I recall nothing about lost licenses. That is actually the one thing that is almost impossible in the US.
Huh. Well, here in the Midwest, you just lose your license for going 25+ over. I know because I almost triggered the revocation when I was younger and stupider. (Actually, I learned about it later in a voluntary points reduction class, but I digress.) They won't arrest you, at least in the midwest, simply for speeding. Of course, if you're going *extremely* fast they may get you for reckless driving. Which, again, is not the same thing as speeding.
So you're saying, you can't go to jail for speeding, unless you're speeding really fast.
Read what I wrote again. And again. And again. Until you get the part about "you're not going to jail". Unless for some reason you think that having your license revoked is the same as going to jail?
That's the reason why if I'm driving too fast in Europe I get a ticket, and if I'm in USA I go to jail.
Why do you people even troll this crap? In fact, where do you even get such an idea? A bad Hollywood movie?
As far as I know, there is absolutely NO way to go to jail simply from speeding. (Reckless or druken driving is a whole other matter.) In every state I've ever been in, you just get a ticket. The only exception to that is if you're going exceptionally high speeds - usually defined as 25+ miles per hour over the speed limit - in which case your license may be automatically revoked.
If someone says they saw a car like yours driving erractically and that caused an accident they can come to your house, dump the system and tell if it was you.
Except for the fact that they need a warrant before they can do that. Illegal Search and Seizure *does* still mean something, despite the Patriot Act and other nonsense. (Although I have a feeling that the Patriot Act would be repealed in a hurry if it was abused by law enforcement.)
How long until my car decides I am driving to fast, and calls the police to mail me a ticket. Before you write this off as never_gonna_happen, consider that many highways now have radar guns attached to cameras, and they mail out tickets in the mail.
It's important to note that this sort of thing is far more ubiquitous in Europe than it is in the US. The US laws about entrapment, personal property, and privacy tend to preclude these sorts of measures.
Absolutely right, the pattern has been derived from the OS design.
Which is, again, completely screwed up. JBoss, NetKernel, Geronimo, etc., all have zip to do with microkernels. They don't manage anything other than loading shared, executable code. Something which, if I may point out, is precisely what a Microkernel *doesn't* do. That's what the servers are for!
Not to mention that the concept *still* fails to commmunicate what precisely the software *does*. When we talk about OS Microkernels, we know that we're talking about the interface between the hardware and software in a multiuser system. When we're talking about a Microkernel in the context of App Server software, it is completely unclear what is meant. At best, we can assume that a central control launches "services". The meaning of this is quite ambiguous and highly confusing. Now if we said "JBoss is an application server based on a JMX Framework where everything in the system is a JMX object," then we know what exactly JBoss is.
It's an abstraction, it looks at what is the essence behind the principle of the OS microkernel and transfers it to other areas.
No, it's not. A Microkernel is a very specific concept, with very little abstract about it. Redefining it to be an abstraction is a problem.
And to make it clear: using this term in that general way is in no way an invention by the poster nor by the JBoss group.
To be clear, I'm merely annoyed at the desire of marketeers to reuse terms like this. It's nothing against JBoss in specific, just the corruption of a perfectly workable dialect.
No, definetely not. "Framework" is far too general and unspecific.
Framework is a starting point. Thanks to the power of the English language, we can be more specific without inventing a new term. "J2EE Framework", "JMX Framework", and "Plugable Server Framework" are all perfectly acceptable, and far more informative than "Microkernel."
Actually, I was trying to be funny myself. I should have put a smiley after that, sorry.:-)
Blogs, however, lend themselves easily to the Dvorak self-righteous, know-it-all style of discourse. Well, it's not discourse so much as soliloquy.
The one difference in blogs (which I'd like to see more of) is that bloggers can argue a point back and forth. While one can do this on a forum, you can only do it for as long as the thread stays active. Blogging allows you a much more open forum, as well as more time to develop a well rounded argument.:-)
The term Microkernel is in no way reduced the the usage in an OS kernel.
"Kernel" in CompSci refers specifically to the low level, hardware/resource control layer of an Operating System. The definition is very clear and distinct. Using the term in places other than OS design is obviously an attempt to liken the new design to a kernel's design. Yet that doesn't make sense, since the two concepts are very different.
The result is that the term gets overloaded and the definition starts to become fuzzy. Which then means that a greater context is required to fully explain the subject.
