In Japan, people are already doing this. Here, we have iPods, cell phones, computers, PSPs, all kinds of toys. In Japan, I remember hearing years ago -- basically, your typical teenage girl there just needs a cell phone, and she'll pretty much just use it for text messaging -- but if she needs music, photos, whatever, it's all in there. In fact, she'll use it for just about anything except a phone (since talking on a cell phone is considered rude).
You make a good point about remote access. However, there are a few issues with that:
The machine has to be on. This is a waste of power if you simply leave it on 24/7, and a significant amount of lag if you have a way to boot it remotely. (I can SSH in to a server and have it etherboot my desktop, but I still need the server on, and I need to wait while my desktop boots.)
You need a network connection. It's nice to be able to do something useful without one, and your network is always going to be slow-ish, particularly if you're using VNC. Even RDP is sluggish compared to, say, ssh, and while ssh is lightning over any working connection, as soon as the connection starts to lag, it's almost not worth using.
You still need your local terminal to be "good enough." This depends on what you want it to do -- I like to watch anime on long trips, for instance, so I like to have a bit of hard drive space and a decent-sized screen -- oh, and a CPU and video card fast enough to actually play the videos at their intended framerate.
I guess the rest depends on what you need all that desktop speed for. I need speed so I can play games. There's really no other reason -- sure, I like my web browsing to be snappy, but if my box was slower, I'd switch away from Firefox. Sure, a compile will take less time, but on a slower box, I wouldn't use Gentoo, and really, with a decent makefile, you're talking about maybe one gcc command that might take a second or two longer for most changes. Boot time matters to me, but only because I dual-boot, and I only keep Windows for games.
But, these are still reasons a fast handheld would be nice -- eventually, stuff gets to the local machine. Even things like web browsing are places where you could get slowed down.
Anyway... I actually agree with you, to a point. But, I do want a laptop with a decent-sized screen and keyboard, which is capable of playing movies, which takes quite a bit more than ssh and VNC. And you know, I didn't need my Powerbook to be 1.67 ghz, but it didn't hurt, either.
It makes a lot of sense, because it gives you a familiar environment, with familiar apps and compatibility.
Which is, incidentally, yet another reason to love Linux. Linux doesn't require x86-compatible hardware or anything remotely resembling it; I can run a full Linux system on an ARM processor. (Or, of course, I can trim the fat and only run what I need.) It's also perfectly possible to remap just about anything, including ctrl+alt+del-style functionality -- my Jornada has the fn key mapped to backtick/tilde (` or ~). So, Linux gets ported to the device, usually by one or two guys in their spare time.
But you don't get source code for Windows, and Microsoft would kind of like to sell Windows Mobile. So in order to run XP/Vista apps on a mobile device, you have to port the device to Windows, not the other way around.
So yes, there's a very good reason for doing it, and yes, it's a hell of a lot more painful than it should be.
Not as "hardwired", I imagine, but after the keyboard is activated, and until you actually boot an OS that knows what to do with a keyboard, CTRL+ALT+DEL will soft-reboot. Simple test: Open your BIOS setup screen (no clue how to do this on your machine; I pres DEL at a certain point during boot, but YMMV a fucking lot). With the BIOS setup screen open, give it the three-fingered salute, and it should soft-reboot IMMEDIATELY.
Of course, these days, it's reduced to mean whatever you want it to mean, if you have an OS flexible enough -- for instance, on Linux, you can set CTRL+ALT+DEL to fire a shell script of your choice. By default it does a graceful soft reboot (as you'd get with shutdown -r), but it can be configured to do whatever you want -- I've configured CTRL+ALT+DEL and other things like pressing the power button to take over the screen and display the message "DON'T TOUCH THAT!!!"
I'm not using hardware RAID controllers. It's a big desktop system, but it's ultimately just a gaming rig. nVidia RAID is fakeraid -- it's BIOS RAID.
I mean, I honestly wouldn't care if it takes 5 extra seconds when booting, having it tell me what key I need to press to configure it would save me a lot of time. I don't want to have to spend twenty minutes of my time looking for documentation for some obscure RAID controller in order to configure it.
If it takes you twenty minutes of looking through documentation for what key to press, you're not trying. Takes me only a few seconds of a Google search to find out what to do with a Powerbook, and it's not obscure at all -- hold T for firewire Target mode. Hold C to boot the Cdrom. Do this once and write it on the machine, if you're that worried.
If it wastes five or ten seconds of the computer's time when it reboots once in a blue moon, I really don't care. If five or ten seconds is a big deal to you as far as uptime goes, then you'll be using something like clustering anyway.
Which is still no excuse. Consider the case where five seconds (times four of five devices, making it closer to 30 seconds) is a fucking HUGE deal, but we might not have the money to build a cluster to cover it? Consider that even if you've got a second machine, every second that first machine is down is a second you could be knocked offline entirely?
More importantly, consider that rebooting happens sometimes on servers (kernel upgrades, at least), and quite a lot on desktops. Because you're too lazy to check in a manual to find out what key to hold to bring up a RAID menu for the maybe one or two times in the lifetime of the machine that you'll have to, we all have to suffer on the once-a-month reboots. Maybe not a big enough deal for me to waste time here -- but hey, it's even more annoying on my desktop machine, where even if I reconfigured the RAID array once a month, it's still making me wait every single day to make that once-a-month easier -- assuming I don't know what I'm doing, which I do.
But again, that assumes I'm talking about servers. I'm not, I'm talking about custom-built desktop machines. Network card capable of booting off the network? There's a chance it'll throw up one of those screens. Have a high-end video card? Maybe it'll give you a splash screen before even letting you into the BIOS. This kind of shit is especially annoying when some off-the-shelf Dell-like boxes just boot -- maybe two seconds from power-on to bootloader.
