But as for that before winning the right he finally to say
My lexical analyser dumped core trying to get through that.
Slashdot editors - allow me to introduce reality
on
Java Is So 90s
·
· Score: 1
as old and out of style as IBM's (IBM) mainframe computers and SAP's corporate applications?
Mainframe machines provide the easiest and most well-tested way of getting the five 9s. If it's out of fashion, how come IBM's Systems and Technology group (which sells the mainframes and other servers but not PCs) earned them $17bn last year, and grew by 14% on the previous year ?
And SAP, that company with an operating income of EUR1.7bn last year.
I wouldn't recommend that anyone follow the "write unmaintainable code to make your job secure". If you're working for a fairly sane company and a boss who knows some of what they are doing, they'll have a pretty good idea of how maintainable your code is whenever they ask other people in your group to work on it. If your company is an outsourcing operation, your customers will more than likely be inspecting the code at random and they'll be peeved if you're doing stupid stuff.
Same time, the article is good in that it's a really good essay about what NOT to do in order to be a successful software developer.
AFAIR, that's incorrect. It's possible to put POSIX mutexes (created with a certain SHARED flag) into shared memory segments and share them between processes that way. W. Richard Stevens uses at least one example of this in his book on IPC. I haven't RTFA to check whether they are doing this though.
I'm not a linux kernel developer, but I am an Open Source developer, and have made significant contributions to at least a few projects.
This whole discussion is about the Linux kernel, so I guess you've shot yourself in the foot.
That binary drivers are bad (see reason #1 why Linus doesn't want an stable binary ABI)
Why are binary drivers bad ? Clearly it is unreasonable to expect kernel developers to debug them or support them - but then again why would I ever expect a developer to support software he hadn't written ?
I don't remember reading Linus saying that he did not want a stable binary ABI. Actually, any time I've followed lkml, Linus has quite correctly been in favour of avoiding internal API and ABI changes which would needlessly break things.
Linus is pretty much on the fence over kernel modules. If there was any suggestion that kernel developers would start suing binary-only driver producers to produce their source (a shaky case as it stands) Linux would quickly see the industry support it has built up run scared. An awful lot of kernel developers have their salaries paid by the industry...
My question is in response to your original question: "What gives you the right to impose that restriction on me ?" The right given to the Copyright holder to restrict your usage of their work outside of the realm of Fair Use.
Dangerous business you're getting into there. Does the author of DVD copying software have the legal right to stop you making illegal copies of movies ? Is that inside or outside the realm of fair use ?
If you don't like the ideological position that Linux is forwarding
What ideological position is Linux forwarding ?
then you're free to use only operating systems that do support binary drivers.
Where do you get the right to dictate to me what operating systems I may or may not use ? Your autocratic language makes Microsoft sound positively benevolent.
What right do you have to tell me what I must do with my code to make you feel better?
I wouldn't have any right, but then again at NO TIME have I told you what to do with your code. Why throw up these silly straw men ?
Tell me, what BIOS do you run ? Do you have the source to the firmware in your IDE disk drive ? In your CD-ROM/DVD-ROM drive ? Do you have the source to your SCSI controller's firmware ?
If you think you have control over your computer you are suffering under a delusion.
It is not welcome. Linux is about Open Source, and allowing people to link-in binary closed drivers goes against this.
I liked your use of the term "allowing people", as though it was something evil. It's funny how you think that Open Source should be about taking away from people the right to link whatever drivers they want. That's not allowing people to be "free" is it ?
If you want to insist on having the source to your drivers, you are free to choose to buy only hardware with an open source driver. What gives you the right to impose that restriction on me ?
AFAIK, solar cells are presently 10% efficient. Theoretically, if one were to double the efficiency of such cells, then a person would only need to purchase half as many in order to provide the same power output, thereby achieving the cost reduction being described ?
If 50% efficiency is ever achieved then we'd be able to buy one-fifth of the number of cells to get the same power output. This all assumes, of course, that enhancing the efficiency of the cells does not make them more expensive to produce.
