Read the story. B & N isn't challenging the patent, they are saying they don't infringe on it because their implementation requires more than one click.
"But your Honour, we have been very careful not to infringe by deliberately being less convenient."
B & N isn't looking to change anything about the patent system. They are playing by the "rules". If they can win, they will probably patent their version.
To put Napster out of business, even if not permanently, would do a serious disservice to the online community's reputation. While the RIAA claims that their clients are being harmed right now, and therefore Napster should be shut down, it is also true that Napster has a right to be providing it services, pending a court verdict. Either way the court decides, action must take place after the verdict, and after the judge has had time to make up his mind.
Napster isn't being put out of business. The injunction says that Napster cannot allow the distribution of copyrighted works with out permission. This doesn't actually change anything, legally, since Napster can't systematically allow it's service to be used for copyright infringment anyway. (There are many precidents stating this explicitly for Internet services. For Napster to run an Internet service and be ignorant of this is simply not believable.) The fact that Napster chooses to represent this as "putting them out of business" simply shows the lie in their assertion that their service is actually used, to any significant degree, for anything other than copyright infringment.
While there are some nice features about components, like anything else it's possible to have too much of a good thing.
. . . .
I believe this is called "DLL Hell" in Windows circles.
You are confusing traditional shared DLLs with components. In MS land, components use COM which supports proper versioning of components - either with multiple components implementing different versions or with a single component implementing backwards compatability.
COM is the foundation of COM+ (which is a marketing term, not a technology as COM is) and the whole platform will evolve into.NET which will allow MS to treat the OS as a comodity when their appeal is concluded and they loose Windows. By then, their platform will be.NET and it will be business as usual.
Meanwhile, Linux has no component model, no distributed transaction services, and no server component runtime. EJB/CORBA isn't nearly as comprehensive as COM+, and isn't real competition. MS is pulling an end run and no one has noticed, yet.
Miguel didn't talk about OOP, he talked about components. OOP does not mean components and components can be built with out OOP. In general OOP is a powerful way to build components, but Component Based Development is conceptually very different from OOP. The two can be complementary, but are not the same.
The fundimental shift from a finite number of TLD's to an infinite number is a major conceptual change. It's going to be a tough sell no matter how simple the technical issues are.
Technically, the current hierarchy is simple and clean. Abandoning that for almost anything else is going to offend a lot of people's sense of order.
The licensee of your software is the operator of the server, not the end user. The operator of the server can modify the software and, as long as they only run it on their own servers, are not re-distributing your software - even though their users are distributed.
Well, it doesn't if you read his bpost a little further. He proposes using the alphabet as a natural hierarchy.
That would require infrastructure changes because currently it's the dots that delimit zones of authority. Changing to using each character of the TLD still breaks the existing infrastructure and therefore isn't a solution.
I rather like that idea.
I like the idea, too. But I still don't think it will (or should be) adopted. DNS changes like this could break a huge piece of the internet really fast. People have tried to implement new TLD naming schemes before, but without entries in the current root servers, noone can find their servers. And the roots aren't about to break a working system.
HOWEVER, still require registrations to consist of domain name + TLD. i.e., you must still sumbit both parts to constitute a single registration application. The TLD itself cannot be registered to anyone. and remains open for anyone to use.
I like the idea, but it breaks the hierarchical nature of DNS. Each "." in a machine name delimits a "zone of authority". With out any cacheing, you have to ask a root server for a server that can answer.org queries. Then you ask that server for a server that can answer.slashdot.org queries. Lastly you ask that server for the address of www.slashdot.org. Normally, most of this data is cached in the lower levels of the hierarchy, giving use reasonable DNS performance as well as managability.
As good as this idea is, it won't be adopted any time soon because of the infrastructure changes needed to support an unlimited number of TLDs.
You appear to have take an "architectural" approach to designing TUX, so I have some architectural questions.
The choice of a kernel space implementation is probably going to be a controversial one. You suggest that HTTP is commonly used enough to go in the kernel just as TCP/IP did years ago. What performance or architectural advantages do you see to moving application protocols into the kernel that cannot be achieved in user space?