The best shot you have is if your publisher decides that your studio needs to create *THE* cross platform engine for X number of games across X studios to use. Otherwise, as far as the consumers are concerned, what does it matter if you're releasing a crazily coded mess?
Because especially when those consumers are coders, we fucking *HATE* crazily coded messes, because they so often don't work, and fail in really stupid ways. And because it costs you money in the long run -- what you've described, with the building on fire, is much more likely to happen, be less manageable, and take longer, unless you do it right in the first place. On the other hand, if you do it right in the first place, you're more likely to be able to calmly weather the storm.
Put another way: Suppose you're doing application development. You could write a whole new OS, or even just windowing system or graphics library, for every app. Or you could write a good, solid, flexible system, or buy someone else's, so that you can just focus on the application.
Frankly, if anyone involved in game logic has to worry about anything related to the underlying graphics API, you're back in the development model where each game had to be shipped with sound drivers. It's like an office application having to know what an IRQ is. I understand that the reality of the market may not give you time to do this, but as soon as you have the time and the resources, you NEED to start doing this right, even if it looks like a black-hole investment, and even if it will never be used outside your studio.
Imagine you're developing a new IP on a new engine- suddenly production and design changes EVERYTHING and you have to restructure- new requirements arise, platform changes mid-production!
Suppose one of those requirements is targeting the 360 instead of (or in addition to) the PS3, or suddenly they want Mac/Linux ports...
If your code is solid at that point, ANY change they want is something you can handle. If your code is not solid, you may find yourself having to tell them, "That's just impossible, it'll push the release date back half a decade."
Major upgrades normally mean having to relearn interfaces. Updates are a different matter, and MS and Apple both provide updates quite regularly.
Let me put it this way: The Linux 2.2 kernel does still occasionally get patches. Most of us would rather be on 2.6, sure, but 2.2 has got to be at least 5-10 years old. Neither MS nor Apple provide regular updates for anything older than a certain point -- I imagine that's Panther for Apple and Win2K for MS.
The other point is, of course, that Vista and Leopard are bragging about all their major new features. Leopard is closest to the Linux way -- bunch of new features added, but most of the old stuff isn't changed arbitrarily, while Vista just rearranges control panels and such for no good reason.
The difference is, of course, you can get all this new stuff gradually and free, instead of all at once for $129.
If it only does the things Windows and MacOS can do, why not just keep using what you have?
Because the only way you could do that is to run Windows in a virtual machine on MacOS, assuming you have a version of Windows which lets itself be run in a virtual machine, and assuming you like wasting RAM. People complain about inconsistent UIs on Linux -- Wine/GTK/QT -- but that's nothing compared to actually trying to run two separate OSes.
Remember: It does both the things Windows does, and the things MacOS does, and some things neither of them do.
Windows drivers are pretty much guaranteed for any hardware.
Except when they aren't. And when they aren't, you can't do much about it.
Also: we're not just talking about some random peripheral. We're talking about ANY hardware. Sure, Windows Mobile runs on ARM devices, but I can run pretty much any desktop Linux environment on a PDA, to the extent that said PDA can actually provide the performance needed. You can claim that I'll have to strip it down to the point where it may as well be Windows Mobile, but that simply isn't true -- Windows Mobile is even more different from Windows XP/Vista as Windows XP is from Windows 98. You can strip XP down, but it will never run on your phone -- Linux will.
Same for old PPC Macs or anything else you want. ANY hardware. And it will look and feel the same, given the same input devices.
Not even that. Aside from doubts as to whether you can actually parallize the hardware initialization, there's the fact that most of these are just stupidly written -- Grub included. Lilo had the right idea -- it can be configured to automatically boot, with no delay, unless you're holding alt. I want everything up to and including Grub to function that way. I honestly don't care about the memory check, and most of these things are disabled on my box, but many controllers will take over the whole screen to give me some tiny piece of information and wait 5-10 seconds in case I might want to configure something.
That's what bugs the hell out of me. Even though my Grub is on a timeout of only five seconds, I very rarely want to do something strange and configure it, and I very rarely want to boot into Windows -- usually when I do, I reboot with a script in Linux which modifies the grub config to default to Windows anyway. So, 95% of the time, I have 5 seconds added to my boot to make it easy for the other 5% of the time when I might have to customize something -- and that's just Grub.
My nVidia RAID screen, like Promise RAID and every other (shitty) RAID I've ever worked with, displays a screen showing me the status of my array and prompting me to press some key if I want to configure it. I know this is an unnecessary pause, because if I hit enter as soon as that screen displays, I'll skip straight to Grub -- but if I leave it alone, it'll wait 5-10 seconds just in case I want to configure my RAID this boot.
Compare this to Macs. Without Yaboot (or Grub, or Bootcamp, but my last Mac was a Powerbook), you can simply power on and it will hit OS X as fast as it possibly can. If you need to do something strange, you hold down a key while booting, and you can get things like boot from CD, firewire target mode, etc. That's the way any boot choice should operate -- boot the default as fast as you possibly can, unless the user is holding down a key. You don't need the RAID configuration to be user-friendly, especially when a person might want to configure their RAID once in their computer's lifetime -- maybe -- even if they do it once a month, you're slowing down ~30 boot cycles to make it easier for that one.
LinuxBIOS is where we might make some real progress here -- it might actually parallize some of this -- but there will always be the odd NIC BIOS that wants to do things its own way.
What you can do is remove startup stuff. I used to do this with start->run->msconfig, but I'm told Vista has a better way of doing it.
