Shuttleworth sure seemed to be talking about breaking applications, not device drivers.
And count yourself lucky that you got an update. Every year when I was working as a network administrator at ABB we would have to throw out innumerable peripherals and cards that were not and never would be supported on new systems, software, and firmware... for everything from microcontrollers to mainframes.
You can't run 16 year old Windows GUI software on modern versions of Windows.
That's not a reason to agree with Mark's suggestion about breaking "old" apps on Ubuntu being OK, unless he's talking about comparably complex apps that are similarly dated. Given the state of the art of Linux in 1992 (this was the year of the Torvalds-Tannenbaum debate triggered by the increasing level of Linux discussin in comp.os.minix) I'm not sure you'll be able to find any.:)
1. Make the application rather than the HTML control solely responsible for the opening of windows, execution of plugins, following links, and so on. This does not necessarily mean a callback to the application, but the API needs to be such that without the application explicitly loading and registering handlers the HTML control itself would be unable to do anything but display the contents of the file it was passed.
2. Having done this, make IE simply another application, remove the links between IE and the desktop, stop using IE as the shell for things like Windows Update.
3. Remove the ability to run ActiveX components directly from IE, with or without "UAC" or similar approval dialogs.
4. Applications based on the HTML control that grant greater rights (such as Windows Update) would still be thin shells around the HTML control, but there would be no mechanism for other applications to be tricked into running any components they had not explicitly registered.
Once this kind of approach is taken for all components and helper applications, one that makes rendering HTML by default a "hard" sandbox rather than the "soft" one it currently is, has been completed... the vast majority of malware attacks on IE would no longer be possible, because once you eliminate security zones "cross zone" attacks are moot.
I can run MS-DOS software from the '80s on Windows XP SP2. I can run some Windows software from 1992, and I can run just about any well behaved application (which rules out things like shell extensions) from 10-15 years ago.
So we know the window of compatibility is less than 16 years.
What's the window of compatibility for binary executables on Linux? Even if they only depend on glibc, and don't pull in any GUI libraries, is it as long as 10 years? When was the last time they broke glibc? If you want to run a 10 year old GUI binary on a recent Linux, would you even know where to find all the back-rev lib*.so files it needs?
For FreeBSD installing compat3x should take you back to 1998, but I don't know if compat3x (let alone compat22) is still usable on FreeBSD 7.
I don't even think the "window of compatibility" for Mac OS is as long as 15 years.
16 years of binary compatibility is pretty damn good, for a desktop OS. Servers, now, you can probably still run VMS 2.0 binaries from 1980, but that's a whole different world.
Shuttleworth says: "And you can't run an old Windows application on a recent Windows version."
There are some applications, particularly ones that are pushing the limits of what you can do on a PC, that can't run on the most recent versions of Windows, but in general that's not true. I've got programs that I've carried around for decades that still work as far as I've been willing to take Windows.
Mind you, Vista might be an exception, but Microsoft has... up to Vista... bent over backwards to ludicrous levels to maintain backwards compatibility. The phrase "the exception that proves the rule" is a cliche, but this is a perfect example of an exception that DOES prove the rule... there's an enormous push-back against Vista simply because it's perceived as being incompatible. It's NOT a model to follow.
Ray Bostock: Governor Icahn, I should have expected to find you holding Balmer's leash. I recognized your foul stench when I was brought onboard. Carl Icahn: Charming to the last. You don't know how hard I found it, signing the order to terminate your employment. Ray Bostock: I'm surprised you had the courage to take the responsibility yourself. Carl Icahn: Chairman Bostock. Before your termination, you will join me at a ceremony that will make this search engine operational. No users will dare abandon Microsoft now. Ray Bostock: The more you tighten your grip, Icahn, the more users will slip through your fingers. Carl Icahn: Not after we demonstrate the power of this search engine. In a way, you have determined the choice of the website that is to be spidered first. Since you are reluctant to provide us with the parameters of the search algorithm, I have chosen to test this engine's destructive power on your home system of Yahoo. Ray Bostock: No! Yahoo is peaceful, we have no weapons. You can't possibly...