Solaris could do anything and everything that Linux could do.
Solaris and Linux are not equivalent in features, but this is not my point. My point was that Solaris' UNIX pedigree does not intrinsically make it better. Operating systems are not monarchies.
Why would Sun switch from Solaris to Linux if Solaris can do everything that Linux can do?
They'd do that if they were very silly. Since they aren't doing that, we can conclude that they are not silly. Linux is part of the strategy for every single IT integrator and vendor out there. Sun would be equally silly if they ignored it.
You should read more about Plan 9. Plan 9 isn't a reinvention of the wheel; it is something completely different.
So is Windows. There are aspects of Windows that are certainly done better than UNIX. But being different does not mean better. I'd much take a known, tried and trusted solution that I know will work with all my existing stuff, than take a redesign that requires me to throw all that away.
And, yes, interoperatability is quite possible with Plan 9
Interoperability is quite possible with Windows, but is such a pain that it's easier not to.
if I were a corporation dropping my Unix product,
There is no evidence that Sun is killing Solaris; quite the contrary. But anyway.
I would rather go one step above the competition (Plan 9) rather than a sell a complete clone of something that I have already sold (Linux).
Plan 9 isn't competitive, that is why nobody uses it. If you tried to ship your machines with that OS, you'd quickly die. If I were a shareholder in your corporation, I'd be asking for your head on a plate.
The SPARC may fit your needs if you want a very powerful workstation, the x86 may fit your needs if you want a PC. That's up to the customer to decide.
The SPARC isn't more powerful than the x86 in the vast majority of cases - that is the issue here. It may have cleaner SMP design, nicer caching and a more sensible instruction set, but that is precisely bugger all use to me if the x86 crunches my numbers twice as fast.
However, I still believe that Sun is better off selling SPARC machines with Solaris than x86 machines with Linux. I have a feeling that Dell would just kill Sun if Sun made that move.
Sun knows the server and enterprise market in a way that Dell just doesn't comprehend. It offers levels of service and support that a souped-up clone builder like Dell can't touch. There are people out there willing to pay top dollar of all the nines. It doesn't matter what hardware it is at the end of the day.
Besides, Solaris is a real Unix (based off AT&T Unix System V), whereas Linux is just a clone.
What meaning does this statement have at the end of the day ? Why do I care about an operating system's pedigree ? Either it does what I want, or it does not. In fact, it seems pretty silly to market something solely on the basis that it still has ten year old bits of code floating around in it.
If Sun were to drop anything, I'd rather Sun adopt Plan 9 rather than Linux. At least Plan 9 is unique and different.
"unique and different" are two words which describe Microsoft. I wonder why their way of doing things appeal to you ? It may have escaped you that people want things that are standard and interoperable. I want to choose between vendors or open source products that just work. I don't want to see the wheel re-invented each go-round.
the SPARC is much better designed
The term "better designed" means nothing. Will it allow me to do my job better, or not ? Sure, x86 is a heap of shit. But if it performs better, why should I care ?
And when we're stuck with the x86 as our only platform, then innovation will slow down, and we might not see better platforms again.
Why are you so convinced that the world will reject a "better platform" ?
We don't need another Linux PC manufacturer.
There is no such thing as too much competition. Sun's hardware is among the best there is.
Hans lives in the real world and has to earn a living for himself. It's alright for Stallman and co, with a charitable foundation to support them. You can love free software as much as you want (I do), but you're merely exceptionally lucky if you can find a way to make it pay the rent.
I concede right off that Linux cannot touch Solaris on SPARC for robustness and scalability (if not raw speed) and it will be at least two or three years before it even has a chance.
But Dell will happily sell you a box with a fully supported and certified Red Hat or SuSE Linux environment. On those certified boxen with the standard install, those commercial distros run very smoothly indeed - they stand up to Solaris/SPARC boxen very well. You don't have to muck about with hardware or drivers or anything else - it all just works, just as it would on a Sun setup.
