Is your position that there should be a total lack of anonymity for all people or just people in power? I always think of the , "With great power comes great responsibilities line." Do you mean that one of those responsiblities is a lose of privacy? What a goverment employee's personal life? The press seems to follow the idea that high level goverment employees (senators, independent councils, etc.) are like celebreties and have choosen a public lifestyle that precludes privacy. Is this fair? If so, does this include Internet transactions like e-mails and web browsing? Do we have a right to know what is in George W.'s bookmarks? I agree that people need to have more information about what their government but perhaps the amount isn't the problem. It is the difficulty in getting it. Who really researches voting records? Who watches CSPAN? All of these things take time. I have always wondered why even during an election, the only way to find a simple summary of a canidate's voting record is to check out his competition and then you only see the bad. This ends up making the decision trying to choose the lesser of two evils. We really need a new style of press that is beholden to no one and gets the answers that people need. The Associated Press is pretty good but all to often many of their stories and others like them are not noted by the major distributers (CNN, MSNBC, FoxNews, ect). We have a public television channel and although it has news we need more. We need a public news channel/web site that has all the information sorted and catagorized for easy searching.
The biggest problem with embedded MS and what will kill them in this market is their high resource demands and relative instability of their OS base. For toys like a cable box, MS can do pretty well. If it crashes, just flip it on and off. For something like a FibreChannel adapter or Gigabit Network adapter (both of which I have worked on) I can't imagine CE or Emb.NT making the grade. RealTime NT is a long way off from being a real choice. Without source access, it isn't customizable. That means it has to be very general in order to meet the demands of all their customers. How can this compete with an open source project where I company can create a stripped down solution that only has what the need in the way that they need it.
I am working on a FibreChannel adapter that uses the I2O (Intelligent I/O) model. We are currently using WindRiver's IxWorks for StrongArm but as part of our next design we are looking at other solutions. As you might imagine, speed is critical. We looked into using a Linux or BSD based system and found that an open Linux/BSD solution was better than a proprietary solution. The only problem was it required so much customization that we found writing our own system from scratch could give us a better, more specific solution in about the same amount of development time. I suspect that in a world where resources are scarce and every clock cycle matters, a lot of companies are going to find that a general (although customizable) solution like Linux will have trouble competing with a completly specific solution. Companies that do not have the technical resources to start from scratch will find Linux attractive but for companies like mine where we have people that have a lot of operating system and specifically mission critical system development experience, Linux/BSD does't provide a real edge. Linux may work well for cable boxes, digital video recorders, and PDAs but can it run a FibreChannel adapter? How about a realtime system to be used in manufacturing? I think that it can do it but each of these situations and many others require so much customization that it begs the question, "Why not just do it from scratch?"
Well...you do a little. I think that you still need competition. Patents are the problem, not the competitive streak that spawned them. Patents by their very nature are anti-competitive and IMHO bad for an economy. You are right in saying that in their current state they create stagnation but you forget that they are used to stiffle competition. How can I make a better "1 click" system if I can't make one at all? They are an example of a company trying to compete by preventing another from doing the same. They need to be changed. I understand a company's need to pay for their years of development but the control needs to be limited. They should not be able to just have total control over any patent infringment. It should be somewhat like phone lines. Whoever builds them gets the right to charge for usage but they must allow others to use them. Energy supplier competition in my home state of PA is like this. Cable lines will hopefully go this way. Patents should only give the holder the right to charge for usage, not to block usage. The only problem is how do you regulate the fee structure without creating a massive amounts of expensive goverment that tax payers have to pay for? This is where most what kills most regulatory issues... the expense to force people to play fair.
Remember the "largest bookstore" lawsuite? Amazon used this line to note the size of their massive list of books. B&N sued because they claimed that they had more books in stock then Amazon. This is what really got the web book wars going.