What you can't do is make it take less time with the bouncing "progress bar" between power on and actually starting up the GUI (and giving you the login page if you actually login). This can go very quickly, but it seems one of the inevitable little things about Windows that deteriorates -- after a few months, for whatever reason (I'm guessing the Registry), this step will take longer than a tuned Linux does to boot entirely to a usable desktop.
Like everything else -- Windows, out of the box, may or may not be closer to what you want. But Windows can only be tuned so far. Same with OS X, or any proprietary system. And when you run into the limit of how far it can be tuned, you have to ask Microsoft or Apple for more power, or to make it better for you. On Linux, you can do it yourself, or fire off a message to an entire community, any of whom can solve your problem for free -- although usually you can tweak quite a bit more before having to dig in source code.
After you build an engine, how much work does it take to keep said engine relevant? Specifically, how much work must be done with the underlying graphics API?
It seems to me like it might be a good investment to create a good, solid, cross-platform engine (using OpenGL), and maintain that indefinitely. It may be harder to create initially, but I imagine it would not be significantly harder to maintain in the long run.
I would consider selling my soul to work on Halo, if I could ever get into Bungie at Microsoft. They hired an orchestra for that little Halo 3 trailer. It's a game I love and enough corporate backing to be able to do it right, even if you lose money.
But I will not have anything to do with Microsoft, because I know that somewhere, at the top of the chain, is an insane monkey boy who frankly embarrasses me enough to be of the same species as him that I could never be in the same company.
Are you seriously trying to tell me that every company that got hit hard by Sasser, Blaster, and other major viruses were staffed by shitty sysadmins?
I'd say a tentative "maybe". Are you seriously going to tell me that blocking web access will do anything to slow down Sasser, Blaster, or any of the other major worms?
Now, if you could please tell me how I can, without blocking any sites, and with virus scanners that we know cannot block everything,
Simple: Run a secure system. Firewalls on indivdiual machines if they're vulnerable, or run systems that generally aren't, like Linux or another Unix. Even if you can only convert your servers, that's a win -- it means a virus wipes out a segment (which it shouldn't -- remember those firewalls?), the server is still up.
Also, keep backups. Disk images of working machines, tape/DVD/network backups of your servers. If you can repair a machine and make it usable and productive for the user again in 10 seconds of your time and maybe 15 minutes of machine time, then you can pretty much let users screw it up as much as they want.
And what you're doing is actually very much equivalent to locking the front door, but leaving the garage door open -- you try to keep stuff from getting into your network, but you have no way of dealing with it once it's there.
You've already likened me to a rapist in a round-about fashion.
As an analogy, to illustrate an attack. And you of course ignored the entire argument and used that analogy to say "Take your meds." Because obviously, anyone who disagrees with you enough to write a long-ish response must be frothing at the mouth.
Really, I'm quite calm.
If you didn't have such a bug up your ass about people actually making money
I have nothing against making money. I just don't think it should be anywhere close to a first priority. And if you think you can take another analogy (gasp!), I don't have anything against sex, but I don't think it should be such a high priority as to justify rape.
Quote me as to where I said that making a profit is an excuse for "anything wrong" one can do.
You only used it as an excuse for this particular action. So, it's not all about you personally. There are plenty of people here who throw this excuse out no matter what a company does. Sony Rootkit? They have to make profit. Oil spills? Oh well, they have to make profit. Don't be evil, but you know that's only so long as they make profit. Steve Jobs telling others to stop doing DRM while it continues to be the only option for iTMS? Apple's a company, they have to profit. MS monopoly? They're a business, they have to make profit. There's also either an implied or stated "What do you expect?" at the end of that.
As for you, I'm logically deriving this from things you did say. If profit is a first priority, than everything else is secondary. I'm trying to get you to admit that there may be other things, even implicit things like following the law, which should come first, even if they mean less profit or none at all. Life savings or not.
I suspect that this is what you meant in the first place, and that you think I'm being pedantic here. But if you can't just fall back on "Well, they have to make a profit," then you'll actually have to explain to me why profit comes first in this case. If you don't care about OpenOffice or open source, or don't think they're important enough here, you could say that, and I'll accept it -- and sadly, a different group of fanatics might come and flame you then. If you honestly think it would kill Dell's business model to distribute OpenOffice, I could say I doubt it, but we'd be done, since at that point, we're really in speculation.
Maybe you're just reading too much into it and went off raving like a mad man and now are trying to save face?
Maybe I'm just doing a reductio ad absurdum argument.
I can't imagine I'd care much about saving face now, really -- look how deep we are into the thread. No mod can hear you scream.
The right thing? How is preinstalling a free piece of software that anyone can download on their own suddenly the morally correct thing to do?
Maybe you haven't been following the discussion about things like preinstalled Linux.
It's not about hating MS. It's about the fact that MS has such a huge monopoly on desktop OSes and office suites that we're even having a discussion about whether Dell can afford to do anything but install a trial Office. It's not that the morally correct thing to do is take up pitchforks and go kill MS, but rather, to not be a part of that machine where it makes sense.
Well, if you'd rather die than have Dell not offer OO so be it.
I'm illustrating a point, and you missed it. Again.
I would rather have Dell die -- the company, not any individual -- than have the Microsoft monopoly last another ten years. I would rather have Dell make less profit and distribute OO than have them make more and distribute trial Office, though I admit that's probably not worth losing a company over.
You are the one who made it black and white here, though. According to you, if MS Office
I don't even want to have to use daemontools, much less some nasty crack. And I'm certainly not going to be a good little consumer and put the game disc in every time I play, scratching it more and more until it's unreadable.