What a lot of rubbish. Linux in the commercial world (which is what we are talking about here) is not about the geeks, it never was. It's about Red Hat, IBM, HP, Novell and all of these other companies who are backing it as a serious enterprise OS.
I doubt the geeks are ever going to jump to Solaris. Linux has too much momentum. Why would anyone ever want to run Solaris on x86, what's the point ? Licensing and community aside, it's the same problem as faced by the BSDs - driver support and vendor backup.
If you want to run Solaris in your business you splash the cash and buy a SPARC. If you want a stable and robust production UNIX environment and do not want to spend an arm and a leg on Sun's boilerplate hardware, you get a nice Dell or HPaq rack and Red Hat Enterprise.
It's taken Linux many years and a lot of hard work to get it to the point where the big-box vendors will install it for you, and where the major applications have been ported to it. Those are really the benchmarks by which we can assess the maturity of a product as a commercial OS.
The large commercial/enterprise applications do of course run on Solaris/SPARC but have they all been ported to Solaris/x86 ? Are Dell going to start offering boxen with Solaris pre-installed ? Have stable and robust device drivers been written for all the popular x86 server devices that people will want Solaris to properly support ? I'm afraid Sun are in for a shock if they think they can suddenly take on Linux just because they've open-sourced their stuff.
Solaris on the latest SPARC hardware is still the most stable and scalable environment you can buy, where money is no object. Unfortunately money is an object for 99% of the folks out there. Sun can still get to win-win on this; the way forward would be for them to realize that they are simply no longer going to win by simply producing boxes and operating systems for money, and re-invent them as a services organization in the same vein as IBM. The first important step into this business would be to properly GPL Solaris and allow it to cross-pollenate with Linux and the other GPL OSs there - short-circuiting IBM who still won't GPL AIX. This would enable speedier porting of essential device drivers into Solaris and allow Linux to cherry pick some of the nicest Solaris features.
You missed the first train so don't beat the dead horse now.
That dead horse is presently earning EMC $91 million. I guess that's the sort of money you "serious computer scientists" (although how misinterpreting an article in front of the geek community on Slashdot is supposed to be "serious" I'll never know) wouldn't bother getting out of bed for.
Disappointingly, it didn't take long for the a slashdotter to confirm my point about not understanding the difference between SMP systems and clustering.
Threads do exist under Linux and there is a library to manage them. The details under the hood about how they are implemented are not relevant from a userspace perspective. On other OSs such as Solaris a thread is also known as a "lightweight process"; I imagine that on most OSs threads are implemented just like a process sharing another process's entire address space. The critical thing that makes Linux interesting is the low overhead involved in context switches due to the way processes allocate their memory. In practice, processes under Linux turn out not to be that much more expensive than threads (but still expensive).
The second part of your post is completely wrong; Windows NT 3.51 had 4-way SMP support (in 1995) - I suspect older versions also supported SMP - long before Alan Cox implemented it for Linux version 2 in June 1996. While Linux's SMP is perfectly good, I sincerely doubt it would scale quite as well as the boilerplate enterprise OSs such as Solaris or the mainframe platforms.
Linux runs on large clusters of machines (clustering is a fundamentally different concern than SMP, unfortunately a lot of slashdotters don't seem to understand this) mainly because it is cheap and easy to do, not because it is the best possible multiprocessing architecture. Theoretically there is nothing stopping this being done with Windows, indeed Windows does include clustering technology. These large commodity research-class machines are a brute force approach to parallel computing.
Slashdot Mirror
You've never heard of the Brit awards ? Or listened to any Britpop ?
Does anyone really see the competition regulators allowing this to go ahead ? I seriously doubt it.
But as for that before winning the right he finally to say
My lexical analyser dumped core trying to get through that.
as old and out of style as IBM's (IBM) mainframe computers and SAP's corporate applications?