It is interesting to see what Jeff had to say but the proof will be in what he does over the coming years. Will he panic and go the way of Apple and Lotus or will he roll up his sleeves and get to work on beating the competition? It seems that all to often companies that start out trying to do just that go either one of two directions. They either give up and become like the rest or they lose but go down fighting. In the first direction is the argument, "Everybody else is doing it so I have to survive." I understand that on a business level. If you are playing a game and someone is cheating and you want to win, you have to cheat, right? Perhaps. I think that playing better, harder, smarter, etc. can still win the day. Sadly, though there is proof of the contrary. Or is there? Netscape's demise is the noted as the perfect example of why to fear trying to "be the good guy". Netscape did all the right things but lost. Was it evil marketing, monopolistic practices, bad software, bad business decisions? IMHO, all of the above. This suggests to me that a good, smart technology company can still beat a "cheating" company. Netscape can not be used as the reason to cheat; although their competition cheated, they made a ton of their own mistakes. Their products became inferior. They cheated a little themselves with their own extensions to HTML. Jeff Bezos should try to lead his company to keep their technical edge. They need to keep their marketing edge. They do not need software patents. It seems that Jeff is a bit afraid that he can't win without cheating. He is a student on test day and although he studies hard he is worried it will not be enough to beat the curve created by cheaters. Come on Jeff, you have set the curve so far; just study harder (hire smart people), worker harder (hire more of them), and you will do fine.
For linux to win the home desktop, it has to win the corporate one. Why did Windows beat Mac? There are many reasons but one is that Windows (or DOS at the time) was used at work. People only want to learn a few things. Multiple operating systems and programs are not one of those things. Sometimes, people want to work on job related stuff at home. Again, you need the work os. Finally, OEMs really have the power. They will put on whatever OS a company wants if they are willing to buy enough systems. Once they start selling Linux to business in serious quantities, they will start to sell Linux to home users. Once people start seeing Linux at work, they will want it at home. The real question is, "How can Linux win the business world?"
What makes Gnome better for sysadmins and Kde better for the masses? I personally think that Kde is a bit ahead of Gnome in some ways but otherwise they do the smae things. What's the difference?
I spent the last 6 months working on a FibreChannel RAID adapter. It could also be run in direct connect mode. I can tell you that with any OS the raid on the controller could always beat the host based software raid. External raid (raid boxes of storage with no controller) are around the same speed, perhaps better under big loads. Software raid really doesn't make sense. All the extra time spent in raid calculations wastes host CPU which could be better spent running the database, web server, etc. Under heavy loads, software raid starts to either effect the rest of the system or vice-versa. Offloading RAID processing on to a controller or external storage system that has dedicated processors is the best solution. Think about video cards...you could do the work on the host processor but why? Host processors are for general, non specific tasks. For specific tasks (only raid, only video, only tcp/ip) that do not change much over time, offloading is a good idea.
Perhaps but I don't see Unisys or any of the employees complaining anytime soon. Most of the them are pretty excited to be working on something different and will enjoy the five year vacation from our mainframe OS. I have avoided joining this group because I have been holding out for a Linux port. Then, Unisys has to play Chinese Wall games with the Linux-CMP guys and the Win2k-CMP guys. Fun.
I think that is was something like $1 Million for NT4 and $1.5 million for Win2k. There were some limits on the number of users. Yep, it means that every employee on the development team who is licensed to see the code can not work on another OS for 5 years. That includes Unisys's mainframe OS. MS let us have the source because we are offing them a chance at real enterprise computing. We make mainframes that have decade long uptimes and are trying to bring some of that stability to Win2k. MS (and Intel to some extent) really needs a hardware vendor to make a super Wintel box. MS can not hope for Alpha anymore so CMP (and a few other similar architectures out there) are their best bet. Big Bill mentioned this program directly at the Win2k release party in SanFran. It is a pretty big deal at MS. They are really hopeing this will be the Unix killer. The again...this thing supports running SCO UnixWare already and Solaris for x86 port is in the works. Even a Linux port is in the discussion phase. I am trying my best to get on that project!
I am working on the storage for the big system so there are really others who could answer this better than I can. I will try anyway, though. Our system supports up to 64 GB of RAM which coincidentally is the limit of addressable space on Intel's 32-bit processors. The normal 32-bit limit is 4 GB but Intel added Physical Address Extensions to give you 36-bit addressing, hence 64 GB, limit. NT4 did not support this but Win2k does. The standard virtual address space for NT4 is 4GB with 2GB for the system and 2GB for user space. NT4 Enterprise allowed you to make that 1 GB system and 3GB user. Win2k's support for PAE (not on by default) gives varying levels of support, depending on the version of Win2k. DataCenter will support the full 64 GB. The support comes in three forms:
Kernel Support - The kernel supports using the memory for itself in a hidden way. For example, this memory can be used to reduce paging. This is done without applications or drivers needed to know about it.