My first "LAN edge" router/firewall, back in the early '90s, was running BSDI, and since Free/Net/OpenBSD came out I've always used an old computer running one of the open source BSDs for this job. You don't need much horsepower for the job... for a long time at ABB I was using old Compaq Desqpro 20e boxes... 386/20s with 4MB RAM... and by the time FreeBSD CURRENT outgrew 4MB I had a stack of old DEC Multias with 12-16 MB sitting around. My current firewall is really overkill, a VIA Mini-ITX box, but it uses less power than the ex-desktop K6 it replaced.
As far as embedded systems and reliability goes, there are always lemons, but I have found that once you get a good network device (router, modem, what have you) if you can keep the ambient temperature reasonable they'll pretty much last forever. If you skimp on the air conditioning you'll pay for it in frequent replacements. Depending on the circumstances that might actually be the cheap option, mind you... spending $100/month extra on A/C to protect a $20 router is probably a false economy... especially if the time value of replacing the router is "oh boy, field trip to Frys!".
Now, should a *distribution* include non-free software?
Depends on the distribution. Debian, for example, is at one end of the spectrum... they won't include anything that isn't rigorously open source. But there's no reason that ALL distributions must be held to the same standard... particularly ones that are intended to operate without live access to the internet.
It seems to me that a default assumption has grown up that purports to demonstrate that protecting the rights of content creators is somehow immoral.
The rights of copyright owners are to enforce a limited monopoly for a limited period of time for the purpose of advancing the art. One of the ways that the art is advanced is by the works of amateurs. Many, in fact I would suspect most, actors, producers and directors, musicians, and other creators of copyrighted works started out as amateurs.
The creators of user-generated content are also copyright owners, and have the same rights. If some copyright owners are using the law to shut down the sites that other copyright-owners use, then it is no longer a matter of the law being "for" or "against" copyright owners, it's a matter of copyright owners against copyright owners. It is, in fact, a matter of established copyright owners against the next generation of creative artists.
"disclaimer: only tested on OS X, not tested on animals, no DLLs were harmed in the making of this message, age 45 and older excluded, milage may vary depending on driving conditions, poster assumes slashdot readers are capable of understanding the concept of 'context', emoticons are optional and may require a surcharge."
Just about every camera I've looked at recently has a bog standard mini-USB connector and didn't require custom software to see the camera's contents as a drive, at least not with OSX.
how come they haven't been able to duplicate the Apple GUI experience?
Because they're trying to duplicate the Windows GUI experience, complete with periodically pissing off half the user base by changing the entire interface for oddball reasons.
They say "the desktop hasn't had a radical redesign in X years!" So what? The command line hasn't had a radical redesign since the Bourne shell, unless you're using Plan 9, and that was about 30 years ago. You don't *need* a radical redesign of things that work well. You don't *need* to break applications and force people to upgrade to a new API, either. Yes I'm looking at YOU, Trolltech... what's the point of using an OO programming language if you don't take advantage of the fact that you can have multiple methods with the same name, so you don't HAVE to remove the old calls when you change the calling sequence?
That's like when Microsoft declared "all new code will be in.NET" and had people hanging on to Visual Studio 6 for years because that was the only way to stay backwards compatible.
(and, no, I don't think Apple's going to get everyone to dump Carbon either)
Yes, you occasionally have to break stuff, but unless you're doing it because of security problems you do it after a transition period, and I don't think (for one random example) "directory.exists(name, TRUE)" counts as a security hole.
Or is there more to it than that, such as difference philosophies or lack of people with good a understanding of user psychology and graphic design principles?
All of the above. Not that Apple's user interface is perfect (god knows it isn't), but it's proof that you don't have to blindly clone everything Microsoft does to produce a great user experience.
The rule of law works because there are checks and balances, meta-laws that laws are subject to, judges and juries to interpret the laws, an appeals process. When these are short-circuited, the law becomes ineffective.
