It is possible, but not (at this point) very plausible that this will work. After all the technology is still pretty young, the thing has been dead for a long time, and it does not help that they are trying to clone a male.
However this does raise the question of whether or not we should be establishing a bank of samples from endangered species. Even if they go extinct, we could still have the possiblity of bringing them back at some point in the future.
Does anyone know of any efforts along this direction?
Sun is bad (not evil, bad) because when they get control of APIs they deliberately play games with them to make their stuff look good and others look bad. If you own the APIs you control the rules and can force others into playing catch-up.
Don't believe that is an issue? Take a look at how Microsoft uses the Windows API to leverage their OS monopoly into further monopolies (eg Office). Sun would love to get the same control...
I know a lot of companies run Perl4. But hey, wake up and smell the coffee. Perl5 has been the stable developed Perl for longer than all previous versions were *combined*.
Besides which the Perl5 syntax is simpler, and it supports lots of neat things. (Taken a look at CPAN recently?)
I agree with your model. In fact what you describe is exactly what Apache does. But if you allow select() or poll() then people can also write the good old select, and then race through their sockets and try to grab it. The problem? There is a race condition. Apache used to do this with the result that occasionally threads would get blocked trying to grab sockets that another thread got. Also waking up 10 processes at once is inefficient.
In other words Sun tries to force Java programmers to program their way, which is safely. As always, doing so has good and bad points. Is it worth it? Perhaps. Is it irritating..?
The "right" architecture is the "wake up one" model that I think is will arrive with Linux 2.4. Here a number of processes do a blocking read on the same descriptor, but the OS knows that when input comes in, it chooses one to wake up and leaves the rest asleep. Why is this better? Well it is simpler to program for, more efficient on resources, and it is easy to support efficiently when you move to an SMP (or clustered) system with inputs coming in many directions. (Just find the first CPU with a waiting process...)
But given the Java design, that option is now precluded entirely in that language until Sun allows it. And if it is something that helps Linux compete with Solaris, do you think that they will change that feature?
If Solaris were free, I would ask what the catch is. Don't get me wrong, they (unlike the Redmond folks) do quality work. However if Sun had the same opportunities as Microsoft, they would be just as bad.
For instance look at Java. When Sun came out with Java they had a simple threading model that they wanted people to use. You want to wait for some IO? Spawn a thread to make a blocking call for the IO. In some ways good, for instance this architecture removes the possibility of writing a lot of possible race conditions. However was it coincidence that it also uses lots of threads, and all of the other forms of Unix out there at the time could not handle large numbers of threads efficiently? How convenient to have a cross-platform language that coincidentally cannot be made to run as well on your main competitor's platforms without major modifications to the OS!
Sun has a history of these games. The current one is Java3D. They have a pretty nice spec for 3D graphics and vector math. There are two possible implementations - one is native (using the video card, etc for extremely good performance) and the other is in pure Java (for the molasses effect). Of course to get permission to even try and implement the native version for a platform you need Sun's permission - and they refuse to give it for Linux.
So if Solaris was made free, here is what I would open up that gift-horse's mouth and look for:
What is the license? Have they tried to retain absolute control with that horrible pseudo-free PoS called the Sun Community licence?
What games are they playing with support? Sun is a hardware vendor. Presumably the aim would be to sell more hardware. One way to do that is guarantee that other people's hardware does not run as well...
What games are they playing with the APIs? Take a look at Java with its API of the day for more on that...
So yes, if Sun released Solaris free, I would almost certainly just stick with my Debian system. Yes, they do quality work. But Sun doesn't do anything that Sun is not the main beneficiary of, which is not unreasonable in and of itself but is unlikely to match my long-term road map. Linux (by a pleasant contrast) has no such hidden agenda to watch out for.
If you are interested in this stuff, look at The history of the Internet for fuller details. But the basic story is that the idea of packet-based networks arose independently in two places. The first was Kleinrock et al at MIT, the second was Baran at RAND. The former group was interested in them as a way of efficiently sharing the same lines between many different computers. The latter was interested in them for creating communication systems that could survive nuclear war. The two groups did not know of each other.