I honestly could care less if the game works offline, though others will certainly want them to work on their laptops even without wireless. I don't even care if it's horribly inefficient -- say, a 5k/s trickle of data -- my gaming rig is always plugged into nice fast DSL, except when it goes out, which is usually when the power goes out anyway.
I'd much rather have a completely DRM-free game. I actually bought Darwinia and Uplink. Both are available in downloadable form, and the downloads, as far as I can tell, have absolutely no copy protection at all, other than the limit of only being able to download it 3 times (which is fine, considering the games are only some 50 megs, I can back that up forever). So, as far as I could tell, even the Windows version should have been easy to find on a filesharing network, but I didn't care, I went and bought them anyway.
But given the choice, fuck it, saturate my connection all you want, send them everything you can find on my computer -- not that you'll find much (my Wine is isolated enough, and my Windows has NOTHING of value on it) -- just don't lamely insist on a CD.
Whenever I see that bit, I tell the people sitting there, "You know, the pirated version doesn't have that ad..."
Let's think about this. They want pirates to stop pirating, so they go and try to beat their message into... their own legitimate customers? To the point where they actually end up driving some of them to piracy, just so they don't have to be called a pirate?
My own compromise has generally been renting and ripping. I think the rental price is fair, especially considering I simply don't have the hard disk storage to keep the movie, and it's not worth the hassle to burn and keep track of a huge archive of DVDs (especially when most movies must either be re-encoded or cut (special features, etc) or split onto two disks).
But yes, that's generally my procedure with any game, whether I buy it or not: If it includes CD-based copy protection, I go download a crack and/or convince it to run from Daemon Tools on Windows or cdemu on Linux. (Haven't gotten the cdemu to work for one of these, yet, but I think I'm getting closer.)
But I was shocked to find how true it was for me. I seriously considered for a moment, and yes, if Sony was doing this to, say, YouTube, I'd probably see it as stealing. Not horrible, not a rootkit, but still kind of a bad thing to do.
Whereas my initial reaction to this piece was more of the kind of "Sony needs to wake up to the realities of the Internet, and if it's not legal to do this, it should be."
I think Searchles is probably pretty cool, but I also think that this is going to end up being a bit like several Linux tools -- for instance, ies4linux. While it's not likely to be an issue for most people, it's probably illegal for them to redistribute IE, and it's certainly not legal to have a copy of IE without a Windows license. However, they probably have avoided trouble because they're so small, they warn you that it might be illegal, and they do it with deep linking, not mirroring -- you actually are downloading IE from Microsoft's website.
Besides: What copy protection, anyway? Sending an unencrypted flash video to a proprietary browser plugin is worse than DVD CSS, and that's saying something!
Moral of the story? I suppose I won't shed any tears for Sony anyway. Sony can do no right, and Google can do no wrong -- but they earned these reputations, and they continually reinforce them. Compare that to, say, desktop Linux, where there are actually far fewer jokes about recompiling your kernel, as people start to realize that it's unlikely you'll have to compile anything while using Linux, ever, unless you want to -- it certainly earned its reputation as a geek operating system, but it's now working to correct that.
Cygwin is slower than Ruby, and that's saying something.
Think about it -- process creation on Windows is an order of magnitude or two higher than process creation on Linux, and that's ignoring the implications of an unoptimized fork. Wine just seems a lot more likely to be fast than Cygwin ever could.
And all of that assumes that there's a point, really. Even if Windows is easier to admin than Linux, I'm not sure Cygwin on Windows has any advantages over the same app on Linux. The reverse is true also -- in fact, I know for a fact that Windows apps under Wine aren't always better than they are under Windows -- sometimes they are, mostly not.
And I'm thinking the best way to crack down would be to force people to be responsible for what happens on their hardware, unless someone signs some sort of waiver first -- and maybe some sort of standards as to what kind of community you have to be to allow that. A difference between something like YouTube, or an ISP, and something like an anonymous, deliberately unprotected FTP server.
I have ssh keys distributed everywhere. If I really wanted to, I could do something like:
for machine in `cat machines`; do ssh root@$machine apt-get update ssh root@$machine apt-get dist-upgrade done
For the record, that's sloppy -- will take twice as long as a smarter command I didn't feel like putting together, which would require only one ssh command. Could also tell ssh to never prompt for a password, just log failures, and I could create a different loop which iterates over every ip on the network, just in case I've forgotten about a box.
Of course, I could just put it in the individual machines' crontabs, and/or use something like cfengine. There are all kinds of ways to do that.
But really, the above commands are pretty close to a one-liner, and that's assuming you're comfortable pushing updates out to thousands of machines without verifying each one -- the verification is where most of the time comes from.
Oh, and let's not forget -- these are not just kernel updates. It's not even comparable to a Windows Update from MS -- every single piece of software on the machine is controlled by package management. Everything -- from a word processor to a web browser to a game to a mailserver to some random webapp that was nice enough to provide a package -- all of those will automatically be updated, and in the same place, with the same local cache (apt-proxy as one example).
Or use a distributed filesystem -- you can run the entire OS off the network. Not a thin client, but any change you make, every config file tweak you do, is instantly propagated over the network. This is actually pretty easy to set up... on Linux. I don't even know if it's possible to boot Windows off a network.
And by the way, 15 minutes of admin time is nothing when you consider other possible time sinks. I have to spend that 15 minutes maybe once a week, if I feel like it, and I really don't have to do much else. How much time do you spend?
If all you're doing is looking at spreadsheets, training will take you ten minutes. ("The spreadsheet is here, the Internet is here, email is here, and the shutdown button is here. Any questions?")
