Enterprise speaking, you would have a technical staff that was competent (or SHOULD). You are not going to have the users messing with systems. The ease of install matters not, when you will be installing essentially ghosted machines. Before you install it, you will be testing it.
In fact, gentoo, is possibly even more suited to the enterprise than RedHat or SuSE. Why? The admins have even more control. They also only have to compile a package once for each group of machines, and can deploy it to all the machines. (and to a few test machines first) Not to mention the "emerge security" which will be coming along in the mainline portage/emerge stable release fairly soon. Which will essentially allow you to fix a system at a certain point (of your own choosing) and have all the security updates.
That honestly isn't true that portage has more 'packages' than other distros. If gentoo used as many subpackages as other distros as opposed to use flags, the only one that could rival it would probably be debian.
See, portage has between 6000-8000 ebuilds in it. There are a few which don't really build anything and are lists of dependancies (see the 'kde' ebuild, it's essentially just an empty thing that requires kdelibs, kdebase, kdepim etc) However, dispite the few of these, almost all ebuilds are a whole program or library that stands on it's own. With debs or rpms, the little customization allowed by them is included in packages such as qt, qt-MySQL, qt-PostgreSQL, etc. With Gentoo there is one ebuild and there are USE flags (mysql, which can add dependencies on mysql to the qt ebuild, but doesn't add a whole extra package)
Speaking as a Gentoo user, I can be emerging a package and play quake3, admittedly I usually don't because it can get slow and laggy.
Certain things don't allow you to do that (ut2004-demo being the worst) because they are writing to the disk a whole lot, or something else. However, the things like that are rather few and far between.
Actually, I would say that that is not the case here: The Constitution specifically give Congress the power to regulate trade with Foreign Nations, and these containers were certainly used in trade.
The "with one voice" doesn't seem to add anything except to enhance the point of Congress' power to regulate Foreign Commerce, which (from the judgment) appears to be the intent of those writing the Constitution.
I see this as a case of a State going up against constitutionally Congressional Power.
It is solvable via licencing. The Inventor or Inventors get the Patent. (as the constitution states) It is likely that they will be employees and that their contract will state that it will be exclusively licenced to the company with the company able to sub-licence, and if the company merges, the licence merges.
And also, the limited time as mentioned in the constitution, should be dependant upon the Inventor MAKING it exclusive, by licencing. (or trying to after they recieve the patent.)
Now, what does this accomplish? 1) Actually follows the constitution's wording. 2) Actually benefits small inventors who apply for patents, I think. 3) It strengthens the position of people working at companies: if they are let go in a way not consistant with their contract, they lose the patent. 4) No more submarine patents.
1st amendment=part of Constitution which is the highest law. DMCA federal law, which unless it gets into the constitution (not even a treaty, which is the 2nd highest), is inferior to the Constitution, and any place where the 1st amendment & the DMCA overlap, the 1st amendment wins.
Even with instantaneous processing, you can't get around being at least one frame too late without a time machine
Bull. All you have to do is have a fast enough video processor, and dear goodness, we have those. Lets say 50fps video: all you have to do is cacluclate it within 1/50th of a second, essentially running it through three passes of filtering. Woah, what's that? Video cards can do many passes per second and achieve frame rates like 75 or 100+ on much more complex data sets. Even 2D work (for example Anti-aliasing being one of the biggest examples) makes you realize how far behind TVs are. Newer Geforce and Radeon cards should have any trouble decoding the video stream (via shaders) and doing 3:2 pulldown. Is it really that much to buy a $50-200 part to put in a $2000-$3000 TV?
Re:Why aren't tech authors into "free as in beer?"
on
Samba 3 By Example
·
· Score: 1
Mind trying out sloppyadm which sets this up (currently a bunch of redhat w/ap t & gentoo specific things, but what mostly needs to be done is a bit of modification to get slack, etc to work) I do need to update it some, but it works great for a lab that has cups+samba+ldap w/windows and linux clients, it even has provisions to install common config files (all distros) and distro-specific config files. Of course, it doesn't have a gui (yet, I am working on it!) (but frankly it's the best one I have come upon (primarily becasue it does what I want it to do, because I wrote it:) )
A friend purchased a 9500 when they got cheap, because he thought they were good enough now. Within a month or two it burned up (died/stopped working) It was under warranty, but they were out of 9500s, so they sent him a 9700. Within a month it also died, Send back, 9600 dead in box.