For example, if I say "Having extensively used DiVX, I must say that I dislike it intensely," it is unclear if I'm referring to DiVX the codec, or DiVX the DVD alternative. And having this statement made in the context of a Video Encoding forum does not necessarily help. Discussing the relative merits of the DVD format vs. the DiVX format is well within the bounds of normal discussion for such a forum. Which means that people often must state, ""Having extensively used DiVX (the disc format, I love the codec), I must say that I dislike it intensely."
This breaks up the flow of the idea, and makes communication more difficult. Why would we want to *intentionally* muck up our communications like that? Especially when the term "Framework" fits perfectly well, and is intended for this precise situation.
As I said, the overloading of the term is confusing, highly annoying, and destructive to communication.
You should try Orion sometime. Everytime I'm forced to work with JBoss, I can't help but wish I was using Orion.;-)
Why is Orion cool? Because it combines simplicity in design with all the features you've come to expect in high end commercial app servers. For example, Orion was the first server that I'm aware of that had *working* hot-deploy for EARs, WARs, and EJB JARs. (JBoss supposedly had hot-deploy for EJB in earlier versions, but it never seemed to work *quite* right.)
No other app server has managed to stay as focused on keeping the essentials working well instead of adding ancillary features out the wazoo.:-)
Sinking, nothing. I'm pointing out an oddity in the review, which *seems* to claim that JBoss is somehow superior for its use of JMX. Since JBoss has nothing to do with the development of the technology, nor does it have a monopoly on the use of the technology, I find that statement odd at best, disingenuous at worst. However, I think you answered the argument after this:
What does set them apart is that they base their architecture on JMX where as others simply expose JMX objects, usually for mundane tasks simply to get the checkbox on feature set. JBoss on the other hand, uses JMX to bootstrap the entier application server, almost every component within the server is configured with a JMX MBean meaning you can use the JMX server infrastructure to easily interrogate and alter the configuration at runtime if desired. This is what sets them apart.
Or in other words, the quality of the JMX implementation sets JBoss apart. Which I can accept quite readily. Thank you for clarifying.:-)
My reading of the post is that the poster is implying either responsibility for or a current monopoly on the technology. Nothing could be further from the truth.
However, I am willing to concede that others (such as yourself) have not read it this way. Which only leaves the point about being "set apart" based on this technology. I find it hard to accept that JBoss is "set apart" by a technology that is in many competing J2EE servers. In which case, I can only assume that the poster either did not understand this, or intended to say that "the quality of the JMX implementation sets JBoss apart."
JBoss doesn't claim to have invented JMX. They merely utilize it in their implementation.
The poster implies that JMX is a JBoss technology, or at least that JBoss has a current monopoly on JMX.
Still, I'm willing to concede the point since no one else seems to be bothered by this. The rest of my point still stands from an informative perspective. JMX is a standard technology used by a variety of J2EE servers.
"Microkernel" refers to the architectural pattern around which JBoss was designed.
Irrelevant. My point was that the use of the term "Microkernel" is confusing, and that its ongoing use to describe shared code environments only serves to heavily overload the term. Don't we have enough confusing terminology in the technology sector without reassigning clear-cut terms?
My quotation states that "JMX sets JBoss apart". How can JBoss be set apart by a technology used by many (if not all) the major vendors? Perhaps the author meant "JMX sets JBoss apart from the other low cost/free/open source J2EE servers?"
Besides the usual advantages of open source and GPL licensing, what sets it apart is its JMX based microkernel, a light-weight framework to run independently developed Java programs within a single JVM.
FYI, JMX is not a JBoss technology, but rather a Sun JSR Specification. Perhaps the most telling point is that JBoss's name doesn't appear anywhere in the working group for the JSR. Claiming or attributing responsibility for such technology is a bit disingenuous. Especially since several other app servers (e.g. WebSphere and Sun J2EE) use the same technology.
Also, am I the only one who's annoyed at the use of the word "microkernel"? While I'm sure that some similarities exist, a J2EE server is not an operating system. It's a shared environment high above the sysetm management level, and as such cannot be classified in the same manner. Using Operating System terms at that level only serves to confuse potential customers about the purpose of the technology.
My pet peeve in this area is the 1060 NetKernel. They get so wrapped up in the "kernel" language, that they forget to tell everyone exactly what their product does. I mean, look at this stuff:
NetKernelTM Microkernel is a REST microkernel. It provides a clean, robust and scalable foundation for the development and delivery of robust, scalable and adaptable systems.
