I registered utexas.edu on August 13, 1985. At that time the DNS was just getting started. The old ARPANET used static host tables. For a while we had to support both systems which was a bit of a pain.
I use rsync over the network to backup my laptop to my local server. I can even do it when I am on the road, since I tunnel rsync using SSH. I do daily gziped cpio archives of the snapshot nightly. So I have a history of changes.
I am glad I did this, since my hardrive on my notebook died last month. Only lost a day of data.
Exactly. A rational response to piracy would be to lower prices. The hidden problem here is Microsoft is loath to lower prices, because doing so would hammer its stock price and lead to an inevtiable downward stock price spiral.
Microsoft is stuck between a rock and a hard place. High prices in the third world inevitably lead to piracy and an added incentive to move to Open Source software. Lowering prices means lower margins and a drastic reduction in market capitalization.
OO does not have to be better than MS Word. It has to be good enough at the right price to erode Word's installed base. Read Clay Christansen's book "The Innovator's Dilemma" to understand this phenomena.
The quid pro quo with AT&T was universal service and regulated rates. AT&T was not allowed to compete in the computer field until after the breakup of the company. In hindsight their computer marketing was so poor, there really was no worry about it monopolizing the computer field at it did telecommunications.
AT&T did exercise strong monopoly powers. I remember when it was illegal to hook up anything to the phone system. You had to lease your phone from AT&T. The phone device and all the wiring belonged to AT&T. To tamper with the phone or the wiring was illegal. This of course sounds awfully similar to some of the DRM legislation being pushed in Congress to forbid tampering with DVDs and other multimedia.
AT&T had the telecommunications strangelhold as a government regulated monopoly with at least a publically stated quid pro quo. It seems that Microsoft wants the benefits of monopoly power without any of that pesky government interference.
I hope legislators see through a lot of Microsoft's FUD and understand that a truly competitive playing field which includes FOSS software is the best environment for software innovation.
I would oppose any deal with Microsoft that limits competition. I do not think such a deal would serve the public interest.
For all my serious document creation, I use LyX ) which is a LaTeX GUI front-end. LyX does have its share of nits, but it does a good job of separating the logical layout from the formatting and still gives you a good visual representation of the eventual document layout.
I built a document control system using LyX as the document processor and had the back-end automatically generate PDF and HTML output files for Web publication with CVS version control thrown in for good measure.
Best of all you can use LyX without having to know a whole lot about LaTeX. So it is good for the LaTeX challenged user.
The GPL does not prevent forking, but it does discourage forking. Previous posters have explained this point quite well, so I will not repeat the arguments here.
Switching to a BSD license will just encourage code forking which is bad. We will just end up in another "Unix Wars" scenario with proprietary versions of GNU/Linux popping up with their own incompatible extensions.
The real beauty of the GPL is that it encourages collaboration and sharing with the side effect that forking is discouraged. GNU/Linux is becoming the commodity OS precisely because of the GPL.
I did a swapon of the root partition on a client's Solaris server about 10 years ago. It does not fail all at once, but gradually the root partition is overlayed with paging data. Commands fail in very strange ways.
I caught this after a few minutes, but the damage had already been done.Fortunately, the client had good system backups. I refunded them a day of my consulting time for the error.
And Tru64 (formerly OSF1). I am looking forward to a cluster filesystem where you can share the root disk. For a Unix system a Tru64 cluster is hard to beat for easy of management. It looks like a single system image.
Supposedly HP is going to port the Trucluster code to HP-UX. I think they would be better off porting it to Linux and GPLing it. Perhaps the Red Hat move will incent them to do so.
Price competition is but one factor in the European interest in OSS. The other is nationalism - they do not want to be beholden to a foreign company for their critical software needs.
The Unix vendors did fragment in the late 80s and early 90s because they were marketing computer systems and not an operating system. Most of the workstation vendors at the time such as Sun (and there were many more that failed) opted to use either a BSD or System 5 base for their OS development efforts. I believe this was driven by cost concerns. It was simply too expensive to write an OS from scratch. The resulting Unices did, unfortunately, diverge as a result.
