GTA SA uses disk spooling, for example, to present its vast (compared to previous games) world.
Maybe we can clarify disk spooling...
When I imagine making a game like GTA, with a vast world, I imagine something close to just creating a 5 gig file or so, and running it through mmap.
HL2 uses almost 1 GB of RAM, and that is for just a single level
How much of said single level is visible onscreen at once? Why must it all be cached in RAM? Can players really walk so fast that dynamically loading things just beyond the horizon can't keep up?
I hate to keep bringing it back to the PS2, but the pilot game Jak & Daxter, all the way through Jak 3, did dynamic loading, and looks as good as or better than most PS2 games. Prince of Persia games did that too, and were praised for their attention to detail. Halo and Halo 2 did it, but most people can't tell, and the people who can usually only see visual artifacts in split-screen.
And we have much more CPU to spare than the PS2 does, and games get installed to real hard drives, so why should desktop games be using huge amounts of RAM when they could cover it with dynamic loading? Even with huge caches/buffers to ensure that you don't run into the edge of the world, I can't see us needing more than a gig for at least another year or two -- and probably considerably less.
No. I am talking about procedurally developed graphics/models that are developed in real-time (i.e. as the game progresses) and stored in memory for later reference;
How is this significantly different than what we do now, loading off the disk and storing in memory? Certainly with dual core processors on the rise, people could have a whole core or two completely idle while they're playing -- why not use that for generation? Not every frame, but on demand. If it really becomes prohibitive, use a disk cache also.
Remember, we only really need enough RAM for what a player can see from any one point in the game, and then maybe twice that to make sure the player can't walk faster than we can load. Now, who actually uses the 512 megs of video RAM that we have now?
What? have you compared PS2 games to PC games lately? PS2 games look previous generation (because they are previous generation). And let's not forget resolutions: PS2 games are usually in 640x240, whereas PC games are 1280x1024 (at least!)...
Fine, then. The Xbox 360 ships with half a gig. The ps3 will ship with 256 megs. We already know the 360 is capable of playing HD games, which are well over 1280x1024.
Half-Life 2 doesn't use a gig of RAM because it needs to, it uses a gig of RAM because they know they can, and they'd rather be spending time on sexier programming (HDR, film grain, facial expressions, physics, gameplay) and actual artwork than reducing RAM usage and load times, especially when they know that their game is such a classic (due to all of the above) that everyone will buy it anyway, and upgrade their system to be able to play it, and STILL tolerate being interrupted every 2 minutes by a loading screen.
Now, if we do actually fill those ridiculous 512 meg cards, then yes, we will likely need more than 4 gigs of RAM. Just spinning in a circle, you'd probably need a gig or two.
To me, it already sounds normal; I am not impressed.
You're living in the future, then. I certainly couldn't afford a system with that much RAM (32 gigs?), and if I could, I wouldn't (yet) know what to do with it, other than use it for more cache.
Oh, I'm sure 64-bit gaming will make it matter, eventually. I just don't think it will matter as soon as you think, but maybe you're right. Lazy programmers have driven upgrade cycles before...
On the other is Linus Torvalds, the father of Linux. He and others in his open-source camp believe that freely sharing code simply produces the best software, but if other people want to hide their code, that's fine, too.
Will someone (preferrably Linus) please tell me why the GPLv2 is any kind of rational choice. From what I keep hearing over and over again, it really seems like Linux could have done just fine under a BSD license.
Of what use is it to have the source code if you can't change a line and recompile, and have it run just the same?
If Linus is really ok with Tivo as it currently is, would he be equally happy with a Tivo with no source code at all, just a penguin sticker on it? Or not even that?
Again: Why isn't Linux under a BSD license if Linus doesn't care about this RMS philosophy?
It really seems to me like his opposition to the GPLv3, however irrelevant (Linux can't reasonably move to GPLv3, as you'd have to contact all copyright holders), seems to stem more from a misunderstanding of GPLv3, rather than actual disagreement. But that was only last I checked, thinks may have changed.
Sun was one. Remember Java, or what it was supposed to be? AJAX 10 years earlier and 10 times faster. No wonder MS had to kill it.
And then there's ActiveX, claimed to be implemented in Internet Explorer to deliberately counter Netscape's plugin architecture. I don't know the history about that well enough to make a case...
More recently, we've got other browsers: Mozilla/Firefox, Konqueror/Safari, and Opera. And by "More recently", I mean 2001-03. MSN.com used to have a CSS flaw that, be it deliberate or accidental, was wholly unnecessary -- they modified the CSS sent out to Opera (using browser detection), when the CSS they send to other browsers (like IE) renders fine in Opera. Done subtly enough that the CSS could look like a typo (but why change the CSS in the first place?), and if you didn't notice the flaw was with MSN, it looked like a bug in Opera.
By making.doc a moving target (for no apparent reason), they deliberately killed off both old versions of Office and competitors. That, or they really do suck that much at following their own "standards".
Speaking of which, what happened to RTF?
After much talk about why XML is slow, will always be slow, will never be good enough, etc ad absurdum, MS releases WordML, which shows why they think so -- WordML is a serialization of Word internal structures, probably roughly equivalent to.DOC, only in XML. They still don't implement ODF, which was actually designed to be cross-implementation.
