Apt-get. Which is an update program, and not a package program. Mandrake has MandrakeUpdate (along with stable and cooker sources), and I'm sure most other rpm based distros have similar tools.
Ever tried to use those tools ??
You are right though - this is not strictly a deb vs rpm issue.
More packages in the standard tree - fewer compiled by other people who are not part of the distribution. Again, squat to do with deb versus rpm. And for that matter, Mandrake's Cooker has most of what I want, being an open submission of Mandrake rpms.
Actually this has a ton to do with reliability. The entire point of having a distribution is that they will ensure that packaging is reliable and consistent. When third party vendors start offering Redhat RPMs, or Mandrake RPMs, or you start using contrib RPMs, package dependencies get less consistent, and the machine is just not as clean.
I get everything I need at debian.org in woody non-us and non-free, all packaged by debian maintainers.
Debconf. Okay now THATs more like it
Debconf is a little scripting conf that allows packages to pop up questions in console windows during the package installation that ask you if you would like to clobber your existing configuration files with new ones or not. The default is always do not clobber. Or, a package can pop up a reminder about a configuration issue associated with a package, and either show or do not show the reminder in the future. Lastly, they have help hints if you cannot resolve a dependency that tell you what to do !! These have helped me a few times.
This is a dramatic change in computer using from Windows/Mac to linux.
First, recognize that linux can do just about any network function possible on a computer. Most of these require some sys admin skills to accomplish, and have HOWTOs included with the distribution.
Other. simpler functions, have a variety of ways to receive help. Usenet is a great resource for linux. So are the help documents for KDE/GNOME. So are man pages. So are web sites.
In general, for linux, you NEVER call a help line, or consult a hard copy. The references are on the computer or on the net. Just search for what you want with Google, check the HOWTOs, acclimatize to using man -k and locate, and enjoy the ride.
You will eventually come to realize that help resources for linux dwarf those available for vendor operating systems, since the help inevitably comes from the user base.
The software installs as encrypted software plus key in your CPRM sector.
Any use will require encryption keys plus encrypted media.
No way in he!! open source will ever be allowed to read CPRM sectors, as that would make the encryption keys copyable.
Totally illegal to defeat this measure under DMCA.
Copying original enrypted media is accepted as commonplace - the protectiion resides in the keys that you cannot even legally see unless you are software provided by the media copyright owners.
That is right - the installation program will give more rights to YOUR hard drive to copyright owners than to you.
That is another thing I like about Debian. Although I can understand why a beginner would dislike it.
The packaging is done with a finer granularity. They often choose a minimal number of packages to, say, have a working emacs. Then you can apt-get all the elisp you need to increase its function.
The same with tetex. There are, I think, three packages that are necessary. Then you can add docs, fonts, and postprocessors to make it tetex on 'roids.
But it is really about choice and flexibility - the linux way. And I think this applies especially to Debian. Of course, flexibility also means more knowledge is required to achieve the functionality you desire.
trying to associate single neurons with particular brain processes is like trying to find a needle in a haystack.
No one was suggesting assigning functions to single neurons - but rather studying the nervous system on the basis of the signals of single neurons. I would suggest that there is an enormous difference. Researchers have used signal detection theory in several well known examples in the labs of Ranulfo Romo and Bill Newsome to demonstrate that signals of some neurons, when averaged together in groups of around 100, can predict the decision of the animal in discrimination trials. These studies on relatively simple percepts can and will be elaborated into more complex behaviors as time progresses.
Using microelectrode technology we have been able to verify that sensory disturbances exist in focal hand dystonia, a sensory-motor disorder that can be understood as a complication of repetitive stress injury. Such detailed observations as we made cannot be made with EEG. However, MEG studies were started after the animal studies to demonstrate recordings that were abnormal and predicted by our hypothesis. (EEG was inappropriate because the area in question is orthogonal to the radial vectors picked up in EEG).
This is just one example of a case in which recordings made in lab animals transferred nearly completely to humans in the wild.