What is your approach to concurence? In particular, you refer to "event driven". What do you mean by that and what did you choose as the core to your event engine? Also, how do you handle threading to scale on SMP machines?
Are there any plans to generalize the infrastructure elements of TUX so that other protocol servers can take advantage of the TUX architecture and also go fast as hell?
The benchmark seems to test concurrency. That is, how many clients can the server serve with out dropping any connections and still maintain decent through put overall.
Concurrency is the major consideration for total system performance as as opposed to individual connection performance. Given appropriate hardware, any crappy server can push 1Gb/sec on a connection. But what happens when there are 1000 connections? 5000? 10000? More? Serving tens of thousands of concurrent connections efficiently is the holy grail of network servers.
It's gotta be custom since you're going to be handling 10k idle threads at any one time.
A thread per connection model is probably not the best architecture for a system support so many users. See Dan Kegel's C10K site for more info on high performance servers supporting many concurrent connections.
I run Squid on a P75 with 32 MB RAM. The workstations are both PII 300's with 64 MB and 128 MB RAM respectively. Netscape is configured to use a RAM cache and no disk cache. Squid has a 100 MB disk cache.
The Squid cache may be slower over 10 Mbps than on a local disk but my Internet connection is a 56K modem, and even the Squid cache is faster than that. Even if you have cable modem or DSL at 1.5 Mbps, Squid will still be faster.
Squid can also be used as part of a firewall, can block Doubleclick.net and other annoying people, and gives you nice HTML page when there are errors. All in all, Squid makes a nice addition to any home network.
Mr. Konrad esentially claims patent on multi-tiered client/server architecture. Obviously, Microsoft has been delaying their Next Generation Windows Services anouncement because of delays in negotiating with Mr. Konrad, not because of the impending breakup of the company.
Seriously, the patent acknowledges so much prior art, including X Windows and Cathode Ray Tube Display, that it can't possibly be considered "inventive" or "non-obvious".
This is basicly the longest description of multi-tiered client/server architecture that I've ever read. The entire software industry is infringing on this one. And has been since the first dumb terminal was connected to a mainframe.
Get this: "The above description is illustrative and not restrictive. Many variations of the invention will become apparent to those of skill in the art upon review of this disclosure. The scope of the invention should, therefore, be determined not with reference to the above description, but instead should be determined with reference to the appended claims along with their full scope of equivalents." In short, the idea may be fifty years old, but I only want to claim future uses of it. Can I patent the business process of riding on the coat tails of real inovators and collecting royalties for doing nothing?
There is some irony here. This database is maintained by Human Resources and used for decision making. Over at Stats Canada, they want to destroy the 1911 census rather than release it as they have previous censuses, including 1901. Their argument is that releasing the original census data - it is literally copies of the forms that where filled out on people's doorsteps - would violate the privacy of the people listed. This is pissing off people who want to do geneological research as censuses are incredibly useful when researching family history. It lets you find out where people were at a given point in time so you can more effectively search other public records - which were not, until now, centralized. With out the census data, we deprive people of the ability to find out where they came from - unless we give them access to this centralized database. Secure access, of course. On the web. Uh huh. Bureaucrats just don't get it.
I recently downloaded a copy of the Borland C++ Compiler v5.5. It's free as in beer. The license contains the paragraph quoted above but if you read the whole thing, you find the following paragraph:
All Inprise libraries, source code, Redistributables and other files remain Inprise's exclusive property. Regardless of any modifications that you make, you may not distribute any files (particularly Inprise source code and other non-executable files) except those that Inprise has expressly designated as Redistributables. Nothing in the License Agreement permits you to derive the source code of files that Inprise has provided to you in executable form only, or to reproduce, modify, use, or distribute the source code of such files. You are not, of course, restricted from distributing source code or byte code that is entirely your own. Source code which you generate with an Inprise source code generator, such as an Application Wizard, is considered by Inprise to be your code.
Very little, likely. Go read the patent. It references 4 other patents on similar systems from Apple, IBM, Samsung and another from MS. There appear to be over 150 other patents on automatic software distribution over networks. Obviously that basic idea is not patentable.
