We have small cubes here (a little under 6x8 feet); many people are scrounging spare cube parts (usually the corner desk section) and adding a second level for storage (and to block Nerf darts, rubber bands, Skittles, and whatever else gets thrown their way). In my ~45 sq. feet of space, I now have 8 computers. They are mostly nice computers (an AlphaServer 1000A, a Sun Ultra 2, a couple of Penguin Altus 3200 dual Opterons, a dual PIII, a dual PII, a PPro, and a diskless VAX X terminal), but they are big, noisy, and generate lots of heat - my cube is often 5-10 degrees warmer than the rest of the office (which has poor HVAC and ventilation in any case).
I decided today I need to find a cube mounted 19" rack.
Dad worked at MSFC in a trailer for a couple of years while they "renovated" 4487. Almost every computer got zapped because the network cables were just laid on the ground between trailers; the water pipes were the same way and froze regularly during the winter. Dad's trailer was right on the end and got hit by a Space Camp tour bus one day. Someone walked out the door of the trailer next to his one day, and the wooden steps/platform just peeled away and collapsed to the ground.
Before they even moved anyone back in to 4487, the roof and windows were leaking. They never really got it fixed before he had to move to another building.
The shuttle orbiters are in grand shape, compared to many of NASA's other work environments.
What this really means is that NASA might see a 1% budget increase instead of a budget cut next year, and after that (after Bush is re-elected or someone else is elected), it'll go back down.
At a time when some of the established Linux companies are turning away from their Open Source roots and progressively abandoning full-time commitment to Open Source Software...
Other than being FUD targeted at Red Hat (please list ONE of these "rules" that is not being followed by Red Hat), why is this news? The only established Linux company turning away from Open Source that I'm aware of is the former Caldera, and they were never a big Open Source company (most of the software they developed was closed source). With Fedora, Red Hat has the most open development model among commercial Linux vendors. It can't touch Debian, but I think it'll come close once they get all the infrastructure in place.
Red Hat's old RHL support policy was to support one major release back. So, 5.x support continued until 7.0 was released, although in many cases, only the last minor release of a major release got updates (i.e. the only 5.x version to get many updates was 5.2). In general, releases happened every 6 months, with 3 minor releases to a major, so you'd see support for a major release for 18 months after the last minor (which was usually a year after the first minor), but that was not really in writing anywhere.
By the old standards, the release of RHL 9 in April would have meant the end of support for RHL 7.x and the release of FC 1 (what more-or-less would have been RHL 10) would have ended support for RHL 8.0, but Red Hat extended support for both to the end of 2003. When RHL 9 was released, they announced the change in support policy; going forward, new releases would get 12 months of support, so RHL 9 will get support until next April.
Despite much FUD, Red Hat has never cut off support prior to the policies at the time of release. The difference is that the old policy wasn't explicitly time based, it was based on future releases (which was confusing, bringing the change back in April).
Red Hat doesn't nag you for money on the download page for their free distribution or require you to pay to join mailing lists. When I went to the Mandrake page, to get to the download link I either had to select "Send me to register" or "I'm already a member or will pay you soon".
I always forget that it is 1W in, not 1W out. Someone at work even called the FCC and talked to them about that to verify it, and he has to correct me every time. D'oh!
If you read the actual pages, they used 1.5W amplifiers. Their "lessons learned" page says "Unsure of FCC regulations"; more like FCC regulations ignored. FCC regulations for unlicensed use of the 2.4GHz spectrum for communications limit you to 1W ERP; with 1.5W amps and high gain antennas, they were well beyond the FCC regulations. Give me a big enough amp and a good antenna and I can transmit 2.4GHz a lot longer distance, at least until the feds track me down.
The "53k" limit was a problem with the way X2 worked. Blaming it on the
FCC is just a marketing scam; the fact is that the US Robotics engineers
couldn't make X2 hit 56k and still work within the pre-defined limits of
the telephone system, so they tried to blame someone else.
Lucent's 56k system could actually do 56k and stay within the limits,
but the v.90 standard didn't use Lucent's technology for that.
As to why nothing is more than 56k: that is all that a standard voice
line (or POTS line, for Plain Old Telephone System) can do. A POTS line
is carried within a DS0 (the base channel of the phone system), and a
DS0 is 64k. You can't get all 64k though, because many voice lines use
"robbed bit" signalling that takes one of every eight bits to handle
switch communication. Getting 56k at all requires that one end be a
digital line (ISDN BRI or PRI or channelized T1); you can't push 56k
through the analog to digital conversion otherwise.
The "what's next" for the telephone system is already here; it is DSL.
DSL uses different frequency bands that are not used for POTS lines but
that can be carried over the same copper reliably (more or less).