A perfectly good graphics library exists in OpenGL. Why, then, does Microsoft implement DirectX? Especially in today's world of SDL, OpenGL, and OpenAL, it's quite easy to develop a cross-platform game, except that everyone develops for DirectX, making it much more difficult to even support Windows games under Wine. I have never, ever heard a convincing argument for DirectX being any better than SDL, except that it is better documented -- which means that there was certainly a time when it would have been considerably less work for them to simply document SDL than to continue to develop DirectX. From a purely technology standpoint, there really doesn't seem to be a significant difference these days, especially with shops like id pushing the major graphics cards to keep OpenGL working.
And as an added bonus, Vista promises to implement OpenGL on top of DirectX, potentially up to 2x slower than it is in XP. Why did MS use DirectX for their flashy new interface, when Apple has proven with Quartz Extreme that OpenGL can do it?
I can see no reason for this other than to maintain their dominance of desktop games. After all, if they were to depricate DirectX in favor of OpenGL and SDL, and developers coming out of gaming colleges started to realize that if you don't do anything stupid, games are almost trivial to port between OSes, then MS might actually have to (Shock! Awe!) compete on a level playing field in the desktop gaming arena. I don't think they're up to the task. Joe Gamer, tweaking every last ounce out of his overclocked, water cooled, assembled-by-hand gaming rig, is not going to risk losing the FPS edge to a few random Windows processes, viruses -- or AntiVirus software (Norton has interrupted more than one game for my roommate) -- so the logical solution is to run Linux for his games, in a minimal desktop environment like Fluxbox or WindowMaker. Or maybe Susie Newbie wants to play a few games -- she's never played before -- but she's a computer newbie, so she likes Macs, because they "just work" -- it would be nice if she could play more than WoW.
I could find more examples, but I think that's enough for now. It's the old "embrace and extend" philosophy at work. Everyone wonders why we all love Google? When AOL implemented an IM system, they made it closed and proprietary. Ditto Yahoo. Ditto Microsoft. Recently, AIM and Yahoo have native Linux versions, but Microsoft only has Windows and Mac. AOL has also aquired ICQ and made AIM talk to ICQ, an
Ruby isn't very fast to begin with, but I strongly suspect that using Ruby on the client-side, with some sort of cross-platform graphics library, would be at least 2x as fast as JavaScript+HTML for the client-side interface.
The point is, Ruby is known for being slow. I'm demonstrating what a stupendously bad idea it is to build an application in AJAX, if you have any alternatives.
On the other hand, some of the concepts that made it so hard for me to get into Linux in the first place are reasons why I'll be defecting back from OS X, after using it on my laptop for over a year. The lack of a decent, well-supported package manager is high on the list. I'm sick of my apps telling me at random times that I have to stop doing what I'm doing and update -- or worse, having to update every app individually. "Does FileMaker have an update? Hmm... Cog? Chicken of the VNC? VLC? Mplayer? Sunbird? Cyberduck? OS X? AbiWord? NeoOffice? Gimp?"
For a "user-friendly" OS, this is a serious pain in the ass, coming from a world of "emerge --sync && emerge -uaDN world" and "apt-get update && apt-get upgrade" and even Ubuntu's system tray app -- "Oh, I have updates for some of the programs on your system -- click here and let me do everything for you."
Getting a particular program to work may be a challenge, but I imagine I waste less time that way than I do making sure I'm up-to-date on my Mac whenever I want to submit a bug report.
What's your point? Why does it matter who added the descriptions or not?
Besides, it's good to be distinctive once people know what you are. For instance, once people know what a web browser is, it really helps to be called something like "Firefox" -- or Konqueror, or Safari, or Opera. So you google for "Web Browser", look for a good one, then you want it everywhere -- when you're at a friend's house, you hate their Internet Explorer, because it's not Firefox. Or something...
Compare this to, say, Outlook vs Outlook Express. Or Netscape versus Internet Explorer -- that's just a recipie for confusing people, as those trained on Netscape will think "Netscape == Internet", only less consciously in that they won't understand the word "Internet", just "net", so they won't know what Internet Explorer is.
The postfixes are a good idea, because that way people will know what kind of program it is, so that if they don't like it, they know what to search for -- instead of "ekiga replacement", they could search for "Linux softphone". And so on.
And really, explain to me why the new user would ignore the postfixes? Explain to me why it matters what name a project actually has, if most users will only ever install anything through package management -- and thus it only matters what the package manager calls it?
Could you clarify for me what your point is, or exactly when a new user is supposed to encounter these names and not understand what they mean? So far, this seems like as old and tired and untrue an argument as the "instructions for running Quake 3 on Linux" or "I shouldn't have to recompile my kernel to get a printer working" arguments.
This is just as retarded as the last comment about memory.
Look, when was the last time a game used anywhere close to 4 gigs of RAM? Yes, we want to stay ahead of the curve, but WTF does this have to do with procedurally generated stuff?
I mean, you are talking about this, right? As I understand it, this does not use significantly more RAM than a static animation -- in fact, it uses significantly less, as rather than reading (and caching) a static death animation, you have a (much smaller) formula for what the death animation should look like, which you generate in real time.
So 64-bit still helps, but it helps because it's faster, not because it can support more RAM. More RAM will help with more detailed models, unless those are also procedurally generated. Remember, consoles have been severely gimped for RAM for ages. Many PS2 games still look amazing, and the entire system has 32 megs of RAM.
Also, remember how stupid you're going to sound when "multigigabytes" will sound to future generations the way "multimegabytes" sounds to us. How do you like the sound of a system with 32 gigs of RAM? That's going to sound to your grandkids the way the PS2 sounds to us today.