The Internet arose out of ARPANET which was based on the work at MIT. The goal was to allow computer resources at different research institutions with different types of computers to be shared. The fact that surviving nuclear war was not a goal can be seen in the fact that the machinery used to set it up had no protection against the electro-magnetic effects of a nuclear warhead. Furthermore the initial set-up heavily relied upon a single back-bone. With no redundancy in your physical network, what good is a redundant protocol?
In fact the initial proof of concept and then proposal for ARPANET was made before the MIT people even heard of the work at RAND. Indeed the two groups found out about each other at a conference where the ARPANET was being proposed. Don't believe me? Check it out for yourself!
What gets reported in the media has more to do with our prejudices than with the prejudices of those present or the actual facts.
This is similar to how rumors spread. Rumors don't spread based on how much factual basis there is for them. Instead they spread based on how much they resonate. People want to believe that the criminals are part of marginalized groups. So the coincidence of clothing makes them "trench coat mafia". It makes them gay. People want to believe that they are bad people, out to target a group. But they hit a lot of different people so they are targeting blacks, Christians, and jocks. (Hmmm...between those groups you can "explain" a *lot* of deaths.)
And people don't want to ask questions. So articles to the contrary (for instance the Rolling Stones article that someone mentioned) get ignored. This one will as well - it does not make good copy.
These are phenomena familiar to all members of disliked minorities. But people really don't want to think about that. And so we accumulate a few more urban legends...
Seriously, people tend to contribute to OSS depending on their specific needs and resources. Having military forces investigate and contribute to the security of OSS would be an...interesting...development.
DOS was a clone of CP/M bought by Microsoft from the person who developed it. And this is your example of innovation? Have any more?
No, Microsoft does not have a record of innovation. It has a history of cloning, polishing, and bundling. But not of innovation. (ESR estimates that Microsoft has put the computer industry back about 10 years.
The license is very clear on that. Are they distributing it? If they are then they have to release modifications. If you really know of such a case, then please follow these instructions.
OTOH if they are not distributing it but are selling a service based on it (for instance they are using a modified version in the back-end for a web-site) then they should be fine.
Is there any GPLed program in the beta distribution that they have modified and not released modifications to?
I am asking this honestly. Most of the parts people mention, like the compiler and kernel, are almost assuredly the same as you can find plenty of other places. There are other pieces of code, like Wine, that are critical to Corel's distribution which they have modified and released patches for. (Wine is not GPLed.) How close are the publically released pieces to what is in the beta? Is it identical for all GPLed programs?
There are certainly pieces in their distribution that are proprietary and will remain that way. Unquestionably the distribution as a whole will not be free.
But when all is said and done, can't they meet the terms of the GPL by having a publically available ftp site with patches and/or full source-code to the modified GPLed programs in their suite?
(And yes, there is a real issue here. If Corel can get good publicity for improvements to a product in their beta while denying other companies the opportunity to match this, the resulting first mover advantage is something that the GPL is intended to prevent.)
Corel's product, their distribution plus whatever extra tools they have, need not be GPLed. Sure, large portions of it have to be. But they can do like SuSE, Caldera, and others have in the past and have a distribution that has non-free pieces and cannot be freely redistributed.
The irony of it. With this purchase Microsoft is selling a branded Unix. Remember what Microsoft keeps on saying about Linux taking marketshare from Unix? Well now that Microsoft is a Unix vendor, I will gleefully admit that that is true for at least one flavor of Unix!
Sorry Microsoft, you can't polish a turd, and you cannot make it into a workhorse by adding a fur coat either!
Those commands do do something special - on 64-bit operating systems - which is why they are carefully #ifdefed out so they cannot be used on 32-bit ones.
Remember that Linux standard libraries have a common source for different operating systems. So just because you can see it in the header files does not mean that you can use it on your machine.
I will believe that the big-name databases truly take Linux seriously for real work when they listen to what Linus has to say on the subject of big files.
If you want a database on Linux with over 2 GB in the database today, you cannot use one of the big commericial guys. Linux has a limit on file-size for 32-bit architectures and it is unlikely to go away. If you are serious about handling more data than that, Linus wants you to get a 64-bit CPU like the Alpha. He does not want to mess up a lot of code for a case that affects relatively few people. One that will be a non-issue as the 32-bit to 64-bit migration continues.