Without the ability to verify the credibility of a source (including editorials, and the people who write them) the information is still worthless.
Ah, I see your point. However, what makes blogs any different than the articles published in something like JDJ? The only credentials these people give are the blurbs that they supply to the magazine. For all you know, they could be making up their PHD or 50 books published. And even if they aren't you have no method of validating the quality of their previous works.
Even more difficult is that the general populace often ignores those who actually have credentials and place their faith in someone who doesn't. Case in point, when Nader declared Plutonium "the most toxic substance in existance." He has no credibility behind those statements, but he is believed over the scientists and engineers who *do* have credentials.
So it is a difficult situation. As a result, bloggers tend to rely on something more important than credentials: Reputation.
Reputation can be built a variety of ways. Not the least of which is by consistently sharing a valid and intelligent opinion.:-)
Having to develop two parallel code branches for years to come?
As long as developers haven't relied on something inherent in the chip (e.g. endianess), there should be no need to maintain separate code branches. All that will happen is that the compiler will produce multi-platform or "fat" binaries. This shouldn't impact most existing code in any way, shape, or form.
Without the ability to identify credible sources (yes, I know it is sometimes possible with blogs) a piece of information is essentially worthless.
As you said, sometimes it is possible to cite credible sources. In the blog that I write (which I actually consider to be a collection of articles), I always include a "Links" area at the bottom of the article. This allows the reader to get up to speed and verify the technologies I have discussed. It also allows me to point to prior art when a new idea is being discussed.
However, such references are not always required. The purpose of an editorial is to provide an intelligent opinion. As a result, there are no references needed except to validate the subject of the topic. Quite often, it is unnecessary for the writer to provide links to the subject since it is considered a subject of which the readers should already be familiar. Thus editorials can be perfectly valid despite the lack of references.
Seriously, I'd like to see more intelligent discourse on the 'net. Far too much of it has been "look at me!", even in the early days. (Anyone remember personal home pages?) Blogging software merely provides an outlet through which public communication can occur. It can be used intelligently, or it can be used frivolously. It's up to the "blogger". In that respect, nothing has changed.
As I said, it isn't always true that a blogger isn't serving other masters. However, I have a hard time believing that Google is one of them. Google doesn't demand any sort of articles out of a blogger. It's completely up to the blogger what he wants to write about. The most that Google might do to a technologist is cause him to "optimize" his language for adwords.
OTOH, someone might look at the "Top 100 Keyword List" and decide to create a blog on that. If he has something interesting to say about viagra or bankruptcy, then he may very well make money on it. Otherwise this is hardly a workable business plan.
At the end of the day though, you have to judge the article on content. A blogger *may* be better because he's not serving other masters, or he may not. If you judge on the content itself, it really doesn't matter in the end, does it?
Nonsense. When it comes to editorials, there's little difference between an "official" shill/idiot like Dvorak, and a valid opinion from a blogger. In fact, the blogger may tend to make more sense because he's not serving other masters. (Note: This isn't ALWAYS true.) In other words, the editorial should be judged on its content, not necessarily who wrote it.
The problem is that blogging has a stigma attached to it. It's seen as a method through which people can voluntarily make their private lives public, making it the oddest form of online voyuerism available. However, the concept of publishing articles on a regular basis is not new, and there is no real division made between "blogging" and "writing regular articles". Thus it can be difficult to tell if a "blog" is actually something that can be safely ignored as a poor form of entertainment, or a reasonable attempt at serious writing.
Jewel (a grocery store primarily in the midwest) currently has a scanner system that works like that. When you come in the door you swipe your Jewel card at the scanner rack. In a "Blake's 5"ish way, the rack then allows you to remove a scanner. The shopping bags are also located right there so you can take a few with you.
Then as you shop, you simply scan the barcode and place the item in a bag. When you're done shopping, you go to a special checkout lane where you scan an "I'm Done" barcode. The system then tallies your order (and allows you to add non-barcoded items as necessary) and takes your money. A receipt prints and you walk out of the store.
It's not perfect, but it's much faster than existing methods.
Slashdot Mirror
I recall nothing about lost licenses. That is actually the one thing that is almost impossible in the US.