I think the situation with Linux is different. With the Linux kernel as a freely distributed base, I am hoping both the legacy Unix vendors and the new Linux commercial vendors recognize the operating system is simply a tool to run applications and the vendors spend some effort revising and extending the POSIX standards.
Combined with the commodization of hardware (think x86 and its clones, ia64, and Power as the only viable computing options in a few years), Linux can become the neck in the hourglass of computing. It can become the standard mediation layer to the underlying hardware and present a common API to application software. This is a huge win to an ISV. They can write applications against a single common API and run it on anybody's hardware platform.
This is a completely different model than the one Microsoft wants. I think MS understands the implications of Linux as a commodity OS and that is why they are fighting so hard to prevent Linux from becoming more mainstream. MS wants to own the neck in the hourglass. If they do, then they can indirectly control the hardware layer and the application layer. If they lose the neck, then their business model quickly falls apart.
A key difference between OpenOffice and MSOffice is the file format. OO's is open. MS's is not. As a business or individual do you really want a third party dictating the file format of your critical business or personal information? It is interesting this was never mentioned in the evaluation as an issue.
Most people think MS's DOC format is a standard It is not and MS keeping it closed is the only way they maintain their Office monopoly. MS in effect has control over your information. It amazes me how so few business people 'get it' when it comes to this issue.
The real trend line Mr. Gates missed is the trend towards computing platforms being simply service delivery vehicles. Savvy business people are already moving to this model.
This is a difficult model for Microsoft to see, since it cuts at the heart of their business model which is based on the IP of their software. Some within MS do get it, but it is hard to turn a big ship.
The real challenge for the FOSS community is to recognize that in order to avoid vendor lockin there must be standard interface protocols for the evolving service delivery models. For example if the US or other governments (or large enterprises for that matter) would adopt OASIS as a common document storage standard, then vendor lockin for document management can be avoided.
The growth and adoption of the Internet is a good analogy in this regard. The Internet did not fragment as some predicted because interoperability was a key driver for the consumers of Internet services.
In the same way the adoption of key standards further up the protocol stack will be a brake to single vendor lockin. It will take the active participation of user to prevent the natural tendencies of software houses to lock customers into their products. What we as users should be demanding is service interoperability.
Just as I can use my Nokia cellphone to talk to someone using a Motorola cell phone, I should not be required to use MS Word in order to send a document to someone who happens to use MS Word.
Microsoft wants the world to be a monoculture where they control the software gene pool. This is very dangerous to the health of the software industry as a whole.
> Linux is built for the network more than printing
I found this point of view a rather interesting, if inaccurate, way to spin Linux versus Windows.
The author goes on to explain the advantage of a command line interface for creating automated tasks. However, the Unix CLI was not originally built with networking in mind. There was no networking in ancient Unix (unless you count UUCP).
As for Windows being designed around printers. I do not think that is correct either. I think the argument here is a bit weak.
Your comments are highly speculative and most mainstream historians with extensive access to Japanese, American and Russian archives from the period would not agree with your conclusions.
I suggest you read "The Last Great Victory" by Stanley Weintraub. It gives a very good and balanced account of the last days of WW II.
A few points:
Stalin knew about the atomic bomb from his spies within Los Alamos. Truman also told him about the bomb at Potsdam before they were dropped.
Stalin was asked by the US and Britain to invade Manchuria. There was great fear that the Japanese Army would not surrender in Manchuria. Only the Soviet Union had the logistical wherewithall to deal with that eventuality.
It is true that the Japanese cabinet was divided as to surrender in the waning days of WW II. We knew what was going on because of our interception of the Japanese diplomatic code.
It is likely in retrospect that the Navy's blockade of Japan would have worked and caused Japan's surrender. But the political realites of the time demanded all and any means be used to effect Japan's surrender. To Truman the atomic bomb was another tool to bring Japan to the surrender table.
As to the casualty figures for the invasion of Japan, most historians agree the figure of 1 million allied casualties is probably way too high. The actual figure for allied casualty estimates ranged as high as 265,000 with an estimate of 76,000 killed in action. This is still a very high figure and not much comfort to the troops already in the Pacific theater and those being transfered from Europe to the Pacific in preparation for Olympic (the invasion of Kyushu).
