with some quirks in my workplace network) regularly at every release, and for the last three years or so I have not touch said configuration.

Tasting the Ice Cream Sandwich

mirabilos — Sun, 24 Jun 2012 11:30:29 +0000

Sorry, I don’t have these applications, so I don’t know about them.
This all also depends on which mksh you’re using.

Basically, mksh looks in "${ENV:-~/mkshrc}", and the path has been
changed to /system/etc/mkshrc in the in-tree AOSP version. There’s
also /etc/profile and ~/.profile…

Tasting the Ice Cream Sandwich

nye — Thu, 07 Jun 2012 16:34:57 +0000

> If you have a run-of-the-mill, garden-variety network configuration, it will not stop working every six months

The idea of what is 'run-of-the-mill' changes too frequently for my liking - for example, the last three Ubuntu releases have not supported wired networking out of the box (with differing varieties of breakage depending on the release and the machine in question).

I suspect the reason that the default configuration doesn't just attempt to make a DHCP connection has something to do with NetworkManager being expected to bring up the interface, but I've managed to get NM to bring up a wired connection on precisely one occasion - and that broke on the next release upgrade; certainly I've never seen it work automatically.

Unfortunately the only machine I have access to that has a wireless connection has some kind of Broadcom chip that I think needs special firmware that I've not bothered to track down, so I can't speak for how well wireless works on supported hardware. (I do apparently have a 'Broadcom STA propretary wireless driver' installed, but it seems that's not enough.)

Actually, now that I think about it there was an Ubuntu release a while back (it was around the release of KDE4.2, so presumably it was 9.04) which did get the wireless device in this machine to work without any special configuration that I can recall; alas the next release came along and hosed it so thoroughly that I couldn't figure out how to get any networking back *at all*, and eventually resorted to reinstalling from scratch.

In contrast, I have a rather more complex setup on my Debian systems which has worked reliably for many years, but they required a reasonable amount of technical knowledge to configure in the first place. But Debian has its own problems of course; there's no single option that won't periodically come with pain.

Tasting the Ice Cream Sandwich

hummassa — Wed, 06 Jun 2012 19:23:05 +0000

> Nobody would seriously try to claim that 'the latest 64-bit version of Windows no longer runs my 16-bit Windows applications from 1992' is in any way equivalent to 'my networking stops working every six months'.

Ok, slow down. If you have a run-of-the-mill, garden-variety network configuration, it will not stop working every six months (mine worked without modification for the last ten years or so). AND if you have some complicated, exotic configuration with strange and mysterious drivers, then every Windows update or hotpatch is an adventure.

I know that for some video configurations, things were far rougher on Linux than on Windows.

> There are probably some other examples of extremely cheap hardware with drivers that work in one Windows release but not the next, however in the vast majority of cases that hardware either doesn't work in Linux *at all*, or works well enough to satisfy a tick-list but not well enough to actually use (eg a webcam that manages 30fps in Windows, but 2 fps in Linux). I'm aware that in such a case it's the manufacturer of that crappy hardware that's at fault, but then to make a fair comparison you need to acknowledge that in the case of Windows as well.

Actually, I had problems on Linux on the "extremely expensive and especialized hardware" range more often than on the "dollar-store hardware" range.

> Statements like yours are a textbook example of why Linux - and Free Software in general - is not taken seriously by normal computer users, since the only thing you are interested in is nursing your damaged pride at all costs. The very idea that a competitor might be better in some way must not be entertained under any circumstances, with the inevitable result that real deficiencies cannot be fixed because they cannot even be acknowledged.

People acknowledge and fix those problems much more often in the Linux world than in the Windows/OSX worlds. Oh, there are lots of hardware best supported on Windows and OSX. The model where the hardware maker is usually also the driver maker works faster, even if it does not work so well.

Tasting the Ice Cream Sandwich

nye — Wed, 06 Jun 2012 15:15:27 +0000

I'll elaborate a little:

The problem is that you are obviously aware that your statement is an attempt to compare entirely unrelated things in entirely unrelated contexts, simply as a means of avoiding making any real point. It is deliberately dismissive and inflammatory, and very blatantly intended as nothing but snide trolling.

The regressions in Windows in successive releases are so trivial in comparison with the regressions in Linux distributions that trying to claim any equivalence displays, at best, a breathtaking level of ignorance. Furthermore, every Windows version is supported for longer than the best support level available in any Linux distribution, while simultaneously making it trivial for any user to use the latest version of any application software they desire.