As long as I can remember compiling my own kernel, there's been the option, on normal 32-bit Linux, to either not support high memory (and be limited to 896 megs of RAM), or support 4 gigs of RAM, or 64 gigs of RAM. I believe individual processes are limited to 4 gigs, though.
It's Windows that's been locked at 4 gigs total RAM, not the 32-bit arch, although I readily admit that 64-bit does it in a much cleaner way. But really, we should be upgrading to 64-bit for the obvious speed increases, at really no cost to us. It's a natural upgrade, and better to get it done sometime in the next 5 years than wait till applications really are hitting that 4 gig limit, or system RAM really is hitting the 64 gig limit.
For what it's worth, this has already been done with Google Browser Sync. It allows you to backup all your personal settings -- passwords, bookmarks, everything -- from your browser to Google's servers. It's a small-ish Firefox extension that so far seems platform independent -- tried on OS X PPC and Linux PPC, x86, and amd64 -- which would suggest that Google has working JavaScript encryption.
So, it would be nice for that to be able to prevent the plain text from hitting Google's servers, but then, that would prevent the kind of advertising they want. Anyway, it wouldn't be useful for Gmail, unless you're willing to use PGP -- and then they still get everything anyone decides to send you in plaintext, and everything you're required to send in plaintext due to the fact that 99% of the world doesn't use PGP. Hell, I try to use GnuPG for my email that runs through my own server, and it still means Google has everything I ever send to/from anyone else's Gmail account -- and you can bet they know how to aggrigate that into my permanent GRecord.
Still, my guess is that most of the truly massive number of people willing to use Gmail are actually, like most people, completely unaware of even the most basic security concepts. Most people seem to have some illusion about security that's some composite of the following:
Any security will be broken, but mine never will, because it never has (unconscious security-blanket warm-fuzzies), and because I'm not important enough for anyone to want what I have. And if I am important enough, then, why, it's not my job to keep myself secure -- the IT department handles that -- so as long as they're doing their job, absolutely everything I do is secure, including Gmail -- in that case, I'm just outsourcing my security to Google. And anyway, anyone who thinks I'm not secure is just paranoid, because there's no way it's as bad as they're saying it is -- or if it is that bad, it doesn't matter anyway, because...
And it goes on and on and on.
So yes, most people would gladly turn their computer into a thin-client to the Google Grid, if it made their lives easier. Actually controlling their own computers is yet another kind of doublethink -- it's unreasonable to ask for control of my computer on the level that Linux/OSS people want, but I do control my computer, absolutely, sortof.
Slashdot is biased one way, Digg is biased another -- but reagrdless, it's purely subjective. You see individual Slashdotters and individual Diggers with significantly different opinions than the "groupthink". I'm beginning to think the only real groupthink here is among the people who always scream "groupthink" when a Slashdotter disagrees with them.
The IE team broken compatiblity with previous versions of IE _just_ to make it more standards compliant.
It's not the intent we care about, it's the end result. Microsoft has broken standards in the past deliberately to kill competitors. It's been done over and over and over again -- Java, HTML, CSS, RTF... That's just what I can remember off the top of my head.
The burden is not on us to prove that Microsoft is doing it again. The burden is on Microsoft to prove they are not. Breaking compatibility with previous versions of IE is the right thing to do, but I'll believe they're actually doing it when I see it.
Writely works fine in Opera if you "fake" it, yet nobody complains about that.
Of course they do. I assume you're reading the threads right here on Slashdot on how to fake it, and those threads do include complaining.
By the way, this is not Opera's fault. If anything, it's Google's fault.
Writely doesn't work at all in Safari, and Google provides some excuse, everyone says OK.
That "excuse" is a feature that is genuinely missing in Safari. If you really think it should work in Safari, what's stopping you? Go ahead and prove them wrong, and implement a decent word processor in Safari.
You don't like IE/Microsoft, that's fine, use something else. But cut the crap with the paranoia.
How is it paranoia to expect Microsoft to do exactly the same thing they've always done? It's only very recently that Microsoft has begun to support any kind of open standards, and it does so half-assedly, usually only when pressured into it. Meanwhile, they've got over 30 years of history to make up for. 30 years of business ethics that are questionable at best.
It's a good idea with a bad implementation. JavaScript and HTML were never designed to implement full applications. Really, we'd be better off with Ruby everywhere -- it might even be faster.
See, this kind of sucks for people who actually care about the ethics of it. Suppose you've got a violent crime, like, say, a murder -- no amount of punishment for the perp will ever bring the victim back. On the other hand, something like this -- especially something that has already been solved -- will, in fact, bring the laptop back.
Now, you could claim that catching a murderer is better for society, because it prevents them from killing again, but really, how many one-shot murders are there? How many serial killers? Besides, it can't be good for society to have cops constantly losing respect by being lazy assholes.
You know there is always a better or faster or cheaper way. With this program it is the same as with a car. There is no 100% protection, but it help's a lot to lock it.
</sarcasm>
Actually, the WebSafe "Website Encryption" is much better for keeping away "prying ices" than U3. At least WebSafe actually does some kind of encryption, even if the decryption algorithm and the keys are right there in the source code for everyone to see. U3, on the other hand, at least appears to claim encryption where there is none. I'll direct you to their website, where they claim:
The U3 platform is designed to leave no trace of the user's data or application usage on the host computer after the smart drive is removed. The U3 platform also supports the creation of security solutions to protect the privacy and security of user data and applications. These solutions include encrypted files and folders, and sign-on and password protection and management.