If the rules of a body do not include equivalent mechanisms, if they *have to* be blindly followed, as if every one of these rules was the equivalent of the mandatory sentencing rules that tie judges hands and prevent them from applying the judgement that oils the wheels of justice, then the rules are broken.
For example, one of the decisions made by the ISO was to permit this standard to have a fast track process. The resulting standard is clearly not ready. Therefore the original decision to allow a fast track process was the wrong decision. This has nothing to do with how many people are unhappy with the decision, it's proven to be the wrong decision by the fact that they do not have a usable standard at the end of the process.
One possibility is that there is no rule that says a fast tracked standard has to be essentially ready for publication.
One possibility is that there is such a rule, and it was ignored.
In the latter case, the rules were not followed.
In the former case, the rules are meaningless. Whether they follow them or not doesn't matter, and the fact that the ISO has approved a standard is no more than informational.
Murray Gell-Mann named the 'quark' after a line in James Joyce's novel Finnegan's Wake because he liked the sound of the word. The quarks themselves come in six 'flavors': up, down, charm, strange, top, and bottom. Only the up and down quarks are stable, which is why it's taken 30 years to create [eta]b Bottomonium.
Meanwhile, astronomers worry about whether Pluto is a planet or not.
Slashdot Mirror
Shuttleworth sure seemed to be talking about breaking applications, not device drivers.
And count yourself lucky that you got an update. Every year when I was working as a network administrator at ABB we would have to throw out innumerable peripherals and cards that were not and never would be supported on new systems, software, and firmware... for everything from microcontrollers to mainframes.
Of course it's a joke. Go ahead, google for 'slood' and 'verbogeny'.
Hell, I didn't mention Linux, you did.
Actually, Mark Shuttleworth did. :)
Take it as a data point.
You can't run 16 year old Windows GUI software on modern versions of Windows.
That's not a reason to agree with Mark's suggestion about breaking "old" apps on Ubuntu being OK, unless he's talking about comparably complex apps that are similarly dated. Given the state of the art of Linux in 1992 (this was the year of the Torvalds-Tannenbaum debate triggered by the increasing level of Linux discussin in comp.os.minix) I'm not sure you'll be able to find any. :)
Praise Loki for small miracles!
1. Make the application rather than the HTML control solely responsible for the opening of windows, execution of plugins, following links, and so on. This does not necessarily mean a callback to the application, but the API needs to be such that without the application explicitly loading and registering handlers the HTML control itself would be unable to do anything but display the contents of the file it was passed.
2. Having done this, make IE simply another application, remove the links between IE and the desktop, stop using IE as the shell for things like Windows Update.
3. Remove the ability to run ActiveX components directly from IE, with or without "UAC" or similar approval dialogs.
4. Applications based on the HTML control that grant greater rights (such as Windows Update) would still be thin shells around the HTML control, but there would be no mechanism for other applications to be tricked into running any components they had not explicitly registered.
Once this kind of approach is taken for all components and helper applications, one that makes rendering HTML by default a "hard" sandbox rather than the "soft" one it currently is, has been completed... the vast majority of malware attacks on IE would no longer be possible, because once you eliminate security zones "cross zone" attacks are moot.
Somehow *I* haven't been convinced about the great MS backwards compatibility.
I'm sorry, but I'm not arguing that Windows binary compatibility is "great".
Rather, I'm pointing out that no matter how bad it is, everyone else is a good deal shonkier.
With the exception of laptops, are there any graphics cards available that won't make my room an inferno when I'm gaming?/I.
GeForce 2 was pretty good, or maybe a Radeon 8500?
How far back do you want to go?
I can run MS-DOS software from the '80s on Windows XP SP2. I can run some Windows software from 1992, and I can run just about any well behaved application (which rules out things like shell extensions) from 10-15 years ago.
So we know the window of compatibility is less than 16 years.
