All of today's stories are being edited by kdawson and Zonk. (I guess the real editors are taking an extra-long weekend.) So expect lots of stories that make you scratch your head.
Jeez people, when you give us a story, please bear in mind that many of us have never heard of the stuff you're talking about. I had to puzzle for 5 minutes to figure out that a "win32-loader" is Linux installer that runs under Windows. ("Loader" is a really stupid choice of words, but that's another issue.) Yeah, yeah, if my time is that valuable, I shouldn't be wasting it on Slashdot. But damn it Rob, can't you find some editors with some basic communication skills?
OK, I stand corrected. The isolation problem was IBM's fault, not Microsoft's. (Another sordid fact: IBM's inferior design became a de-facto standard simply because it was IBM's.) Obviously IBM thought that they could get away with this design because they planned to migrate to more carefully thought out systems later. What they didn't grasp was that they had invented the first commodity computer and that its APIs had created a degree of lockin that not even they could fight. Hence the failure of the PS/2.
On the other hand, Microsoft did make those nasty mistakes in the design of MS-DOS. Or rather, they made the mistake of basing MS-DOS on QDOS, an "OS" designed by a guy who had no idea what an OS actually is! This mistake created problems that persisted to the very last version of DOS-based Windows, which was only EOLed last year!
And whether the mistakes were Microsoft's or IBM's they created a degree of lock-in that made Bill the richest man on the planet. Not superior technology, not even undersellng the competition, just a matter of making the right mistakes! Ultimate proof that it's better to be born lucky than smart.
I work on the hardware side, so I'm in no position to address that issue. I am curious as to what you mean by "native" authentication.
In any case, I suspect "yesterday" is not feasible. There's lots of stuff Sun should have done, but just at the moment we're concentrating on what we will do.
The author of that article knows jack. We (I'm the documentation lead for a couple of Sun x64 boxes) have been selling and supporting Windows servers for some time. We have a fair number of people working on Windows-related software, QA, support, and documentation (including me). We've even contributed some source code to a couple of open-source products in order to make them work better on Windows.
What we haven't been doing is selling servers with Windows pre-installed, or providing install discs with our drivers already on them. We couldn't do these things without an OEM agreement. Now we can. That will mean less work for me and various other Sun people, and (much more important) fewer headaches for our customers.
Next time I see Jonathan Schwartz (no, we don't know each other, but we eat in the same cafeteria) I'll have to resist the urge to prostrate myself. I just hope he's working on similar deals with our other OS partners.
Don't get me wrong, I love Solaris. It's a beautiful OS. We'll always support it. (In fact, the x86/x64 version is a lot better supported than it was 8 years ago.) But our job is to meet our customers needs, not force our favorite technology down their throat.
Get it through your heads, folks: the Sun-Microsoft feud is over. And good riddance. It was bad for both companies.
Microsoft didn't grab its market share by low pricing. Microsoft didn't even set out to be an OS vendor. By pure blind luck, they ended up owning the officially supported OS for the IBM PC. (Bill Gates himself advised IBM to use Digital Research's CP/M, which was the leading micro OS at the time; only DR's unwillingness to sign the necessary NDAs prevented this.) That meant that any computer that pretended to be IBM-compatible had to run MS-DOS. That's the "sordid reality" I was referring to.
I don't know why QNX was so expensive. Perhaps they didn't understand the new mass market for computers. Or maybe they just realized that they couldn't compete head-to-head with Microsoft, and decided to go for niche markets that needed more functionality than MS-DOS could provide. Either way, Microsoft didn't triumph just because of its prices: there have always been plenty of low-cost alternatives to MS-DOS and Windows.
I think you've got it backwards. Everyone here is a believer in RMS and RMS is a believer in microkernels. Ergo, microkernels are kewl.
Perhaps you're thinking of the Tanenbaum-Torvalds flamefest, where Tanenbaum argued that the Linux kernel was obsolete from day 1, because it was monolithic. So if you believe that Linus is God (or at least the Flying Spaghetti Monster), then you hate microkernels. On the other hand, monolithic kernels are old technology and microkernels are (relatively) new technology; so no self-respecting geek will have anything to do with monolithic kernels or other paleolithic technology!
It's a very confusing problem. Of course, you can always cop out and say, "I use what works." But what's the fun of that?
I find the history of QNX very frustrating. When I first heard of it in the mid 80s, it was advertised as a simple Unix-like OS with very low hardware requirements. It was network-aware, supported distributed computing, and had a nice microkernel architecture.