Kernel Driver Support - Kernel drivers can note that they support 64-bit addressing (althought they will not see more than 36) and avoid Win2k having to double-buffer reads and writes to high memory (above 4GB)
Application Support - Apps can use this memory directly using Address Windowing Extensions. This lets applications allocate physical un-paged memory and create several virtual views of this memory.
Give that, our hardware can limit the OS's "view" of memory and make some OS changes to allow for physical memory reservations. This system of limiting and reserving is part of what allows us to move memory between systems. Also, even the "physical" address may not be real addresses in this system. This means that there could be a Win2k virtual to physical mapping and then a lower physical to real physical mapping. There are also designs to allow OSs to communicate through memory while running at the same time on this system. A lot of the Dynamic Partitioning stuff is still being designed by us. Since it involves Windows source, Microsoft has to approve of our designs and the changes we make.
Companies can already license Windows source code. We did it at Unisys for our CMP group. They need to hack away at NT4/Win2k to get it to support Dynamic Partitioning (where memory, hard drives, processors, etc. can be moved to other OSs running at the same time on the same system), multi-processor support with linear performance increases, and other main frame like features. The big drawback to their current license is that you can not work on another OS for five years. Ouch. There are several other restrictive aspects of the non-disclosure agreement but this one is the toughest to swallow. It is the one that MS would have to give up to satisfy the goverment.
PHP should at least get a thought. It has many of the features of Perl/Python but is embedded in the html (like asp). It has all the flexibility of Perl. A PHP programmer can write in whatver style works best for him/her.
That's really what this thing comes down to. In most cases, to get the services that you want, you have to give up some privacy. You want the goverment to give you Social Security; then you have to have a number attached to you. You want a credit card company to loan you money; then you have let them know about every purchase you make. If you want to have MP3.com handle all your music, then you have to let them know what music you like. That's just the way things go. Although there are often some insidious reasons for collecting user data, the biggest reason is usually because it is either integral to the service or it makes it work much better. For example,/. has a feature to remember your user name and password. It is pretty insecure but it makes getting access easier. In MP3.com's case, some of the information is needed, some of it may make improve the service, and a some of it may turn out to be nefarious. The consumers can dictate what they want by either using or not using the service. That is part of the beauty of a free market. Consumers can dictate the forms of new products and services with their buying power. Companies will not offer what people do not want.
I am working on a FibreChannel card now but before I worked on a dual-port fast ethernet card. I asked MS engineers about supporting multi-port cards and they had no ideas about how. The old system (under NT4) let us install multiple NICs in NT's mind while our driver knew that it was one card with two ports. Uder the new model, Win2k tells the driver where the devices are and this model didn't imagine two ports, each with it's own IP on one card. Whoops. To hack around that was quite painfull.
I have been using every weekly build since before Beta2. I have been judging Win2k on the latest code available. I know for a fact that there were pretty big bugs in the kernel that were discovered at the Windows Plugfest that did not go into the gold release. The gold release is build 2195, cut after the first night of the plugfest. All the other bugs discovered at the plugfest will not make it in until SP1. All bugs discovered after the end of december will nto be fixed until SP2 at the earliest. There is a very long delay from when a bug is discovered until it is fixed. Meanwhile, Linux bugs are often fixed in days or sometimes, hours. New builds, the equivalent of Windows SPs, come out many times a month. Linux users have the right to demand that testing be done with latest release; MS wouldn't settle for not using the latest SP.