Nobody would seriously try to claim that 'the latest 64-bit version of Windows no longer runs my 16-bit Windows applications from 1992' is in any way equivalent to 'my networking stops working every six months'.

Pretty much the only legitimately comparable example is that there are a number of printers for which the existing drivers haven't worked in new OS releases - and Ubuntu has that problem periodically in minor (non-release) updates, so doesn't exactly come out ahead.

There are probably some other examples of extremely cheap hardware with drivers that work in one Windows release but not the next, however in the vast majority of cases that hardware either doesn't work in Linux *at all*, or works well enough to satisfy a tick-list but not well enough to actually use (eg a webcam that manages 30fps in Windows, but 2 fps in Linux). I'm aware that in such a case it's the manufacturer of that crappy hardware that's at fault, but then to make a fair comparison you need to acknowledge that in the case of Windows as well.

Statements like yours are a textbook example of why Linux - and Free Software in general - is not taken seriously by normal computer users, since the only thing you are interested in is nursing your damaged pride at all costs. The very idea that a competitor might be better in some way must not be entertained under any circumstances, with the inevitable result that real deficiencies cannot be fixed because they cannot even be acknowledged.

Tasting the Ice Cream Sandwich

nye — Wed, 06 Jun 2012 14:48:28 +0000

I'm sorry, but I can't assume good faith when you make such an obviously ludicrous statement.

Tasting the Ice Cream Sandwich

anselm — Sat, 02 Jun 2012 12:18:54 +0000

But, in real life, I don't think there are as many fashion victims as you say. Used Mac hardware is still outrageously expensive, even previous-gen iPhones and iPads without cameras. That seems to directly contradict your theory.

No, it doesn't. Used Apple stuff is still Apple stuff, i.e., cool and fashionable. If you want Apple stuff in preference to other stuff but can't really afford to buy it new then you buy it used. Many people seem to want even the used Apple stuff so there is a lot of demand. Demand keeps the prices for used Apple stuff up.

The new Apple stuff is quite expensive in order to cream off those people who must have the new up-to-date Apple stuff and are ready to pay for it. Do note that, e.g., the iPad 2 is still available new but priced somewhat lower than it used to be when it was the top-of-the-line model. This makes it more accessible to people who want new (as in, unused) Apple stuff but not the very new Apple stuff at the premium price. It takes a trained eye to distinguish the iPad 2 from the »new iPad« but it still has an Apple logo, so the »lifestyle incentive« of being seen with cool Apple stuff is the same. (Also, it's supposedly not a bad tablet at all.)

Tasting the Ice Cream Sandwich

bronson — Fri, 01 Jun 2012 16:45:35 +0000

If people are foolish enough to spend $600 merely to be »with it« then more power to Apple. They should be encouraged to take advantage of that revenue stream.

But, in real life, I don't think there are as many fashion victims as you say. Used Mac hardware is still outrageously expensive, even previous-gen iPhones and iPads without cameras. That seems to directly contradict your theory.

"Last year's phone"

rschroev — Fri, 01 Jun 2012 09:17:39 +0000

Interesting.

Mine says:
Display 45%
Cell standby 18%
Dolphin Browser HD 9%
Phone idle 7%
Wi-Fi 6%
Android OS 5%
Android System 5%
...

The screen is most certainly not on most of the time. Also I didn't realize Dolphin uses that much power; from the stats you'd think I do nothing but surfing the web all the time, which is not true at all.

The phone is a Samsung Galaxy S (GT-I9000) running Cyanogenmod 7.1 (but I seem to remember that the start where more or less the same when the phone still had the stock software).

Tasting the Ice Cream Sandwich

dlang — Thu, 31 May 2012 18:55:45 +0000

every windows upgrade (including service packs) break things on windows systems as well. somehow people manage

"Last year's phone"

nye — Thu, 31 May 2012 15:49:21 +0000

> battery life is dominated by the backlight anyway.

What? How can that be? Are you using your phone with the display on constantly, while still using nearly zero CPU?

Have you actually tested this, or is it just an intuitive assumption? If the former, what kind of screen does your device have? If the latter, your intuition is incorrect.

Unless I'm using the power-guzzing Google navigation, 'cell standby' is always at the top of my battery usage list, followed by 'phone idle', followed by 'Wi-Fi' if I've had it enabled, usually followed by 'Android system', with perhaps a few other entries, and 'display' dead last - and that's if I've had it on for the hour or so that it takes to even show up in the list (I think there's a threshold of 1 or 2 percent).