Not to mention that it'll be free forever. When MS releases whatever their next OS is and charges $300 for it, and Apple releases Leopard and charges $130 for it (and another one the year after, and the year after that), if you've done your Linux right, you'll run one command and your systems will automagically upgrade. All of them, for free.
You could netboot them, or any of a number of other ways.
What is it you're missing from Active Directory? I bet I can do it with "a cludge of logon scripts, spit, and duct tape" in less than ten minutes. You can probably do it in ten seconds with Active Directory -- and I'll gain those ten minutes a hundred times over when your network gets hit with a simple worm.
Slashdot Mirror
In Japan, people are already doing this. Here, we have iPods, cell phones, computers, PSPs, all kinds of toys. In Japan, I remember hearing years ago -- basically, your typical teenage girl there just needs a cell phone, and she'll pretty much just use it for text messaging -- but if she needs music, photos, whatever, it's all in there. In fact, she'll use it for just about anything except a phone (since talking on a cell phone is considered rude).
You make a good point about remote access. However, there are a few issues with that:
The machine has to be on. This is a waste of power if you simply leave it on 24/7, and a significant amount of lag if you have a way to boot it remotely. (I can SSH in to a server and have it etherboot my desktop, but I still need the server on, and I need to wait while my desktop boots.)
You need a network connection. It's nice to be able to do something useful without one, and your network is always going to be slow-ish, particularly if you're using VNC. Even RDP is sluggish compared to, say, ssh, and while ssh is lightning over any working connection, as soon as the connection starts to lag, it's almost not worth using.
You still need your local terminal to be "good enough." This depends on what you want it to do -- I like to watch anime on long trips, for instance, so I like to have a bit of hard drive space and a decent-sized screen -- oh, and a CPU and video card fast enough to actually play the videos at their intended framerate.
I guess the rest depends on what you need all that desktop speed for. I need speed so I can play games. There's really no other reason -- sure, I like my web browsing to be snappy, but if my box was slower, I'd switch away from Firefox. Sure, a compile will take less time, but on a slower box, I wouldn't use Gentoo, and really, with a decent makefile, you're talking about maybe one gcc command that might take a second or two longer for most changes. Boot time matters to me, but only because I dual-boot, and I only keep Windows for games.
But, these are still reasons a fast handheld would be nice -- eventually, stuff gets to the local machine. Even things like web browsing are places where you could get slowed down.
Anyway... I actually agree with you, to a point. But, I do want a laptop with a decent-sized screen and keyboard, which is capable of playing movies, which takes quite a bit more than ssh and VNC. And you know, I didn't need my Powerbook to be 1.67 ghz, but it didn't hurt, either.
It makes a lot of sense, because it gives you a familiar environment, with familiar apps and compatibility.
Which is, incidentally, yet another reason to love Linux. Linux doesn't require x86-compatible hardware or anything remotely resembling it; I can run a full Linux system on an ARM processor. (Or, of course, I can trim the fat and only run what I need.) It's also perfectly possible to remap just about anything, including ctrl+alt+del-style functionality -- my Jornada has the fn key mapped to backtick/tilde (` or ~). So, Linux gets ported to the device, usually by one or two guys in their spare time.
But you don't get source code for Windows, and Microsoft would kind of like to sell Windows Mobile. So in order to run XP/Vista apps on a mobile device, you have to port the device to Windows, not the other way around.
So yes, there's a very good reason for doing it, and yes, it's a hell of a lot more painful than it should be.
Not as "hardwired", I imagine, but after the keyboard is activated, and until you actually boot an OS that knows what to do with a keyboard, CTRL+ALT+DEL will soft-reboot. Simple test: Open your BIOS setup screen (no clue how to do this on your machine; I pres DEL at a certain point during boot, but YMMV a fucking lot). With the BIOS setup screen open, give it the three-fingered salute, and it should soft-reboot IMMEDIATELY.
Of course, these days, it's reduced to mean whatever you want it to mean, if you have an OS flexible enough -- for instance, on Linux, you can set CTRL+ALT+DEL to fire a shell script of your choice. By default it does a graceful soft reboot (as you'd get with shutdown -r), but it can be configured to do whatever you want -- I've configured CTRL+ALT+DEL and other things like pressing the power button to take over the screen and display the message "DON'T TOUCH THAT!!!"
I'm not using hardware RAID controllers. It's a big desktop system, but it's ultimately just a gaming rig. nVidia RAID is fakeraid -- it's BIOS RAID.
If it takes you twenty minutes of looking through documentation for what key to press, you're not trying. Takes me only a few seconds of a Google search to find out what to do with a Powerbook, and it's not obscure at all -- hold T for firewire Target mode. Hold C to boot the Cdrom. Do this once and write it on the machine, if you're that worried.
Which is still no excuse. Consider the case where five seconds (times four of five devices, making it closer to 30 seconds) is a fucking HUGE deal, but we might not have the money to build a cluster to cover it? Consider that even if you've got a second machine, every second that first machine is down is a second you could be knocked offline entirely?
More importantly, consider that rebooting happens sometimes on servers (kernel upgrades, at least), and quite a lot on desktops. Because you're too lazy to check in a manual to find out what key to hold to bring up a RAID menu for the maybe one or two times in the lifetime of the machine that you'll have to, we all have to suffer on the once-a-month reboots. Maybe not a big enough deal for me to waste time here -- but hey, it's even more annoying on my desktop machine, where even if I reconfigured the RAID array once a month, it's still making me wait every single day to make that once-a-month easier -- assuming I don't know what I'm doing, which I do.