As a device driver developer, I have been using every weekly build of Win2k since before Beta2. I can tell you for a fact that Win2k is buggy and unstable. I was at the Microsoft Plugfest, where system vendors and device vendors get together and try running their stuff together under Win2k and WinMe (Windows Millenium). Build 2195, the build that went gold was cut after the first day of the plugfest, due to a major bug that had to be fixed. Lots of bugs were reported during the following days of testing. NONE of these low level, at the core of OS, in the kernel type of bugs were fixed for the gold release. We were told that they would go into SP1. In fact, the cut off date to get a fix into SP1 was the end of december. My group has already submitted Plug 'N Play issues that will not be fixed until SP2 at the earliest. This thing is not ready for prime time! At the plugfest, Microsoft's engineers were often stumped with problems that only a small hotel full of only three days or so of testing; imagine what millions of users in months of continuous running will find. Win2k's bug list is so large that you have to search for your problem at their site rather than all the known issues being made public through a definitive list. I for one would want to read that list before I bet my e-business site on it. Try running a check build of Win2k and ready the output from WinDbg. Note all of the errors flying by, filling WinDbg's 30,000 line buffer size. Try running WinDbg, the main graphical kernel debugging tool Microsoft ships. It is perhaps the worst piece of software ever made. Every version fixes one bug but creates another. MS's own pplugfest engineers would not use it. If it is buggy than what kind of drivers will it lead to? How about Visual C++, which all of Windows is build with? How can an OS be stable when the development environment that created it needed three service packs?! As for living in a heterogeneous lan, Win2k's Active Directory uses Dynamic DNS, which most other systems, including NT4, do not support. Although, you can get DDNS for Unix/Linux systems, it requires you changing all your other machines to work with Win2k. Win2k is very bloated! Look at the size of all the running modules in a base Win2k Pro installation; it is massive! Check out the minimum requirements; they are unbelievable! This isn't anti-microsoft, brainwashed by/. FUD; it is the result of years of low level study.
This thing makes a lot of claims at their site. This article then goes crazy with it. Their site claims that it is possible to make things with this technology that could go at 1 gigabyte per second. I say perhaps. Maybe. But probably not. This thing is vapor ware and the author of this article is totally irresponsible, if not unethical, in hyping this thing. The greatest thing since WWW opened shop? Please! He then goes on to take the someday numbers from the company's site and assume that this technology that is different from CD-ROMs will increase in performance at the same rate as CD-ROMs. He ignores the issues of a controller and a BUS for this thing. PCI sure isn't going to do it. Perhaps one of its successors will handle it. Oh...and we will use it instead of RAM. Sure. Perhaps it will also be used for primary cache on the processor? Give me a break! If I want big and portable, I will go with DVD. If I want big and fast, I will go with a RAID over Fibre Channel. Only if I wanted vapor would I go with this!
That is pretty damn fast. I am working on an Fibre Channel adapter and I know that blows any single port FC adapter on the market out of the water. Also, that kind of speed is going to need a tight driver so that it doesn't just suck up all the processor. I wonder how many I/Os per second per processor percentage they can get? What about megabytes per second per processor percentage? This is the real question. It is one thing to be fast but it is another to be able to run well. For example, some EIDE drives are nearly as fast as slow SCSI drives, but EIDE controllers require much more host cpu power.
I still have my Chester County Interlink account. I have had it for over five years. It is a great freenet and has some great people associated with it. A great gift to Chester County from ESR. Thanks! As for the drivers Mr. Anon E. Mus, they are going very well. You are now looked at a post from the co-creator of one of the fastest FibreChannel card/driver around!
I live in Chester County; according to his site ESR currently lives in Malvern, about 15-20 minutes from where I live. There are tons of tech jobs out here. Our market is one of the top five in the US. I am getting ready to look for a new job and with all the possibilities job hunting becomes fun. Philadelphia is trying to attract more technology companies. So is Pittsburg. Who would have thought that Penn's Woods would become a technology haven?
You are totally correct. Sales tax is kind of like someone making you pay for a service twice. In fact, that is exactly what it is. I don't understand why many businesses are able to provide similar services in some areas as the government without using all sorts of strange double payments. I just wish I had one tax (I really wish that I that didn't have any taxes...one can dream can't he?;-) ) that I could figure out with pen and paper in about 2 minutes. What did I make? X. What do I owe? Y% * X. That's all. No loop holes, tax brackets, marriage penalties, limits or anything like that. Just one simple bill. I can handle that. Tax code? The tax code should be an Algebra 1 problem. That's all.
Let me get this straight...you don't want more sales? You don't want to the ability to discount your product without losing any money on it? You don't want to be competitive with your neighbor down south? Being busy at work is a good thing. If people at your company are not busy they fire people until the remaining people are busy. That is not good thing.