But the most important thing was that it was a real OS, with the ability to multitask and to effectively isolate hardware from software. Contrast this with MS-DOS 3.0, which had only the most primitve, kludgy excuse for background processing. (Patterson knew zilch about os design when he set out to clone CP/M; it never occurred to him that OS code needed to be reentrant. And MS-DOS did a really lousy job of isolating hardware from software. Ironically, this fuckup assured lockin of the IBM-compatible/PC combination: software written for this platform was essential impossible to port to other platforms.
What was particularly tantalizing was that QNX claimed to run well even on very limited hardware — even 8088 systems were said to run robustly. And it shared some key features with CTOS an first-rate OS that was then dying off, due to its dependence on proprietary hardware.
The problem with QNX was that commercial license fees were very high; that's why I never played with it. It did become popular at universities (cheap academic licenses) and among certain kinds of embedded application developers (because of its nice feature set and minimal hardware requirements. I'm told that by the late 80s, most video stores used POS systems based on QNX.
Then MS-DOS/Windows started grabbing more and more of the market and QNX was forced to specialize. So for a long time now they've advertised themselves as a real-time operating system. And yes, their real-time features are very good — but they're just one part of a really good general-purpose OS.
Now, much too late to do me any good, there's an open-source version of QNX. I wish the QNX OSS community well, but there's just no place for it in the world I work in. Hopefully, embedded application developers will keep QNX alive. But I'll always be sad that QNX never found a following among common PC users — which it surely would have if the marketplace were driven by technical excellence instead of various sordid realities. This is one of the great lost opportunities in computing history. And should be a lesson to Linux advocates who think they can easily displace Microsoft.
Second Life does claim to give you another life. That's the whole point of the name. No, they're not promising to resurrect you when you die. But "life" has other meanings. As in "get a life" — they're getting you a second one.
ajaxWindows isn't an OS, not even a "virtual OS" (whatever that means) because it doesn't do most of the things that an OS does. It's a GUI, and a GUI is just one tiny part of an OS, even though it's often the most conspicuous.
I remember when MUDs were all the rage, people who were into them called them "virtual reality games". (VR was also big, and had more pop culture presence.) Now, a MUD shares does share some features with VR (both are virtual, and both are about fake reality), but anybody who really knows what VR is would find the comparison is laughable. By the same token, calling ajaxWindows an "operating system" is laughable to anybody who knows what an OS is, even if most people don't know the difference between a GUI and an OS.
This isn't about free parking. (If it were, they'd be talking about more than 3 meters!) This is about Apple not wanting its pretty, high-tech store sullied by the proximity of ugly, low-tech parking meters. Typical Apple. First because they care so much about looking kewl (their packaging has more fancy "design" vibe than most products). Second because they're willing to spend a lot of money to get that kewlness: they're notorious for buying off (at great expense) people who own the code names they want to use — or claim to.
OK, you're not an idiot, you're just reading out of context. The subthread started when I stated that VB skills weren't transferrable to Borland Pascal because the languages and their programming culture were very different. Someone said, no, Pascal, C, Fortran, and Ada are very similar. I responded with examples that differentiated each language, and mentioned delegates and OOP in connection with Borland Pascal. There was no suggestion that Borland was first or only with delegates and OOP in Pascal; I don't know about delegates, but I've used two other Pascal dialects that had objects.
No wait a minute. (Looks over thread.) First you debunked my "claim" that Borland Pascal was the only Pascal with objects. (Didn't say that.) Then you debunked my "claim" that Apple Pascal didn't do objects. (Didn't say that either.) There might be something to your third accusation, but it's a little stupid to jump through so many logical hoops just to avoid admitting that you misread my original post.
This was a marketing ploy that always bothered me. I was the only one at Borland who actually had serious VB experience, and the notion that skills in that environment were transferable to Object Pascal was absurd.
I couldn't get simple projects I tried out myself to run without it crashing on a regular basis. I would have stuck with it, but Borland gave up even faster than I did.
Your experience was different from mine. I'm very curious what was different about your system, but I don't suppose there's any way of figuring it out.
I don't think Borland gave up all the quickly. We went through three full development cycles, and even did a C++ version. The product didn't so much end early as start late. The 1.0 release was many months behind schedule, mainly because Borland's developers just didn't believe in schedules.