If I do nothing after unplugging the phone but tap the screen every now and then to keep it on, 'display' will go up to 20%, with 'Android' system in the 60s, so if that were more efficient it would have a substantial impact on battery life (although in the real world case 'cell standby', 'phone idle', and 'Wi-Fi' tend to come to around 90% of the total battery use).

Tasting the Ice Cream Sandwich

nye — Thu, 31 May 2012 15:18:59 +0000

>more likely that's how it will work in RTM.

(Except when secure boot is enabled, obviously, since that would entirely defeat the point of secure boot)

Tasting the Ice Cream Sandwich

nye — Thu, 31 May 2012 15:15:07 +0000

I don't know about Windows 7 or Server 2008, but I do have experience of Windows 8 CP.

One of the advanced reboot options is to disable driver signing enforcement for the next boot. You can then install unsigned drivers by clicking through a scary warning as in previous versions of Windows. Once that driver is installed, you can reboot in normal mode and continue using it. I'm not certain if there's a boot flag that can be set *permanently* to keep enforcement disabled, but in practice that's only going to be a problem if you need to install unsigned drivers on a frequent basis, and to be honest I can't really fault MS for not considering that a high-priority use case.

Since the advanced reboot options menu is entirely new to Windows 8, I doubt it is a left-over from old versions that they're planning to remove in the final release; more likely that's how it will work in RTM.

Tasting the Ice Cream Sandwich

nye — Thu, 31 May 2012 12:52:56 +0000

>Which itself proves that if Linux would come preinstalled on one half of desktops (the less expensive half), and Windows/OSX on the other half, most of the people would chose Linux desktops and stay with them

I thought that too at one time, but it simply can't be true.

As you use your computer day-by-day, month-by-month, try to think about all the things you do without thinking. I mean things like subconsciously learning that if you do certain things, something crashes or the computer grinds to a halt.

Or things like upgrading your distribution and needing a simple one-line fix to get something working again. Or getting new hardware and needing to spend even just a few minutes Googling for how to get it to work.

For most of us here these obstacles are insignificant. We don't really even notice that we need to spend five minutes now-and-then fixing things, because it's so easy.

But consider the perspective of someone who actually *can't* do those things. That's not a five-minute interruption in exchange for new cool versions of software; it's a complete inability to do something you used to be able to, in exchange for...nothing.

When was the last time Ubuntu made a new release that didn't require you to a) learn a new way of interacting with your computer, b) fix something that broke, or c) both of the above? I've tried every Ubuntu release, at least very briefly. All of them. And one of those things has happened *every* *single* *time*. It's not only on release upgrades even. My partner uses Ubuntu and has learned not to accept new release upgrades so long as the current release is still supported, because of the inevitable breakage, but periodically a high-priority update comes along and her printer will stop working; fortunately she's usually able to solve it with some Googling, because countless other people had exactly the same problem and managed to figure out the magical incantation to fix it.

Back to those things you learn to subconsciously ignore: try actively looking for minor bugs - the kind of things they're calling 'paper cuts'. When you start paying attention, you start to realise that you encounter *dozens* every day. Around the KDE 4.2 time (IIRC) I actually wrote down a list of all the minor bugs I experienced within the first 5 minutes. It wasn't a small list. I filed bugs for some of them; others had already been reported. In some cases there was some flaming about worrying over small things, but *small things add up*.

Here's an example: http://dl.dropbox.com/u/4957647/calendar.png
What's gone wrong with the rendering there? No idea. I'd have to start with a fresh KDE profile, then start bisecting all my configuration changes to find out. And what's the point, when more little problems like that will appear with the next dist-upgrade?

By the way, the red 'fail' icon in that shot is because so far I've only spent about 45 minutes trying to work out how to get wireless networking to work. It only took me a couple of minutes to get wired networking working because I have enough Debian experience to know that I needed to add 'auto eth0 inet dhcp' to /etc/network/interfaces. Good luck figuring that out if you're a new user. Why doesn't wired networking work out-of-the-box on a default install of the last two Ubuntu releases? No idea, but that's exactly the kind of thing I've come to expect (to be fair, in many cases the network connection does work; it just refuses to perform any name resolution - to a non-technnical user that's the same thing).

A solution that's permanently 90% finished is not 90% as good; that last 10% is utterly crucial for a good user experience.