Perhaps you would if were someone who wants a PC but does not have the time, inclination, or ability to build it from parts. Computers should not be just for geeks; they should be for all to use. Computers should not be cheap only for geeks; they should be inexpensive for all. Give power to the people and let's see where they go with it.
Slashdot Mirror
Is your position that there should be a total lack of anonymity for all people or just people in power? I always think of the , "With great power comes great responsibilities line." Do you mean that one of those responsiblities is a lose of privacy? What a goverment employee's personal life? The press seems to follow the idea that high level goverment employees (senators, independent councils, etc.) are like celebreties and have choosen a public lifestyle that precludes privacy. Is this fair? If so, does this include Internet transactions like e-mails and web browsing? Do we have a right to know what is in George W.'s bookmarks?
I agree that people need to have more information about what their government but perhaps the amount isn't the problem. It is the difficulty in getting it. Who really researches voting records? Who watches CSPAN? All of these things take time. I have always wondered why even during an election, the only way to find a simple summary of a canidate's voting record is to check out his competition and then you only see the bad. This ends up making the decision trying to choose the lesser of two evils.
We really need a new style of press that is beholden to no one and gets the answers that people need. The Associated Press is pretty good but all to often many of their stories and others like them are not noted by the major distributers (CNN, MSNBC, FoxNews, ect). We have a public television channel and although it has news we need more. We need a public news channel/web site that has all the information sorted and catagorized for easy searching.
The biggest problem with embedded MS and what will kill them in this market is their high resource demands and relative instability of their OS base. For toys like a cable box, MS can do pretty well. If it crashes, just flip it on and off. For something like a FibreChannel adapter or Gigabit Network adapter (both of which I have worked on) I can't imagine CE or Emb.NT making the grade. RealTime NT is a long way off from being a real choice. Without source access, it isn't customizable. That means it has to be very general in order to meet the demands of all their customers. How can this compete with an open source project where I company can create a stripped down solution that only has what the need in the way that they need it.
I am working on a FibreChannel adapter that uses the I2O (Intelligent I/O) model. We are currently using WindRiver's IxWorks for StrongArm but as part of our next design we are looking at other solutions. As you might imagine, speed is critical. We looked into using a Linux or BSD based system and found that an open Linux/BSD solution was better than a proprietary solution. The only problem was it required so much customization that we found writing our own system from scratch could give us a better, more specific solution in about the same amount of development time. I suspect that in a world where resources are scarce and every clock cycle matters, a lot of companies are going to find that a general (although customizable) solution like Linux will have trouble competing with a completly specific solution. Companies that do not have the technical resources to start from scratch will find Linux attractive but for companies like mine where we have people that have a lot of operating system and specifically mission critical system development experience, Linux/BSD does't provide a real edge.
Linux may work well for cable boxes, digital video recorders, and PDAs but can it run a FibreChannel adapter? How about a realtime system to be used in manufacturing? I think that it can do it but each of these situations and many others require so much customization that it begs the question, "Why not just do it from scratch?"
Well...you do a little. I think that you still need competition. Patents are the problem, not the competitive streak that spawned them. Patents by their very nature are anti-competitive and IMHO bad for an economy. You are right in saying that in their current state they create stagnation but you forget that they are used to stiffle competition. How can I make a better "1 click" system if I can't make one at all? They are an example of a company trying to compete by preventing another from doing the same.
They need to be changed. I understand a company's need to pay for their years of development but the control needs to be limited. They should not be able to just have total control over any patent infringment. It should be somewhat like phone lines. Whoever builds them gets the right to charge for usage but they must allow others to use them. Energy supplier competition in my home state of PA is like this. Cable lines will hopefully go this way. Patents should only give the holder the right to charge for usage, not to block usage. The only problem is how do you regulate the fee structure without creating a massive amounts of expensive goverment that tax payers have to pay for? This is where most what kills most regulatory issues... the expense to force people to play fair.
Remember the "largest bookstore" lawsuite? Amazon used this line to note the size of their massive list of books. B&N sued because they claimed that they had more books in stock then Amazon. This is what really got the web book wars going.
It is interesting to see what Jeff had to say but the proof will be in what he does over the coming years. Will he panic and go the way of Apple and Lotus or will he roll up his sleeves and get to work on beating the competition?