EEG-type monitoring also does not measure the level of anything - it is only sensitive to changes in vector averaged voltages. Whereas signals carried by neurons have bandwidths of around 1-4 kHz, EEG hardly detects any energy above 100 Hz. All you get is a grossly averaged signal envelope that represents some aspects of action potentials and synaptic currents - in hypothesis only. If all the neurons in a cortical area had very high firing rates, but were largely asynchronized, EEG would measure nothing.
Even knowing the neural signals exactly it is very difficult to predict the EEG with a high precision.
Even so, we have studied field potentials - a kind of greatly amplified EEG - in depth in very simple cerebral cortical systems. It is far from obvious, even in such a simple case, how one can go from field potentials to neural signals.
The signal processing of the brain represents such a challenge because the purest signals can be measured only by microelectrodes with tip exposures of 5 microns or less - as originally described by Galambos in 1943. That presents an enormous technical difficulty that challenges investigators still.
If you've done EEG you know the limitations, and the variance, and the problems of uneven electrical conductance interfaces, and the studies to hypothesize what synaptic and action potential derived currents contribute to signals. I speak with EEG researchers regularly, and respect their work immensely. I don't think they would suggest a helmet such as the one described in the article would be adequate for much.
And the gamma band (25-40 Hz)?? What about the gamma band ??
The one really invoked in attention ??
Truly, claims that anyone understands if or how a toy like this might work are mere pie in the sky. We'd need to understand how attention worked before we could train it using EEG waves that demonstrate at best a very weak noisy reflection of SOME but certainly not MOST brain events,
In animal experiments people recorded EEG in the 1930s and 1940s as soon as amplifier technology began being applied to Neuroscience. Then in 1943 a researcher named Galambos (famous for co-discovering echo-location in bats) saw that using very small electrode tip exposures allowed recording from single neurons.
This breakthrough led the animal researchers to all but throw away EEG as a useful tool, although there are tons of human data still. But the problems with the EEG are several. First, it only reflects a vector averaging of many million neurons and synaptic currents. And it only reflects the dot product of that average with a radial vector.
Researchers estimate about 100 neurons in one area of the brain would be sufficient to carry the information in a percept, with perhaps 5, possilby as many as 10, brain areas involved. Such things are immeasurable by EEG. The things that are measurable are the oscillating features of brain processing, which could for all we know be epiphenomenonological. Or not. We really don't know yet. But the studies of the signals carried by single neurons clearly bear close relation to brain processing, and have been very difficult to relate to EEG-type monitoring.
I think its great what they're doing. But, Isn't debian mostly alpha stuff. I mean it's always in the testing phases so what's the real point here?
Try running potato - the stable distro.
Woody is unstable. If you apt-get upgrade your woody regularly you will occasionally have some fun manual work to do. Like that upgrade last week that broke perl and thus debconf.
But the debian releases are in general more stable and MUCH MUCH easier to maintain than any RPM based distro. The packaging is just clean.
The reason the packaging is clean is simple. There are 644 Debian packagers, most of them system administrators running Debian for a living. It is human nature to be lazy, so these administrators each do their job on Debian so that their day jobs are easy and they can play quake while RedHat, Mandrake, and SuSE admins are resolving package dependencies and compiling unsupported packages from tarballs.
As Ralph Nader would say, the federal government consists of people handing out more and more to the major corporations, and stomping all over consumers' rights. It is really a shame that we have evolved into a government of the corporations, and not a government of the people.
Major corporate mergers like this one should be seriously challenged. Instead, once a REPUBLICAN goes into office, the merger goes through unanimously. I don't think it would have taken too much longer had Gore been president-elect, though.
Another Naderism "The only difference between Gore and Bush is how fast their knees hit the floor when big business walks in the room."
It's time for RMS & ESR to behave in the same fashion, IMO.
RMS lives in a world of free software. He dislikes M$, but largely he preaches free software to the masses. He practices what he preaches, and is not involved in commercialism.
ESR is a poster boy for Open Source commercial movement.
Lumping them together is doing a large disservice to both of them.
Big businesses like accountability, someone they can point a finger at and say 'Make it work'.