But again, that assumes I'm talking about servers. I'm not, I'm talking about custom-built desktop machines. Network card capable of booting off the network? There's a chance it'll throw up one of those screens. Have a high-end video card? Maybe it'll give you a splash screen before even letting you into the BIOS. This kind of shit is especially annoying when some off-the-shelf Dell-like boxes just boot -- maybe two seconds from power-on to bootloader.
Next commercial, if they can find a hotter chick, it would be the perfect time to say "I'm Linux, too!"
Be a perfect introduction to distros and why they're a good thing.
Because especially when those consumers are coders, we fucking *HATE* crazily coded messes, because they so often don't work, and fail in really stupid ways. And because it costs you money in the long run -- what you've described, with the building on fire, is much more likely to happen, be less manageable, and take longer, unless you do it right in the first place. On the other hand, if you do it right in the first place, you're more likely to be able to calmly weather the storm.
Put another way: Suppose you're doing application development. You could write a whole new OS, or even just windowing system or graphics library, for every app. Or you could write a good, solid, flexible system, or buy someone else's, so that you can just focus on the application.
Frankly, if anyone involved in game logic has to worry about anything related to the underlying graphics API, you're back in the development model where each game had to be shipped with sound drivers. It's like an office application having to know what an IRQ is. I understand that the reality of the market may not give you time to do this, but as soon as you have the time and the resources, you NEED to start doing this right, even if it looks like a black-hole investment, and even if it will never be used outside your studio.
Suppose one of those requirements is targeting the 360 instead of (or in addition to) the PS3, or suddenly they want Mac/Linux ports...
If your code is solid at that point, ANY change they want is something you can handle. If your code is not solid, you may find yourself having to tell them, "That's just impossible, it'll push the release date back half a decade."
Let me put it this way: The Linux 2.2 kernel does still occasionally get patches. Most of us would rather be on 2.6, sure, but 2.2 has got to be at least 5-10 years old. Neither MS nor Apple provide regular updates for anything older than a certain point -- I imagine that's Panther for Apple and Win2K for MS.
The other point is, of course, that Vista and Leopard are bragging about all their major new features. Leopard is closest to the Linux way -- bunch of new features added, but most of the old stuff isn't changed arbitrarily, while Vista just rearranges control panels and such for no good reason.
The difference is, of course, you can get all this new stuff gradually and free, instead of all at once for $129.
Because the only way you could do that is to run Windows in a virtual machine on MacOS, assuming you have a version of Windows which lets itself be run in a virtual machine, and assuming you like wasting RAM. People complain about inconsistent UIs on Linux -- Wine/GTK/QT -- but that's nothing compared to actually trying to run two separate OSes.
Remember: It does both the things Windows does, and the things MacOS does, and some things neither of them do.
Except when they aren't. And when they aren't, you can't do much about it.
Also: we're not just talking about some random peripheral. We're talking about ANY hardware. Sure, Windows Mobile runs on ARM devices, but I can run pretty much any desktop Linux environment on a PDA, to the extent that said PDA can actually provide the performance needed. You can claim that I'll have to strip it down to the point where it may as well be Windows Mobile, but that simply isn't true -- Windows Mobile is even more different from Windows XP/Vista as Windows XP is from Windows 98. You can strip XP down, but it will never run on your phone -- Linux will.
Same for old PPC Macs or anything else you want. ANY hardware. And it will look and feel the same, given the same input devices.
Not even that. Aside from doubts as to whether you can actually parallize the hardware initialization, there's the fact that most of these are just stupidly written -- Grub included. Lilo had the right idea -- it can be configured to automatically boot, with no delay, unless you're holding alt. I want everything up to and including Grub to function that way. I honestly don't care about the memory check, and most of these things are disabled on my box, but many controllers will take over the whole screen to give me some tiny piece of information and wait 5-10 seconds in case I might want to configure something.
That's what bugs the hell out of me. Even though my Grub is on a timeout of only five seconds, I very rarely want to do something strange and configure it, and I very rarely want to boot into Windows -- usually when I do, I reboot with a script in Linux which modifies the grub config to default to Windows anyway. So, 95% of the time, I have 5 seconds added to my boot to make it easy for the other 5% of the time when I might have to customize something -- and that's just Grub.
My nVidia RAID screen, like Promise RAID and every other (shitty) RAID I've ever worked with, displays a screen showing me the status of my array and prompting me to press some key if I want to configure it. I know this is an unnecessary pause, because if I hit enter as soon as that screen displays, I'll skip straight to Grub -- but if I leave it alone, it'll wait 5-10 seconds just in case I want to configure my RAID this boot.
Compare this to Macs. Without Yaboot (or Grub, or Bootcamp, but my last Mac was a Powerbook), you can simply power on and it will hit OS X as fast as it possibly can. If you need to do something strange, you hold down a key while booting, and you can get things like boot from CD, firewire target mode, etc. That's the way any boot choice should operate -- boot the default as fast as you possibly can, unless the user is holding down a key. You don't need the RAID configuration to be user-friendly, especially when a person might want to configure their RAID once in their computer's lifetime -- maybe -- even if they do it once a month, you're slowing down ~30 boot cycles to make it easier for that one.
LinuxBIOS is where we might make some real progress here -- it might actually parallize some of this -- but there will always be the odd NIC BIOS that wants to do things its own way.
What you can do is remove startup stuff. I used to do this with start->run->msconfig, but I'm told Vista has a better way of doing it.
What you can't do is make it take less time with the bouncing "progress bar" between power on and actually starting up the GUI (and giving you the login page if you actually login). This can go very quickly, but it seems one of the inevitable little things about Windows that deteriorates -- after a few months, for whatever reason (I'm guessing the Registry), this step will take longer than a tuned Linux does to boot entirely to a usable desktop.