Face facts, people. Unless you have the permission of the artists and their publishers, the music you share on Napster is stolen. You can try to justify it and Jon Katz can write long winded hogwash, but that doesn't change anything. You can argue that the publishing companies make most of the money, but that ignores the fact that the artists make their living from selling their art. When you steal music on Napster, you are stealing from the artists you claim to like and support so much. Grow up and behave responsibly. This isn't about freedom - it's about wanting something for nothing.
You've limited you solutions by assuming the application must be monolithic. That need not be the case. All applications can be logically divided into three parts: UI, Business and Data. In this case, business may be "Update customer address" or it may be "Get URL". Data may "Execute stored procedure" or "Make TCP/IP connection".
If this business piece is written as a library or as a server process, any number of specialized UI's can use it. Each UI will be percieved as a different application by it's users, but the real work (everything that isn't about interacting with the user) will be common across all UI's. (There are a number of X Windows utilities that are strictly UI and use command line tools to do their real work.)
This is esentially what all the "N-Tier" hullaballoo is about.
Lets face it machines can't fuck up half as badly as politicians have mangaged to do over the last 100 years.
Not on their own. That would require human intervention.;) What kind of security would we need to ensure that no one altered the machine's programming or data? I can see it now: The Electoral Computer Review Board. The question is, do we trust people on this board or not? The logical end of this journey is to have humans serve the needs of the machines. No thank you.
Sure, but how will that help Slashdot? The software has to run in a specific environment and meet specific needs. If they make the software truly open source, they will be inundated with dozens of patches to either generalize it or make it work in some other specific environment. This would be more a hinderance than a help to Slashdot, increasing their maintenance. Sure they would have lots of help, but if you think co-ordinating that help is trivial, you're really out to lunch.
The people who want the Slashdot software as open source must have pretty selfish motivations, since there is little they can do to help anyone but themselves. All they can do is take the ideas that have made Slashdot successful on the technology front or show off their "brilliance" by "fixing" what they, in their wisdom, view as bugs.
Get used to the idea that CmdrTaco and Hemos have control of Slashdot, or go away.
You need to decide how much you want to bleed. These specs use a NEW motherboard chipset (840), a NEW memory architecture (RAMBUS), and a NEW CPU technology (Coppermine). If you want to pursue this technology, do so with your eyes wide open. This stuff is too new to be really stable. There are bound to be gotchas hiding all over the place. Remember that when it comes to computers, Murphy was an optimist.
Consider a Xeon based system if you really need the raw horsepower, but it sounds like you really want maximum I/O. To get that, look at SCSI RAID controlers, or a non-Intel architecture with a decent motherboard I/O bus. Alpha motherboards have 128 (or even 256?) bit buses. Don't get too hung up on CPU; that's seldom the bottleneck on a new system.
What ESR is addressing is that both Quake and ecommerce clients exist in this same environment, and that when open source principles are applied from the get-go, a secure solution can be provided.
ESR's assertion that Quake would be more secure if it had been developed as Open Source really doesn't wash. To get the performance nessecary over slow links requires that you trust the client. This is never secure. Period. No well designed e-Commerce app would ever trust the client; Open or Closed Source. Therefore, the comparison isn't valid. The environments are not the same.
ESR's most important point is that opening the source now hasn't created any new cheating opportunities. This is the message that needs to get to the people that think opening the source to Quake has created a new problem. Unfortunately, this message will be obscured by ESR's conclusions. This issue didn't require any new insight; just some facts.
If the OS's are not running concurrently, would it be necessary to emulate the hardware? For instance, would it be possible to have a very thin layer that allowed me to switch between 'running' OS's, similar to switching between virtual terminals in linux? The basic concept is that the background OS's would be suspended and only the foreground OS would actually be running on the system. If this were the case, I see no reason why anything would have to be emulated because each OS would be running natively, just not concurrently.
To do what you are suggesting would essentially be a microkernel in concept. It may be possible to use the APCI (not sure of the acronym) "Suspend to RAM" (also called "Instant On" or "OnNow") function to signal the OS that it is suspended, take a snapshot of the RAM and dump it to disk, and then switch to the other OS. The problem would likely be the "state" of the system. That is, PNP settings and IO strategies. One OS may be using PIO and the other DMA.