However, DSL is not a switched circuit like a modem connection; the DSL
frequencies are pulled off the line (by a DSLAM, DSL Access Multiplexer)
before the line connects to the regular phone network. So, you can't
"dial" a different DSL provider or your friend's house; you can only be
connected to one service (and any changes require a call to the DSLAM
owner, usually the phone company).
The other "what's next" was ISDN, which would give you the full 64k
channel (because signalling is always done on a separate dedicated
channel with ISDN), or 128k if you use both channels (the base ISDN line
is a BRI, which has 2 64k data channels plus a signalling channel).
However, ISDN use was slowed because it was complicated to configure
(you couldn't just plug a phone in and use it), required all new
equipment, and even the telcos really never understood it well (so when
there was a problem, it could take weeks to get it fixed).
It was originally planned that at the end of its life, Hubble would be brought back to Earth in a shuttle to be put in a museum. However, the increased inspections and repair plans now being put in place for the shuttle orbiter thermal protection system require the shuttle to go to the space station. There will still probably be one more flight to Hubble, but that will be it most likely.
We don't want another Skylab, with the whole world wondering where it will crash. Hubble is a rather large satellite (nothing like Skylab, but still quite large), and NASA doesn't want it falling on a populated area. Electronics wear out (especially in the harsh environment of space), batteries die, etc. If it is going to be brought down safely, it must be done while it is still functioning, so the de-orbit can be controlled.
Even before Columbia, there were only a couple of more Hubble servicing missions planned, before Hubble was decommissioned and replaced by the Webb telescope. The service missions have now been reduced to one, and they'll get everything that they think is reasonably possible out of it, but then they need to give up on Hubble and move on.
Telescopes on the far side of the moon would give us a new
view of the universe uninterrupted by light
You do realize that the Sun does shine on the far side of the Moon,
don't you? Just because we can't see it from here doesn't meen it is
dark all the time. The lunar day is a lot longer than 24 hours, but the
Moon still has day and night.
Many of those cards are not really RAID cards. They add additional ATA interfaces and then have a BIOS and software drivers that do software RAID. That is why they are so cheap.
If you want RAID under Linux, just use the motherboard ATA interfaces and software RAID; you'll get the same results with no driver hassles. IIRC Win2K (and maybe XP Pro) support software mirroring as well.
If von Braunn had had his way, we'd never have built the Saturn V. Instead, we'd have built the Nova, and we'd have gone to Mars.
The Saturn V could lift to Earth orbit only; the command module had an engine to go to the Moon. The design for the Nova was to be a direct to Lunar orbit vehicle; the first stage would have had nearly twice the thrust of the Saturn V. The unspoken goal of the Nova was that it would also be able to lift a vehicle to Earth orbit that was then capable of going to Mars.
I've read that the longer a project is extimated to take, the greater the likelihood of it running overtime, and the greater the degree of this overrun. Plotting expected versus actual project length looked logarithmic. I think it was in Scientific American a few years ago but I can't find the reference.
Try "The Mythical Man Month", originally published in 1975 (an expanded second edition was published in 1995). It is based on lessons learned during the development of the IBM System/360 and OS/360. It should be required reading for any CS degree (it was for me) and for any manager for any large project.
Not that I want to support SCO in ANY way, but two things:
The license under which the "historical" versions of Unix are released is a BSD style license with the advertising clause. The advertising clause makes it incompatible with the GPL, according to the license comparison at the FSF site. So, it is not legal to pull code from the historical Unix versions into the Linux kernel, unless the historical Unix code in question was licensed from UCB under the BSD license (UCB retroactively removed the advertising clause from their code).
DMR worked for AT&T on the code is now owned by SCO. So SCO does have legal ownership of that code and is allowed to control how it is released. Much of it (including the code in question) was released under the license mentioned above, but that license conflicts with the GPL as used by the Linux kernel.
So, if DMR's old Unix code was used verbatim in the Linux kernel AND it is not available under a license different than the historical Unix license, then the kernel is violating SCO's license terms.
Both my original Series1 TiVo and my newer Series2 have a section in the manual with the GNU GPL and the URL for their kernel and GNU utilities source. They've been legal from the beginning, unlike Linksys.
TiVo has been a leader in releasing the required source and a little more; they also provide the compiler toolchain used to build the kernel (which is not required, but a nice touch, since it allows users to easily build additional binaries with the same toolchain).
When you use Brightmail to sideline spam, each user opted-in gets a periodic report (default is daily, how often is determined by admin) that lists the subject, sender, date, and size of all new sidelined messages. Then you can login to a website, view the messages, redirect them to your regular mailbox if they were false positives (and also send them to Brightmail for analysis so they can fix what triggered on the message).