What's the window of compatibility for binary executables on Linux? Even if they only depend on glibc, and don't pull in any GUI libraries, is it as long as 10 years? When was the last time they broke glibc? If you want to run a 10 year old GUI binary on a recent Linux, would you even know where to find all the back-rev lib*.so files it needs?
For FreeBSD installing compat3x should take you back to 1998, but I don't know if compat3x (let alone compat22) is still usable on FreeBSD 7.
I don't even think the "window of compatibility" for Mac OS is as long as 15 years.
16 years of binary compatibility is pretty damn good, for a desktop OS. Servers, now, you can probably still run VMS 2.0 binaries from 1980, but that's a whole different world.
The code is there to look at and study in order to program something that will bypass or disable it.
Anyone who is going to bother would simply reformat the drive anyway.
Google says 'Results 1 - 10 of about 21,000 for slood', human.
That's more hits than 'verbogeny' gets!
Why don't you sit down in an unused cubicle and use THEIR desktop.
All your important stuff is on the network, isn't it?
Isn't it?
Shuttleworth says: "And you can't run an old Windows application on a recent Windows version."
There are some applications, particularly ones that are pushing the limits of what you can do on a PC, that can't run on the most recent versions of Windows, but in general that's not true. I've got programs that I've carried around for decades that still work as far as I've been willing to take Windows.
Mind you, Vista might be an exception, but Microsoft has... up to Vista... bent over backwards to ludicrous levels to maintain backwards compatibility. The phrase "the exception that proves the rule" is a cliche, but this is a perfect example of an exception that DOES prove the rule... there's an enormous push-back against Vista simply because it's perceived as being incompatible. It's NOT a model to follow.
Like humans have no word for bissonomy, tubso, or slood?
Ray Bostock: Governor Icahn, I should have expected to find you holding Balmer's leash. I recognized your foul stench when I was brought onboard.
Carl Icahn: Charming to the last. You don't know how hard I found it, signing the order to terminate your employment.
Ray Bostock: I'm surprised you had the courage to take the responsibility yourself.
Carl Icahn: Chairman Bostock. Before your termination, you will join me at a ceremony that will make this search engine operational. No users will dare abandon Microsoft now.
Ray Bostock: The more you tighten your grip, Icahn, the more users will slip through your fingers.
Carl Icahn: Not after we demonstrate the power of this search engine. In a way, you have determined the choice of the website that is to be spidered first. Since you are reluctant to provide us with the parameters of the search algorithm, I have chosen to test this engine's destructive power on your home system of Yahoo.
Ray Bostock: No! Yahoo is peaceful, we have no weapons. You can't possibly...
My first "LAN edge" router/firewall, back in the early '90s, was running BSDI, and since Free/Net/OpenBSD came out I've always used an old computer running one of the open source BSDs for this job. You don't need much horsepower for the job... for a long time at ABB I was using old Compaq Desqpro 20e boxes ... 386/20s with 4MB RAM ... and by the time FreeBSD CURRENT outgrew 4MB I had a stack of old DEC Multias with 12-16 MB sitting around. My current firewall is really overkill, a VIA Mini-ITX box, but it uses less power than the ex-desktop K6 it replaced.
As far as embedded systems and reliability goes, there are always lemons, but I have found that once you get a good network device (router, modem, what have you) if you can keep the ambient temperature reasonable they'll pretty much last forever. If you skimp on the air conditioning you'll pay for it in frequent replacements. Depending on the circumstances that might actually be the cheap option, mind you... spending $100/month extra on A/C to protect a $20 router is probably a false economy... especially if the time value of replacing the router is "oh boy, field trip to Frys!".
Now, should a *distribution* include non-free software?
Depends on the distribution. Debian, for example, is at one end of the spectrum... they won't include anything that isn't rigorously open source. But there's no reason that ALL distributions must be held to the same standard... particularly ones that are intended to operate without live access to the internet.
It seems to me that a default assumption has grown up that purports to demonstrate that protecting the rights of content creators is somehow immoral.
