you'll trust them that much eh? I trust backstabbing friends LESS than honest enemies myself!
Between this and the crappy Ultra 10 hardware that all seem to break down at about 12-18 months old, Sun isn't the same. It's not superfast, reliable, slick Solaris and Sparc. It's bloated, slow, unreliable Java and Hype!
Of course, I shed a single tear at this unhappy change of events. Just one.
you can read ESR stuff (like the OSI paper) for your own edification. If you give it to anyone else, it might backfire, that's because while he can make a good point, ESR is crazy and this has a way of coming through in his writing. Kind of like the software engineer you know is valuable for his skills, but whome has to be seated far away from everyone else because they don't understand the tourettes and libertarian mania.
But I love my eccentric geek luminaries! I'm just saying.
basically Stallman is right in that the OS was supposed to be called GNU. He's wrong in that Linus didn't make GNU... how could he? That's the Stallman/FSF project.
So stallman, I believe you, the OS is GNU... not GNU/Linux, not Linux, just GNU.
So make a distribution and call it GNU! That was your plan, finish it! If you can't stand having a linux kernel in GNU, use the HURD. If the HURD isn't ready, GNU isn't ready. The world has not seen GNU, just a bunch of the parts of GNU. If you want people to understand the kernel in a given version of GNU, go ahead and call it GNU/Linux... but for god's sake remember you don't have to tell people what to call THEIR distribution! It's free and in liberation software, make the distro yourself and call it what you want.
Honestly, I think there is some kind of mental block here. Why doesn't the FSF have a distro? They are envious AND too good to use the Linux kernel? I don't know, but it's something psychological because as a practical matter it would be nothing for them to roll a distro and call it GNU, according to the original plan (more or less).
In fact, now would be a good time since the FSF is MUCH more careful about copyright than any other OSS effort, clearly, and that could be a serious value-add for a linux distro.
The most amazing part is where he convinced the OS really is called GNU. Not GNU/Linux, but GNU. GNU with a linux kernel. GNU with the HURD kernel, whatever. The OS is GNU.
The GNU OS. He's right and it's unfscking believable! Right again, you dastardly villian.
NEXT UP: RMS proves right in the xemacs vs. emacs split... you really do need to collect your copyrights carefully if you ever hope to enforce the GPL, and if not... just use public domain in the first place! He wasn't paranoid... ok, well, he was... but he was right, it really works like that.
I don't think it's possible to pervert your code, that's put so emotionally, and of course it'll be used as intended unless it's so screwed up I have to reverse engineer it to get things to work, something C++ would let me do.
You can learn best practices. The language doesn't have to know them.
The create and destroy (new/delete, malloc/free) are important events, programming is all about placing them. It's analagous to opening and closing a context. Would there be a movement to obliterate the ending brace? Remember where you open and close, where you create and destroy, is fundamental, that is, you WANT to think about that in order to get the implimentation right.
Similarly bad moves seem to have been made about synchronization. I cannot think of a single other thing harder to generically solve that optimized synchronization. Hiding that in objects and letting that be entirely under the covers is not good. I want synchronization methods besides mutex, essentially. And if someone would like to point me at the other methods, I'll thank them and merely say, well there, that's my point, they're needed.
I think it's as if two thing collided when it comes to Java performance. (1) it's as if they thought Moore's Law would just come to the rescue. (2) they relied to heavily on that old canard of always optimize last.
(1) someone need to do the math, "When will machines run this design fast if computers double their capacities every 18 months?" scribble scribble "Five Thousand Years".
(2) You can't optimize design decisions. You have to anticipate what you will want to optimize.
I'm not believing that.NET is any better until I see it. VM software works for limited domains, why? Because the VM is the program. A virtual machine to run Quake C in Quake can be high speed, because Quake does all the work. Sierra's old SCI language was fast enough to do then-intensive graphics and sound because it too understood the domain the "scripts" were to work in. Java as an applet language IS fast enough. Little applets, run fine. It makes sense to program devices like the proverbial digital fridge and toaster, so that developers are learning one language to program all these devices. Java is overgrown as a general purpose idea.