Mainframe machines provide the easiest and most well-tested way of getting the five 9s. If it's out of fashion, how come IBM's Systems and Technology group (which sells the mainframes and other servers but not PCs) earned them $17bn last year, and grew by 14% on the previous year ?
And SAP, that company with an operating income of EUR1.7bn last year.
Is it out of style to earn pots of money ?
I wouldn't recommend that anyone follow the "write unmaintainable code to make your job secure". If you're working for a fairly sane company and a boss who knows some of what they are doing, they'll have a pretty good idea of how maintainable your code is whenever they ask other people in your group to work on it. If your company is an outsourcing operation, your customers will more than likely be inspecting the code at random and they'll be peeved if you're doing stupid stuff.
Same time, the article is good in that it's a really good essay about what NOT to do in order to be a successful software developer.
AFAIR, that's incorrect. It's possible to put POSIX mutexes (created with a certain SHARED flag) into shared memory segments and share them between processes that way. W. Richard Stevens uses at least one example of this in his book on IPC. I haven't RTFA to check whether they are doing this though.
I'm not a linux kernel developer, but I am an Open Source developer, and have made significant contributions to at least a few projects.
...
This whole discussion is about the Linux kernel, so I guess you've shot yourself in the foot.
That binary drivers are bad (see reason #1 why Linus doesn't want an stable binary ABI)
Why are binary drivers bad ? Clearly it is unreasonable to expect kernel developers to debug them or support them - but then again why would I ever expect a developer to support software he hadn't written ?
I don't remember reading Linus saying that he did not want a stable binary ABI. Actually, any time I've followed lkml, Linus has quite correctly been in favour of avoiding internal API and ABI changes which would needlessly break things.
Linus is pretty much on the fence over kernel modules. If there was any suggestion that kernel developers would start suing binary-only driver producers to produce their source (a shaky case as it stands) Linux would quickly see the industry support it has built up run scared. An awful lot of kernel developers have their salaries paid by the industry
My question is in response to your original question: "What gives you the right to impose that restriction on me ?" The right given to the Copyright holder to restrict your usage of their work outside of the realm of Fair Use.
Dangerous business you're getting into there. Does the author of DVD copying software have the legal right to stop you making illegal copies of movies ? Is that inside or outside the realm of fair use ?
Is it more important to have control than it is to have hardware ?
Um... I write the code?
Which code did you write ?
If you don't like the ideological position that Linux is forwarding
What ideological position is Linux forwarding ?
then you're free to use only operating systems that do support binary drivers.
Where do you get the right to dictate to me what operating systems I may or may not use ? Your autocratic language makes Microsoft sound positively benevolent.
What right do you have to tell me what I must do with my code to make you feel better?
I wouldn't have any right, but then again at NO TIME have I told you what to do with your code. Why throw up these silly straw men ?
But we retain control over our computers.
Tell me, what BIOS do you run ? Do you have the source to the firmware in your IDE disk drive ? In your CD-ROM/DVD-ROM drive ? Do you have the source to your SCSI controller's firmware ?
If you think you have control over your computer you are suffering under a delusion.
It is not welcome. Linux is about Open Source, and allowing people to link-in binary closed drivers goes against this.
I liked your use of the term "allowing people", as though it was something evil. It's funny how you think that Open Source should be about taking away from people the right to link whatever drivers they want. That's not allowing people to be "free" is it ?
If you want to insist on having the source to your drivers, you are free to choose to buy only hardware with an open source driver. What gives you the right to impose that restriction on me ?
AFAIK, solar cells are presently 10% efficient. Theoretically, if one were to double the efficiency of such cells, then a person would only need to purchase half as many in order to provide the same power output, thereby achieving the cost reduction being described ?
If 50% efficiency is ever achieved then we'd be able to buy one-fifth of the number of cells to get the same power output. This all assumes, of course, that enhancing the efficiency of the cells does not make them more expensive to produce.
What a shock it is where people can express their opinions freely. Stop them! Arrest them! This outrage must be curtailed forwith!