Oh, I get it. They "support the creation" of encryption, when actually, if you look at their smart drive page, the word "encryption" is nowhere to be found. Instead, it's all about "Password Management" -- so they keep themselves clean, but it's obviously confusing enough to fool customers, especially when others claim "Secure data encryption" on what they call a "U3 Smart Drive", although I can't figure out whether Verbatim is wrong/lying or whether they've simply taken the existing U3 software and actually added encryption.
Or maybe there's some other loophole. But even if I wasn't planning on using the encryption, I wouldn't do business with these jokers. (U3, not necessarily Verbatim.) It's clearly designed to fool people into thinking they're getting something they're not, which really makes them no better than the WebSafe moron -- and perhaps significantly worse, as the WebSafe guy may actually still believe his product is worth something.
Unless I'm very wrong, brute-forcing can be pretty easily averted by simply using a long enough password. Last I checked, 8 chars is secure.
Remember, if it's a standard USB drive, then as I understand it, any software mechanism to force things like time delays would be easily circumvented by simply not using that software. But then, I hear things like "doesn't work on Mac/Linux", so that makes me think it's not quite standard, so maybe they could force something like this in hardware?
Is there any discount involved in getting the blank HD version? If not, don't bother -- you're paying the MS tax anyway, you may as well get an MS license. You might even try EULA tricks to see if you can get a refund.
If there is such a discount, I applaud Lenovo for this choice -- I would rather see them preload it, but personally, I buy all my desktops with a blank HD anyway, and there's no way I'm letting anyone else install my Linux. And it's a way to not pay the MS tax on a laptop, which is pretty rare.
I learned by doing -- by reading manpages. But you don't have to RTFM -- reading comments is much different than TFM, because comments are right there next to the option.
And dabbling below the surface of a Linux desktop, just about anything that would be graphical at all is already handled. What's left boils down to whether it really makes a meaningful difference for people to have a GUI versus a config file, for a program which has no GUI. So, for instance, you're saying we need a GUI for fstab?
Actually, I kind of agree about fstab, but most of the other ones are at the point where if you gave them a toy GUI to mess with a toy config file for a toy program, and gave them 5 minutes with that, they should be able to read through most real config files and grok the comments. For most purposes, I really can't see this as being useful in everyday life, unless it's somehow more convenient (doubtful), or unless we have a situation of people who are terrified of text, or people who only very, very occasionally have to tweak a setting...
Maybe you make a good point though. Kids. Kids have reasonably flexible minds. Adults were raised on GUIs...
Yes, it was a fabrication -- you missed the humor in it.
Were there a god, how would this god know he wasn't created by some other, more powerful god?
All-knowing, but that's beside the point. Do you claim that there cannot possibly be a God? I'm capitalizing the G for a reason -- we're talking about a monotheistic God.
As you say, a house could have cookies in it.
Depending what definition we use, God can either be necessarily true, or just another deity, or not true at all. Suppose, for a moment, we're talking about just another deity, that really can't tell if he was or wasn't created by another, more powerful god. In this case, there most certainly can be a Christian God, even if it's some Stargate-like God -- just like any of us, but He has better technology.
We simply do not know.
Let me ask you another question, while you're on that topic: Do you believe in extraterrestrial life? Why or why not? Do we have any more evidence for that than we do for the evidence of a sky-deity? What about the reverse: Do we have any more evidence against the sky-deity than we do against extraterrestrial life?
We have a complete and utter lack of evidence, one way or the other, at least about the sky-deity. With ET, we only have statistics -- lies, damn lies, and statistics. You know, there is a statistical argument that suggests that if we will ever have fully immersive, Matrix-like VR, then we are very probably there already -- and that, in itself, would allow a God, a Heaven, and a Hell, in just about every meaningful sense.
But let's not even get started about moral relativism.
I always wondered whether Google ever considered just throwing a link on their homepage occasionally, when they really want people to read something. I mean, sure, it would be immeasurably worse than a Slashdotting, but even something like the Net Neutrality stuff.
I doubt they would actually do it, though. A large advantage Google has over the competition is that they are at least perceived as a commons -- anyone can buy Google adspace, and it has nothing to do with their relationship with Google and everything to do with statistical analysis -- PageRank.
My original argument was that we should be spending more time fixing Wine and less time doing VMs, because Wine is inherently a better solution when it works.
If it's a server, I usually have a lot more choice, so moving config files is pretty much everything. Library hell does not exist when you use a good package manager, and mine (Gentoo's Portage) has a config file which says which packages I want installed. Thus, all I have to do is copy config files to the new box, run "emerge world", and all the software is installed and configured. So I really am limited to doing things specific to the hardware -- for instance, I moved from an x86 box to an amd64 one, so I did have to tweak some things -- kernel build scripts now need to know I want x86_64, for one -- but a huge amount of things (I'll refrain from pulling a number out of my ass, since you take them too literally) just worked.
Good point about modeling networks, though. I wish I had a nice, simple way of doing that. As it is, I'm more careful and I set up logging rules first, so I know how traffic is going to hit that particular rule before I implement it. Although a single OS on a single machine does work for having multiple IP addresses, I think most packet filters place a bit too much implicit trust on locally-originating traffic for that to be useful.
I don't think I said heterogeneous. What I don't know is if I can resume on a system with identical hardware, except inescapable differences like Mac address.
And I know I've seen XP hibernate in less time than it could possibly take to flush the entire system. 2000 takes a constant amount of time, based on how much RAM you have, XP seems to depend on how much you're actually using.