Compiled languages: if you have a problem with them, fix them, they are the real languages. The VM -is- the only program, you are configuring that program. You configure the VM to do what you want. That's why it's hard to make a VM that can perform, it has to anticipate every possible desire. With a compiled language the program is in fact constructed directly to the domain, you get exactly what you put into the program and use.
it's not that easy to make a VM on a chip that'll be any faster. There are design issues. The Virtual Machine is not like a Real Machine. Software engineers know what a machine should be like! Those pesky hardware engineers have to deal with the laws of physics, imagine the indignity!
I think it's more about the natural process of commoditization.
You can't charge 150% markup on Thyme and Sage forever, you know.
Also, most software engineering takes place for people that use software, not people that sell it. Those people have good reason to share a bedrock of tools.
But your average whiney capitalist think they ought to collect six figures for software written in 1985 just forever --- that's the fair thing. Oh well, it's a battle, you'll win or lose and right and wrong will have little to do with it.
>That indicates a failure of the patch-based open source process. CVS couldn't be fixed within the process; it was necessary to start a new project and rewrite.
it doesn't, this is it's advantage. With commercial software you write the same solution over and over. I cannot use a solution I've written somewhere else, so I have to start from scratch. Instead of take a system to it's logical conclusion and limits, I make the same system again, because at least it's known what I'll get when I've built it.
With OSS systems, you follow a system to it's logical conclusion. Then, only when the design is exhausted, do you need to start over. You only replace a system when you need to. In commercial software you are always reproducing systems.
For nearly all my carreer I've used a lot of commercial tools, and they are simply not as reliable in the long haul as free software is for these sorts of reasons.
If we get people used to this, to heck with all of you... I'll be able to create a portable library that I can use the rest of my life wherever I work, whatever I do! sounds good to me.
yes, except that with royalty generating IP in linux the GPL doesn't apply. I would think they would have to stop using it, but if not literally, in effect, as they could not legally make a copy to, e.g., install on a new server.
But anyway, I agree, it's SCO that is the clear violator and legitimate target for GPL based actions.
What it actually says is that the only thing that gives you the ongoing right to duplicate other people's IP is to ship with the GPL license attached. It doesn't say that you have to mean it, nor that you can't stop shipping (thereby stopping your own violation of other peoples' rights, and SCO are slowly getting there) and then renege on your GPL license and prosecute the duplication and distribution of your copyrighted code.
yes of course, I agree, you can not mean it, and then GPL doesn't apply. You have to stop and have a good story about accidently shipping it in the first place.
The topic of this story was SCO further sanctioning the use of linux by their own customers. They know they have no license now by which to have given those customers Linux, but they are not recalling those distributions, they are actively promoting that their customers keep Linux, promising support, and standing behind what they have done.
What they have done accidently or by design is ship their code under the GPL. So in other words, by sanctioning what they did, even if they stop further shipments, they have said loud and clear, "We DO mean it... our IP, under the GPL, merged with linux, boot up and be merry".
It's as if there is a mind bubble and some people do not realize the other linux code is not public domain, that it's fully copyrighted material. Not just any hypothetical AT&T code in there, but all the other stuff, the other parts of the kernel that linus or whomever added.
SCO must be sued by OSS copyright holders. Specifically those into which SCO code is currently merged.
You have a point, but the fact is, business expect you to sue. If you use Sun or MS IP against their license, they WILL sue. And most businessmen are aware that actually, companies like Sun and Microsoft are constantly in litigation as plaintiffs and defendants, so I don't think you are right.
In fact, I think if they don't sue over these violations, they will lose credibility and the GPL will be used like public domain. It must be enforced for a business person to take it seriously.
You make a good point about the IBM v. SCO nature not involving the GPL, I'm thinking about the fact that SCO has GPLed their code however.
They distributed it under GPL.
That can be inadvertent, so they immediately cease shipping linux. By indemnifying their customers, however, they have said to the court "make no mistake about it, we know we've shipped our code with GPL code, and please understand we are fine with that, we have told our customers this is fine." Right. Assuming their lawyers have read the GPL, their code is now GPLed.
OR: they are in violation of the GPL and can be sued. It's not a part of the IBM case, but it could be, there is no reason that IBM, as a linux copyright holder cannot counter sue SCO now.
Sorry if that wasn't clear. I agree the post you were replying too was the most wrong of all, implying that using SCO at anytime would indemnify you from getting sued, when clearly they mean "we won't sue you for using SCO IP that we sold you."
but let's assume we're back to an honest disagreement. You understand and I understand but we disagree on interpretation and on our predictions.