And so, if the database vendors are serious about using Linux for databases, they have to offer us a port to a 64-bit architecture. Well Linux on the Alpha is 64-bit, here, and popular, where are the databases for it?
I know that Jay Sulzberger of LXNY thinks that having Gates and Microsoft get tax deductions for donations to schools tied with restrictive clauses is grounds for a lawsuit.
You never know with Jay how to take what he says, he is serious but he is not known as...put it this way...the most calm and rational individual in the world. OTOH he is pretty intelligent and does generally know his stuff.
Face facts, one of the biggest way that Linux sneaks in the back door is because of Samba. Once it is there it grows, but Samba in many cases is the necessary first step without which Linux either would not be considered, or would have been thrown out because it could not integrate itself in the existing environment..
I therefore think that Andrew Tridgell deserves this award (and more pizza).
I know about Windows NT Server 4.0 vs Unix but it would be handy to have something targeted to this situation. A study that showed a realistic comparison of support costs, network usage, security issues, and hardware costs between Windows and Unix.
Why?
So that when Microsoft wants to donate to a university with strings attached, people can evaluate whether or not the Microsoft solution with the handout included will cost more than a Unix solution. If it (as it often will) does then it becomes much easier to make the argument that the offer is not worthwhile.
Otherwise cash in hand is a very hard thing to argue with...
This is truly hearsay, but I heard that the decision was made by a sub-contractor. But still it is a nice decision and Red Hat should certainly see whether they can get the standard, By appointment to... line approved. (At least for their British sales.)
Slashdot Mirror
The expert they were quoting pointed out that cloning success rates are higher with females than males. The individual they are working with is male.
Cheers,
Ben
It is possible, but not (at this point) very plausible that this will work. After all the technology is still pretty young, the thing has been dead for a long time, and it does not help that they are trying to clone a male.
However this does raise the question of whether or not we should be establishing a bank of samples from endangered species. Even if they go extinct, we could still have the possiblity of bringing them back at some point in the future.
Does anyone know of any efforts along this direction?
Ben
Seriously, go to http://www.debian.org, follow their directions, and fill it out as a wishlist item for apt.
Do you think that the apt developers will notice a suggestion tossed out deep in a thread on slashdot about Red Hat's new release? I thought not!
Cheers,
Ben
Sun is bad (not evil, bad) because when they get control of APIs they deliberately play games with them to make their stuff look good and others look bad. If you own the APIs you control the rules and can force others into playing catch-up.
Don't believe that is an issue? Take a look at how Microsoft uses the Windows API to leverage their OS monopoly into further monopolies (eg Office). Sun would love to get the same control...
Cheers,
Ben
I know a lot of companies run Perl4. But hey, wake up and smell the coffee. Perl5 has been the stable developed Perl for longer than all previous versions were *combined*.
Besides which the Perl5 syntax is simpler, and it supports lots of neat things. (Taken a look at CPAN recently?)
Cheers,
Ben
I agree with your model. In fact what you describe is exactly what Apache does. But if you allow select() or poll() then people can also write the good old select, and then race through their sockets and try to grab it. The problem? There is a race condition. Apache used to do this with the result that occasionally threads would get blocked trying to grab sockets that another thread got. Also waking up 10 processes at once is inefficient.
In other words Sun tries to force Java programmers to program their way, which is safely. As always, doing so has good and bad points. Is it worth it? Perhaps. Is it irritating..?
The "right" architecture is the "wake up one" model that I think is will arrive with Linux 2.4. Here a number of processes do a blocking read on the same descriptor, but the OS knows that when input comes in, it chooses one to wake up and leaves the rest asleep. Why is this better? Well it is simpler to program for, more efficient on resources, and it is easy to support efficiently when you move to an SMP (or clustered) system with inputs coming in many directions. (Just find the first CPU with a waiting process...)
But given the Java design, that option is now precluded entirely in that language until Sun allows it. And if it is something that helps Linux compete with Solaris, do you think that they will change that feature?
Not a chance.