Well, the good news is, if you try to take all of that into account, and make a very conservative estimate to begin with, and you follow Doug's Law, then you end up with something that is pretty damn accurate, or at least close enough that your employers/beancounters/customers won't doubt it. Then, if you really can get it done in a fraction of the time, it's that much better for you.
Oh, I'm going to add my own amendment to Doug's Law: Doug's Law only covers unknowns. Anything you know -- time cost by miscommunication, time spent polishing the product -- even if you don't know how much time it will take, if you know what to call it, you must add something to your pre-Doug estimate.
The point is, you're going to have to give them some number, and it can't be too insane or you'll never get the go ahead. "When it's done" won't cut it. But it's far, far better to be ludicrously early than slightly late.
Slashdot Mirror
Maybe we can clarify disk spooling...
When I imagine making a game like GTA, with a vast world, I imagine something close to just creating a 5 gig file or so, and running it through mmap.
How much of said single level is visible onscreen at once? Why must it all be cached in RAM? Can players really walk so fast that dynamically loading things just beyond the horizon can't keep up?
I hate to keep bringing it back to the PS2, but the pilot game Jak & Daxter, all the way through Jak 3, did dynamic loading, and looks as good as or better than most PS2 games. Prince of Persia games did that too, and were praised for their attention to detail. Halo and Halo 2 did it, but most people can't tell, and the people who can usually only see visual artifacts in split-screen.
And we have much more CPU to spare than the PS2 does, and games get installed to real hard drives, so why should desktop games be using huge amounts of RAM when they could cover it with dynamic loading? Even with huge caches/buffers to ensure that you don't run into the edge of the world, I can't see us needing more than a gig for at least another year or two -- and probably considerably less.
How is this significantly different than what we do now, loading off the disk and storing in memory? Certainly with dual core processors on the rise, people could have a whole core or two completely idle while they're playing -- why not use that for generation? Not every frame, but on demand. If it really becomes prohibitive, use a disk cache also.
Remember, we only really need enough RAM for what a player can see from any one point in the game, and then maybe twice that to make sure the player can't walk faster than we can load. Now, who actually uses the 512 megs of video RAM that we have now?
Fine, then. The Xbox 360 ships with half a gig. The ps3 will ship with 256 megs. We already know the 360 is capable of playing HD games, which are well over 1280x1024.
Half-Life 2 doesn't use a gig of RAM because it needs to, it uses a gig of RAM because they know they can, and they'd rather be spending time on sexier programming (HDR, film grain, facial expressions, physics, gameplay) and actual artwork than reducing RAM usage and load times, especially when they know that their game is such a classic (due to all of the above) that everyone will buy it anyway, and upgrade their system to be able to play it, and STILL tolerate being interrupted every 2 minutes by a loading screen.
Now, if we do actually fill those ridiculous 512 meg cards, then yes, we will likely need more than 4 gigs of RAM. Just spinning in a circle, you'd probably need a gig or two.
You're living in the future, then. I certainly couldn't afford a system with that much RAM (32 gigs?), and if I could, I wouldn't (yet) know what to do with it, other than use it for more cache.
Oh, I'm sure 64-bit gaming will make it matter, eventually. I just don't think it will matter as soon as you think, but maybe you're right. Lazy programmers have driven upgrade cycles before...
Will someone (preferrably Linus) please tell me why the GPLv2 is any kind of rational choice. From what I keep hearing over and over again, it really seems like Linux could have done just fine under a BSD license.
Of what use is it to have the source code if you can't change a line and recompile, and have it run just the same?
If Linus is really ok with Tivo as it currently is, would he be equally happy with a Tivo with no source code at all, just a penguin sticker on it? Or not even that?
Again: Why isn't Linux under a BSD license if Linus doesn't care about this RMS philosophy?
It really seems to me like his opposition to the GPLv3, however irrelevant (Linux can't reasonably move to GPLv3, as you'd have to contact all copyright holders), seems to stem more from a misunderstanding of GPLv3, rather than actual disagreement. But that was only last I checked, thinks may have changed.
Sun was one. Remember Java, or what it was supposed to be? AJAX 10 years earlier and 10 times faster. No wonder MS had to kill it.
.doc a moving target (for no apparent reason), they deliberately killed off both old versions of Office and competitors. That, or they really do suck that much at following their own "standards".
.DOC, only in XML. They still don't implement ODF, which was actually designed to be cross-implementation.
And then there's ActiveX, claimed to be implemented in Internet Explorer to deliberately counter Netscape's plugin architecture. I don't know the history about that well enough to make a case...
More recently, we've got other browsers: Mozilla/Firefox, Konqueror/Safari, and Opera. And by "More recently", I mean 2001-03. MSN.com used to have a CSS flaw that, be it deliberate or accidental, was wholly unnecessary -- they modified the CSS sent out to Opera (using browser detection), when the CSS they send to other browsers (like IE) renders fine in Opera. Done subtly enough that the CSS could look like a typo (but why change the CSS in the first place?), and if you didn't notice the flaw was with MSN, it looked like a bug in Opera.
By making
Speaking of which, what happened to RTF?
After much talk about why XML is slow, will always be slow, will never be good enough, etc ad absurdum, MS releases WordML, which shows why they think so -- WordML is a serialization of Word internal structures, probably roughly equivalent to
A perfectly good graphics library exists in OpenGL. Why, then, does Microsoft implement DirectX? Especially in today's world of SDL, OpenGL, and OpenAL, it's quite easy to develop a cross-platform game, except that everyone develops for DirectX, making it much more difficult to even support Windows games under Wine. I have never, ever heard a convincing argument for DirectX being any better than SDL, except that it is better documented -- which means that there was certainly a time when it would have been considerably less work for them to simply document SDL than to continue to develop DirectX. From a purely technology standpoint, there really doesn't seem to be a significant difference these days, especially with shops like id pushing the major graphics cards to keep OpenGL working.
