I think in case of SuSE they can state that
it is SuSE as long as it does contain YAST
(which is not open source and it is a core
part of the distribution).
Remember that Linus
choice to create a Linux, even though Minix and FreeBSD already
existed.
Linus decided to create Linux because minix did not satisfy his needs. BSD for x86
(386BSD, not even speaking of FreeBSD) did not exist at that time. I am former user of 386BSD
0.1 and NetBSD up to 0.9. But when I discovered
Linux, it was way ahead of NetBSD (speaking about kernels around 0.99.11 and SLS
distribution): shared libraries, much better
stability, and wider hardware support.
Linux DOES
NOT handle 24 processors just now... we need to fix this!)
There is (almost) nothing to fix. Linux kernel
does indeed handle more than 24 processors.
Check some recent post on linux-kernel (I can
remember some post about a 128GB-equipped alpha).
There is no explicit limit to the # of CPUs
Linux kernel can handle. It is just that it is not
known or proven tha Linux scales well on this amout of CPUs. But the work on improving scalability is in progress (NUMA memory allocators, etc). Stay tuned:-)
I detect which users would moderate post in accordance with other users and then give them a higher weighting.
This can be abused quite easily - user can just display posts with rating +5, and moderate them up. You would give him credit based on nothing more than the ability of read the article rating.
The reason is that the driver interfaces haven't stabilized to the point where it makes sense to define a binary interface for drivers. The interfaces are making signficant and sometimes radical changes that are necessary to keep the kernel improving. Once the kernel has developed "enough", the driver interface (at least for certain types of drivers) will probably stabilize and become de facto standard or be blessed by Linus as standard.
The situation is more different than you would think. Linus (and many other kernel hackers) thinks that there is no need for supporting binary-only drivers and thus make no effort of keeping the driver interface stable over time.
Many binary-only drivers even violate the GPL by using the inline functions from the (GPLed) kernel header files. I think Linus gave only one explicit exception from this rule to some company (etinc, I think).
IMHO it is hard (if not impossible) to write (and especially maintain) the Linux driver not covered by GPL. You basically end up duplicating the changes in the kernel develoment yourself.
Speaking from the point of view of kernel driver author/maintainer it is the best thing for maintainance to have your driver included in the mainstream kernel tree. Core kernel hackers often modify drivers in the kernel tree to reflect changes of the driver interface, so you don't have to track closely the kernel development.
-Yenya -Yenya --
Re:It doesn't look like this will happen but..
on
Akopia Buys Minivend
·
· Score: 1
The key in my previous response is make your proprietary extension. If the closed-source vendor does this and the extension gets popular, you are out of luck and have to re-implement the same extension. This way GPL is superior in terms that it prevents unnecessary project forking and duplicating of work.
See also a discussion on slashdot on this topic from November 18.
Of course, it depends on what is your primary goal. If you develop the software for yourself and for fun, BSD license can be appropriate, because you don't care on what will happen with your code. OTOH, when you develop software for others, you may want to ensure they will have the same rights as you and that the project can continue without you with the same terms. Thus GPL is more about protection of users (as opposed to programmers).
The above text is not meant to start a license flame war, it is just my opinion.
-Yenya --
Re:It doesn't look like this will happen but..
on
Akopia Buys Minivend
·
· Score: 1
How is that the beauty of the *GPL*? The BSD license is the same as are other open source licences.
Nope. You can get BSD-licensed code, make your proprietary extension, and release the result under non-BSD license (even without releasing the source.
Re:Thread support in BSD?
on
Java 2 For BSD
·
· Score: 2
I don't see why there would be a difference [...]
The kernel-supported threads do matter when you have multiple processors or you want to do stuff like asynchronous I/O. In traditional UNIX, processes cab be blocked inside I/O operations (such as waiting for a page to be read into a mmap-ped region). If you have a kernel-supported thread model, you can run another thread even when the first one is waiting in page fault. And you can take advantage of multiple CPUs as well.
So there really can be a difference between running JVM with kernel-supported thread model and JVM with user-space threads.
Alex de Joode (the openssh.org owner) is not a domain squatter. The domain was registered before the OpenSSH project was announced (see the discussion at/. for details).
Not related to this, this guy runs one of the best archives of crypto/security-related software for RedHat, the ftp.zedz.net (which used to be ftp.replay.com before they sold the domain).