A big problem I see with running an audit service is critical mass, doing financial checkups takes time/money and some types of audit require either a lot of cooperation from the company being audited or the use of force.
Good point. On the other hand, there's lots of details that are easy to find out that would blow away a lot of smoke. For example, if you go to the web site of most hosting/colo providers, you'll see pictures and descriptions of of "their" facility, designed to give the impression of a big, stable company. In reality, the pictures and descriptions are of the the data center in which the provider rents a few racks, and the provider is often just one guy with a pager.
As you point out, it's hard to squeeze financial data out of people. But more accessible data like head count can tell you a lot about what kind of company you're dealing with.
Also, if customers start taking the audit concept seriously, providers will start cooperating to avoid losing business.
But now that I think about it, customers are never going to take the audit concept seriously. Because most customers are geeks, and that just isn't part of geek culture.
but the hopes of catching a single, solitary nuclear device on the move is a crapshot at best.
Not even a crapshoot. A nuke isn't a terribly distinctive object.
But you're not looking for a solitary nuke. You've looking for the huge effort that goes into manufacturing and testing the thing. Things like materiel going into suspected labs and factories, or the local army cordoning off the test site. That's a lot harder to hide.
Not impossible, of course. But it's not easy, either. And knowing when the spy satellite passes over makes all the difference.
I don't see the pervasive corruption you do. Yes we've had a lot scandals, but mostly among political appointees. Besides, scandals are a good sign: they show that people give a shit.
Even so, an ISP audit bureau should probably come from the private sector. When it comes to deciding on "best practices", you don't want the politics and bureaucracy of a government agency.
The question is, how does it get started? I was tempted to try something after I left that ISPs like a web site with questionnaires filled out voluntarily by the ISPs. More rigorous methods could he employed as the operation grew.
But I'm the wrong person to run something like this. Inferior business and organization skills. Anyone?
Well, it worked for me. And for our QA people. And for the testers. And most of the customers, judging from the feedback we got. Care to share what you mean by "couldn't get it to work"?
Of course, it really was a doomed product. Not because it didn't work, but because there was no demand for desktop Linux apps, and thus no demand for a Linux IDE.
Uh, dude, Pascal, C, Fortran, and Ada are very different languages. Borland's Pascal is the only one of these with delegates and built in OOP support. C is the only one with type-aware pointers. Fortran is a relic of the 50s. And Ada is a world onto itself.
They may be more like each other than Lisp, Forth, or Haskell (they're all basically procedural for one thing) but they're still very different and have very different programmer cultures. That's why only one language on your list (C) is still widely used.
Slashdot Mirror
All of today's stories are being edited by kdawson and Zonk. (I guess the real editors are taking an extra-long weekend.) So expect lots of stories that make you scratch your head.
Jeez people, when you give us a story, please bear in mind that many of us have never heard of the stuff you're talking about. I had to puzzle for 5 minutes to figure out that a "win32-loader" is Linux installer that runs under Windows. ("Loader" is a really stupid choice of words, but that's another issue.) Yeah, yeah, if my time is that valuable, I shouldn't be wasting it on Slashdot. But damn it Rob, can't you find some editors with some basic communication skills?
OK, I stand corrected. The isolation problem was IBM's fault, not Microsoft's. (Another sordid fact: IBM's inferior design became a de-facto standard simply because it was IBM's.) Obviously IBM thought that they could get away with this design because they planned to migrate to more carefully thought out systems later. What they didn't grasp was that they had invented the first commodity computer and that its APIs had created a degree of lockin that not even they could fight. Hence the failure of the PS/2.
On the other hand, Microsoft did make those nasty mistakes in the design of MS-DOS. Or rather, they made the mistake of basing MS-DOS on QDOS, an "OS" designed by a guy who had no idea what an OS actually is! This mistake created problems that persisted to the very last version of DOS-based Windows, which was only EOLed last year!
And whether the mistakes were Microsoft's or IBM's they created a degree of lock-in that made Bill the richest man on the planet. Not superior technology, not even undersellng the competition, just a matter of making the right mistakes! Ultimate proof that it's better to be born lucky than smart.
I work on the hardware side, so I'm in no position to address that issue. I am curious as to what you mean by "native" authentication.
In any case, I suspect "yesterday" is not feasible. There's lots of stuff Sun should have done, but just at the moment we're concentrating on what we will do.