IBM, or RedHat, or VA Linux, or dozens of other companies will support linux and fix problems for you, allowing you to hold them accountable.
I still think this is a straw argument - You can CHOOSE anyone in the world, and they would have source code access that allows the fixing of ANY bug. That will NEVER work with a closed source vendor.
Microsoft wants to drive more traffic to its web site. Its security postings are one mechanism to do so. That takes precendent over things like full disclosure, or serving the security community.
Web traffic is $$.
Don't even think Microsoft cares about security - they don't except for its ability to make them look bad. If they can market something as secure, it really doesn't matter whether it is or not.
And this is a direct attempt to hit BugTraq squarely in the wallet by taking most of their web traffic, and having them click through to Microsoft.
Re:Does this potentially kill Debian?
on
An RPM Port Of APT
·
· Score: 5
Apt draws volunteers to Debian because it makes Debian better than most of the other distros. But if the value of Debian is replicated somewhere else, w/out all of the inconveniences of Debian, then your volunteer force diminishes.
Here begins some serious speculation.
1) Apt on other distros will not work. Apt depends on package dependencies being done very cleanly, and this is simply not true in any other distro to the same extent that it is in Debian. The other distros need not only apt, but they also need a packaging policy.
2) Debian is self-supporting. People who find Debian and enjoy it because it is done for the benefit of its volunteers generically enjoy the distro. This is not going away any time soon. One might argue that Debian is competitive with develops with other distros, but I don't think that is true. Other distros pay their supporters, and Debian is still a distro of volunteers.
3) Debian developers are among the most stringent Free Software supporters. They are in it to create the best Free Software distribution possible. Many people think they already have it. There are literally no challengers - distributions with strict Free Software guidelines.
4) Debian is an active development environment. Apt is just one example - it is not a killer app in and of itself. Debian initscripts are better (IMHO) than those of the other distros I have checked. Debian security is up there as well.
5) Debian has more packages than Redhat or Mandrake or SuSe. They have these packages in their 'official' distribution - not available packaged by someone else at rpmfind.
Basically, I think the care in packaging of Debian is about 18+ months ahead of anyone else. And for that reason I can see no reason to even consider another distribution for my boxes.
We use Redhat at work. I get called a weenie for supporting Debian. But administering my Debian boxes takes 1/10th the time it takes me to administer my Redhat boxes. And that is the only reason I need to stay with Debian. YMMV.
Re:Does this potentially kill Debian?
on
An RPM Port Of APT
·
· Score: 2
Don't get me wrong, I like Debian for a lot of reasons, but apt is the absolute killer app that keeps me on Debian. I now find myself wondering, if RedHat adopts apt, what would happen to Debian.
The beauty of Debian is how cleanly it is packaged - not just apt. It is a distribution done by system administrators largely for themselves. It is not going anywhere. As long as it still benefits the package maintainers, Debian will thrive.
Other distributions are packaged for profit. That brings in other motives. Distributions will sell themselves as packaging things more rapidly or packaging things for 586 instead of 386 (like Mandrake). Distributions will fund a lot of development and sell themselves as containing the latest compiler (Redhat). Debian is packaged because it works the way a bunch of system administrators think Free Software should work.
And there will always be plenty of room for that. If you and/or others go running to Redhat once apt is working well with rpm, so be it. Debian will continue to thrive because it cannot die. It can't lose its funding because it hardly has any. Debian works because volunteers make it work (and a very small but growing number of paid developers).
This is a great next step for rpm based distributions. However, the cleanliness of debian packaging is only part apt.
Most of it comes from thorough policy and packaging guidelines from The Debian Documentation Project. Until other distributions develop such comprehensize packaging policies, the package will not interrelate as well (read - dependency problems will screw up apt). Debian maintainers spent a lot of time thinking out these compehnsive guidelines.
I can rarely upgrade Redhat distributions cleanly without tons of rpm commands ignoring dependencies - however, I find this trivially simple with debian. And the capabilities of dpkg and rpm offer no advantages. But the packaging policies do.