Tasting the Ice Cream Sandwich

djao — Sun, 27 May 2012 04:04:34 +0000

Sure, Android includes a non-GNU userland and does not include the GNU userland, but the GNU userland (including even regular old glibc) will still work if you put it on the phone yourself, even though your phone doesn't come with it. There are even Play Store apps that download and install an entire Debian/Ubuntu ARM distribution in chroot on Android. I wouldn't want static binaries for everything but a shell is critical enough that I'll make an exception.

Thanks for the mkshrc script. It works fine in adb shell from /system/etc/mkshrc but doesn't seem to work in apps like Terminal Emulator (?) or ConnectBot's local terminal.

Tasting the Ice Cream Sandwich

mirabilos — Sat, 26 May 2012 19:37:24 +0000

> I don't understand why you say Android is not Linux.

No, Android is different userland on top of the Linux kernel, instead of eglibc or µClibc userland. In this case, mksh’s got the ability to run directly using the Android bionic libc.

As for the .mkshrc file – it depends on which mksh you use. The one included in AOSP, Android-x86 and Google Android 3 and up looks for this file in /system/etc/mkshrc and standard mksh builds in ~/.mkshrc (but the location can be overridden by setting $ENV properly).

A suitable mkshrc file for Android follows:

tg@tglase-amd64:~ $ cat android/external/mksh/mkshrc
# Copyright (c) 2010, 2012
# Thorsten Glaser <t.glaser@tarent.de>
# This file is provided under the same terms as mksh.
#-
# Minimal /system/etc/mkshrc for Android
#
# Support: https://launchpad.net/mksh

: ${TERM:=vt100} ${HOME:=/data} ${MKSH:=/system/bin/sh} ${HOSTNAME:=$(getprop ro.product.device)}
: ${SHELL:=$MKSH} ${USER:=$(typeset x=$(id); x=${x#*$}; print -r -- ${x%%$*})} ${HOSTNAME:=android}
if (( USER_ID )); then PS1='$'; else PS1='#'; fi
function precmd {
typeset e=$?

(( e )) && print -n "$e|"
}
PS1='$(precmd)$USER@$HOSTNAME:${PWD:-?} '"$PS1 "
export HOME HOSTNAME MKSH SHELL TERM USER
alias l='ls'
alias la='l -a'
alias ll='l -l'
alias lo='l -a -l'

function more {
local dummy line llen curlin=0

cat "$@" | while IFS= read -r line; do
llen=${%line}
(( llen == -1 )) && llen=${#line}
(( llen = llen ? (llen + COLUMNS - 1) / COLUMNS : 1 ))
if (( (curlin += llen) >= LINES )); then
print -n -- '\033[7m--more--\033[0m'
read -u1 dummy
[[ $dummy = [Qq]* ]] && return 0
curlin=$llen
fi
print -r -- "$line"
done
}

function setenv {
eval export $1'="$2"'
}

for p in ~/.bin; do
[[ -d $p/. ]] || continue
[[ :$PATH: = *:$p:* ]] || PATH=$p:$PATH
done

unset p

: place customisations above this line

Tasting the Ice Cream Sandwich

djao — Sat, 26 May 2012 19:18:12 +0000

I don't understand why you say Android is not Linux. Android is a very lightly forked Linux kernel and the kernel/userspace API is almost fully compatible with GNU/Linux. In any case I used eglibc. It even works fine for system scripts that use /system/bin/sh.

I would love to make mksh work. Where do I put the .mkshrc file on my Android phone in order to make it load whenever I start the shell?

"Last year's phone"

job — Sat, 26 May 2012 19:12:56 +0000

It's actually a lot worse. Without iTunes you can't install apps at all on the iPhone. On Android you can at least install the apk-files directly from the web browser or a file manager. The only problem is getting hold of the package files, as many authors do not distribute their software outside Play, but for those that do (including all free software) it's totally painless.

Tasting the Ice Cream Sandwich

mirabilos — Sat, 26 May 2012 18:58:53 +0000

Statically against bionic (what I don’t believe)
or statically against µClibc or eglibc (which would be not using Android but running stock Linux/ARM code on your device)?

Just try mksh for once. (Do make sure that a ~/.mkshrc file exists, if you do this on your desktop OS.) You won’t miss much, and it’s faster.

File transfer

djao — Sat, 26 May 2012 18:18:11 +0000

Thanks for posting this. It's really much better than the standard mtpfs.

Tasting the Ice Cream Sandwich

djao — Sat, 26 May 2012 18:15:45 +0000

I installed bash on my Galaxy Nexus since I've grown too accustomed to it on Linux. It works fine if you build it statically.