It seems that all to often companies that start out trying to do just that go either one of two directions. They either give up and become like the rest or they lose but go down fighting.
In the first direction is the argument, "Everybody else is doing it so I have to survive." I understand that on a business level. If you are playing a game and someone is cheating and you want to win, you have to cheat, right? Perhaps. I think that playing better, harder, smarter, etc. can still win the day. Sadly, though there is proof of the contrary. Or is there?
Netscape's demise is the noted as the perfect example of why to fear trying to "be the good guy". Netscape did all the right things but lost. Was it evil marketing, monopolistic practices, bad software, bad business decisions? IMHO, all of the above. This suggests to me that a good, smart technology company can still beat a "cheating" company. Netscape can not be used as the reason to cheat; although their competition cheated, they made a ton of their own mistakes. Their products became inferior. They cheated a little themselves with their own extensions to HTML.
Jeff Bezos should try to lead his company to keep their technical edge. They need to keep their marketing edge. They do not need software patents. It seems that Jeff is a bit afraid that he can't win without cheating. He is a student on test day and although he studies hard he is worried it will not be enough to beat the curve created by cheaters. Come on Jeff, you have set the curve so far; just study harder (hire smart people), worker harder (hire more of them), and you will do fine.
For linux to win the home desktop, it has to win the corporate one. Why did Windows beat Mac? There are many reasons but one is that Windows (or DOS at the time) was used at work. People only want to learn a few things. Multiple operating systems and programs are not one of those things. Sometimes, people want to work on job related stuff at home. Again, you need the work os. Finally, OEMs really have the power. They will put on whatever OS a company wants if they are willing to buy enough systems. Once they start selling Linux to business in serious quantities, they will start to sell Linux to home users. Once people start seeing Linux at work, they will want it at home.
The real question is, "How can Linux win the business world?"
What makes Gnome better for sysadmins and Kde better for the masses? I personally think that Kde is a bit ahead of Gnome in some ways but otherwise they do the smae things. What's the difference?
I spent the last 6 months working on a FibreChannel RAID adapter. It could also be run in direct connect mode. I can tell you that with any OS the raid on the controller could always beat the host based software raid. External raid (raid boxes of storage with no controller) are around the same speed, perhaps better under big loads. Software raid really doesn't make sense. All the extra time spent in raid calculations wastes host CPU which could be better spent running the database, web server, etc. Under heavy loads, software raid starts to either effect the rest of the system or vice-versa. Offloading RAID processing on to a controller or external storage system that has dedicated processors is the best solution. Think about video cards...you could do the work on the host processor but why? Host processors are for general, non specific tasks. For specific tasks (only raid, only video, only tcp/ip) that do not change much over time, offloading is a good idea.
Perhaps but I don't see Unisys or any of the employees complaining anytime soon. Most of the them are pretty excited to be working on something different and will enjoy the five year vacation from our mainframe OS.
I have avoided joining this group because I have been holding out for a Linux port. Then, Unisys has to play Chinese Wall games with the Linux-CMP guys and the Win2k-CMP guys. Fun.
I think that is was something like $1 Million for NT4 and $1.5 million for Win2k. There were some limits on the number of users.
Yep, it means that every employee on the development team who is licensed to see the code can not work on another OS for 5 years. That includes Unisys's mainframe OS.
MS let us have the source because we are offing them a chance at real enterprise computing. We make mainframes that have decade long uptimes and are trying to bring some of that stability to Win2k. MS (and Intel to some extent) really needs a hardware vendor to make a super Wintel box. MS can not hope for Alpha anymore so CMP (and a few other similar architectures out there) are their best bet. Big Bill mentioned this program directly at the Win2k release party in SanFran. It is a pretty big deal at MS. They are really hopeing this will be the Unix killer.
The again...this thing supports running SCO UnixWare already and Solaris for x86 port is in the works. Even a Linux port is in the discussion phase. I am trying my best to get on that project!