The author of that article knows jack. We (I'm the documentation lead for a couple of Sun x64 boxes) have been selling and supporting Windows servers for some time. We have a fair number of people working on Windows-related software, QA, support, and documentation (including me). We've even contributed some source code to a couple of open-source products in order to make them work better on Windows.
What we haven't been doing is selling servers with Windows pre-installed, or providing install discs with our drivers already on them. We couldn't do these things without an OEM agreement. Now we can. That will mean less work for me and various other Sun people, and (much more important) fewer headaches for our customers.
Next time I see Jonathan Schwartz (no, we don't know each other, but we eat in the same cafeteria) I'll have to resist the urge to prostrate myself. I just hope he's working on similar deals with our other OS partners.
Don't get me wrong, I love Solaris. It's a beautiful OS. We'll always support it. (In fact, the x86/x64 version is a lot better supported than it was 8 years ago.) But our job is to meet our customers needs, not force our favorite technology down their throat.
Get it through your heads, folks: the Sun-Microsoft feud is over. And good riddance. It was bad for both companies.
When they say "minimal hardware" they mean "minimal hardware"!
Microsoft didn't grab its market share by low pricing. Microsoft didn't even set out to be an OS vendor. By pure blind luck, they ended up owning the officially supported OS for the IBM PC. (Bill Gates himself advised IBM to use Digital Research's CP/M, which was the leading micro OS at the time; only DR's unwillingness to sign the necessary NDAs prevented this.) That meant that any computer that pretended to be IBM-compatible had to run MS-DOS. That's the "sordid reality" I was referring to.
I don't know why QNX was so expensive. Perhaps they didn't understand the new mass market for computers. Or maybe they just realized that they couldn't compete head-to-head with Microsoft, and decided to go for niche markets that needed more functionality than MS-DOS could provide. Either way, Microsoft didn't triumph just because of its prices: there have always been plenty of low-cost alternatives to MS-DOS and Windows.
Now that is hard to accept, since very few Slashdotters seem to know about the <blockquote> tag.
I think you've got it backwards. Everyone here is a believer in RMS and RMS is a believer in microkernels. Ergo, microkernels are kewl.
Perhaps you're thinking of the Tanenbaum-Torvalds flamefest, where Tanenbaum argued that the Linux kernel was obsolete from day 1, because it was monolithic. So if you believe that Linus is God (or at least the Flying Spaghetti Monster), then you hate microkernels. On the other hand, monolithic kernels are old technology and microkernels are (relatively) new technology; so no self-respecting geek will have anything to do with monolithic kernels or other paleolithic technology!
It's a very confusing problem. Of course, you can always cop out and say, "I use what works." But what's the fun of that?
I find the history of QNX very frustrating. When I first heard of it in the mid 80s, it was advertised as a simple Unix-like OS with very low hardware requirements. It was network-aware, supported distributed computing, and had a nice microkernel architecture.
But the most important thing was that it was a real OS, with the ability to multitask and to effectively isolate hardware from software. Contrast this with MS-DOS 3.0, which had only the most primitve, kludgy excuse for background processing. (Patterson knew zilch about os design when he set out to clone CP/M; it never occurred to him that OS code needed to be reentrant. And MS-DOS did a really lousy job of isolating hardware from software. Ironically, this fuckup assured lockin of the IBM-compatible/PC combination: software written for this platform was essential impossible to port to other platforms.
What was particularly tantalizing was that QNX claimed to run well even on very limited hardware — even 8088 systems were said to run robustly. And it shared some key features with CTOS an first-rate OS that was then dying off, due to its dependence on proprietary hardware.
The problem with QNX was that commercial license fees were very high; that's why I never played with it. It did become popular at universities (cheap academic licenses) and among certain kinds of embedded application developers (because of its nice feature set and minimal hardware requirements. I'm told that by the late 80s, most video stores used POS systems based on QNX.
Then MS-DOS/Windows started grabbing more and more of the market and QNX was forced to specialize. So for a long time now they've advertised themselves as a real-time operating system. And yes, their real-time features are very good — but they're just one part of a really good general-purpose OS.
Now, much too late to do me any good, there's an open-source version of QNX. I wish the QNX OSS community well, but there's just no place for it in the world I work in. Hopefully, embedded application developers will keep QNX alive. But I'll always be sad that QNX never found a following among common PC users — which it surely would have if the marketplace were driven by technical excellence instead of various sordid realities. This is one of the great lost opportunities in computing history. And should be a lesson to Linux advocates who think they can easily displace Microsoft.