"Last year's phone"

djao — Sat, 26 May 2012 17:41:18 +0000

Apple iOS is no better in this regard. It's impossible to install apps from the Apple App Store unless you have an Apple ID.

Tasting the Ice Cream Sandwich

job — Thu, 24 May 2012 11:31:15 +0000

Perhaps there is some middle ground between not communicating with security researchers for *months* about open holes, and running a fully secured system?

I don't think it's asking too much to plug wide open holes that are already public. It's not like it's the first time.

"Last year's phone"

Fowl — Thu, 24 May 2012 10:28:14 +0000

Yes, Google Maps/Navigation is quite amazing, it would be great if they could ship and maintain the browser in the same way!

"Last year's phone"

bawjaws — Thu, 24 May 2012 08:50:59 +0000

I think the "update" argument is somewhat unfortunately framed to favour one business model over another.

You seem quite pleased to have got some portion of the new features of iOS 4 and 5, yet the last time I checked my cheap ZTE Blade was still getting updates to all the core apps direct from Google. Youtube, Market/Play Store, Maps being just three apps that have received *major* overhauls since I bought the device, with no need (or desire) for ZTE or my carrier to be involved in the rollout. I would have thought this decoupling would be hailed as a good thing, but it seems if you don't increment the OS version number then no-one cares about new user visible features.

You're also lucky you had the 3GS, I had the 3G and switched to the ZTE Blade after the iOS4 update rendered it nearly unusable. Clearly not enought planning and effort went into providing that cut down version.

Tasting the Ice Cream Sandwich

njwhite — Wed, 23 May 2012 21:54:54 +0000

> I'm more comfortable with Debian's security, because I believe that the program author and the packaging developer are on my side as opposed to most apps where the developer is wasting my bandwidth and CPU time downloading ads, and we're hoping that's all he's doing.

Exactly. The relationship between developer and user is a very important factor indeed, and one that is far too often undervalued.

Tasting the Ice Cream Sandwich

Cyberax — Tue, 22 May 2012 22:00:12 +0000

I don't have 2008 Server right now, but it definitely doesn't work on my up-to-date Windows 7 and Windows Vista. I've just re-checked to be sure I'm not going completely mad.

Google suggests that several Microsoft updates break it:
http://www.microsoft-questions.com/microsoft/Windows-Upda... so your OS is probably not completely up-to-date.

Tasting the Ice Cream Sandwich

mirabilos — Tue, 22 May 2012 19:56:54 +0000

It actually is.

But trying to avoid the GPL isn’t so bad. Most BSDs do so, too. Less restrictions, less worries, you know, free spirits. (At least on the BSD side. On the other hand, Android/Google do give back to the OSS projects they add, even if the latter are BSD licenced, and even if Google is slightly evil.)

Tasting the Ice Cream Sandwich

mirabilos — Tue, 22 May 2012 19:52:19 +0000

Sure, but mksh is much more versatile than any of the shells busybox offers. Just saying. (And it’s been in AOSP even before 2.3 was released, but 2.3 was cut off the Google internal master tree without merging mksh in.)

Tasting the Ice Cream Sandwich

Cyberax — Tue, 22 May 2012 17:40:18 +0000

I've somehow set up my Cyanogen to use busybox's sh automatically. I don't remember how :)

But you can do it manually, by running "busybox sh" in the default shell.

Tasting the Ice Cream Sandwich

mirabilos — Tue, 22 May 2012 16:35:21 +0000

Don’t forget that the default shell (when using “adb shell” or a local terminal emulator) now has things like tab completion, as /system/bin/sh has been mksh – https://www.mirbsd.org/mksh.htm – for a while in AOSP already. (Sadly, Cyanogen appears to never have it picked up, while Android-x86 was an early adopter, going back to 1.6 days.) In fact, other than ash, it’s the only shell that can be supported by Android bionic libc out-of-the-box.

Tasting the Ice Cream Sandwich

mpr22 — Tue, 22 May 2012 14:54:27 +0000

Unfortunately, seL4 isn't ready for desktop use.

Tasting the Ice Cream Sandwich

djao — Mon, 21 May 2012 17:40:20 +0000

And undocumented DDISABLE_INTEGRITY_CHECKS is disabled in final releases of Microsoft OSes (it's enabled in previews). You can try it yourself.

I did try it, just now, not more than 10 minutes ago, on my retail release version of Windows Server 2008 R2. And here is the result. As you can see, it works. You do not have to press F8 every time you reboot; the screenshot was taken from a clean reboot done without user interaction.