Cheers,
Ben
For instance look at Java. When Sun came out with Java they had a simple threading model that they wanted people to use. You want to wait for some IO? Spawn a thread to make a blocking call for the IO. In some ways good, for instance this architecture removes the possibility of writing a lot of possible race conditions. However was it coincidence that it also uses lots of threads, and all of the other forms of Unix out there at the time could not handle large numbers of threads efficiently? How convenient to have a cross-platform language that coincidentally cannot be made to run as well on your main competitor's platforms without major modifications to the OS!
Sun has a history of these games. The current one is Java3D. They have a pretty nice spec for 3D graphics and vector math. There are two possible implementations - one is native (using the video card, etc for extremely good performance) and the other is in pure Java (for the molasses effect). Of course to get permission to even try and implement the native version for a platform you need Sun's permission - and they refuse to give it for Linux.
So if Solaris was made free, here is what I would open up that gift-horse's mouth and look for:
So yes, if Sun released Solaris free, I would almost certainly just stick with my Debian system. Yes, they do quality work. But Sun doesn't do anything that Sun is not the main beneficiary of, which is not unreasonable in and of itself but is unlikely to match my long-term road map. Linux (by a pleasant contrast) has no such hidden agenda to watch out for.
Sincerely,
Ben Tilly
If you are interested in this stuff, look at The history of the Internet for fuller details. But the basic story is that the idea of packet-based networks arose independently in two places. The first was Kleinrock et al at MIT, the second was Baran at RAND. The former group was interested in them as a way of efficiently sharing the same lines between many different computers. The latter was interested in them for creating communication systems that could survive nuclear war. The two groups did not know of each other.
The Internet arose out of ARPANET which was based on the work at MIT. The goal was to allow computer resources at different research institutions with different types of computers to be shared. The fact that surviving nuclear war was not a goal can be seen in the fact that the machinery used to set it up had no protection against the electro-magnetic effects of a nuclear warhead. Furthermore the initial set-up heavily relied upon a single back-bone. With no redundancy in your physical network, what good is a redundant protocol?
In fact the initial proof of concept and then proposal for ARPANET was made before the MIT people even heard of the work at RAND. Indeed the two groups found out about each other at a conference where the ARPANET was being proposed. Don't believe me? Check it out for yourself!
Cheers,
Ben Tilly
What gets reported in the media has more to do with our prejudices than with the prejudices of those present or the actual facts.
This is similar to how rumors spread. Rumors don't spread based on how much factual basis there is for them. Instead they spread based on how much they resonate. People want to believe that the criminals are part of marginalized groups. So the coincidence of clothing makes them "trench coat mafia". It makes them gay. People want to believe that they are bad people, out to target a group. But they hit a lot of different people so they are targeting blacks, Christians, and jocks. (Hmmm...between those groups you can "explain" a *lot* of deaths.)
And people don't want to ask questions. So articles to the contrary (for instance the Rolling Stones article that someone mentioned) get ignored. This one will as well - it does not make good copy.
These are phenomena familiar to all members of disliked minorities. But people really don't want to think about that. And so we accumulate a few more urban legends...
Regards,
Ben
Seriously, people tend to contribute to OSS depending on their specific needs and resources. Having military forces investigate and contribute to the security of OSS would be an...interesting...development.
Cheers,
Ben
DOS was a clone of CP/M bought by Microsoft from the person who developed it. And this is your example of innovation? Have any more?
No, Microsoft does not have a record of innovation. It has a history of cloning, polishing, and bundling. But not of innovation. (ESR estimates that Microsoft has put the computer industry back about 10 years.
Ben
No, he is not in the business of selling cheap alibis, these hypocrites are! (Click on the UK link to see what they do.)
There is a difference...
Ben
OTOH if they are not distributing it but are selling a service based on it (for instance they are using a modified version in the back-end for a web-site) then they should be fine.
Sincerely,
Ben
Is there any GPLed program in the beta distribution that they have modified and not released modifications to?
I am asking this honestly. Most of the parts people mention, like the compiler and kernel, are almost assuredly the same as you can find plenty of other places. There are other pieces of code, like Wine, that are critical to Corel's distribution which they have modified and released patches for. (Wine is not GPLed.) How close are the publically released pieces to what is in the beta? Is it identical for all GPLed programs?
