There was a discussion, but I thought that it was inconclusive, in as much as there was not universal agreement upon the type of use of kernel headers that you mention.
"This was not the final word, and several people pointed out the while Richard's point might be valid, it still did not change the status of GPLed code. There was no real resolution though... "
IANAL, but, Patents are always granted to (and filed by) individuals. Corporations receive rights to patents through agreements with the individuals who file them (generally part of the employment contract). Thus, it's quite possible for these patents to have been filed by the individuals responsible for the "innovations" before 1998 (when company was incorporated) and the company can have since acquired the rights to any eventually-granted patents.
I've burned two Debian ISO images to CDR (on Debian 2.1) without any difficulty. The process is a little more involved, but Debian is clear in pointing out that they make use of rsync so as not to tax ISO servers when they already have port servers that can handle most of the load. You _ARE_ able to ensure that you have the canonical ISO images, since after the process is complete you can run md5sum on the actual finished ISO image and compare it to the published ISO image reference.
Considering that Debian is a non-profit, I think this isn't too much additional legwork to get an image.
Something I learned from a dear departed cousin of mine is a useful tip in this type of situation (to prevent abuse of monetary gifts, ie for alcohol, etc.)
He used to buy a package of gift certificates for McDonalds/BurgerKing and could hand these out, promising warm food to those in need.
Generally what happens is that MIT's PLTO (Patent/Licensing Transfer Office, or something like that) usually owns the patent, but then gives an exclusive license to the company founded by the researchers working at MIT (relatively typical in academia). Usually works out that of the proceeds from this patent license (and they can be significant), 1/3 goes to university proper, 1/3 to founder's department (physics or mat-sci eng.), 1/3 to founders themselves.
So, it's not really a case of state-supported research where the investigators are making undue profit.
Also of interest, might be the following web-page concerning the research group of one of the principal investigators (John Joannopoulos of MIT): ab-initio.mit.edu.
Actually, when I was at MIT, I took one of Joannopoulos's graduate courses in solid-state physics and can vouch for his teaching abilities in addition to his well-known reputation within the field of electronic structure calculations.
Also of interest might be the webpage of Prof. Tomas Arias at Cornell (whom I work for now), who was a collaborator of John's at MIT up till last year: www.ccmr.cornell.edu/~muchomas.
For a little more background:
Many of the computational calculations that are used by these investigators, in situations like the one where the perfect mirror was postulated, fall into the category of "ab-initio electronic structure calculations". The "ab-initio" part, latin for "from first principles", denotes that the calculations attempt to simulate actual resultant macroscopic behavior from the much more fundamental precepts of the quantum mechanical interactions between the atoms and electrons in the material under investigation. This has some very interesting advantages, not the least of which is that the resulting calculations do not have to justify higher level assumptions, whose applicability might be less assured. That's not to imply that no assumptions are used in this process (if NO assumptions were used, even most modern supercomputers would be unable to calculate the resultant quantities of interest for any more than 4 or so atoms). As it is, typical experiments generally are able to consider 100-150 atoms, which is usually sufficient to determine many properties of interest. The main approximations that are still necessary are the free electron approximation (which mandates that atomic nuclei and core non-valence electrons are immobile compared to the much lighter valence electrons which are important for conduction) and the independent electron approximation (which stipulate that the potential felt by a valence electron is not specifically dependent on the impact of every other electron [as it would be ideally], but is instead affected by a sort-of mean-field approximation of all the other electrons' potentials). However, this independent electron approximation necessitates that the resulting Hamiltonians (energies of the system) must be found by iterative self-consistent methods, whereby each successive output is computationally fed into the algorithm as input until the result converges within certain error limits. The independent electron approximation is usually implemented in terms of either the Hartree or Hartree-Fock theories (in case you want to search for more info).
Anyway, that's all I have the energy to write about, but the websites I spoke of above, probably give links to lots more material. They also have some amazing photos of the ab-initio simulations.
I tend to agree with the above comment that the most important asset of SCO is probably its large and loyal base of VAR's (Value-added-resellers) and ISV's (Independent-software-vendors). These are regionally-based solution providers who generally have intimate relationships with their customers to provide complete focused solutions (hardware, off-the-shelf + custom software, training, support, upgrades) --- be it medical, legal, industrial, or other custom applications. The customers might not even realize that it's SCO UNIX running underneath their solution (maybe Caldera Linux in the future?), all they want is the solution and the VAR/ISV support.
Therefore, this VAR/ISV sales-force will be able to push the Linux solution of whomever acquires SCO.
I'm not affiliated with Razorfish whatsoever, but I used to casually know one of the executives there, and I think your estimate of $40k-100k for the website is awfully low. It _is_ a crime how much some of these firms are charging, but the guy I knew at Razorfish led me to believe their average fee is closer to $1 million, and it's very hard to do much of anything with them for less than $400-600k. YMMV, though.
It has always been developed and licensed under the BSD license. However, I would agree that most competition in the Free Software / Open Source RDBMS market will be between MySQL and PGSql.
I've been using Postgres v.7.0 for 1-2 months now (having previously only used Oracle), and really enjoy it.
This move on weather.com's part is not incredibly surprising (although auspicious for open source/free software's continued growth) as the parent company of weather.com is Landmark Communications, which has relatively deep connections to the open source community. The chairman of Landmark Communications, Frank Batten Jr., was personally an early angel investor in Red Hat, and now his company has funded ($25 million) a subsidiary, Great Bridge LLC to provide commercial support for the advanced BSD-licensed PostgreSQL. The press releases detailing the connection between these companies can be found here.
Slashdot Mirror
There was a discussion, but I thought that it was inconclusive, in as much as there was not universal agreement upon the type of use of kernel headers that you mention.