Huh. Well, here in the Midwest, you just lose your license for going 25+ over. I know because I almost triggered the revocation when I was younger and stupider. (Actually, I learned about it later in a voluntary points reduction class, but I digress.) They won't arrest you, at least in the midwest, simply for speeding. Of course, if you're going *extremely* fast they may get you for reckless driving. Which, again, is not the same thing as speeding.
Then you're not getting arrested for speeding then, are you? :-)
So you're saying, you can't go to jail for speeding, unless you're speeding really fast.
Read what I wrote again. And again. And again. Until you get the part about "you're not going to jail". Unless for some reason you think that having your license revoked is the same as going to jail?
That's the reason why if I'm driving too fast in Europe I get a ticket, and if I'm in USA I go to jail.
Why do you people even troll this crap? In fact, where do you even get such an idea? A bad Hollywood movie?
As far as I know, there is absolutely NO way to go to jail simply from speeding. (Reckless or druken driving is a whole other matter.) In every state I've ever been in, you just get a ticket. The only exception to that is if you're going exceptionally high speeds - usually defined as 25+ miles per hour over the speed limit - in which case your license may be automatically revoked.
If someone says they saw a car like yours driving erractically and that caused an accident they can come to your house, dump the system and tell if it was you.
Except for the fact that they need a warrant before they can do that. Illegal Search and Seizure *does* still mean something, despite the Patriot Act and other nonsense. (Although I have a feeling that the Patriot Act would be repealed in a hurry if it was abused by law enforcement.)
How long until my car decides I am driving to fast, and calls the police to mail me a ticket. Before you write this off as never_gonna_happen, consider that many highways now have radar guns attached to cameras, and they mail out tickets in the mail.
It's important to note that this sort of thing is far more ubiquitous in Europe than it is in the US. The US laws about entrapment, personal property, and privacy tend to preclude these sorts of measures.
Absolutely right, the pattern has been derived from the OS design.
Which is, again, completely screwed up. JBoss, NetKernel, Geronimo, etc., all have zip to do with microkernels. They don't manage anything other than loading shared, executable code. Something which, if I may point out, is precisely what a Microkernel *doesn't* do. That's what the servers are for!
Not to mention that the concept *still* fails to commmunicate what precisely the software *does*. When we talk about OS Microkernels, we know that we're talking about the interface between the hardware and software in a multiuser system. When we're talking about a Microkernel in the context of App Server software, it is completely unclear what is meant. At best, we can assume that a central control launches "services". The meaning of this is quite ambiguous and highly confusing. Now if we said "JBoss is an application server based on a JMX Framework where everything in the system is a JMX object," then we know what exactly JBoss is.
It's an abstraction, it looks at what is the essence behind the principle of the OS microkernel and transfers it to other areas.
No, it's not. A Microkernel is a very specific concept, with very little abstract about it. Redefining it to be an abstraction is a problem.
And to make it clear: using this term in that general way is in no way an invention by the poster nor by the JBoss group.
To be clear, I'm merely annoyed at the desire of marketeers to reuse terms like this. It's nothing against JBoss in specific, just the corruption of a perfectly workable dialect.
No, definetely not. "Framework" is far too general and unspecific.
Framework is a starting point. Thanks to the power of the English language, we can be more specific without inventing a new term. "J2EE Framework", "JMX Framework", and "Plugable Server Framework" are all perfectly acceptable, and far more informative than "Microkernel."
Close. Oracle licensed Orion for the basis of Oracle 9i. They didn't actually acquire them, however.
Actually, I was trying to be funny myself. I should have put a smiley after that, sorry. :-)
:-)
Blogs, however, lend themselves easily to the Dvorak self-righteous, know-it-all style of discourse. Well, it's not discourse so much as soliloquy.
The one difference in blogs (which I'd like to see more of) is that bloggers can argue a point back and forth. While one can do this on a forum, you can only do it for as long as the thread stays active. Blogging allows you a much more open forum, as well as more time to develop a well rounded argument.
No, it is quite irrelevant.
The term Microkernel is in no way reduced the the usage in an OS kernel.
"Kernel" in CompSci refers specifically to the low level, hardware/resource control layer of an Operating System. The definition is very clear and distinct. Using the term in places other than OS design is obviously an attempt to liken the new design to a kernel's design. Yet that doesn't make sense, since the two concepts are very different.