Although interesting, I don't think your idea would provide sufficient benefits over VMWare's strategy to justify the effort.
Slashdot Mirror
Read the story. B & N isn't challenging the patent, they are saying they don't infringe on it because their implementation requires more than one click.
B & N isn't looking to change anything about the patent system. They are playing by the "rules". If they can win, they will probably patent their version.
To put Napster out of business, even if not permanently, would do a serious disservice to the online community's reputation. While the RIAA claims that their clients are being harmed right now, and therefore Napster should be shut down, it is also true that Napster has a right to be providing it services, pending a court verdict. Either way the court decides, action must take place after the verdict, and after the judge has had time to make up his mind.
Napster isn't being put out of business. The injunction says that Napster cannot allow the distribution of copyrighted works with out permission. This doesn't actually change anything, legally, since Napster can't systematically allow it's service to be used for copyright infringment anyway. (There are many precidents stating this explicitly for Internet services. For Napster to run an Internet service and be ignorant of this is simply not believable.) The fact that Napster chooses to represent this as "putting them out of business" simply shows the lie in their assertion that their service is actually used, to any significant degree, for anything other than copyright infringment.
While there are some nice features about components, like anything else it's possible to have too much of a good thing.
. . . .
I believe this is called "DLL Hell" in Windows circles.
You are confusing traditional shared DLLs with components. In MS land, components use COM which supports proper versioning of components - either with multiple components implementing different versions or with a single component implementing backwards compatability.
COM is the foundation of COM+ (which is a marketing term, not a technology as COM is) and the whole platform will evolve into .NET which will allow MS to treat the OS as a comodity when their appeal is concluded and they loose Windows. By then, their platform will be .NET and it will be business as usual.
Meanwhile, Linux has no component model, no distributed transaction services, and no server component runtime. EJB/CORBA isn't nearly as comprehensive as COM+, and isn't real competition. MS is pulling an end run and no one has noticed, yet.
Miguel didn't talk about OOP, he talked about components. OOP does not mean components and components can be built with out OOP. In general OOP is a powerful way to build components, but Component Based Development is conceptually very different from OOP. The two can be complementary, but are not the same.
Two things, I think.
The licensee of your software is the operator of the server, not the end user. The operator of the server can modify the software and, as long as they only run it on their own servers, are not re-distributing your software - even though their users are distributed.
Well, it doesn't if you read his bpost a little further. He proposes using the alphabet as a natural hierarchy.
That would require infrastructure changes because currently it's the dots that delimit zones of authority. Changing to using each character of the TLD still breaks the existing infrastructure and therefore isn't a solution.
I rather like that idea.
I like the idea, too. But I still don't think it will (or should be) adopted. DNS changes like this could break a huge piece of the internet really fast. People have tried to implement new TLD naming schemes before, but without entries in the current root servers, noone can find their servers. And the roots aren't about to break a working system.
Allow anything to be used as a TLD.
HOWEVER, still require registrations to consist of domain name + TLD. i.e., you must still sumbit both parts to constitute a single registration application. The TLD itself cannot be registered to anyone. and remains open for anyone to use.
I like the idea, but it breaks the hierarchical nature of DNS. Each "." in a machine name delimits a "zone of authority". With out any cacheing, you have to ask a root server for a server that can answer .org queries. Then you ask that server for a server that can answer .slashdot.org queries. Lastly you ask that server for the address of www.slashdot.org. Normally, most of this data is cached in the lower levels of the hierarchy, giving use reasonable DNS performance as well as managability.
As good as this idea is, it won't be adopted any time soon because of the infrastructure changes needed to support an unlimited number of TLDs.
You appear to have take an "architectural" approach to designing TUX, so I have some architectural questions.
The benchmark seems to test concurrency. That is, how many clients can the server serve with out dropping any connections and still maintain decent through put overall.