Hey, I wonder if IBM is actually gonna write a NUMA layer for linux? I mean, if they don't then all you end up with a buncha 4-way rack mounted linux boxes.. for $365,000 apiece.
The (preliminary) NUMA support for page allocations is already there, written by Kanoj Sarcar of SGI. Look at linux/mm/numa.c
Free/NetBSD have been around longer than Linux, but they didn't get the attention because they're more concerned with refining the code than writing press releases and speaking at conferences.
I don't want to jump into an OS flamewar, but I have to correct the above statement. Both NetBSD and FreeBSD projects were founded later than Linux. The BSD UN*X has been there longer than Linux for sure. 386BSD and Linux are about the same age, but the FreeBSD/NetBSD projects as well as the OpenBSD project are younger than Linux.
I think (no flames, please) the success of Linux should be attributed to its license and to its more open development model instead of speaking at the conferences and writing the press releases (Do you remember any press release by Linus except for the press releases for Linux 1.0, 1.2, 2.0 and 2.2?).
It is interesting to see that the monopoly in the telecommunications and problems related to it is the same problem in the post-communist country such as Czech Republic as it is in relatively rich country - Australia.
We in the Czech republic have the monopoly in the voice calls held by the Czech Telecom, the only voice calls operator here (not including the cell phones). The market in the data connections has been de-monopolized for a while, and this of course caused the rapid lowering of prices of the data connections (leased lines, frame relay etc).
The monopoly in the area of the voice calls still lasts here. The monopoly guaranteed by law was expected to end by the end of this year, but the government moved the factical end of the monopoly two years further by proposing the change of the law. The only purpose of this was to increase the value of the Czech Telecom (the only voice calls operator here, not including cell phones), because they want to sell it. The bad thing is that this increase of the value is in fact paid by customers of the Czech Telecom.
There is an image checkbox that does the same thing! Which blocks images (read ads) from servers that aren't the originating server.
It is not useful (at least until it is possible to fine tune it), because many of the web sites (like/.) have a separate server for images, highly tuned for a static data (khttpd?). So with these sites it would be the same like disabling loading the images at all.
I think there should be a more fine-grained solution. And why implement any in browser, when there is a separate one (JunkBuster, available even in RPM format). --
However, even serial console support isn't perfect. After all, how do you send the three-finger salute over a serial line?
I think Sun uses the serial line break to get you to the PROM monitor (the "ok"-prompt). I think we can use the same for ctrl-alt-del or SysRQ.
However, this behaviour can be quite annoying, because the break signal occurs also when you power off the serial terminal. We have a Sun E5k here connected to the Linux console server with Cyclades serial board, and it falls into the PROM monitor prompt when we power off the console server.
I am a maintainer of ftp.fi.muni.cz (aka ftp.linux.cz), which is one of the official mirrors of RedHat Linux. On the RedHat mirror-list they asked us (mirror maintainers) to mirror the 6.1 release, but to not make it available until Monday 4pm UTC.
But: They left their FTP servers open for public downloads of the RH6.1, which does not make sense for me (they have a separate host rh-mirror.redhat.com for mirrors downloading).
Slashdot Mirror
I think in case of SuSE they can state that it is SuSE as long as it does contain YAST (which is not open source and it is a core part of the distribution).
-Yenya
--
Linus decided to create Linux because minix did not satisfy his needs. BSD for x86 (386BSD, not even speaking of FreeBSD) did not exist at that time. I am former user of 386BSD 0.1 and NetBSD up to 0.9. But when I discovered Linux, it was way ahead of NetBSD (speaking about kernels around 0.99.11 and SLS distribution): shared libraries, much better stability, and wider hardware support.
-Yenya
--
There is (almost) nothing to fix. Linux kernel does indeed handle more than 24 processors. Check some recent post on linux-kernel (I can remember some post about a 128GB-equipped alpha).
There is no explicit limit to the # of CPUs Linux kernel can handle. It is just that it is not known or proven tha Linux scales well on this amout of CPUs. But the work on improving scalability is in progress (NUMA memory allocators, etc). Stay tuned :-)
-Yenya
--
http://decibel.fi.muni.cz/phantom/.
-Yenya
--
This can be abused quite easily - user can just display posts with rating +5, and moderate them up. You would give him credit based on nothing more than the ability of read the article rating.
-Yenya
-Yenya
--
The situation is more different than you would think. Linus (and many other kernel hackers) thinks that there is no need for supporting binary-only drivers and thus make no effort of keeping the driver interface stable over time.