Apt will help a lot though.
This also shows that competition in Free Software is good. If debian innovates, the innovations can be copied to rpm based distros. And everyone wins.
FFS is basically a log-structured file system (like ReiserFS) that also uses soft updates. This is a way that the file system can determine if the files are intact or not at every point in time.
For the development TUX2 (which I've read more about), the file system writes a tree of inodes to disk. They are organized, and have a sort of 'completeness' bit. The last thing in the tree written is the completeness bit, which then allows the file system to be updated atomically without a journal. Basically, you write a bunch of files to disk, which are recognizable as being grouped. Then, the last thing you do is write a bit on disk that allows you to see that the group has been written intact. So in a recovery, you merely check groups one by one to verify their completeness bits are written. In more advanced schemes the file system can also tell you which of several groups are potentially not intact to begin with. Recovery from crashes are as fast or faster than journalled file system crashes. But file systems still need fscks. You cannot guarantee the integrity of data simply because the file system didn't crash.
I've seen numerous tests of FFS in massive compiles in which the compilation is just allowed to end when the power goes off. FFS does extremely well.
However, for the name server, an administrator wants uniformity. I recently had a situation where one group had name servers on Bind 4 OpenBSD, and everyone else was on Bind 8
Compiling BIND 8 is not exactly rocket science. It is what I do for my name servers anyway. But stop hyping on name servers - they are a little task suitable to a 386 in a closet running linux kernel 2.0. Setting them up on a large number of computers in a small area is just opening potential doors. Named is the number one mechanism of computer breakins in the last 5 years.
You will also note that Red Hat installs Bind 8 as a caching server only, and I've seen networks with hundreds of UNIX workstations, all running such caching servers - named can be considered client software, although I understand that nscd can now serve a similar purpose.
I find this to be quite silly. A reasonable named can serve a few hundred very active computers without issue. Matter of fact, our UCSF campus uses two on the entire campus serving tens of thousands of machines. Redhat has no business setting up named by default in any capacity - they caused hundreds if not thousands of break-ins by doing this in Redhat 5.x releases. If you want named to run, you ought to be able to figure it out. Named is not a garden variety server task. Any server task should be considered seriously as a potential breakin mechanism. If it is not really necessary, it should be closed.
Unless, of course, you relish reinstalling the OS.
As far as mature memory management goes (if I understand your point correctly), I think an administrator will try to avoid swap/paging on a mission-critical system. And how many high-capacity, heavy-load system run with a single processor?
This is much more pervasive. A single large compile will push almost any machine to the limits of its memory management. You ought to be able to retain use of the mouse at the same time you compile. This is MUCH more robust on the BSDs. This is something I would notice on any machine I use on my desktop - and I am a very conservative users of resources.
Things have got to scale. Right now, both Linux and BSD fall short on this point. I look forward to the day when this is no longer the case.
Linux outscales almost all commercial Unices using the 2.4 kernels with respect to load scaling and maximal TCP/IP throughput. That was entirely the point of the development series. So it is coming.
Actually, ReiserFS is a good deal faster than ext2,
This is sort of a loaded statement, as it depends on the nature of the data. In particular, if the blocksize is small or comparable to the average file sizes, then it is pretty much a wash. In cases with lots of files smaller than the ext2 blocksize, Reiserfs will have an edge.
and FreeBSD's FS is inbetween ext2 and ReiserFS in speed. I don't know why people say journeling FSs are slow
Maybe they benchmarked them ?? Seriously, FFS is a really solid file system, different from ext2 or ReiserfFS.
the issue with journalling is that you need to keep your journal synchronous on disk. This creates a bit of inflexibility in the VFS used by the kernel. Soft updates allow a lot more flexibility, as does tux2, another soft update system using atomic updates of phase trees. The issue is whether restructuring the file system to allow atomic updates causes you more speed loss than adding a journal. Generally journalling will lose, although ReiserFS has other reasons that make it fast. Those will largely be copied into soft updates FSs like Tux2, and you will be able to choose a faster soft update file system or a slower journalled system.