Concurrency is the major consideration for total system performance as as opposed to individual connection performance. Given appropriate hardware, any crappy server can push 1Gb/sec on a connection. But what happens when there are 1000 connections? 5000? 10000? More? Serving tens of thousands of concurrent connections efficiently is the holy grail of network servers.
It's gotta be custom since you're going to be handling 10k idle threads at any one time.
A thread per connection model is probably not the best architecture for a system support so many users. See Dan Kegel's C10K site for more info on high performance servers supporting many concurrent connections.
I run Squid on a P75 with 32 MB RAM. The workstations are both PII 300's with 64 MB and 128 MB RAM respectively. Netscape is configured to use a RAM cache and no disk cache. Squid has a 100 MB disk cache.
The Squid cache may be slower over 10 Mbps than on a local disk but my Internet connection is a 56K modem, and even the Squid cache is faster than that. Even if you have cable modem or DSL at 1.5 Mbps, Squid will still be faster.
Squid can also be used as part of a firewall, can block Doubleclick.net and other annoying people, and gives you nice HTML page when there are errors. All in all, Squid makes a nice addition to any home network.
Mr. Konrad esentially claims patent on multi-tiered client/server architecture. Obviously, Microsoft has been delaying their Next Generation Windows Services anouncement because of delays in negotiating with Mr. Konrad, not because of the impending breakup of the company.
Seriously, the patent acknowledges so much prior art, including X Windows and Cathode Ray Tube Display, that it can't possibly be considered "inventive" or "non-obvious".
This is basicly the longest description of multi-tiered client/server architecture that I've ever read. The entire software industry is infringing on this one. And has been since the first dumb terminal was connected to a mainframe.
Get this: "The above description is illustrative and not restrictive. Many variations of the invention will become apparent to those of skill in the art upon review of this disclosure. The scope of the invention should, therefore, be determined not with reference to the above description, but instead should be determined with reference to the appended claims along with their full scope of equivalents." In short, the idea may be fifty years old, but I only want to claim future uses of it. Can I patent the business process of riding on the coat tails of real inovators and collecting royalties for doing nothing?
There is some irony here. This database is maintained by Human Resources and used for decision making. Over at Stats Canada, they want to destroy the 1911 census rather than release it as they have previous censuses, including 1901. Their argument is that releasing the original census data - it is literally copies of the forms that where filled out on people's doorsteps - would violate the privacy of the people listed. This is pissing off people who want to do geneological research as censuses are incredibly useful when researching family history. It lets you find out where people were at a given point in time so you can more effectively search other public records - which were not, until now, centralized. With out the census data, we deprive people of the ability to find out where they came from - unless we give them access to this centralized database. Secure access, of course. On the web. Uh huh. Bureaucrats just don't get it.
I recently downloaded a copy of the Borland C++ Compiler v5.5. It's free as in beer. The license contains the paragraph quoted above but if you read the whole thing, you find the following paragraph:
All Inprise libraries, source code, Redistributables and other files remain Inprise's exclusive property. Regardless of any modifications that you make, you may not distribute any files (particularly Inprise source code and other non-executable files) except those that Inprise has expressly designated as Redistributables. Nothing in the License Agreement permits you to derive the source code of files that Inprise has provided to you in executable form only, or to reproduce, modify, use, or distribute the source code of such files. You are not, of course, restricted from distributing source code or byte code that is entirely your own. Source code which you generate with an Inprise source code generator, such as an Application Wizard, is considered by Inprise to be your code.
Talk about going off half cocked.
What consequenses will this have?
Very little, likely. Go read the patent. It references 4 other patents on similar systems from Apple, IBM, Samsung and another from MS. There appear to be over 150 other patents on automatic software distribution over networks. Obviously that basic idea is not patentable.
Here is Apple's patent: http://www.patents.ibm.com/detail s?&pn10=US05732275.
Hang on - gotta get my asbestos undies on.
Ok . . . Lets' go.
Face facts, people. Unless you have the permission of the artists and their publishers, the music you share on Napster is stolen. You can try to justify it and Jon Katz can write long winded hogwash, but that doesn't change anything. You can argue that the publishing companies make most of the money, but that ignores the fact that the artists make their living from selling their art. When you steal music on Napster, you are stealing from the artists you claim to like and support so much. Grow up and behave responsibly. This isn't about freedom - it's about wanting something for nothing.