My point is that it hardly matters how many slashdotters would buy an MS linux, but all the Microsoft customers that would.
Microsoft has decided to fight. I'm sure there is a line where they will go with it (originally they thought they would fight the internet too! (ps: I was an original MSN beta tester)), but up to that point they are fighting. I'm hoping they miscalculate their timing because if MS made a distribution and pushed it, other distro's would HAVE to be compatible.
But one other thing: Microsoft cannot destroy linux, that is what they are trying to do. But Microsoft COULD embrace and extend linux, or at least I think so. They havn't because it's much more lucrative to destroy it (except for the impossible part).
You seem to forget, even if SCO is a copyright holder (doubtful at the moment), they are not the only copyright holder! If you lose Linux because the GPL doesn't hold --- SCO CAN'T GIVE IT BACK TO YOU.
It's not clear to me if the customers can be sued, but SCO certainly can, they are knowingly and wontonly violating the license on other people's IP, which is what these kind of cases hinge on, the intentionality of it.
The GPL specifically states that if you ship the code, the GPL applies, and that if you ship your proprietary IP it has to be available for a royalty free non-terminable licence.
So SCO is clearly in violation of the GPL to be doing this.
Further, since it's distributed under the GPL and they have confirmed this now, their IP can be redistributed, as the GPL allows, by their customers.
I do not share you confidence in SCO's lawyers, so I'll call your SCO lawyers and rais you a pack of IBM lawyers.
Slashdot Mirror
you'll trust them that much eh? I trust backstabbing friends LESS than honest enemies myself!
Between this and the crappy Ultra 10 hardware that all seem to break down at about 12-18 months old, Sun isn't the same. It's not superfast, reliable, slick Solaris and Sparc. It's bloated, slow, unreliable Java and Hype!
Of course, I shed a single tear at this unhappy change of events. Just one.
you can read ESR stuff (like the OSI paper) for your own edification. If you give it to anyone else, it might backfire, that's because while he can make a good point, ESR is crazy and this has a way of coming through in his writing. Kind of like the software engineer you know is valuable for his skills, but whome has to be seated far away from everyone else because they don't understand the tourettes and libertarian mania.
But I love my eccentric geek luminaries! I'm just saying.
basically Stallman is right in that the OS was supposed to be called GNU. He's wrong in that Linus didn't make GNU... how could he? That's the Stallman/FSF project.
So stallman, I believe you, the OS is GNU... not GNU/Linux, not Linux, just GNU.
So make a distribution and call it GNU! That was your plan, finish it! If you can't stand having a linux kernel in GNU, use the HURD. If the HURD isn't ready, GNU isn't ready. The world has not seen GNU, just a bunch of the parts of GNU. If you want people to understand the kernel in a given version of GNU, go ahead and call it GNU/Linux... but for god's sake remember you don't have to tell people what to call THEIR distribution! It's free and in liberation software, make the distro yourself and call it what you want.
Honestly, I think there is some kind of mental block here. Why doesn't the FSF have a distro? They are envious AND too good to use the Linux kernel? I don't know, but it's something psychological because as a practical matter it would be nothing for them to roll a distro and call it GNU, according to the original plan (more or less).
In fact, now would be a good time since the FSF is MUCH more careful about copyright than any other OSS effort, clearly, and that could be a serious value-add for a linux distro.
If there is something more petty than Stallman wanting the OS to be called GNU like he originally named it... this plan you outline is it.
> a car without an engine or a transmission isn't a car.
a car without a transmission IS a car. It won't drive, but it's still a car. One that needs a transmission.
The most amazing part is where he convinced the OS really is called GNU. Not GNU/Linux, but GNU. GNU with a linux kernel. GNU with the HURD kernel, whatever. The OS is GNU.
The GNU OS. He's right and it's unfscking believable! Right again, you dastardly villian.
NEXT UP: RMS proves right in the xemacs vs. emacs split... you really do need to collect your copyrights carefully if you ever hope to enforce the GPL, and if not... just use public domain in the first place! He wasn't paranoid... ok, well, he was... but he was right, it really works like that.
Of course, xemacs is still better. But still.