No estimate was given for Japanese casualties, although they would likely be far higher given the experience with the invasion of Okinawa. But be mindful that the primary focus of allied war planners was on allied casualties.
Even though this more realistic casualty figure is lower than the 1 million which has achieved an almost mythic dimension, it is still a large number and Truman was determined to use any means at his disposal to shorten the war and the casualty figure.
As for the use of the atomic bomb as a "demonstration" to deter Stalin, this is simply revisionist history. It is clear from the records of the time, the bombs were used to shorten the war against Japan, not to frighten the Soviet Union. We only had 2 bombs available in any event at the time. And we used both of them.
Some of your comments are right on target. The first time I saw Plan 9 (in the early 90s) I was very impressed. And I agree there is a lot in Unix which are hacks. But for the most part Unix has stood the test of time remarkably well and remains an elegant operating system.
You are a bit unfair to the BSD crowd and the networking API. Go back and read "The Design and Implementation of the 4.3BSD Unix Operating System." You will see some very specific design goals for the socket based network interface API, one of which was portability. They were also influenced by a TCP/IP implementation from Bolt Beranek and Newman which ran on 4.1 BSD. So the interface was not necessarily Unix specific.
I do agree that the streams based interface developed by AT&T is much more elegant and more in line with the Unix philosophy. But by the time streams was deployed, the BSD socket API dominated the field and became the standard API for TCP/IP networking.
As for sendmail, it certainly could have been better, but at the time it was trying to solve interoperability between very disparate email systems. Today the email routing problem is a lot simpler than it was in the early 80s. Back then DNS was in a nascent stage and you had to deal with UUCP, BITNET, CSNET and other mail networks in a very heterogeneous way. At the time, sendmail was the only MTA which could effectively deal with all those email systems within a single system.
Slashdot Mirror
I registered utexas.edu on August 13, 1985. At that time the DNS was just getting started. The old ARPANET used static host tables. For a while we had to support both systems which was a bit of a pain.
I saw the IMP at the University of Texas at Austin in 1982. It was a single roll around cabinent with a door and rubber seals.
I use rsync over the network to backup my laptop to my local server. I can even do it when I am on the road, since I tunnel rsync using SSH. I do daily gziped cpio archives of the snapshot nightly. So I have a history of changes.
I am glad I did this, since my hardrive on my notebook died last month. Only lost a day of data.
Exactly. A rational response to piracy would be to lower prices. The hidden problem here is Microsoft is loath to lower prices, because doing so would hammer its stock price and lead to an inevtiable downward stock price spiral.
Microsoft is stuck between a rock and a hard place. High prices in the third world inevitably lead to piracy and an added incentive to move to Open Source software. Lowering prices means lower margins and a drastic reduction in market capitalization.
OO does not have to be better than MS Word. It has to be good enough at the right price to erode Word's installed base. Read Clay Christansen's book "The Innovator's Dilemma" to understand this phenomena.
The quid pro quo with AT&T was universal service and regulated rates. AT&T was not allowed to compete in the computer field until after the breakup of the company. In hindsight their computer marketing was so poor, there really was no worry about it monopolizing the computer field at it did telecommunications.
AT&T did exercise strong monopoly powers. I remember when it was illegal to hook up anything to the phone system. You had to lease your phone from AT&T. The phone device and all the wiring belonged to AT&T. To tamper with the phone or the wiring was illegal. This of course sounds awfully similar to some of the DRM legislation being pushed in Congress to forbid tampering with DVDs and other multimedia.
AT&T had the telecommunications strangelhold as a government regulated monopoly with at least a publically stated quid pro quo. It seems that Microsoft wants the benefits of monopoly power without any of that pesky government interference.
I hope legislators see through a lot of Microsoft's FUD and understand that a truly competitive playing field which includes FOSS software is the best environment for software innovation.
I would oppose any deal with Microsoft that limits competition. I do not think such a deal would serve the public interest.