You've limited you solutions by assuming the application must be monolithic. That need not be the case. All applications can be logically divided into three parts: UI, Business and Data. In this case, business may be "Update customer address" or it may be "Get URL". Data may "Execute stored procedure" or "Make TCP/IP connection".
If this business piece is written as a library or as a server process, any number of specialized UI's can use it. Each UI will be percieved as a different application by it's users, but the real work (everything that isn't about interacting with the user) will be common across all UI's. (There are a number of X Windows utilities that are strictly UI and use command line tools to do their real work.)
This is esentially what all the "N-Tier" hullaballoo is about.
Lets face it machines can't fuck up half as badly as politicians have mangaged to do over the last 100 years.
Not on their own. That would require human intervention. ;) What kind of security would we need to ensure that no one altered the machine's programming or data? I can see it now: The Electoral Computer Review Board. The question is, do we trust people on this board or not? The logical end of this journey is to have humans serve the needs of the machines. No thank you.
Sure, but how will that help Slashdot? The software has to run in a specific environment and meet specific needs. If they make the software truly open source, they will be inundated with dozens of patches to either generalize it or make it work in some other specific environment. This would be more a hinderance than a help to Slashdot, increasing their maintenance. Sure they would have lots of help, but if you think co-ordinating that help is trivial, you're really out to lunch.
The people who want the Slashdot software as open source must have pretty selfish motivations, since there is little they can do to help anyone but themselves. All they can do is take the ideas that have made Slashdot successful on the technology front or show off their "brilliance" by "fixing" what they, in their wisdom, view as bugs.
Get used to the idea that CmdrTaco and Hemos have control of Slashdot, or go away.
You need to decide how much you want to bleed. These specs use a NEW motherboard chipset (840), a NEW memory architecture (RAMBUS), and a NEW CPU technology (Coppermine). If you want to pursue this technology, do so with your eyes wide open. This stuff is too new to be really stable. There are bound to be gotchas hiding all over the place. Remember that when it comes to computers, Murphy was an optimist.
Consider a Xeon based system if you really need the raw horsepower, but it sounds like you really want maximum I/O. To get that, look at SCSI RAID controlers, or a non-Intel architecture with a decent motherboard I/O bus. Alpha motherboards have 128 (or even 256?) bit buses. Don't get too hung up on CPU; that's seldom the bottleneck on a new system.
What ESR is addressing is that both Quake and ecommerce clients exist in this same environment, and that when open source principles are applied from the get-go, a secure solution can be provided.
ESR's assertion that Quake would be more secure if it had been developed as Open Source really doesn't wash. To get the performance nessecary over slow links requires that you trust the client. This is never secure. Period. No well designed e-Commerce app would ever trust the client; Open or Closed Source. Therefore, the comparison isn't valid. The environments are not the same.
ESR's most important point is that opening the source now hasn't created any new cheating opportunities. This is the message that needs to get to the people that think opening the source to Quake has created a new problem. Unfortunately, this message will be obscured by ESR's conclusions. This issue didn't require any new insight; just some facts.
Sperm don't go UP, they go IN.
If the OS's are not running concurrently, would it be necessary to emulate the hardware? For instance, would it be possible to have a very thin layer that allowed me to switch between 'running' OS's, similar to switching between virtual terminals in linux? The basic concept is that the background OS's would be suspended and only the foreground OS would actually be running on the system. If this were the case, I see no reason why anything would have to be emulated because each OS would be running natively, just not concurrently.
To do what you are suggesting would essentially be a microkernel in concept. It may be possible to use the APCI (not sure of the acronym) "Suspend to RAM" (also called "Instant On" or "OnNow") function to signal the OS that it is suspended, take a snapshot of the RAM and dump it to disk, and then switch to the other OS. The problem would likely be the "state" of the system. That is, PNP settings and IO strategies. One OS may be using PIO and the other DMA.
Although interesting, I don't think your idea would provide sufficient benefits over VMWare's strategy to justify the effort.