Bored now.
Second Life does claim to give you another life. That's the whole point of the name. No, they're not promising to resurrect you when you die. But "life" has other meanings. As in "get a life" — they're getting you a second one.
ajaxWindows isn't an OS, not even a "virtual OS" (whatever that means) because it doesn't do most of the things that an OS does. It's a GUI, and a GUI is just one tiny part of an OS, even though it's often the most conspicuous.
I remember when MUDs were all the rage, people who were into them called them "virtual reality games". (VR was also big, and had more pop culture presence.) Now, a MUD shares does share some features with VR (both are virtual, and both are about fake reality), but anybody who really knows what VR is would find the comparison is laughable. By the same token, calling ajaxWindows an "operating system" is laughable to anybody who knows what an OS is, even if most people don't know the difference between a GUI and an OS.
This isn't about free parking. (If it were, they'd be talking about more than 3 meters!) This is about Apple not wanting its pretty, high-tech store sullied by the proximity of ugly, low-tech parking meters. Typical Apple. First because they care so much about looking kewl (their packaging has more fancy "design" vibe than most products). Second because they're willing to spend a lot of money to get that kewlness: they're notorious for buying off (at great expense) people who own the code names they want to use — or claim to.
OK, you're not an idiot, you're just reading out of context. The subthread started when I stated that VB skills weren't transferrable to Borland Pascal because the languages and their programming culture were very different. Someone said, no, Pascal, C, Fortran, and Ada are very similar. I responded with examples that differentiated each language, and mentioned delegates and OOP in connection with Borland Pascal. There was no suggestion that Borland was first or only with delegates and OOP in Pascal; I don't know about delegates, but I've used two other Pascal dialects that had objects.
No wait a minute. (Looks over thread.) First you debunked my "claim" that Borland Pascal was the only Pascal with objects. (Didn't say that.) Then you debunked my "claim" that Apple Pascal didn't do objects. (Didn't say that either.) There might be something to your third accusation, but it's a little stupid to jump through so many logical hoops just to avoid admitting that you misread my original post.
Sorry, you are an idiot after all.
Jeez, learn to read. That's your second "response" to something I didn't say.
Amusing that my experience has such a direct parallel. But no, I was working at Hurricane Electric.
I don't think Borland gave up all the quickly. We went through three full development cycles, and even did a C++ version. The product didn't so much end early as start late. The 1.0 release was many months behind schedule, mainly because Borland's developers just didn't believe in schedules.
As you point out, it's hard to squeeze financial data out of people. But more accessible data like head count can tell you a lot about what kind of company you're dealing with.
Also, if customers start taking the audit concept seriously, providers will start cooperating to avoid losing business.
But now that I think about it, customers are never going to take the audit concept seriously. Because most customers are geeks, and that just isn't part of geek culture.
But you're not looking for a solitary nuke. You've looking for the huge effort that goes into manufacturing and testing the thing. Things like materiel going into suspected labs and factories, or the local army cordoning off the test site. That's a lot harder to hide.
Not impossible, of course. But it's not easy, either. And knowing when the spy satellite passes over makes all the difference.
I don't see the pervasive corruption you do. Yes we've had a lot scandals, but mostly among political appointees. Besides, scandals are a good sign: they show that people give a shit.
Even so, an ISP audit bureau should probably come from the private sector. When it comes to deciding on "best practices", you don't want the politics and bureaucracy of a government agency.
The question is, how does it get started? I was tempted to try something after I left that ISPs like a web site with questionnaires filled out voluntarily by the ISPs. More rigorous methods could he employed as the operation grew.
But I'm the wrong person to run something like this. Inferior business and organization skills. Anyone?
Well, it worked for me. And for our QA people. And for the testers. And most of the customers, judging from the feedback we got. Care to share what you mean by "couldn't get it to work"?
Of course, it really was a doomed product. Not because it didn't work, but because there was no demand for desktop Linux apps, and thus no demand for a Linux IDE.
Uh, dude, Pascal, C, Fortran, and Ada are very different languages. Borland's Pascal is the only one of these with delegates and built in OOP support. C is the only one with type-aware pointers. Fortran is a relic of the 50s. And Ada is a world onto itself.
They may be more like each other than Lisp, Forth, or Haskell (they're all basically procedural for one thing) but they're still very different and have very different programmer cultures. That's why only one language on your list (C) is still widely used.