Like everything else -- Windows, out of the box, may or may not be closer to what you want. But Windows can only be tuned so far. Same with OS X, or any proprietary system. And when you run into the limit of how far it can be tuned, you have to ask Microsoft or Apple for more power, or to make it better for you. On Linux, you can do it yourself, or fire off a message to an entire community, any of whom can solve your problem for free -- although usually you can tweak quite a bit more before having to dig in source code.
I think the word you're looking for is "abstracting". If so, what's stopping you from abstracting similarly on top of OpenGL?
(Sound and I/O is another problem, for another library. SDL isn't great, but it's there, and OpenAL is looking better and better.)
After you build an engine, how much work does it take to keep said engine relevant? Specifically, how much work must be done with the underlying graphics API?
It seems to me like it might be a good investment to create a good, solid, cross-platform engine (using OpenGL), and maintain that indefinitely. It may be harder to create initially, but I imagine it would not be significantly harder to maintain in the long run.
I would consider selling my soul to work on Halo, if I could ever get into Bungie at Microsoft. They hired an orchestra for that little Halo 3 trailer. It's a game I love and enough corporate backing to be able to do it right, even if you lose money.
But I will not have anything to do with Microsoft, because I know that somewhere, at the top of the chain, is an insane monkey boy who frankly embarrasses me enough to be of the same species as him that I could never be in the same company.
I'd say a tentative "maybe". Are you seriously going to tell me that blocking web access will do anything to slow down Sasser, Blaster, or any of the other major worms?
Simple: Run a secure system. Firewalls on indivdiual machines if they're vulnerable, or run systems that generally aren't, like Linux or another Unix. Even if you can only convert your servers, that's a win -- it means a virus wipes out a segment (which it shouldn't -- remember those firewalls?), the server is still up.
Also, keep backups. Disk images of working machines, tape/DVD/network backups of your servers. If you can repair a machine and make it usable and productive for the user again in 10 seconds of your time and maybe 15 minutes of machine time, then you can pretty much let users screw it up as much as they want.
And what you're doing is actually very much equivalent to locking the front door, but leaving the garage door open -- you try to keep stuff from getting into your network, but you have no way of dealing with it once it's there.
Viacom should be the one required to review them, and the little guys (and Google) should be able to sue Viacom for these mistakes.
As an analogy, to illustrate an attack. And you of course ignored the entire argument and used that analogy to say "Take your meds." Because obviously, anyone who disagrees with you enough to write a long-ish response must be frothing at the mouth.
Really, I'm quite calm.
I have nothing against making money. I just don't think it should be anywhere close to a first priority. And if you think you can take another analogy (gasp!), I don't have anything against sex, but I don't think it should be such a high priority as to justify rape.
You only used it as an excuse for this particular action. So, it's not all about you personally. There are plenty of people here who throw this excuse out no matter what a company does. Sony Rootkit? They have to make profit. Oil spills? Oh well, they have to make profit. Don't be evil, but you know that's only so long as they make profit. Steve Jobs telling others to stop doing DRM while it continues to be the only option for iTMS? Apple's a company, they have to profit. MS monopoly? They're a business, they have to make profit. There's also either an implied or stated "What do you expect?" at the end of that.
As for you, I'm logically deriving this from things you did say. If profit is a first priority, than everything else is secondary. I'm trying to get you to admit that there may be other things, even implicit things like following the law, which should come first, even if they mean less profit or none at all. Life savings or not.
I suspect that this is what you meant in the first place, and that you think I'm being pedantic here. But if you can't just fall back on "Well, they have to make a profit," then you'll actually have to explain to me why profit comes first in this case. If you don't care about OpenOffice or open source, or don't think they're important enough here, you could say that, and I'll accept it -- and sadly, a different group of fanatics might come and flame you then. If you honestly think it would kill Dell's business model to distribute OpenOffice, I could say I doubt it, but we'd be done, since at that point, we're really in speculation.
Maybe I'm just doing a reductio ad absurdum argument.
I can't imagine I'd care much about saving face now, really -- look how deep we are into the thread. No mod can hear you scream.
Maybe you haven't been following the discussion about things like preinstalled Linux.
It's not about hating MS. It's about the fact that MS has such a huge monopoly on desktop OSes and office suites that we're even having a discussion about whether Dell can afford to do anything but install a trial Office. It's not that the morally correct thing to do is take up pitchforks and go kill MS, but rather, to not be a part of that machine where it makes sense.
I'm illustrating a point, and you missed it. Again.
I would rather have Dell die -- the company, not any individual -- than have the Microsoft monopoly last another ten years. I would rather have Dell make less profit and distribute OO than have them make more and distribute trial Office, though I admit that's probably not worth losing a company over.
You are the one who made it black and white here, though. According to you, if MS Office
I don't even want to have to use daemontools, much less some nasty crack. And I'm certainly not going to be a good little consumer and put the game disc in every time I play, scratching it more and more until it's unreadable.
I honestly could care less if the game works offline, though others will certainly want them to work on their laptops even without wireless. I don't even care if it's horribly inefficient -- say, a 5k/s trickle of data -- my gaming rig is always plugged into nice fast DSL, except when it goes out, which is usually when the power goes out anyway.
I'd much rather have a completely DRM-free game. I actually bought Darwinia and Uplink. Both are available in downloadable form, and the downloads, as far as I can tell, have absolutely no copy protection at all, other than the limit of only being able to download it 3 times (which is fine, considering the games are only some 50 megs, I can back that up forever). So, as far as I could tell, even the Windows version should have been easy to find on a filesharing network, but I didn't care, I went and bought them anyway.