Many binary-only drivers even violate the GPL by using the inline functions from the (GPLed) kernel header files. I think Linus gave only one explicit exception from this rule to some company (etinc, I think).
IMHO it is hard (if not impossible) to write (and especially maintain) the Linux driver not covered by GPL. You basically end up duplicating the changes in the kernel develoment yourself.
Speaking from the point of view of kernel driver author/maintainer it is the best thing for maintainance to have your driver included in the mainstream kernel tree. Core kernel hackers often modify drivers in the kernel tree to reflect changes of the driver interface, so you don't have to track closely the kernel development.
-Yenya
-Yenya
--
See also a discussion on slashdot on this topic from November 18.
Of course, it depends on what is your primary goal. If you develop the software for yourself and for fun, BSD license can be appropriate, because you don't care on what will happen with your code. OTOH, when you develop software for others, you may want to ensure they will have the same rights as you and that the project can continue without you with the same terms. Thus GPL is more about protection of users (as opposed to programmers).
The above text is not meant to start a license flame war, it is just my opinion.
-Yenya
--
Nope. You can get BSD-licensed code, make your proprietary extension, and release the result under non-BSD license (even without releasing the source.
-Yenya
--
-Yenya
--
The kernel-supported threads do matter when you have multiple processors or you want to do stuff like asynchronous I/O. In traditional UNIX, processes cab be blocked inside I/O operations (such as waiting for a page to be read into a mmap-ped region). If you have a kernel-supported thread model, you can run another thread even when the first one is waiting in page fault. And you can take advantage of multiple CPUs as well.
So there really can be a difference between running JVM with kernel-supported thread model and JVM with user-space threads.
-Yenya
--
Not related to this, this guy runs one of the best archives of crypto/security-related software for RedHat, the ftp.zedz.net (which used to be ftp.replay.com before they sold the domain).
Alex de Joode is definitely NOT a cybersquatter.
-Yenya
--
The (preliminary) NUMA support for page allocations is already there, written by Kanoj Sarcar of SGI. Look at linux/mm/numa.c
-Yenya
--
I don't want to jump into an OS flamewar, but I have to correct the above statement. Both NetBSD and FreeBSD projects were founded later than Linux. The BSD UN*X has been there longer than Linux for sure. 386BSD and Linux are about the same age, but the FreeBSD/NetBSD projects as well as the OpenBSD project are younger than Linux.
I think (no flames, please) the success of Linux should be attributed to its license and to its more open development model instead of speaking at the conferences and writing the press releases (Do you remember any press release by Linus except for the press releases for Linux 1.0, 1.2, 2.0 and 2.2?).
-Yenya
--
su auth required pam_wheel.so
Welcome to the flexible and configurable world of UNIX.
--
We in the Czech republic have the monopoly in the voice calls held by the Czech Telecom, the only voice calls operator here (not including the cell phones). The market in the data connections has been de-monopolized for a while, and this of course caused the rapid lowering of prices of the data connections (leased lines, frame relay etc).
The monopoly in the area of the voice calls still lasts here. The monopoly guaranteed by law was expected to end by the end of this year, but the government moved the factical end of the monopoly two years further by proposing the change of the law. The only purpose of this was to increase the value of the Czech Telecom (the only voice calls operator here, not including cell phones), because they want to sell it. The bad thing is that this increase of the value is in fact paid by customers of the Czech Telecom.
-Yenya
--
It is not useful (at least until it is possible to fine tune it), because many of the web sites (like /.) have a separate server for images, highly tuned for a static data (khttpd?). So with these sites it would be the same like disabling loading the images at all.
I think there should be a more fine-grained solution. And why implement any in browser, when there is a separate one (JunkBuster, available even in RPM format).
--
I think Sun uses the serial line break to get you to the PROM monitor (the "ok"-prompt). I think we can use the same for ctrl-alt-del or SysRQ.
However, this behaviour can be quite annoying, because the break signal occurs also when you power off the serial terminal. We have a Sun E5k here connected to the Linux console server with Cyclades serial board, and it falls into the PROM monitor prompt when we power off the console server.
--
But: They left their FTP servers open for public downloads of the RH6.1, which does not make sense for me (they have a separate host rh-mirror.redhat.com for mirrors downloading).
Does anybody have an explanation for this?
-Yenya