And as an added bonus, Vista promises to implement OpenGL on top of DirectX, potentially up to 2x slower than it is in XP. Why did MS use DirectX for their flashy new interface, when Apple has proven with Quartz Extreme that OpenGL can do it?
I can see no reason for this other than to maintain their dominance of desktop games. After all, if they were to depricate DirectX in favor of OpenGL and SDL, and developers coming out of gaming colleges started to realize that if you don't do anything stupid, games are almost trivial to port between OSes, then MS might actually have to (Shock! Awe!) compete on a level playing field in the desktop gaming arena. I don't think they're up to the task. Joe Gamer, tweaking every last ounce out of his overclocked, water cooled, assembled-by-hand gaming rig, is not going to risk losing the FPS edge to a few random Windows processes, viruses -- or AntiVirus software (Norton has interrupted more than one game for my roommate) -- so the logical solution is to run Linux for his games, in a minimal desktop environment like Fluxbox or WindowMaker. Or maybe Susie Newbie wants to play a few games -- she's never played before -- but she's a computer newbie, so she likes Macs, because they "just work" -- it would be nice if she could play more than WoW.
I could find more examples, but I think that's enough for now. It's the old "embrace and extend" philosophy at work. Everyone wonders why we all love Google? When AOL implemented an IM system, they made it closed and proprietary. Ditto Yahoo. Ditto Microsoft. Recently, AIM and Yahoo have native Linux versions, but Microsoft only has Windows and Mac. AOL has also aquired ICQ and made AIM talk to ICQ, an
Ruby isn't very fast to begin with, but I strongly suspect that using Ruby on the client-side, with some sort of cross-platform graphics library, would be at least 2x as fast as JavaScript+HTML for the client-side interface.
The point is, Ruby is known for being slow. I'm demonstrating what a stupendously bad idea it is to build an application in AJAX, if you have any alternatives.
You just prove my point. There's no point in brute-forcing, and you haven't convinced me that it would be possible to brute force it.
What, just like that?
Where's the solution?
On the other hand, some of the concepts that made it so hard for me to get into Linux in the first place are reasons why I'll be defecting back from OS X, after using it on my laptop for over a year. The lack of a decent, well-supported package manager is high on the list. I'm sick of my apps telling me at random times that I have to stop doing what I'm doing and update -- or worse, having to update every app individually. "Does FileMaker have an update? Hmm... Cog? Chicken of the VNC? VLC? Mplayer? Sunbird? Cyberduck? OS X? AbiWord? NeoOffice? Gimp?"
For a "user-friendly" OS, this is a serious pain in the ass, coming from a world of "emerge --sync && emerge -uaDN world" and "apt-get update && apt-get upgrade" and even Ubuntu's system tray app -- "Oh, I have updates for some of the programs on your system -- click here and let me do everything for you."
Getting a particular program to work may be a challenge, but I imagine I waste less time that way than I do making sure I'm up-to-date on my Mac whenever I want to submit a bug report.
What's your point? Why does it matter who added the descriptions or not?
Besides, it's good to be distinctive once people know what you are. For instance, once people know what a web browser is, it really helps to be called something like "Firefox" -- or Konqueror, or Safari, or Opera. So you google for "Web Browser", look for a good one, then you want it everywhere -- when you're at a friend's house, you hate their Internet Explorer, because it's not Firefox. Or something...
Compare this to, say, Outlook vs Outlook Express. Or Netscape versus Internet Explorer -- that's just a recipie for confusing people, as those trained on Netscape will think "Netscape == Internet", only less consciously in that they won't understand the word "Internet", just "net", so they won't know what Internet Explorer is.
The postfixes are a good idea, because that way people will know what kind of program it is, so that if they don't like it, they know what to search for -- instead of "ekiga replacement", they could search for "Linux softphone". And so on.
And really, explain to me why the new user would ignore the postfixes? Explain to me why it matters what name a project actually has, if most users will only ever install anything through package management -- and thus it only matters what the package manager calls it?
Could you clarify for me what your point is, or exactly when a new user is supposed to encounter these names and not understand what they mean? So far, this seems like as old and tired and untrue an argument as the "instructions for running Quake 3 on Linux" or "I shouldn't have to recompile my kernel to get a printer working" arguments.
This is just as retarded as the last comment about memory.
Look, when was the last time a game used anywhere close to 4 gigs of RAM? Yes, we want to stay ahead of the curve, but WTF does this have to do with procedurally generated stuff?
I mean, you are talking about this, right? As I understand it, this does not use significantly more RAM than a static animation -- in fact, it uses significantly less, as rather than reading (and caching) a static death animation, you have a (much smaller) formula for what the death animation should look like, which you generate in real time.
So 64-bit still helps, but it helps because it's faster, not because it can support more RAM. More RAM will help with more detailed models, unless those are also procedurally generated. Remember, consoles have been severely gimped for RAM for ages. Many PS2 games still look amazing, and the entire system has 32 megs of RAM.
Also, remember how stupid you're going to sound when "multigigabytes" will sound to future generations the way "multimegabytes" sounds to us. How do you like the sound of a system with 32 gigs of RAM? That's going to sound to your grandkids the way the PS2 sounds to us today.
This is so retarded.