Different Assemblers (MSI being one responsible for 2 of the cards), put into different computers, etc. Now nvidia's assemblers may pull some crappy tricks (64-bit memory path on 5200s from Albatron), but I have never had a NVIDIA card fail except two as-is Geforce 2mx cards, I bought for something like $20 (and one works, but I believe the bios is corrupt, it's a pny who seem to use non-standard bioses)
If Matrox decides to licence their cards, so they aren't almost as expensive as a 6800, I can see an alternative for people, but until then I honestly recommend staying away from ATI, leaving only one at the moment: NVidia.
Actually it isn't that bad on integer, fp could use work. Neither is it's main selling point, it simply does decently.
Overall, it is likely cheaper to have a cluster running on c3s rather than xeons/p4s/opterons/athlons/g5 simply because of the lack of huge power reqirements (10 1GHz c3s vs a 3ghz p4, on a clusterable job will almost certainly see the p4 blasted), not to mention initial cost, which can be lower than $100 per board + processor.
(and any speed c3 with a nehemiah core will murder most anything else on common encryption protocols, except something like an ibm server with a hardware encrypt assist.) Which leads to an interesting idea for encrypting things for which the server is unable to do it's job + encryption, having a c3 act like a transparent proxy for another server.
And is it just me, or are there others who feel that developers should have to have workstation machines that are slow?
"The major bottleneck to the use of quantum computers, once they are designed and built,
is the paucity of algorithms that can make use of this power. At present there are only two:
Shor's factorization [3] and Grover's data base search [4]."
I can ask one of the co-authors of the paper if you need more information.
That's the idea, BUT what it will end up being is states, and a quantum computer might be able to break that instantly.
Some of you might remember when crypto was introduced, it was said that it couldn't be broken without millions and millions of hours on the fastest computers available (I believe it was in the centuries). It was broken in less than one or two years as I recall, based on a seti-like project well before seti existed.
As long as it resolves to ones and zeros, it can be broken. Perhaps not easily, but it can be, if it is based on any arithmatic. In the end, it may prove that traditional crypto is more secure, using huge keys. Not to mention that these quantum crypto systems ONLY work for transiting data, as aside for the 'registers' of quantum computers, there really isn't any reliable way of storing quantum data. And we don't fully understand quantum physics, thus we can't be sure there isn't a way to generate something as funny sounding as an Anti-Heisenberg (insert star trek technical sounding blubbering) field.:) I highly suspect it will turn out to be just like any other product that has promised absolute security: it is either vaporware or becomes a tool in an already large toolset.
You do know that quantum computers are likely going to be for most things slower, if they are able to do them at all, than traditional computers?
There are only a few algorithms that work on quantum computers. I know of people working on being able to 'train' neural nets to figure out the algorithm necessary, but aside from a couple of specific things that they should be able to do Real Darn Fast, they suck compared to traditional computers.
Ctrl/Alt/+- is another tool for people to use. Maybe they are happy with it. I was happy with it, and still used it up to the time when I got two monitors (recently).
It's just like a GUI and a Command Line, some tasks are more efficient in each, neither is always better.
You don't use the mouse for every action do you? Most likely people using Windows or MacOS know at least copy/cut/paste, and don't rely on the mouse all the time, because the shortcut is faster, just as the Ctrl/Alt/+- is. (Actually it really isn't a pure shortcut, because it does something different, (sarcasm) it's an EXTRA feature that Windows/MacOS don't have (/sarcasm), of course given what people consider features these days, useful to only a small percentage of the potential userbase, but touted as the next big thing (for examples see Passport/.NET, Ctrl/Alt/+-, and all sorts of other things).)
Which is better a pen or a pencil? They both allow you to write on paper, sometimes one is more useful (eg pencil when you want to erase, and pen when you don't want it to be erasible.) Just like for the games that aren't on Linux, Windows is a better platform, for the games that are on both Linux is usually faster. Different tools for different tasks.
Overall though, the new X extensions are very good. The KDE applet sits in my tray, and more often than not gets used when I wish to change screen resolutions over Ctrl/Alt/+-, but I still use it.
Did Sun GPL all the code they created for the distribution (as most but TurboLinux/SuSE/Caldera\SCO do)?
What I meant was that most distributions contribute all their code to the community under GPL (it may be distro-specific and use distro-specific things, such as anaconda, but it's GPL.) This means that things like XFree (non-gpl) don't count, nor do other things that they distribute, but didn't create. In the case of Sun, I doubt it (but hope I am wrong).
USB mouse support in linux has been as robust or more robust than Windows for quite a while. What usually is the problem is the distribution's config tools.