I will try anyway, though. Our system supports up to 64 GB of RAM which coincidentally is the limit of addressable space on Intel's 32-bit processors. The normal 32-bit limit is 4 GB but Intel added Physical Address Extensions to give you 36-bit addressing, hence 64 GB, limit. NT4 did not support this but Win2k does. The standard virtual address space for NT4 is 4GB with 2GB for the system and 2GB for user space. NT4 Enterprise allowed you to make that 1 GB system and 3GB user. Win2k's support for PAE (not on by default) gives varying levels of support, depending on the version of Win2k. DataCenter will support the full 64 GB. The support comes in three forms:
Give that, our hardware can limit the OS's "view" of memory and make some OS changes to allow for physical memory reservations. This system of limiting and reserving is part of what allows us to move memory between systems. Also, even the "physical" address may not be real addresses in this system. This means that there could be a Win2k virtual to physical mapping and then a lower physical to real physical mapping. There are also designs to allow OSs to communicate through memory while running at the same time on this system.
A lot of the Dynamic Partitioning stuff is still being designed by us. Since it involves Windows source, Microsoft has to approve of our designs and the changes we make.
Companies can already license Windows source code. We did it at Unisys for our CMP group. They need to hack away at NT4/Win2k to get it to support Dynamic Partitioning (where memory, hard drives, processors, etc. can be moved to other OSs running at the same time on the same system), multi-processor support with linear performance increases, and other main frame like features. The big drawback to their current license is that you can not work on another OS for five years. Ouch. There are several other restrictive aspects of the non-disclosure agreement but this one is the toughest to swallow. It is the one that MS would have to give up to satisfy the goverment.
PHP should at least get a thought. It has many of the features of Perl/Python but is embedded in the html (like asp). It has all the flexibility of Perl. A PHP programmer can write in whatver style works best for him/her.
That's really what this thing comes down to. In most cases, to get the services that you want, you have to give up some privacy. You want the goverment to give you Social Security; then you have to have a number attached to you. You want a credit card company to loan you money; then you have let them know about every purchase you make. If you want to have MP3.com handle all your music, then you have to let them know what music you like. That's just the way things go. /. has a feature to remember your user name and password. It is pretty insecure but it makes getting access easier. In MP3.com's case, some of the information is needed, some of it may make improve the service, and a some of it may turn out to be nefarious. The consumers can dictate what they want by either using or not using the service. That is part of the beauty of a free market. Consumers can dictate the forms of new products and services with their buying power. Companies will not offer what people do not want.
Although there are often some insidious reasons for collecting user data, the biggest reason is usually because it is either integral to the service or it makes it work much better. For example,
I am working on a FibreChannel card now but before I worked on a dual-port fast ethernet card. I asked MS engineers about supporting multi-port cards and they had no ideas about how. The old system (under NT4) let us install multiple NICs in NT's mind while our driver knew that it was one card with two ports. Uder the new model, Win2k tells the driver where the devices are and this model didn't imagine two ports, each with it's own IP on one card. Whoops. To hack around that was quite painfull.
I have been using every weekly build since before Beta2. I have been judging Win2k on the latest code available. I know for a fact that there were pretty big bugs in the kernel that were discovered at the Windows Plugfest that did not go into the gold release. The gold release is build 2195, cut after the first night of the plugfest. All the other bugs discovered at the plugfest will not make it in until SP1. All bugs discovered after the end of december will nto be fixed until SP2 at the earliest. There is a very long delay from when a bug is discovered until it is fixed. Meanwhile, Linux bugs are often fixed in days or sometimes, hours. New builds, the equivalent of Windows SPs, come out many times a month. Linux users have the right to demand that testing be done with latest release; MS wouldn't settle for not using the latest SP.
As a device driver developer, I have been using every weekly build of Win2k since before Beta2. I can tell you for a fact that Win2k is buggy and unstable. I was at the Microsoft Plugfest, where system vendors and device vendors get together and try running their stuff together under Win2k and WinMe (Windows Millenium). Build 2195, the build that went gold was cut after the first day of the plugfest, due to a major bug that had to be fixed. Lots of bugs were reported during the following days of testing. NONE of these low level, at the core of OS, in the kernel type of bugs were fixed for the gold release. We were told that they would go into SP1. In fact, the cut off date to get a fix into SP1 was the end of december. My group has already submitted Plug 'N Play issues that will not be fixed until SP2 at the earliest. This thing is not ready for prime time! /. FUD; it is the result of years of low level study.