I don't think it's possible to pervert your code, that's put so emotionally, and of course it'll be used as intended unless it's so screwed up I have to reverse engineer it to get things to work, something C++ would let me do.
You can learn best practices. The language doesn't have to know them.
It does not and cannot enforce best practices!
In fact, I think that it's likely no language can -enforce- best practices. JNI proves that even attempts at censorship break down.
it was meant to be better than C++ (and honestly, how hard is that!?)
pardong me!?
sir, a duel... at dawn!
absolutely.
.NET is any better until I see it. VM software works for limited domains, why? Because the VM is the program. A virtual machine to run Quake C in Quake can be high speed, because Quake does all the work. Sierra's old SCI language was fast enough to do then-intensive graphics and sound because it too understood the domain the "scripts" were to work in. Java as an applet language IS fast enough. Little applets, run fine. It makes sense to program devices like the proverbial digital fridge and toaster, so that developers are learning one language to program all these devices. Java is overgrown as a general purpose idea.
The create and destroy (new/delete, malloc/free) are important events, programming is all about placing them. It's analagous to opening and closing a context. Would there be a movement to obliterate the ending brace? Remember where you open and close, where you create and destroy, is fundamental, that is, you WANT to think about that in order to get the implimentation right.
Similarly bad moves seem to have been made about synchronization. I cannot think of a single other thing harder to generically solve that optimized synchronization. Hiding that in objects and letting that be entirely under the covers is not good. I want synchronization methods besides mutex, essentially. And if someone would like to point me at the other methods, I'll thank them and merely say, well there, that's my point, they're needed.
I think it's as if two thing collided when it comes to Java performance. (1) it's as if they thought Moore's Law would just come to the rescue. (2) they relied to heavily on that old canard of always optimize last.
(1) someone need to do the math, "When will machines run this design fast if computers double their capacities every 18 months?" scribble scribble "Five Thousand Years".
(2) You can't optimize design decisions. You have to anticipate what you will want to optimize.
I'm not believing that
Compiled languages: if you have a problem with them, fix them, they are the real languages. The VM -is- the only program, you are configuring that program. You configure the VM to do what you want. That's why it's hard to make a VM that can perform, it has to anticipate every possible desire. With a compiled language the program is in fact constructed directly to the domain, you get exactly what you put into the program and use.
hmmm, I wonder if that sounded like a rant?
Java is like Platonism. It's perfect in the ideal sphere, but in the real world we can have only imperfect copies.
it's not that easy to make a VM on a chip that'll be any faster. There are design issues. The Virtual Machine is not like a Real Machine. Software engineers know what a machine should be like! Those pesky hardware engineers have to deal with the laws of physics, imagine the indignity!
... it should be modded up.
well, both meanings sort of fit, that's the beauty.
hmmm, how about a openoffice.com name-fork... new name
"Ms. Office". Have a nice woman as the emblem, she can be carrying a beach towel of four colors, that you use for the icon.
Purely a coincidence.
really?
I think it's more about the natural process of commoditization.
You can't charge 150% markup on Thyme and Sage forever, you know.
Also, most software engineering takes place for people that use software, not people that sell it. Those people have good reason to share a bedrock of tools.
But your average whiney capitalist think they ought to collect six figures for software written in 1985 just forever --- that's the fair thing. Oh well, it's a battle, you'll win or lose and right and wrong will have little to do with it.
>That indicates a failure of the patch-based open source process. CVS couldn't be fixed within the process; it was necessary to start a new project and rewrite.
it doesn't, this is it's advantage. With commercial software you write the same solution over and over. I cannot use a solution I've written somewhere else, so I have to start from scratch. Instead of take a system to it's logical conclusion and limits, I make the same system again, because at least it's known what I'll get when I've built it.
With OSS systems, you follow a system to it's logical conclusion. Then, only when the design is exhausted, do you need to start over. You only replace a system when you need to. In commercial software you are always reproducing systems.
For nearly all my carreer I've used a lot of commercial tools, and they are simply not as reliable in the long haul as free software is for these sorts of reasons.
If we get people used to this, to heck with all of you... I'll be able to create a portable library that I can use the rest of my life wherever I work, whatever I do! sounds good to me.
yes, except that with royalty generating IP in linux the GPL doesn't apply. I would think they would have to stop using it, but if not literally, in effect, as they could not legally make a copy to, e.g., install on a new server.