So have you bothered to file a bug report about it with SuSE? Or just decide to cluelessly bash Linux on/.?
Check/etc/X11/XF86Config for which device the mouse is using, and you have two options: /dev/input/mice - ALL mice (USB, PS/2, busmouse) connected to the system. This is what most people will want. (With some kernels there is a delay before Xfree starts to get the input.) /dev/input/mouseX (where X is a number starting at 0) - a specific mouse. Which may be what SuSE set up, and if so and it disconnected for any reason, the number would go up. (This is useful say when a trackpoint goes out on a laptop, to specify only the trackpad/external mouse, also the AllowMouseOpenFail option is a good option when using this)
With Windows, USB mouse support seems flaky. I haven't extensively used XP (seems not to support Compaq Presario 1700s well at all: blue screen city.), but 2000 certainly doesn't detect all mice when you plug them in. But then I suppose Windows USB mouse support is a WORKSFORYOU resolution if it were in bugzilla.:)
Pretty much everone except SuSE and TurboLinux (and Caldera, when they were not lb)
Unless you mean free as in: if it's not our Definition of free, we don't include it, then Debian is the only one (though Gentoo could be made to (all ebuilds have what the licence is, and portage should be able to filter out all those you don't want.))
Everyone's base system is pretty much the same though, in terms of it being GPL.
Can you hook up 35 printers to your PC? How many terminals does it support? Can you keep four operators busy mounting tapes on the drives connected to it?
With regards to the printers: 128 at the moment via local connections (though 127 sharing 12Mbps of bandwidth, if you include networked postscript printers, effectively unlimited) Terminals: serial (atm, just 2, though I have a couple of multi-port pci serial cables), network(effectively unlimited, and already has netbooted machines (sparcs and x86)), console(could go two console), or combination? Given that it doesn't have tapes, no.
Computers today could do it. It's just a matter of having the extra parts to handle it. Of course most mainframes are going to beat the crap out of single channel scsi raid, 2 64-bit pci busses, etc for IO. Though compared to this thing's processors, the word creamed comes to mind for cpu-intensive tasks.
Not that I disagree about the fast cpu (which is likely unable to even pull the memory speeds reqired.) Comparitively, a RS/6000 (workstations & servers) from 1997 has 4 256-bit memory paths of PC50, which puts it at the equivelent of a PC200 DIMM, and if they interlaced the banks themselves (of which I am not sure, and doubt) it would be equivelent of a PC800 DIMM, only dual-channel DDR400 can match that, from a 7 year old machine.
Essentially a Mainframe is a machine designed to be utterly reliable and have fast IO. A Supercomputer (should be) designed to (be reliable and) have as much processor power as it can, for number crunching.
Now, on certain jobs, a mainframe will beat a clustered supercomputer any day, and on other jobs the clustered supercomputer will pound the "Big Iron" into the ground.
Essentially it comes down to what they are intended for, supercomputers = number crunching, so processor power, mainframes = data storage, so IO.
Slashdot Mirror
In fact, gentoo, is possibly even more suited to the enterprise than RedHat or SuSE. Why? The admins have even more control. They also only have to compile a package once for each group of machines, and can deploy it to all the machines. (and to a few test machines first) Not to mention the "emerge security" which will be coming along in the mainline portage/emerge stable release fairly soon. Which will essentially allow you to fix a system at a certain point (of your own choosing) and have all the security updates.
See, portage has between 6000-8000 ebuilds in it. There are a few which don't really build anything and are lists of dependancies (see the 'kde' ebuild, it's essentially just an empty thing that requires kdelibs, kdebase, kdepim etc) However, dispite the few of these, almost all ebuilds are a whole program or library that stands on it's own. With debs or rpms, the little customization allowed by them is included in packages such as qt, qt-MySQL, qt-PostgreSQL, etc. With Gentoo there is one ebuild and there are USE flags (mysql, which can add dependencies on mysql to the qt ebuild, but doesn't add a whole extra package)
Certain things don't allow you to do that (ut2004-demo being the worst) because they are writing to the disk a whole lot, or something else. However, the things like that are rather few and far between.
Otherwise, what would all the people who post "It's /.ed" have to do with their days?
No plot at all, unless you count things like the t-mode messages. Goldmine for a movie, better as one of the oldest multiplayer games in existance!
The "with one voice" doesn't seem to add anything except to enhance the point of Congress' power to regulate Foreign Commerce, which (from the judgment) appears to be the intent of those writing the Constitution.