I just installed an upgraded spam filter server at the ISP I work for,
and we are now filtering out almost 70% of inbound mail as spam (with
basically zero false positive complaints). We combine
Brightmail with the three main
MAPS lists (RBL, DUL, and RSS),
as well as the basic DNS based checks (for valid domains, etc.) built
into the mail server, with Brightmail catching the most by far.
Our "big" servers here are Alphas running Tru64 Unix and TruCluster. TruCluster is a nice single system image cluster for high availability and load sharing. One of the nicest things about it though is that it is _integrated_ with the OS; I don't have to futz around and find this kernel patch and that application patch, etc.
One other nice thing about Tru64 is the Advanced FileSystem, AdvFS. It is like having a journaling filesystem integrated with some logical volume management. When moving from one set of disks to another, it took two commands: "addvol <new disk>; rmvol <old disk>". The filesystem automatically resizes on the fly as well, so if you do an addvol, the additional space is available instantly. And AdvFS is an integral part of TruCluster, where it becomes Cluter FileSystem, CFS. All members of a cluster have read-write access to all mounted filesystems (unless you don't want that - then you can partition each FS off).
TruCluster also has a "rolling upgrade" mode, where you can apply patch kits or even install OS upgrades without the cluster as a whole being shut down.
In general, Linux is lacking the necessary support for real 24x7 operation. Don't get me wrong, I do have Linux servers with long uptimes; one of them has been up over three years, but when I need to upgrade it, it will be down for the duration of the upgrade. Some of the problem because much of the hardware required is proprietary. There are specialized distributions available that bring some of the software components together, but then you are "locked in" to that distribution. And since no one distribution, especially not a special target one, has the market force of Solaris for example, you always have to worry about your preferred distribution going away.
Incorrect. Intel licensed some of the Alpha technology (and bought the fab that Compaq used for their Alpha CPUs), but Digital, then Compaq, and now HP owns the Alpha. This is not an unexpected release; all along Compaq and HP have said they were committed to one more full generation of Alpha CPUs (the EV7 generation). Supposedly, the third generation Itanium will incorporate some of the Intel licensed Alpha technology, and then it is supposed to "catch up" with the Alpha (so there would be no EV8 generation).
IIRC, some of the associated technologies like the switching architecture and some of the NUMA features were not licensed but held by Compaq for their Itanium servers (to give them an edge).
Slashdot Mirror
As for FAT, from what I've read the patent (patents?) doesn't cover the way Linux uses a FAT filesystem.
I decided today I need to find a cube mounted 19" rack.
Before they even moved anyone back in to 4487, the roof and windows were leaking. They never really got it fixed before he had to move to another building.
The shuttle orbiters are in grand shape, compared to many of NASA's other work environments.
What this really means is that NASA might see a 1% budget increase instead of a budget cut next year, and after that (after Bush is re-elected or someone else is elected), it'll go back down.
By the old standards, the release of RHL 9 in April would have meant the end of support for RHL 7.x and the release of FC 1 (what more-or-less would have been RHL 10) would have ended support for RHL 8.0, but Red Hat extended support for both to the end of 2003. When RHL 9 was released, they announced the change in support policy; going forward, new releases would get 12 months of support, so RHL 9 will get support until next April.
Despite much FUD, Red Hat has never cut off support prior to the policies at the time of release. The difference is that the old policy wasn't explicitly time based, it was based on future releases (which was confusing, bringing the change back in April).
Red Hat doesn't nag you for money on the download page for their free distribution or require you to pay to join mailing lists. When I went to the Mandrake page, to get to the download link I either had to select "Send me to register" or "I'm already a member or will pay you soon".
I always forget that it is 1W in, not 1W out. Someone at work even called the FCC and talked to them about that to verify it, and he has to correct me every time. D'oh!
If you read the actual pages, they used 1.5W amplifiers. Their "lessons learned" page says "Unsure of FCC regulations"; more like FCC regulations ignored. FCC regulations for unlicensed use of the 2.4GHz spectrum for communications limit you to 1W ERP; with 1.5W amps and high gain antennas, they were well beyond the FCC regulations. Give me a big enough amp and a good antenna and I can transmit 2.4GHz a lot longer distance, at least until the feds track me down.
Lucent's 56k system could actually do 56k and stay within the limits, but the v.90 standard didn't use Lucent's technology for that.
As to why nothing is more than 56k: that is all that a standard voice line (or POTS line, for Plain Old Telephone System) can do. A POTS line is carried within a DS0 (the base channel of the phone system), and a DS0 is 64k. You can't get all 64k though, because many voice lines use "robbed bit" signalling that takes one of every eight bits to handle switch communication. Getting 56k at all requires that one end be a digital line (ISDN BRI or PRI or channelized T1); you can't push 56k through the analog to digital conversion otherwise.