For all my serious document creation, I use LyX ) which is a LaTeX GUI front-end. LyX does have its share of nits, but it does a good job of separating the logical layout from the formatting and still gives you a good visual representation of the eventual document layout.
I built a document control system using LyX as the document processor and had the back-end automatically generate PDF and HTML output files for Web publication with CVS version control thrown in for good measure.
Best of all you can use LyX without having to know a whole lot about LaTeX. So it is good for the LaTeX challenged user.
The GPL does not prevent forking, but it does discourage forking. Previous posters have explained this point quite well, so I will not repeat the arguments here.
Switching to a BSD license will just encourage code forking which is bad. We will just end up in another "Unix Wars" scenario with proprietary versions of GNU/Linux popping up with their own incompatible extensions.
The real beauty of the GPL is that it encourages collaboration and sharing with the side effect that forking is discouraged. GNU/Linux is becoming the commodity OS precisely because of the GPL.
This behavior is illegal under the US anti-trust settlement. Probably one of the few good things in the settlement.
I did a swapon of the root partition on a client's Solaris server about 10 years ago. It does not fail all at once, but gradually the root partition is overlayed with paging data. Commands fail in very strange ways.
I caught this after a few minutes, but the damage had already been done.Fortunately, the client had good system backups. I refunded them a day of my consulting time for the error.
And Tru64 (formerly OSF1). I am looking forward to a cluster filesystem where you can share the root disk. For a Unix system a Tru64 cluster is hard to beat for easy of management. It looks like a single system image.
Supposedly HP is going to port the Trucluster code to HP-UX. I think they would be better off porting it to Linux and GPLing it. Perhaps the Red Hat move will incent them to do so.
Price competition is but one factor in the European interest in OSS. The other is nationalism - they do not want to be beholden to a foreign company for their critical software needs.
The Unix vendors did fragment in the late 80s and early 90s because they were marketing computer systems and not an operating system. Most of the workstation vendors at the time such as Sun (and there were many more that failed) opted to use either a BSD or System 5 base for their OS development efforts. I believe this was driven by cost concerns. It was simply too expensive to write an OS from scratch. The resulting Unices did, unfortunately, diverge as a result.
I think the situation with Linux is different. With the Linux kernel as a freely distributed base, I am hoping both the legacy Unix vendors and the new Linux commercial vendors recognize the operating system is simply a tool to run applications and the vendors spend some effort revising and extending the POSIX standards.
Combined with the commodization of hardware (think x86 and its clones, ia64, and Power as the only viable computing options in a few years), Linux can become the neck in the hourglass of computing. It can become the standard mediation layer to the underlying hardware and present a common API to application software. This is a huge win to an ISV. They can write applications against a single common API and run it on anybody's hardware platform.
This is a completely different model than the one Microsoft wants. I think MS understands the implications of Linux as a commodity OS and that is why they are fighting so hard to prevent Linux from becoming more mainstream. MS wants to own the neck in the hourglass. If they do, then they can indirectly control the hardware layer and the application layer. If they lose the neck, then their business model quickly falls apart.
A key difference between OpenOffice and MSOffice is the file format. OO's is open. MS's is not. As a business or individual do you really want a third party dictating the file format of your critical business or personal information? It is interesting this was never mentioned in the evaluation as an issue.
Most people think MS's DOC format is a standard It is not and MS keeping it closed is the only way they maintain their Office monopoly. MS in effect has control over your information. It amazes me how so few business people 'get it' when it comes to this issue.
The real trend line Mr. Gates missed is the trend towards computing platforms being simply service delivery vehicles. Savvy business people are already moving to this model.
This is a difficult model for Microsoft to see, since it cuts at the heart of their business model which is based on the IP of their software. Some within MS do get it, but it is hard to turn a big ship.
The real challenge for the FOSS community is to recognize that in order to avoid vendor lockin there must be standard interface protocols for the evolving service delivery models. For example if the US or other governments (or large enterprises for that matter) would adopt OASIS as a common document storage standard, then vendor lockin for document management can be avoided.