The result is that the term gets overloaded and the definition starts to become fuzzy. Which then means that a greater context is required to fully explain the subject.
For example, if I say "Having extensively used DiVX, I must say that I dislike it intensely," it is unclear if I'm referring to DiVX the codec, or DiVX the DVD alternative. And having this statement made in the context of a Video Encoding forum does not necessarily help. Discussing the relative merits of the DVD format vs. the DiVX format is well within the bounds of normal discussion for such a forum. Which means that people often must state, ""Having extensively used DiVX (the disc format, I love the codec), I must say that I dislike it intensely."
This breaks up the flow of the idea, and makes communication more difficult. Why would we want to *intentionally* muck up our communications like that? Especially when the term "Framework" fits perfectly well, and is intended for this precise situation.
As I said, the overloading of the term is confusing, highly annoying, and destructive to communication.
You should try Orion sometime. Everytime I'm forced to work with JBoss, I can't help but wish I was using Orion. ;-)
:-)
Why is Orion cool? Because it combines simplicity in design with all the features you've come to expect in high end commercial app servers. For example, Orion was the first server that I'm aware of that had *working* hot-deploy for EARs, WARs, and EJB JARs. (JBoss supposedly had hot-deploy for EJB in earlier versions, but it never seemed to work *quite* right.)
No other app server has managed to stay as focused on keeping the essentials working well instead of adding ancillary features out the wazoo.
You're sinking fast.
:-)
Sinking, nothing. I'm pointing out an oddity in the review, which *seems* to claim that JBoss is somehow superior for its use of JMX. Since JBoss has nothing to do with the development of the technology, nor does it have a monopoly on the use of the technology, I find that statement odd at best, disingenuous at worst. However, I think you answered the argument after this:
What does set them apart is that they base their architecture on JMX where as others simply expose JMX objects, usually for mundane tasks simply to get the checkbox on feature set. JBoss on the other hand, uses JMX to bootstrap the entier application server, almost every component within the server is configured with a JMX MBean meaning you can use the JMX server infrastructure to easily interrogate and alter the configuration at runtime if desired. This is what sets them apart.
Or in other words, the quality of the JMX implementation sets JBoss apart. Which I can accept quite readily. Thank you for clarifying.
My reading of the post is that the poster is implying either responsibility for or a current monopoly on the technology. Nothing could be further from the truth.
However, I am willing to concede that others (such as yourself) have not read it this way. Which only leaves the point about being "set apart" based on this technology. I find it hard to accept that JBoss is "set apart" by a technology that is in many competing J2EE servers. In which case, I can only assume that the poster either did not understand this, or intended to say that "the quality of the JMX implementation sets JBoss apart."
JBoss doesn't claim to have invented JMX. They merely utilize it in their implementation.
The poster implies that JMX is a JBoss technology, or at least that JBoss has a current monopoly on JMX.
Still, I'm willing to concede the point since no one else seems to be bothered by this. The rest of my point still stands from an informative perspective. JMX is a standard technology used by a variety of J2EE servers.
"Microkernel" refers to the architectural pattern around which JBoss was designed.
Irrelevant. My point was that the use of the term "Microkernel" is confusing, and that its ongoing use to describe shared code environments only serves to heavily overload the term. Don't we have enough confusing terminology in the technology sector without reassigning clear-cut terms?
My quotation states that "JMX sets JBoss apart". How can JBoss be set apart by a technology used by many (if not all) the major vendors? Perhaps the author meant "JMX sets JBoss apart from the other low cost/free/open source J2EE servers?"
Oh, and besides their Market Speak(TM) being confusing and redundant, their Market Speak(TM) is confusing and redundant.
FYI, JMX is not a JBoss technology, but rather a Sun JSR Specification. Perhaps the most telling point is that JBoss's name doesn't appear anywhere in the working group for the JSR. Claiming or attributing responsibility for such technology is a bit disingenuous. Especially since several other app servers (e.g. WebSphere and Sun J2EE) use the same technology.
Also, am I the only one who's annoyed at the use of the word "microkernel"? While I'm sure that some similarities exist, a J2EE server is not an operating system. It's a shared environment high above the sysetm management level, and as such cannot be classified in the same manner. Using Operating System terms at that level only serves to confuse potential customers about the purpose of the technology.
My pet peeve in this area is the 1060 NetKernel. They get so wrapped up in the "kernel" language, that they forget to tell everyone exactly what their product does. I mean, look at this stuff:
I'm sure I'm not alone when I say, HUH?