Solaris could do anything and everything that Linux could do.
Solaris and Linux are not equivalent in features, but this is not my point. My point was that Solaris' UNIX pedigree does not intrinsically make it better. Operating systems are not monarchies.
Why would Sun switch from Solaris to Linux if Solaris can do everything that Linux can do?
They'd do that if they were very silly. Since they aren't doing that, we can conclude that they are not silly. Linux is part of the strategy for every single IT integrator and vendor out there. Sun would be equally silly if they ignored it.
You should read more about Plan 9. Plan 9 isn't a reinvention of the wheel; it is something completely different.
So is Windows. There are aspects of Windows that are certainly done better than UNIX. But being different does not mean better. I'd much take a known, tried and trusted solution that I know will work with all my existing stuff, than take a redesign that requires me to throw all that away.
And, yes, interoperatability is quite possible with Plan 9
Interoperability is quite possible with Windows, but is such a pain that it's easier not to.
if I were a corporation dropping my Unix product,
There is no evidence that Sun is killing Solaris; quite the contrary. But anyway.
I would rather go one step above the competition (Plan 9) rather than a sell a complete clone of something that I have already sold (Linux).
Plan 9 isn't competitive, that is why nobody uses it. If you tried to ship your machines with that OS, you'd quickly die. If I were a shareholder in your corporation, I'd be asking for your head on a plate.
The SPARC may fit your needs if you want a very powerful workstation, the x86 may fit your needs if you want a PC. That's up to the customer to decide.
The SPARC isn't more powerful than the x86 in the vast majority of cases - that is the issue here. It may have cleaner SMP design, nicer caching and a more sensible instruction set, but that is precisely bugger all use to me if the x86 crunches my numbers twice as fast.
However, I still believe that Sun is better off selling SPARC machines with Solaris than x86 machines with Linux. I have a feeling that Dell would just kill Sun if Sun made that move.
Sun knows the server and enterprise market in a way that Dell just doesn't comprehend. It offers levels of service and support that a souped-up clone builder like Dell can't touch. There are people out there willing to pay top dollar of all the nines. It doesn't matter what hardware it is at the end of the day.
Besides, Solaris is a real Unix (based off AT&T Unix System V), whereas Linux is just a clone.
What meaning does this statement have at the end of the day ? Why do I care about an operating system's pedigree ? Either it does what I want, or it does not. In fact, it seems pretty silly to market something solely on the basis that it still has ten year old bits of code floating around in it.
If Sun were to drop anything, I'd rather Sun adopt Plan 9 rather than Linux. At least Plan 9 is unique and different.
"unique and different" are two words which describe Microsoft. I wonder why their way of doing things appeal to you ? It may have escaped you that people want things that are standard and interoperable. I want to choose between vendors or open source products that just work. I don't want to see the wheel re-invented each go-round.
the SPARC is much better designed
The term "better designed" means nothing. Will it allow me to do my job better, or not ? Sure, x86 is a heap of shit. But if it performs better, why should I care ?
And when we're stuck with the x86 as our only platform, then innovation will slow down, and we might not see better platforms again.
Why are you so convinced that the world will reject a "better platform" ?
We don't need another Linux PC manufacturer.
There is no such thing as too much competition. Sun's hardware is among the best there is.
How incredible - "2005 Google" is clearly visible all over the ground and buildings. Perhaps the extraterrestrials are trying to tell us something..
Hans lives in the real world and has to earn a living for himself. It's alright for Stallman and co, with a charitable foundation to support them. You can love free software as much as you want (I do), but you're merely exceptionally lucky if you can find a way to make it pay the rent.
The POINT: stability
Arse - Solaris' stability is mainly to do with the custom hardware it runs on. That advantage exists on Sun SPARC hardware. It does not exist on x86.
Dell's hardware sucks ass and their customer service doubly so.
I've got ten of their servers here running Linux and Windows, we've got zero problems with them.
I concede right off that Linux cannot touch Solaris on SPARC for robustness and scalability (if not raw speed) and it will be at least two or three years before it even has a chance.