As long as I can remember compiling my own kernel, there's been the option, on normal 32-bit Linux, to either not support high memory (and be limited to 896 megs of RAM), or support 4 gigs of RAM, or 64 gigs of RAM. I believe individual processes are limited to 4 gigs, though.
It's Windows that's been locked at 4 gigs total RAM, not the 32-bit arch, although I readily admit that 64-bit does it in a much cleaner way. But really, we should be upgrading to 64-bit for the obvious speed increases, at really no cost to us. It's a natural upgrade, and better to get it done sometime in the next 5 years than wait till applications really are hitting that 4 gig limit, or system RAM really is hitting the 64 gig limit.
For what it's worth, this has already been done with Google Browser Sync. It allows you to backup all your personal settings -- passwords, bookmarks, everything -- from your browser to Google's servers. It's a small-ish Firefox extension that so far seems platform independent -- tried on OS X PPC and Linux PPC, x86, and amd64 -- which would suggest that Google has working JavaScript encryption.
So, it would be nice for that to be able to prevent the plain text from hitting Google's servers, but then, that would prevent the kind of advertising they want. Anyway, it wouldn't be useful for Gmail, unless you're willing to use PGP -- and then they still get everything anyone decides to send you in plaintext, and everything you're required to send in plaintext due to the fact that 99% of the world doesn't use PGP. Hell, I try to use GnuPG for my email that runs through my own server, and it still means Google has everything I ever send to/from anyone else's Gmail account -- and you can bet they know how to aggrigate that into my permanent GRecord.
Still, my guess is that most of the truly massive number of people willing to use Gmail are actually, like most people, completely unaware of even the most basic security concepts. Most people seem to have some illusion about security that's some composite of the following:
And it goes on and on and on.
So yes, most people would gladly turn their computer into a thin-client to the Google Grid, if it made their lives easier. Actually controlling their own computers is yet another kind of doublethink -- it's unreasonable to ask for control of my computer on the level that Linux/OSS people want, but I do control my computer, absolutely, sortof.
Sad, really, but what can you do?
Slashdot is biased one way, Digg is biased another -- but reagrdless, it's purely subjective. You see individual Slashdotters and individual Diggers with significantly different opinions than the "groupthink". I'm beginning to think the only real groupthink here is among the people who always scream "groupthink" when a Slashdotter disagrees with them.
It's not the intent we care about, it's the end result. Microsoft has broken standards in the past deliberately to kill competitors. It's been done over and over and over again -- Java, HTML, CSS, RTF... That's just what I can remember off the top of my head.
The burden is not on us to prove that Microsoft is doing it again. The burden is on Microsoft to prove they are not. Breaking compatibility with previous versions of IE is the right thing to do, but I'll believe they're actually doing it when I see it.
Of course they do. I assume you're reading the threads right here on Slashdot on how to fake it, and those threads do include complaining.
By the way, this is not Opera's fault. If anything, it's Google's fault.
That "excuse" is a feature that is genuinely missing in Safari. If you really think it should work in Safari, what's stopping you? Go ahead and prove them wrong, and implement a decent word processor in Safari.
How is it paranoia to expect Microsoft to do exactly the same thing they've always done? It's only very recently that Microsoft has begun to support any kind of open standards, and it does so half-assedly, usually only when pressured into it. Meanwhile, they've got over 30 years of history to make up for. 30 years of business ethics that are questionable at best.
I think it says a lot that Microsoft's very first product was vaporware.
So please, say what you will about Microsoft's recent apparent goodwill. But understand our honest skepticism -- and don't call it "paranoia".
It's a good idea with a bad implementation. JavaScript and HTML were never designed to implement full applications. Really, we'd be better off with Ruby everywhere -- it might even be faster.
See, this kind of sucks for people who actually care about the ethics of it. Suppose you've got a violent crime, like, say, a murder -- no amount of punishment for the perp will ever bring the victim back. On the other hand, something like this -- especially something that has already been solved -- will, in fact, bring the laptop back.
Now, you could claim that catching a murderer is better for society, because it prevents them from killing again, but really, how many one-shot murders are there? How many serial killers? Besides, it can't be good for society to have cops constantly losing respect by being lazy assholes.
You know there is always a better or faster or cheaper way. With this program it is the same as with a car. There is no 100% protection, but it help's a lot to lock it.
</sarcasm>
Actually, the WebSafe "Website Encryption" is much better for keeping away "prying ices" than U3. At least WebSafe actually does some kind of encryption, even if the decryption algorithm and the keys are right there in the source code for everyone to see. U3, on the other hand, at least appears to claim encryption where there is none. I'll direct you to their website, where they claim:
Oh, I get it. They "support the creation" of encryption, when actually, if you look at their smart drive page, the word "encryption" is nowhere to be found. Instead, it's all about "Password Management" -- so they keep themselves clean, but it's obviously confusing enough to fool customers, especially when others claim "Secure data encryption" on what they call a "U3 Smart Drive", although I can't figure out whether Verbatim is wrong/lying or whether they've simply taken the existing U3 software and actually added encryption.
Or maybe there's some other loophole. But even if I wasn't planning on using the encryption, I wouldn't do business with these jokers. (U3, not necessarily Verbatim.) It's clearly designed to fool people into thinking they're getting something they're not, which really makes them no better than the WebSafe moron -- and perhaps significantly worse, as the WebSafe guy may actually still believe his product is worth something.
Unless I'm very wrong, brute-forcing can be pretty easily averted by simply using a long enough password. Last I checked, 8 chars is secure.