Tasting the Ice Cream Sandwich

job — Mon, 21 May 2012 16:43:17 +0000

That does not make sense. It is somehow Linux developer's responsibility to lock ourselves out of hardware? Palladium and its ilk was a bad idea. Bad ideas should be criticized.

Tasting the Ice Cream Sandwich

job — Mon, 21 May 2012 16:37:34 +0000

That's not what I said. Your anecdote (singular) is useless as a data point, as much as the fact that I won't recommend an insecure operating system to anyone these days.

Tasting the Ice Cream Sandwich

paulj — Mon, 21 May 2012 16:09:59 +0000

And yet Android's Bluetooth support depends on Bluez, which is GPL. Google seem to believe that wedging in some IPC is enough to "wash" away the GPL...

Tasting the Ice Cream Sandwich

anselm — Mon, 21 May 2012 13:32:20 +0000

Nobody is talking about »forced upgrades«. The original iPad didn't have a camera, so when the iPad 2 came out, many owners of the first iPad thought that would be a cool feature to have and got the new version (after all, being able to video-chat with other people using FaceTime must count for something). How difficult would it have been for Apple to put a camera into the first iPad? Virtually all the mobile phones at the time – including the iPhone – had at least one already, so it's not as if the requisite pieces weren't out there and mass-producible already.

The reason this works is that smartphones and tablets are considered lifestyle products rather than computers. You carry them around in public and people – even complete strangers – see you using them. Having (and being seen as having) the latest Apple stuff to play with is an important part of many people's lives, so it makes sense for Apple to introduce piecemeal upgrades to skim off that part of their customer base who must have everything just because it is new. If the original iPad had had the two cameras already, then fewer people would have felt the need to get a new one a year after the original one came out. (Do note that the new iPad is called exactly that - »the new iPad«. This strongly suggests that to be »with it«, you want that version and not the »old« one.) Even people who would hang on to their computers for a comparatively long time do not seem to mind replacing their phones every two years or so, and even quicker if the new phone seems better to them.

Tasting the Ice Cream Sandwich

Cyberax — Mon, 21 May 2012 09:29:40 +0000

>The link that I provided in the comment to which you replied contains exactly a description of how to permanently disable driver signing checks on both 32-bit and 64-bit versions of Windows OSes. Did you bother to read the page that I linked? The whole page?

I can ask you the same. Have YOU read it?

>You can’t permanently disable the use of signed drivers in the 64-bit version of Windows Server 2008 — at least, not using any Microsoft-recognized technique.

And undocumented DDISABLE_INTEGRITY_CHECKS is disabled in final releases of Microsoft OSes (it's enabled in previews). You can try it yourself.

But what do I know? After all, I'm only writing Windows drivers.

Tasting the Ice Cream Sandwich

djao — Mon, 21 May 2012 06:58:02 +0000

Right, now go read my post above where I point out, with quotes, the part in the Windows Hardware Certification Requirements where it states that users on Intel PC systems must be able to disable secure boot in order for the system to be compliant with the certification requirements.

Secure boot and driver signing on ARM is a genuine obstacle for Linux, because users can't turn it off. Secure boot and driver signing on Intel PCs is not a problem right now, because users can turn it off. It may become a problem in the future and I will be the first to complain if it does. But at the moment I believe it is a legitimate tradeoff to restrict what unsophisticated computer users can do on PCs in the name of security. I'm sick and tired of dealing with Windows botnets and I can't possibly be the only one.

No one is talking about the benefits side of the cost-benefit equation. Secure boot isn't just purely an antagonistic move on Microsoft's part to screw over Linux users. It has some legitimate benefits to offset its costs, benefits which will be appreciated even by Linux users. The key issue is whether advanced users can turn it off. If they can, then I don't have a problem with it.

Tasting the Ice Cream Sandwich

djao — Mon, 21 May 2012 06:48:05 +0000

The link that I provided in the comment to which you replied contains exactly a description of how to permanently disable driver signing checks on both 32-bit and 64-bit versions of Windows OSes. Did you bother to read the page that I linked? The whole page?

I think it's hardly fair to blame Microsoft for Blu-ray not working. Does Blu-ray work in Linux? No. Blu-ray is the fault of the entertainment companies.

All I'm proposing is the very modest suggestion that Microsoft is not 100% at fault for absolutely every single one of Linux's problems. Apparently this claim is too radical for some around here.