I see this as a case of a State going up against constitutionally Congressional Power.
It is solvable via licencing. The Inventor or Inventors get the Patent. (as the constitution states) It is likely that they will be employees and that their contract will state that it will be exclusively licenced to the company with the company able to sub-licence, and if the company merges, the licence merges. And also, the limited time as mentioned in the constitution, should be dependant upon the Inventor MAKING it exclusive, by licencing. (or trying to after they recieve the patent.) Now, what does this accomplish? 1) Actually follows the constitution's wording. 2) Actually benefits small inventors who apply for patents, I think. 3) It strengthens the position of people working at companies: if they are let go in a way not consistant with their contract, they lose the patent. 4) No more submarine patents.
At least, that's how it's supposed to work.
Bull. All you have to do is have a fast enough video processor, and dear goodness, we have those. Lets say 50fps video: all you have to do is cacluclate it within 1/50th of a second, essentially running it through three passes of filtering. Woah, what's that? Video cards can do many passes per second and achieve frame rates like 75 or 100+ on much more complex data sets. Even 2D work (for example Anti-aliasing being one of the biggest examples) makes you realize how far behind TVs are. Newer Geforce and Radeon cards should have any trouble decoding the video stream (via shaders) and doing 3:2 pulldown. Is it really that much to buy a $50-200 part to put in a $2000-$3000 TV?
Mind trying out sloppyadm which sets this up (currently a bunch of redhat w/ap t & gentoo specific things, but what mostly needs to be done is a bit of modification to get slack, etc to work) I do need to update it some, but it works great for a lab that has cups+samba+ldap w/windows and linux clients, it even has provisions to install common config files (all distros) and distro-specific config files. Of course, it doesn't have a gui (yet, I am working on it!) (but frankly it's the best one I have come upon (primarily becasue it does what I want it to do, because I wrote it:) )
A friend purchased a 9500 when they got cheap, because he thought they were good enough now. Within a month or two it burned up (died/stopped working) It was under warranty, but they were out of 9500s, so they sent him a 9700. Within a month it also died, Send back, 9600 dead in box.
Different Assemblers (MSI being one responsible for 2 of the cards), put into different computers, etc. Now nvidia's assemblers may pull some crappy tricks (64-bit memory path on 5200s from Albatron), but I have never had a NVIDIA card fail except two as-is Geforce 2mx cards, I bought for something like $20 (and one works, but I believe the bios is corrupt, it's a pny who seem to use non-standard bioses)
If Matrox decides to licence their cards, so they aren't almost as expensive as a 6800, I can see an alternative for people, but until then I honestly recommend staying away from ATI, leaving only one at the moment: NVidia.
Overall, it is likely cheaper to have a cluster running on c3s rather than xeons/p4s/opterons/athlons/g5 simply because of the lack of huge power reqirements (10 1GHz c3s vs a 3ghz p4, on a clusterable job will almost certainly see the p4 blasted), not to mention initial cost, which can be lower than $100 per board + processor.
(and any speed c3 with a nehemiah core will murder most anything else on common encryption protocols, except something like an ibm server with a hardware encrypt assist.) Which leads to an interesting idea for encrypting things for which the server is unable to do it's job + encryption, having a c3 act like a transparent proxy for another server.
And is it just me, or are there others who feel that developers should have to have workstation machines that are slow?
I c3 worthless posts, including this one!
"The major bottleneck to the use of quantum computers, once they are designed and built, is the paucity of algorithms that can make use of this power. At present there are only two: Shor's factorization [3] and Grover's data base search [4]."
I can ask one of the co-authors of the paper if you need more information.
Some of you might remember when crypto was introduced, it was said that it couldn't be broken without millions and millions of hours on the fastest computers available (I believe it was in the centuries). It was broken in less than one or two years as I recall, based on a seti-like project well before seti existed.
As long as it resolves to ones and zeros, it can be broken. Perhaps not easily, but it can be, if it is based on any arithmatic. In the end, it may prove that traditional crypto is more secure, using huge keys. Not to mention that these quantum crypto systems ONLY work for transiting data, as aside for the 'registers' of quantum computers, there really isn't any reliable way of storing quantum data. And we don't fully understand quantum physics, thus we can't be sure there isn't a way to generate something as funny sounding as an Anti-Heisenberg (insert star trek technical sounding blubbering) field. :) I highly suspect it will turn out to be just like any other product that has promised absolute security: it is either vaporware or becomes a tool in an already large toolset.