At the plugfest, Microsoft's engineers were often stumped with problems that only a small hotel full of only three days or so of testing; imagine what millions of users in months of continuous running will find. Win2k's bug list is so large that you have to search for your problem at their site rather than all the known issues being made public through a definitive list. I for one would want to read that list before I bet my e-business site on it.
Try running a check build of Win2k and ready the output from WinDbg. Note all of the errors flying by, filling WinDbg's 30,000 line buffer size. Try running WinDbg, the main graphical kernel debugging tool Microsoft ships. It is perhaps the worst piece of software ever made. Every version fixes one bug but creates another. MS's own pplugfest engineers would not use it. If it is buggy than what kind of drivers will it lead to? How about Visual C++, which all of Windows is build with? How can an OS be stable when the development environment that created it needed three service packs?!
As for living in a heterogeneous lan, Win2k's Active Directory uses Dynamic DNS, which most other systems, including NT4, do not support. Although, you can get DDNS for Unix/Linux systems, it requires you changing all your other machines to work with Win2k.
Win2k is very bloated! Look at the size of all the running modules in a base Win2k Pro installation; it is massive! Check out the minimum requirements; they are unbelievable!
This isn't anti-microsoft, brainwashed by
This thing makes a lot of claims at their site. This article then goes crazy with it. Their site claims that it is possible to make things with this technology that could go at 1 gigabyte per second. I say perhaps. Maybe. But probably not. This thing is vapor ware and the author of this article is totally irresponsible, if not unethical, in hyping this thing. The greatest thing since WWW opened shop? Please! He then goes on to take the someday numbers from the company's site and assume that this technology that is different from CD-ROMs will increase in performance at the same rate as CD-ROMs. He ignores the issues of a controller and a BUS for this thing. PCI sure isn't going to do it. Perhaps one of its successors will handle it. Oh...and we will use it instead of RAM. Sure. Perhaps it will also be used for primary cache on the processor? Give me a break!
If I want big and portable, I will go with DVD. If I want big and fast, I will go with a RAID over Fibre Channel. Only if I wanted vapor would I go with this!
That is pretty damn fast. I am working on an Fibre Channel adapter and I know that blows any single port FC adapter on the market out of the water. Also, that kind of speed is going to need a tight driver so that it doesn't just suck up all the processor. I wonder how many I/Os per second per processor percentage they can get? What about megabytes per second per processor percentage? This is the real question. It is one thing to be fast but it is another to be able to run well. For example, some EIDE drives are nearly as fast as slow SCSI drives, but EIDE controllers require much more host cpu power.
I still have my Chester County Interlink account. I have had it for over five years. It is a great freenet and has some great people associated with it. A great gift to Chester County from ESR. Thanks!
As for the drivers Mr. Anon E. Mus, they are going very well. You are now looked at a post from the co-creator of one of the fastest FibreChannel card/driver around!
I live in Chester County; according to his site ESR currently lives in Malvern, about 15-20 minutes from where I live. There are tons of tech jobs out here. Our market is one of the top five in the US. I am getting ready to look for a new job and with all the possibilities job hunting becomes fun. Philadelphia is trying to attract more technology companies. So is Pittsburg.
Who would have thought that Penn's Woods would become a technology haven?
You are totally correct. Sales tax is kind of like someone making you pay for a service twice. In fact, that is exactly what it is. I don't understand why many businesses are able to provide similar services in some areas as the government without using all sorts of strange double payments. I just wish I had one tax (I really wish that I that didn't have any taxes...one can dream can't he? ;-) ) that I could figure out with pen and paper in about 2 minutes. What did I make? X. What do I owe? Y% * X. That's all. No loop holes, tax brackets, marriage penalties, limits or anything like that. Just one simple bill. I can handle that. Tax code? The tax code should be an Algebra 1 problem. That's all.
Let me get this straight...you don't want more sales? You don't want to the ability to discount your product without losing any money on it? You don't want to be competitive with your neighbor down south?
Being busy at work is a good thing. If people at your company are not busy they fire people until the remaining people are busy. That is not good thing.
Perhaps you would if were someone who wants a PC but does not have the time, inclination, or ability to build it from parts. Computers should not be just for geeks; they should be for all to use. Computers should not be cheap only for geeks; they should be inexpensive for all. Give power to the people and let's see where they go with it.