You can see some of it at kernel traffic. They say:
"This was not the final word, and several people pointed out the while Richard's point might be valid, it still did not change the status of GPLed code. There was no real resolution though... "
Take care,
Daniel
IANAL, but, Patents are always granted to (and filed by) individuals. Corporations receive rights to patents through agreements with the individuals who file them (generally part of the employment contract). Thus, it's quite possible for these patents to have been filed by the individuals responsible for the "innovations" before 1998 (when company was incorporated) and the company can have since acquired the rights to any eventually-granted patents.
-Dan
I've burned two Debian ISO images to CDR (on Debian 2.1) without any difficulty. The process is a little more involved, but Debian is clear in pointing out that they make use of rsync so as not to tax ISO servers when they already have port servers that can handle most of the load. You _ARE_ able to ensure that you have the canonical ISO images, since after the process is complete you can run md5sum on the actual finished ISO image and compare it to the published ISO image reference.
Considering that Debian is a non-profit, I think this isn't too much additional legwork to get an image.
-Dan
Hi,
Something I learned from a dear departed cousin of mine is a useful tip in this type of situation (to prevent abuse of monetary gifts, ie for alcohol, etc.)
He used to buy a package of gift certificates for McDonalds/BurgerKing and could hand these out, promising warm food to those in need.
I enjoyed your post! -Daniel
Generally what happens is that MIT's PLTO (Patent/Licensing Transfer Office, or something like that) usually owns the patent, but then gives an exclusive license to the company founded by the researchers working at MIT (relatively typical in academia). Usually works out that of the proceeds from this patent license (and they can be significant), 1/3 goes to university proper, 1/3 to founder's department (physics or mat-sci eng.), 1/3 to founders themselves.
So, it's not really a case of state-supported research where the investigators are making undue profit.
-Daniel
Also of interest, might be the following web-page concerning the research group of one of the principal investigators (John Joannopoulos of MIT): ab-initio.mit.edu.
Actually, when I was at MIT, I took one of Joannopoulos's graduate courses in solid-state physics and can vouch for his teaching abilities in addition to his well-known reputation within the field of electronic structure calculations.
Also of interest might be the webpage of Prof. Tomas Arias at Cornell (whom I work for now), who was a collaborator of John's at MIT up till last year: www.ccmr.cornell.edu/~muchomas.
For a little more background:
Many of the computational calculations that are used by these investigators, in situations like the one where the perfect mirror was postulated, fall into the category of "ab-initio electronic structure calculations". The "ab-initio" part, latin for "from first principles", denotes that the calculations attempt to simulate actual resultant macroscopic behavior from the much more fundamental precepts of the quantum mechanical interactions between the atoms and electrons in the material under investigation. This has some very interesting advantages, not the least of which is that the resulting calculations do not have to justify higher level assumptions, whose applicability might be less assured. That's not to imply that no assumptions are used in this process (if NO assumptions were used, even most modern supercomputers would be unable to calculate the resultant quantities of interest for any more than 4 or so atoms). As it is, typical experiments generally are able to consider 100-150 atoms, which is usually sufficient to determine many properties of interest. The main approximations that are still necessary are the free electron approximation (which mandates that atomic nuclei and core non-valence electrons are immobile compared to the much lighter valence electrons which are important for conduction) and the independent electron approximation (which stipulate that the potential felt by a valence electron is not specifically dependent on the impact of every other electron [as it would be ideally], but is instead affected by a sort-of mean-field approximation of all the other electrons' potentials). However, this independent electron approximation necessitates that the resulting Hamiltonians (energies of the system) must be found by iterative self-consistent methods, whereby each successive output is computationally fed into the algorithm as input until the result converges within certain error limits. The independent electron approximation is usually implemented in terms of either the Hartree or Hartree-Fock theories (in case you want to search for more info).
Anyway, that's all I have the energy to write about, but the websites I spoke of above, probably give links to lots more material. They also have some amazing photos of the ab-initio simulations.
-Daniel
I tend to agree with the above comment that the most important asset of SCO is probably its large and loyal base of VAR's (Value-added-resellers) and ISV's (Independent-software-vendors). These are regionally-based solution providers who generally have intimate relationships with their customers to provide complete focused solutions (hardware, off-the-shelf + custom software, training, support, upgrades) --- be it medical, legal, industrial, or other custom applications. The customers might not even realize that it's SCO UNIX running underneath their solution (maybe Caldera Linux in the future?), all they want is the solution and the VAR/ISV support.
Therefore, this VAR/ISV sales-force will be able to push the Linux solution of whomever acquires SCO.
I'm not affiliated with Razorfish whatsoever, but I used to casually know one of the executives there, and I think your estimate of $40k-100k for the website is awfully low. It _is_ a crime how much some of these firms are charging, but the guy I knew at Razorfish led me to believe their average fee is closer to $1 million, and it's very hard to do much of anything with them for less than $400-600k. YMMV, though.
PostgreSQL was originally developed at Berkeley.
It has never been released under a GPL license.
It has always been developed and licensed under the BSD license. However, I would agree that most competition in the Free Software / Open Source RDBMS market will be between MySQL and PGSql.
I've been using Postgres v.7.0 for 1-2 months now (having previously only used Oracle), and really enjoy it.
This move on weather.com's part is not incredibly surprising (although auspicious for open source/free software's continued growth) as the parent company of weather.com is Landmark Communications, which has relatively deep connections to the open source community. The chairman of Landmark Communications, Frank Batten Jr., was personally an early angel investor in Red Hat, and now his company has funded ($25 million) a subsidiary, Great Bridge LLC to provide commercial support for the advanced BSD-licensed PostgreSQL. The press releases detailing the connection between these companies can be found here.