The growth and adoption of the Internet is a good analogy in this regard. The Internet did not fragment as some predicted because interoperability was a key driver for the consumers of Internet services.
In the same way the adoption of key standards further up the protocol stack will be a brake to single vendor lockin. It will take the active participation of user to prevent the natural tendencies of software houses to lock customers into their products. What we as users should be demanding is service interoperability.
Just as I can use my Nokia cellphone to talk to someone using a Motorola cell phone, I should not be required to use MS Word in order to send a document to someone who happens to use MS Word.
Microsoft wants the world to be a monoculture where they control the software gene pool. This is very dangerous to the health of the software industry as a whole.
> Linux is built for the network more than printing
I found this point of view a rather interesting, if inaccurate, way to spin Linux versus Windows.
The author goes on to explain the advantage of a command line interface for creating automated tasks. However, the Unix CLI was not originally built with networking in mind. There was no networking in ancient Unix (unless you count UUCP).
As for Windows being designed around printers. I do not think that is correct either. I think the argument here is a bit weak.
Your comments are highly speculative and most mainstream historians with extensive access to Japanese, American and Russian archives from the period would not agree with your conclusions.
I suggest you read "The Last Great Victory" by Stanley Weintraub. It gives a very good and balanced account of the last days of WW II.
A few points:
Stalin knew about the atomic bomb from his spies within Los Alamos. Truman also told him about the bomb at Potsdam before they were dropped.
Stalin was asked by the US and Britain to invade Manchuria. There was great fear that the Japanese Army would not surrender in Manchuria. Only the Soviet Union had the logistical wherewithall to deal with that eventuality.
It is true that the Japanese cabinet was divided as to surrender in the waning days of WW II. We knew what was going on because of our interception of the Japanese diplomatic code.
It is likely in retrospect that the Navy's blockade of Japan would have worked and caused Japan's surrender. But the political realites of the time demanded all and any means be used to effect Japan's surrender. To Truman the atomic bomb was another tool to bring Japan to the surrender table.
As to the casualty figures for the invasion of Japan, most historians agree the figure of 1 million allied casualties is probably way too high. The actual figure for allied casualty estimates ranged as high as 265,000 with an estimate of 76,000 killed in action. This is still a very high figure and not much comfort to the troops already in the Pacific theater and those being transfered from Europe to the Pacific in preparation for Olympic (the invasion of Kyushu).
No estimate was given for Japanese casualties, although they would likely be far higher given the experience with the invasion of Okinawa. But be mindful that the primary focus of allied war planners was on allied casualties.
Even though this more realistic casualty figure is lower than the 1 million which has achieved an almost mythic dimension, it is still a large number and Truman was determined to use any means at his disposal to shorten the war and the casualty figure.
As for the use of the atomic bomb as a "demonstration" to deter Stalin, this is simply revisionist history. It is clear from the records of the time, the bombs were used to shorten the war against Japan, not to frighten the Soviet Union. We only had 2 bombs available in any event at the time. And we used both of them.
Some of your comments are right on target. The first time I saw Plan 9 (in the early 90s) I was very impressed. And I agree there is a lot in Unix which are hacks. But for the most part Unix has stood the test of time remarkably well and remains an elegant operating system.
You are a bit unfair to the BSD crowd and the networking API. Go back and read "The Design and Implementation of the 4.3BSD Unix Operating System." You will see some very specific design goals for the socket based network interface API, one of which was portability. They were also influenced by a TCP/IP implementation from Bolt Beranek and Newman which ran on 4.1 BSD. So the interface was not necessarily Unix specific.
I do agree that the streams based interface developed by AT&T is much more elegant and more in line with the Unix philosophy. But by the time streams was deployed, the BSD socket API dominated the field and became the standard API for TCP/IP networking.
As for sendmail, it certainly could have been better, but at the time it was trying to solve interoperability between very disparate email systems. Today the email routing problem is a lot simpler than it was in the early 80s. Back then DNS was in a nascent stage and you had to deal with UUCP, BITNET, CSNET and other mail networks in a very heterogeneous way. At the time, sendmail was the only MTA which could effectively deal with all those email systems within a single system.