There are certainly pieces in their distribution that are proprietary and will remain that way. Unquestionably the distribution as a whole will not be free.
But when all is said and done, can't they meet the terms of the GPL by having a publically available ftp site with patches and/or full source-code to the modified GPLed programs in their suite?
(And yes, there is a real issue here. If Corel can get good publicity for improvements to a product in their beta while denying other companies the opportunity to match this, the resulting first mover advantage is something that the GPL is intended to prevent.)
Cheers,
Ben
Corel's product, their distribution plus whatever extra tools they have, need not be GPLed. Sure, large portions of it have to be. But they can do like SuSE, Caldera, and others have in the past and have a distribution that has non-free pieces and cannot be freely redistributed.
Regards,
Ben
Hmmm...
Who did the StrongARM port of Linux?
Corel.
Who has been contributing to Wine in a big way?
Corel.
Who has a big mouth?
Stephen Bamattre.
Who is, judging from the irritating smart quotes, using Windows while playing holier than thou about Linux?
Stephen Bamattre.
Enough said I think.
Ben
Ever tried opening a Unix formatted text file in Notepad? Most other editors have no problem...
Cheers,
Ben
The irony of it. With this purchase Microsoft is selling a branded Unix. Remember what Microsoft keeps on saying about Linux taking marketshare from Unix? Well now that Microsoft is a Unix vendor, I will gleefully admit that that is true for at least one flavor of Unix!
Sorry Microsoft, you can't polish a turd, and you cannot make it into a workhorse by adding a fur coat either!
Cheers,
Ben
Those commands do do something special - on 64-bit operating systems - which is why they are carefully #ifdefed out so they cannot be used on 32-bit ones.
Remember that Linux standard libraries have a common source for different operating systems. So just because you can see it in the header files does not mean that you can use it on your machine.
Cheers,
Ben
I will believe that the big-name databases truly take Linux seriously for real work when they listen to what Linus has to say on the subject of big files.
If you want a database on Linux with over 2 GB in the database today, you cannot use one of the big commericial guys. Linux has a limit on file-size for 32-bit architectures and it is unlikely to go away. If you are serious about handling more data than that, Linus wants you to get a 64-bit CPU like the Alpha. He does not want to mess up a lot of code for a case that affects relatively few people. One that will be a non-issue as the 32-bit to 64-bit migration continues.
And so, if the database vendors are serious about using Linux for databases, they have to offer us a port to a 64-bit architecture. Well Linux on the Alpha is 64-bit, here, and popular, where are the databases for it?
Regards,
Ben
I know that Jay Sulzberger of LXNY thinks that having Gates and Microsoft get tax deductions for donations to schools tied with restrictive clauses is grounds for a lawsuit.
You never know with Jay how to take what he says, he is serious but he is not known as...put it this way...the most calm and rational individual in the world. OTOH he is pretty intelligent and does generally know his stuff.
Cheers,
Ben Tilly
Face facts, one of the biggest way that Linux sneaks in the back door is because of Samba. Once it is there it grows, but Samba in many cases is the necessary first step without which Linux either would not be considered, or would have been thrown out because it could not integrate itself in the existing environment..
I therefore think that Andrew Tridgell deserves this award (and more pizza).
Cheers,
Ben Tilly
I know about Windows NT Server 4.0 vs Unix but it would be handy to have something targeted to this situation. A study that showed a realistic comparison of support costs, network usage, security issues, and hardware costs between Windows and Unix.
Why?
So that when Microsoft wants to donate to a university with strings attached, people can evaluate whether or not the Microsoft solution with the handout included will cost more than a Unix solution. If it (as it often will) does then it becomes much easier to make the argument that the offer is not worthwhile.
Otherwise cash in hand is a very hard thing to argue with...
Cheers,
Ben
That would be the current version of the old saying.
...because God doesn't trust an Englishman in the dark.
Incidentally back in my grandfather's day the wags used to add to that
:-)
Cheers,
Ben
This is truly hearsay, but I heard that the decision was made by a sub-contractor. But still it is a nice decision and Red Hat should certainly see whether they can get the standard, By appointment to... line approved. (At least for their British sales.)
:-)
Cheers,
Ben