But Dell will happily sell you a box with a fully supported and certified Red Hat or SuSE Linux environment. On those certified boxen with the standard install, those commercial distros run very smoothly indeed - they stand up to Solaris/SPARC boxen very well. You don't have to muck about with hardware or drivers or anything else - it all just works, just as it would on a Sun setup.
What a lot of rubbish. Linux in the commercial world (which is what we are talking about here) is not about the geeks, it never was. It's about Red Hat, IBM, HP, Novell and all of these other companies who are backing it as a serious enterprise OS.
I doubt the geeks are ever going to jump to Solaris. Linux has too much momentum. Why would anyone ever want to run Solaris on x86, what's the point ? Licensing and community aside, it's the same problem as faced by the BSDs - driver support and vendor backup.
If you want to run Solaris in your business you splash the cash and buy a SPARC. If you want a stable and robust production UNIX environment and do not want to spend an arm and a leg on Sun's boilerplate hardware, you get a nice Dell or HPaq rack and Red Hat Enterprise.
It's taken Linux many years and a lot of hard work to get it to the point where the big-box vendors will install it for you, and where the major applications have been ported to it. Those are really the benchmarks by which we can assess the maturity of a product as a commercial OS.
The large commercial/enterprise applications do of course run on Solaris/SPARC but have they all been ported to Solaris/x86 ? Are Dell going to start offering boxen with Solaris pre-installed ? Have stable and robust device drivers been written for all the popular x86 server devices that people will want Solaris to properly support ? I'm afraid Sun are in for a shock if they think they can suddenly take on Linux just because they've open-sourced their stuff.
Solaris on the latest SPARC hardware is still the most stable and scalable environment you can buy, where money is no object. Unfortunately money is an object for 99% of the folks out there. Sun can still get to win-win on this; the way forward would be for them to realize that they are simply no longer going to win by simply producing boxes and operating systems for money, and re-invent them as a services organization in the same vein as IBM. The first important step into this business would be to properly GPL Solaris and allow it to cross-pollenate with Linux and the other GPL OSs there - short-circuiting IBM who still won't GPL AIX. This would enable speedier porting of essential device drivers into Solaris and allow Linux to cherry pick some of the nicest Solaris features.
You missed the first train so don't beat the dead horse now.
That dead horse is presently earning EMC $91 million. I guess that's the sort of money you "serious computer scientists" (although how misinterpreting an article in front of the geek community on Slashdot is supposed to be "serious" I'll never know) wouldn't bother getting out of bed for.
Oops. Hoisted by my own petard. It appears that I'm completely wrong. I'll get my coat.
Disappointingly, it didn't take long for the a slashdotter to confirm my point about not understanding the difference between SMP systems and clustering.
Threads do exist under Linux and there is a library to manage them. The details under the hood about how they are implemented are not relevant from a userspace perspective. On other OSs such as Solaris a thread is also known as a "lightweight process"; I imagine that on most OSs threads are implemented just like a process sharing another process's entire address space. The critical thing that makes Linux interesting is the low overhead involved in context switches due to the way processes allocate their memory. In practice, processes under Linux turn out not to be that much more expensive than threads (but still expensive).
The second part of your post is completely wrong; Windows NT 3.51 had 4-way SMP support (in 1995) - I suspect older versions also supported SMP - long before Alan Cox implemented it for Linux version 2 in June 1996. While Linux's SMP is perfectly good, I sincerely doubt it would scale quite as well as the boilerplate enterprise OSs such as Solaris or the mainframe platforms.
Linux runs on large clusters of machines (clustering is a fundamentally different concern than SMP, unfortunately a lot of slashdotters don't seem to understand this) mainly because it is cheap and easy to do, not because it is the best possible multiprocessing architecture. Theoretically there is nothing stopping this being done with Windows, indeed Windows does include clustering technology. These large commodity research-class machines are a brute force approach to parallel computing.