There are only a few algorithms that work on quantum computers. I know of people working on being able to 'train' neural nets to figure out the algorithm necessary, but aside from a couple of specific things that they should be able to do Real Darn Fast, they suck compared to traditional computers.
It's just like a GUI and a Command Line, some tasks are more efficient in each, neither is always better.
You don't use the mouse for every action do you? Most likely people using Windows or MacOS know at least copy/cut/paste, and don't rely on the mouse all the time, because the shortcut is faster, just as the Ctrl/Alt/+- is. (Actually it really isn't a pure shortcut, because it does something different, (sarcasm) it's an EXTRA feature that Windows/MacOS don't have (/sarcasm), of course given what people consider features these days, useful to only a small percentage of the potential userbase, but touted as the next big thing (for examples see Passport/.NET, Ctrl/Alt/+-, and all sorts of other things).)
Which is better a pen or a pencil? They both allow you to write on paper, sometimes one is more useful (eg pencil when you want to erase, and pen when you don't want it to be erasible.) Just like for the games that aren't on Linux, Windows is a better platform, for the games that are on both Linux is usually faster. Different tools for different tasks.
Overall though, the new X extensions are very good. The KDE applet sits in my tray, and more often than not gets used when I wish to change screen resolutions over Ctrl/Alt/+-, but I still use it.
Works and I have personally used it on mice other than the Explorer with more buttons.
What I meant was that most distributions contribute all their code to the community under GPL (it may be distro-specific and use distro-specific things, such as anaconda, but it's GPL.) This means that things like XFree (non-gpl) don't count, nor do other things that they distribute, but didn't create. In the case of Sun, I doubt it (but hope I am wrong).
So have you bothered to file a bug report about it with SuSE? Or just decide to cluelessly bash Linux on /.?
Check /etc/X11/XF86Config for which device the mouse is using, and you have two options:
/dev/input/mice - ALL mice (USB, PS/2, busmouse) connected to the system. This is what most people will want. (With some kernels there is a delay before Xfree starts to get the input.)
/dev/input/mouseX (where X is a number starting at 0) - a specific mouse. Which may be what SuSE set up, and if so and it disconnected for any reason, the number would go up. (This is useful say when a trackpoint goes out on a laptop, to specify only the trackpad/external mouse, also the AllowMouseOpenFail option is a good option when using this)
With Windows, USB mouse support seems flaky. I haven't extensively used XP (seems not to support Compaq Presario 1700s well at all: blue screen city.), but 2000 certainly doesn't detect all mice when you plug them in. But then I suppose Windows USB mouse support is a WORKSFORYOU resolution if it were in bugzilla. :)
Unless you mean free as in: if it's not our Definition of free, we don't include it, then Debian is the only one (though Gentoo could be made to (all ebuilds have what the licence is, and portage should be able to filter out all those you don't want.))
Everyone's base system is pretty much the same though, in terms of it being GPL.
With regards to the printers: 128 at the moment via local connections (though 127 sharing 12Mbps of bandwidth, if you include networked postscript printers, effectively unlimited) Terminals: serial (atm, just 2, though I have a couple of multi-port pci serial cables), network(effectively unlimited, and already has netbooted machines (sparcs and x86)), console(could go two console), or combination? Given that it doesn't have tapes, no.
Computers today could do it. It's just a matter of having the extra parts to handle it. Of course most mainframes are going to beat the crap out of single channel scsi raid, 2 64-bit pci busses, etc for IO. Though compared to this thing's processors, the word creamed comes to mind for cpu-intensive tasks.
Not that I disagree about the fast cpu (which is likely unable to even pull the memory speeds reqired.) Comparitively, a RS/6000 (workstations & servers) from 1997 has 4 256-bit memory paths of PC50, which puts it at the equivelent of a PC200 DIMM, and if they interlaced the banks themselves (of which I am not sure, and doubt) it would be equivelent of a PC800 DIMM, only dual-channel DDR400 can match that, from a 7 year old machine.
Essentially a Mainframe is a machine designed to be utterly reliable and have fast IO. A Supercomputer (should be) designed to (be reliable and) have as much processor power as it can, for number crunching. Now, on certain jobs, a mainframe will beat a clustered supercomputer any day, and on other jobs the clustered supercomputer will pound the "Big Iron" into the ground. Essentially it comes down to what they are intended for, supercomputers = number crunching, so processor power, mainframes = data storage, so IO.
Probably be easier to list those that should have exec permissions: /bin /sbin /usr /opt /lib
and /usr/local (depending upon policies)
Also don't forget noexec on the /tmp directory.