The rights of copyright owners are to enforce a limited monopoly for a limited period of time for the purpose of advancing the art. One of the ways that the art is advanced is by the works of amateurs. Many, in fact I would suspect most, actors, producers and directors, musicians, and other creators of copyrighted works started out as amateurs.
The creators of user-generated content are also copyright owners, and have the same rights. If some copyright owners are using the law to shut down the sites that other copyright-owners use, then it is no longer a matter of the law being "for" or "against" copyright owners, it's a matter of copyright owners against copyright owners. It is, in fact, a matter of established copyright owners against the next generation of creative artists.
That's only a problem if you don't heat the queso before you dip the chips in it.
"disclaimer: only tested on OS X, not tested on animals, no DLLs were harmed in the making of this message, age 45 and older excluded, milage may vary depending on driving conditions, poster assumes slashdot readers are capable of understanding the concept of 'context', emoticons are optional and may require a surcharge."
Maybe I have more money than you and can look at more cameras?
Probably. I'm just a code monkey these days.
Just about every camera I've looked at recently has a bog standard mini-USB connector and didn't require custom software to see the camera's contents as a drive, at least not with OSX.
how come they haven't been able to duplicate the Apple GUI experience?
Because they're trying to duplicate the Windows GUI experience, complete with periodically pissing off half the user base by changing the entire interface for oddball reasons.
They say "the desktop hasn't had a radical redesign in X years!" So what? The command line hasn't had a radical redesign since the Bourne shell, unless you're using Plan 9, and that was about 30 years ago. You don't *need* a radical redesign of things that work well. You don't *need* to break applications and force people to upgrade to a new API, either. Yes I'm looking at YOU, Trolltech... what's the point of using an OO programming language if you don't take advantage of the fact that you can have multiple methods with the same name, so you don't HAVE to remove the old calls when you change the calling sequence?
That's like when Microsoft declared "all new code will be in .NET" and had people hanging on to Visual Studio 6 for years because that was the only way to stay backwards compatible.
(and, no, I don't think Apple's going to get everyone to dump Carbon either)
Yes, you occasionally have to break stuff, but unless you're doing it because of security problems you do it after a transition period, and I don't think (for one random example) "directory.exists(name, TRUE)" counts as a security hole.
Or is there more to it than that, such as difference philosophies or lack of people with good a understanding of user psychology and graphic design principles?
All of the above. Not that Apple's user interface is perfect (god knows it isn't), but it's proof that you don't have to blindly clone everything Microsoft does to produce a great user experience.
The rule of law works because there are checks and balances, meta-laws that laws are subject to, judges and juries to interpret the laws, an appeals process. When these are short-circuited, the law becomes ineffective.
If the rules of a body do not include equivalent mechanisms, if they *have to* be blindly followed, as if every one of these rules was the equivalent of the mandatory sentencing rules that tie judges hands and prevent them from applying the judgement that oils the wheels of justice, then the rules are broken.
For example, one of the decisions made by the ISO was to permit this standard to have a fast track process. The resulting standard is clearly not ready. Therefore the original decision to allow a fast track process was the wrong decision. This has nothing to do with how many people are unhappy with the decision, it's proven to be the wrong decision by the fact that they do not have a usable standard at the end of the process.
One possibility is that there is no rule that says a fast tracked standard has to be essentially ready for publication.
One possibility is that there is such a rule, and it was ignored.
In the latter case, the rules were not followed.
In the former case, the rules are meaningless. Whether they follow them or not doesn't matter, and the fact that the ISO has approved a standard is no more than informational.
Post it on the Internet and someone will steal it.
You just noticed this?
Murray Gell-Mann named the 'quark' after a line in James Joyce's novel Finnegan's Wake because he liked the sound of the word. The quarks themselves come in six 'flavors': up, down, charm, strange, top, and bottom. Only the up and down quarks are stable, which is why it's taken 30 years to create [eta]b Bottomonium.
Meanwhile, astronomers worry about whether Pluto is a planet or not.
Hail Eris!