Remember, if it's a standard USB drive, then as I understand it, any software mechanism to force things like time delays would be easily circumvented by simply not using that software. But then, I hear things like "doesn't work on Mac/Linux", so that makes me think it's not quite standard, so maybe they could force something like this in hardware?
Wow. +500 Informative, I never properly understood HT until now.
Nice sig, too.
Is there any discount involved in getting the blank HD version? If not, don't bother -- you're paying the MS tax anyway, you may as well get an MS license. You might even try EULA tricks to see if you can get a refund.
If there is such a discount, I applaud Lenovo for this choice -- I would rather see them preload it, but personally, I buy all my desktops with a blank HD anyway, and there's no way I'm letting anyone else install my Linux. And it's a way to not pay the MS tax on a laptop, which is pretty rare.
No, the UK has just figured out that it's impossible for most techs, let alone patent clerks, to figure out what's "obvious" in software.
It was turned down because it's about software, not because it's obvious.
I learned by doing -- by reading manpages. But you don't have to RTFM -- reading comments is much different than TFM, because comments are right there next to the option.
And dabbling below the surface of a Linux desktop, just about anything that would be graphical at all is already handled. What's left boils down to whether it really makes a meaningful difference for people to have a GUI versus a config file, for a program which has no GUI. So, for instance, you're saying we need a GUI for fstab?
Actually, I kind of agree about fstab, but most of the other ones are at the point where if you gave them a toy GUI to mess with a toy config file for a toy program, and gave them 5 minutes with that, they should be able to read through most real config files and grok the comments. For most purposes, I really can't see this as being useful in everyday life, unless it's somehow more convenient (doubtful), or unless we have a situation of people who are terrified of text, or people who only very, very occasionally have to tweak a setting...
Maybe you make a good point though. Kids. Kids have reasonably flexible minds. Adults were raised on GUIs...
Yes, it was a fabrication -- you missed the humor in it.
All-knowing, but that's beside the point. Do you claim that there cannot possibly be a God? I'm capitalizing the G for a reason -- we're talking about a monotheistic God.
As you say, a house could have cookies in it.
Depending what definition we use, God can either be necessarily true, or just another deity, or not true at all. Suppose, for a moment, we're talking about just another deity, that really can't tell if he was or wasn't created by another, more powerful god. In this case, there most certainly can be a Christian God, even if it's some Stargate-like God -- just like any of us, but He has better technology.
We simply do not know.
Let me ask you another question, while you're on that topic: Do you believe in extraterrestrial life? Why or why not? Do we have any more evidence for that than we do for the evidence of a sky-deity? What about the reverse: Do we have any more evidence against the sky-deity than we do against extraterrestrial life?
We have a complete and utter lack of evidence, one way or the other, at least about the sky-deity. With ET, we only have statistics -- lies, damn lies, and statistics. You know, there is a statistical argument that suggests that if we will ever have fully immersive, Matrix-like VR, then we are very probably there already -- and that, in itself, would allow a God, a Heaven, and a Hell, in just about every meaningful sense.
But let's not even get started about moral relativism.
I always wondered whether Google ever considered just throwing a link on their homepage occasionally, when they really want people to read something. I mean, sure, it would be immeasurably worse than a Slashdotting, but even something like the Net Neutrality stuff.
I doubt they would actually do it, though. A large advantage Google has over the competition is that they are at least perceived as a commons -- anyone can buy Google adspace, and it has nothing to do with their relationship with Google and everything to do with statistical analysis -- PageRank.
My original argument was that we should be spending more time fixing Wine and less time doing VMs, because Wine is inherently a better solution when it works.
If it's a server, I usually have a lot more choice, so moving config files is pretty much everything. Library hell does not exist when you use a good package manager, and mine (Gentoo's Portage) has a config file which says which packages I want installed. Thus, all I have to do is copy config files to the new box, run "emerge world", and all the software is installed and configured. So I really am limited to doing things specific to the hardware -- for instance, I moved from an x86 box to an amd64 one, so I did have to tweak some things -- kernel build scripts now need to know I want x86_64, for one -- but a huge amount of things (I'll refrain from pulling a number out of my ass, since you take them too literally) just worked.
Good point about modeling networks, though. I wish I had a nice, simple way of doing that. As it is, I'm more careful and I set up logging rules first, so I know how traffic is going to hit that particular rule before I implement it. Although a single OS on a single machine does work for having multiple IP addresses, I think most packet filters place a bit too much implicit trust on locally-originating traffic for that to be useful.
I don't think I said heterogeneous. What I don't know is if I can resume on a system with identical hardware, except inescapable differences like Mac address.
And I know I've seen XP hibernate in less time than it could possibly take to flush the entire system. 2000 takes a constant amount of time, based on how much RAM you have, XP seems to depend on how much you're actually using.
Well, the good news is, if you try to take all of that into account, and make a very conservative estimate to begin with, and you follow Doug's Law, then you end up with something that is pretty damn accurate, or at least close enough that your employers/beancounters/customers won't doubt it. Then, if you really can get it done in a fraction of the time, it's that much better for you.
Oh, I'm going to add my own amendment to Doug's Law: Doug's Law only covers unknowns. Anything you know -- time cost by miscommunication, time spent polishing the product -- even if you don't know how much time it will take, if you know what to call it, you must add something to your pre-Doug estimate.
The point is, you're going to have to give them some number, and it can't be too insane or you'll never get the go ahead. "When it's done" won't cut it. But it's far, far better to be ludicrously early than slightly late.