The "what's next" for the telephone system is already here; it is DSL. DSL uses different frequency bands that are not used for POTS lines but that can be carried over the same copper reliably (more or less). However, DSL is not a switched circuit like a modem connection; the DSL frequencies are pulled off the line (by a DSLAM, DSL Access Multiplexer) before the line connects to the regular phone network. So, you can't "dial" a different DSL provider or your friend's house; you can only be connected to one service (and any changes require a call to the DSLAM owner, usually the phone company).
The other "what's next" was ISDN, which would give you the full 64k channel (because signalling is always done on a separate dedicated channel with ISDN), or 128k if you use both channels (the base ISDN line is a BRI, which has 2 64k data channels plus a signalling channel). However, ISDN use was slowed because it was complicated to configure (you couldn't just plug a phone in and use it), required all new equipment, and even the telcos really never understood it well (so when there was a problem, it could take weeks to get it fixed).
We don't want another Skylab, with the whole world wondering where it will crash. Hubble is a rather large satellite (nothing like Skylab, but still quite large), and NASA doesn't want it falling on a populated area. Electronics wear out (especially in the harsh environment of space), batteries die, etc. If it is going to be brought down safely, it must be done while it is still functioning, so the de-orbit can be controlled.
Even before Columbia, there were only a couple of more Hubble servicing missions planned, before Hubble was decommissioned and replaced by the Webb telescope. The service missions have now been reduced to one, and they'll get everything that they think is reasonably possible out of it, but then they need to give up on Hubble and move on.
What's next to your home key?
...
My car key, my office key, my desk key,
If you want RAID under Linux, just use the motherboard ATA interfaces and software RAID; you'll get the same results with no driver hassles. IIRC Win2K (and maybe XP Pro) support software mirroring as well.
The Saturn V could lift to Earth orbit only; the command module had an engine to go to the Moon. The design for the Nova was to be a direct to Lunar orbit vehicle; the first stage would have had nearly twice the thrust of the Saturn V. The unspoken goal of the Nova was that it would also be able to lift a vehicle to Earth orbit that was then capable of going to Mars.
- The license under which the "historical" versions of Unix are released is a BSD style license with the advertising clause. The advertising clause makes it incompatible with the GPL, according to the license comparison at the FSF site. So, it is not legal to pull code from the historical Unix versions into the Linux kernel, unless the historical Unix code in question was licensed from UCB under the BSD license (UCB retroactively removed the advertising clause from their code).
- DMR worked for AT&T on the code is now owned by SCO. So SCO does have legal ownership of that code and is allowed to control how it is released. Much of it (including the code in question) was released under the license mentioned above, but that license conflicts with the GPL as used by the Linux kernel.
So, if DMR's old Unix code was used verbatim in the Linux kernel AND it is not available under a license different than the historical Unix license, then the kernel is violating SCO's license terms.TiVo has been a leader in releasing the required source and a little more; they also provide the compiler toolchain used to build the kernel (which is not required, but a nice touch, since it allows users to easily build additional binaries with the same toolchain).
This has no relevance according to Novell.
In other news, /. editors consider searching for
dupes before posting.
When you use Brightmail to sideline spam, each user opted-in gets a
periodic report (default is daily, how often is determined by admin)
that lists the subject, sender, date, and size of all new sidelined
messages. Then you can login to a website, view the messages, redirect
them to your regular mailbox if they were false positives (and also send
them to Brightmail for analysis so they can fix what triggered on the
message).
You can see our mail stats here.
One other nice thing about Tru64 is the Advanced FileSystem, AdvFS. It is like having a journaling filesystem integrated with some logical volume management. When moving from one set of disks to another, it took two commands: "addvol <new disk>; rmvol <old disk>". The filesystem automatically resizes on the fly as well, so if you do an addvol, the additional space is available instantly. And AdvFS is an integral part of TruCluster, where it becomes Cluter FileSystem, CFS. All members of a cluster have read-write access to all mounted filesystems (unless you don't want that - then you can partition each FS off).
TruCluster also has a "rolling upgrade" mode, where you can apply patch kits or even install OS upgrades without the cluster as a whole being shut down.
In general, Linux is lacking the necessary support for real 24x7 operation. Don't get me wrong, I do have Linux servers with long uptimes; one of them has been up over three years, but when I need to upgrade it, it will be down for the duration of the upgrade. Some of the problem because much of the hardware required is proprietary. There are specialized distributions available that bring some of the software components together, but then you are "locked in" to that distribution. And since no one distribution, especially not a special target one, has the market force of Solaris for example, you always have to worry about your preferred distribution going away.
DivX at 11.
IIRC, some of the associated technologies like the switching architecture and some of the NUMA features were not licensed but held by Compaq for their Itanium servers (to give them an edge).