But given the choice, fuck it, saturate my connection all you want, send them everything you can find on my computer -- not that you'll find much (my Wine is isolated enough, and my Windows has NOTHING of value on it) -- just don't lamely insist on a CD.
Whenever I see that bit, I tell the people sitting there, "You know, the pirated version doesn't have that ad..."
Let's think about this. They want pirates to stop pirating, so they go and try to beat their message into... their own legitimate customers? To the point where they actually end up driving some of them to piracy, just so they don't have to be called a pirate?
My own compromise has generally been renting and ripping. I think the rental price is fair, especially considering I simply don't have the hard disk storage to keep the movie, and it's not worth the hassle to burn and keep track of a huge archive of DVDs (especially when most movies must either be re-encoded or cut (special features, etc) or split onto two disks).
But yes, that's generally my procedure with any game, whether I buy it or not: If it includes CD-based copy protection, I go download a crack and/or convince it to run from Daemon Tools on Windows or cdemu on Linux. (Haven't gotten the cdemu to work for one of these, yet, but I think I'm getting closer.)
But I was shocked to find how true it was for me. I seriously considered for a moment, and yes, if Sony was doing this to, say, YouTube, I'd probably see it as stealing. Not horrible, not a rootkit, but still kind of a bad thing to do.
Whereas my initial reaction to this piece was more of the kind of "Sony needs to wake up to the realities of the Internet, and if it's not legal to do this, it should be."
I think Searchles is probably pretty cool, but I also think that this is going to end up being a bit like several Linux tools -- for instance, ies4linux. While it's not likely to be an issue for most people, it's probably illegal for them to redistribute IE, and it's certainly not legal to have a copy of IE without a Windows license. However, they probably have avoided trouble because they're so small, they warn you that it might be illegal, and they do it with deep linking, not mirroring -- you actually are downloading IE from Microsoft's website.
Besides: What copy protection, anyway? Sending an unencrypted flash video to a proprietary browser plugin is worse than DVD CSS, and that's saying something!
Moral of the story? I suppose I won't shed any tears for Sony anyway. Sony can do no right, and Google can do no wrong -- but they earned these reputations, and they continually reinforce them. Compare that to, say, desktop Linux, where there are actually far fewer jokes about recompiling your kernel, as people start to realize that it's unlikely you'll have to compile anything while using Linux, ever, unless you want to -- it certainly earned its reputation as a geek operating system, but it's now working to correct that.
Cygwin is slower than Ruby, and that's saying something.
Think about it -- process creation on Windows is an order of magnitude or two higher than process creation on Linux, and that's ignoring the implications of an unoptimized fork. Wine just seems a lot more likely to be fast than Cygwin ever could.
And all of that assumes that there's a point, really. Even if Windows is easier to admin than Linux, I'm not sure Cygwin on Windows has any advantages over the same app on Linux. The reverse is true also -- in fact, I know for a fact that Windows apps under Wine aren't always better than they are under Windows -- sometimes they are, mostly not.
And I'm thinking the best way to crack down would be to force people to be responsible for what happens on their hardware, unless someone signs some sort of waiver first -- and maybe some sort of standards as to what kind of community you have to be to allow that. A difference between something like YouTube, or an ISP, and something like an anonymous, deliberately unprotected FTP server.
I have ssh keys distributed everywhere. If I really wanted to, I could do something like:
for machine in `cat machines`; do
ssh root@$machine apt-get update
ssh root@$machine apt-get dist-upgrade
done
For the record, that's sloppy -- will take twice as long as a smarter command I didn't feel like putting together, which would require only one ssh command. Could also tell ssh to never prompt for a password, just log failures, and I could create a different loop which iterates over every ip on the network, just in case I've forgotten about a box.
Of course, I could just put it in the individual machines' crontabs, and/or use something like cfengine. There are all kinds of ways to do that.
But really, the above commands are pretty close to a one-liner, and that's assuming you're comfortable pushing updates out to thousands of machines without verifying each one -- the verification is where most of the time comes from.
Stick some apt-get commands in cron. You're done.
Oh, and let's not forget -- these are not just kernel updates. It's not even comparable to a Windows Update from MS -- every single piece of software on the machine is controlled by package management. Everything -- from a word processor to a web browser to a game to a mailserver to some random webapp that was nice enough to provide a package -- all of those will automatically be updated, and in the same place, with the same local cache (apt-proxy as one example).
Or use a distributed filesystem -- you can run the entire OS off the network. Not a thin client, but any change you make, every config file tweak you do, is instantly propagated over the network. This is actually pretty easy to set up... on Linux. I don't even know if it's possible to boot Windows off a network.
And by the way, 15 minutes of admin time is nothing when you consider other possible time sinks. I have to spend that 15 minutes maybe once a week, if I feel like it, and I really don't have to do much else. How much time do you spend?
If all you're doing is looking at spreadsheets, training will take you ten minutes. ("The spreadsheet is here, the Internet is here, email is here, and the shutdown button is here. Any questions?")
Not to mention that it'll be free forever. When MS releases whatever their next OS is and charges $300 for it, and Apple releases Leopard and charges $130 for it (and another one the year after, and the year after that), if you've done your Linux right, you'll run one command and your systems will automagically upgrade. All of them, for free.
You could netboot them, or any of a number of other ways.
What is it you're missing from Active Directory? I bet I can do it with "a cludge of logon scripts, spit, and duct tape" in less than ten minutes. You can probably do it in ten seconds with Active Directory -- and I'll gain those ten minutes a hundred times over when your network gets hit with a simple worm.