Without the ability to verify the credibility of a source (including editorials, and the people who write them) the information is still worthless.
:-)
Ah, I see your point. However, what makes blogs any different than the articles published in something like JDJ? The only credentials these people give are the blurbs that they supply to the magazine. For all you know, they could be making up their PHD or 50 books published. And even if they aren't you have no method of validating the quality of their previous works.
Even more difficult is that the general populace often ignores those who actually have credentials and place their faith in someone who doesn't. Case in point, when Nader declared Plutonium "the most toxic substance in existance." He has no credibility behind those statements, but he is believed over the scientists and engineers who *do* have credentials.
So it is a difficult situation. As a result, bloggers tend to rely on something more important than credentials: Reputation.
Reputation can be built a variety of ways. Not the least of which is by consistently sharing a valid and intelligent opinion.
Having to develop two parallel code branches for years to come?
As long as developers haven't relied on something inherent in the chip (e.g. endianess), there should be no need to maintain separate code branches. All that will happen is that the compiler will produce multi-platform or "fat" binaries. This shouldn't impact most existing code in any way, shape, or form.
i find it funny you have a blog.
:-)
Did you read it? Did you find that it supported my point or detracted from it?
Its very existence says nothing unless measured against the point I was attempting to make.
Without the ability to identify credible sources (yes, I know it is sometimes possible with blogs) a piece of information is essentially worthless.
As you said, sometimes it is possible to cite credible sources. In the blog that I write (which I actually consider to be a collection of articles), I always include a "Links" area at the bottom of the article. This allows the reader to get up to speed and verify the technologies I have discussed. It also allows me to point to prior art when a new idea is being discussed.
However, such references are not always required. The purpose of an editorial is to provide an intelligent opinion. As a result, there are no references needed except to validate the subject of the topic. Quite often, it is unnecessary for the writer to provide links to the subject since it is considered a subject of which the readers should already be familiar. Thus editorials can be perfectly valid despite the lack of references.
Was that intended as a personal attack?
Seriously, I'd like to see more intelligent discourse on the 'net. Far too much of it has been "look at me!", even in the early days. (Anyone remember personal home pages?) Blogging software merely provides an outlet through which public communication can occur. It can be used intelligently, or it can be used frivolously. It's up to the "blogger". In that respect, nothing has changed.
As I said, it isn't always true that a blogger isn't serving other masters. However, I have a hard time believing that Google is one of them. Google doesn't demand any sort of articles out of a blogger. It's completely up to the blogger what he wants to write about. The most that Google might do to a technologist is cause him to "optimize" his language for adwords.
OTOH, someone might look at the "Top 100 Keyword List" and decide to create a blog on that. If he has something interesting to say about viagra or bankruptcy, then he may very well make money on it. Otherwise this is hardly a workable business plan.
At the end of the day though, you have to judge the article on content. A blogger *may* be better because he's not serving other masters, or he may not. If you judge on the content itself, it really doesn't matter in the end, does it?
Nonsense. When it comes to editorials, there's little difference between an "official" shill/idiot like Dvorak, and a valid opinion from a blogger. In fact, the blogger may tend to make more sense because he's not serving other masters. (Note: This isn't ALWAYS true.) In other words, the editorial should be judged on its content, not necessarily who wrote it.
The problem is that blogging has a stigma attached to it. It's seen as a method through which people can voluntarily make their private lives public, making it the oddest form of online voyuerism available. However, the concept of publishing articles on a regular basis is not new, and there is no real division made between "blogging" and "writing regular articles". Thus it can be difficult to tell if a "blog" is actually something that can be safely ignored as a poor form of entertainment, or a reasonable attempt at serious writing.
Jewel (a grocery store primarily in the midwest) currently has a scanner system that works like that. When you come in the door you swipe your Jewel card at the scanner rack. In a "Blake's 5"ish way, the rack then allows you to remove a scanner. The shopping bags are also located right there so you can take a few with you.
Then as you shop, you simply scan the barcode and place the item in a bag. When you're done shopping, you go to a special checkout lane where you scan an "I'm Done" barcode. The system then tallies your order (and allows you to add non-barcoded items as necessary) and takes your money. A receipt prints and you walk out of the store.
It's not perfect, but it's much faster than existing methods.