But anyway, I agree, it's SCO that is the clear violator and legitimate target for GPL based actions.
What it actually says is that the only thing that gives you the ongoing right to duplicate other people's IP is to ship with the GPL license attached. It doesn't say that you have to mean it, nor that you can't stop shipping (thereby stopping your own violation of other peoples' rights, and SCO are slowly getting there) and then renege on your GPL license and prosecute the duplication and distribution of your copyrighted code.
yes of course, I agree, you can not mean it, and then GPL doesn't apply. You have to stop and have a good story about accidently shipping it in the first place.
The topic of this story was SCO further sanctioning the use of linux by their own customers. They know they have no license now by which to have given those customers Linux, but they are not recalling those distributions, they are actively promoting that their customers keep Linux, promising support, and standing behind what they have done.
What they have done accidently or by design is ship their code under the GPL. So in other words, by sanctioning what they did, even if they stop further shipments, they have said loud and clear, "We DO mean it... our IP, under the GPL, merged with linux, boot up and be merry".
It's as if there is a mind bubble and some people do not realize the other linux code is not public domain, that it's fully copyrighted material. Not just any hypothetical AT&T code in there, but all the other stuff, the other parts of the kernel that linus or whomever added.
SCO must be sued by OSS copyright holders. Specifically those into which SCO code is currently merged.
You have a point, but the fact is, business expect you to sue. If you use Sun or MS IP against their license, they WILL sue. And most businessmen are aware that actually, companies like Sun and Microsoft are constantly in litigation as plaintiffs and defendants, so I don't think you are right.
In fact, I think if they don't sue over these violations, they will lose credibility and the GPL will be used like public domain. It must be enforced for a business person to take it seriously.
I did read those posts.
You make a good point about the IBM v. SCO nature not involving the GPL, I'm thinking about the fact that SCO has GPLed their code however.
They distributed it under GPL.
That can be inadvertent, so they immediately cease shipping linux. By indemnifying their customers, however, they have said to the court "make no mistake about it, we know we've shipped our code with GPL code, and please understand we are fine with that, we have told our customers this is fine." Right. Assuming their lawyers have read the GPL, their code is now GPLed.
OR: they are in violation of the GPL and can be sued. It's not a part of the IBM case, but it could be, there is no reason that IBM, as a linux copyright holder cannot counter sue SCO now.
Sorry if that wasn't clear. I agree the post you were replying too was the most wrong of all, implying that using SCO at anytime would indemnify you from getting sued, when clearly they mean "we won't sue you for using SCO IP that we sold you."
but let's assume we're back to an honest disagreement. You understand and I understand but we disagree on interpretation and on our predictions.
My point is that it hardly matters how many slashdotters would buy an MS linux, but all the Microsoft customers that would.
Microsoft has decided to fight. I'm sure there is a line where they will go with it (originally they thought they would fight the internet too! (ps: I was an original MSN beta tester)), but up to that point they are fighting. I'm hoping they miscalculate their timing because if MS made a distribution and pushed it, other distro's would HAVE to be compatible.
But one other thing: Microsoft cannot destroy linux, that is what they are trying to do. But Microsoft COULD embrace and extend linux, or at least I think so. They havn't because it's much more lucrative to destroy it (except for the impossible part).
You seem to forget, even if SCO is a copyright holder (doubtful at the moment), they are not the only copyright holder! If you lose Linux because the GPL doesn't hold --- SCO CAN'T GIVE IT BACK TO YOU.
It's not clear to me if the customers can be sued, but SCO certainly can, they are knowingly and wontonly violating the license on other people's IP, which is what these kind of cases hinge on, the intentionality of it.
The GPL specifically states that if you ship the code, the GPL applies, and that if you ship your proprietary IP it has to be available for a royalty free non-terminable licence.
So SCO is clearly in violation of the GPL to be doing this.
Further, since it's distributed under the GPL and they have confirmed this now, their IP can be redistributed, as the GPL allows, by their customers.
I do not share you confidence in SCO's lawyers, so I'll call your SCO lawyers and rais you a pack of IBM lawyers.
No, it'll be pronounced Ibum/Linux
GNU/BM would just be wrong.