I've never seriously used it, but BSD: lacks a journaled file system
But has soft updates, which are a substantial improvement over ext2 with respect to recovery from crashes. More seriously, what advantage does journalling have over soft (atomic) updates, and what advantages do soft updates have over journalling ?? I think reasonable arguments can be made that soft updates are faster, and at least as crash tolerant as journalling, and a heck of a lot easier to program and maintain wrt the VFS layer.
only recently migrated to the ELF format
So???
still uses Bind 4 (OpenBSD specifically)
And this one is really relevant for those 0.01% of machines begin used as name servers.
Does BSD have a multi-threaded IP stack? How does BSD perform on Mindcraft? Linux has been playing catch-up in this space for some time, and may have a big lead.
Actually, *BSD does much better with their IP stack than linux 2.2.* and 2.0.*. *BSD also does much better under heavy loads because it has a more mature memory management scheme. Linux is supposed to work on this in the next devel series now that more fine grained SMP locking is present.
But seriously, how much advantage do you reckon a multi-threaded IP stack makes on a single processor machine ??
The free BSDs are a very fine choice for a kernel and base utilities. There are some areas in which linux is better, and others in which the BSDs are better. Generally common server tasks work out better for *BSD than linux, and application availability and marketing are stronger for linux.
Linux's big recent push has largely oriented around big hardware - SMP, multiple NICs,... but they've actually lost some utility in memory management that is more relevant for workstation users. So you could argue linux is now really good for big iron tasks, but the BSDs have really clean memory management that allows them to be heavily loaded and perform well.
Well, linux has already outnumbered *BSDs and OSX combined.
But that is largely irrelevant. With the introduction of autoconf, the open source components of these technologies will complement each other. Openssh was taken from openbsd to *BSD and linux. GNOME and KDE are largely linux developed, but work fine on *BSD.
The largest linux companies look at big business UNIX and Microsoft as the competition for different markets. *BSD and linux will both continue to grow at the expense of Microsoft and mainframe Unices. The market dynamics may have a few people going from linux to *BSD (and fewer going the other direction), but the changes in user base for linux and *BSD are coming not from each other but from Microsoft and mainframe Unix.
The media loves to play up battles, like KDE/GNOME, Redhat/Mandrake... but the reality is that KDE and GNOME help each other more than they hurt each other through competition. The same is true of Mandrake and Redhat. Any improvements made by open source companies in software lead to strengthening of all open source companies' software.
However, I clearly think that if Gore continues to go ahead with his lawyers in front of Democrat judges (who already have rewritten the law, in effect changing the rules of the game after the ball has been put in play), he's going to destroy his party.
The judges said the law was inconsistent. Therefore it needed interpretation. The interpretation was that the counties could recount if they so chose, and the certification deadline needed to be moved to accomodate. That is hardly changing the rules. So the counties that had large problems with ballots recounted manually. Or, rather, one of them did. Another was harassed by people flown in courtesy of the Republican party to harass vote recounters in Democratic counties. Another couldn't finish in time. None of the Republican counties chose to recount manually.
Those were choices made on a county by county basis. Really the secretary of state should've
1) set standards for manual recounting
2) had the entire state manually recount
No matter what happened in that case, I think it would have achieved maximal trust in the process in Florida and the US. Arguing that a piece of crap election process should be allowed to stand as is forms arguments of patent lunacy.
Let's not forget the accuracy of the count in Florida is at least 10 times worse than the vote tally difference. The entire Florida election is one big ugly mess. The voting and counting process is horribly inaccurate. In the case of a nearly tied vote, the only appropriate thing to do is to work as hard as possible given time constraints to improve the accuracy. Unfortunately Bush's Florida campaign manager is in charge of the process, the Florida legislature is Republican, and the State Supreme Court is stacked Democrat. Any moves made by either side would immediately be interpreted as partisan and destructive - yet improved voting accuracy is essential to our trust of the election process. Gore is fighting mainly for his votes, and Bush is fighting to force acceptance of the piece of crap. I'd rather see efforts made to achieve maximal accuracy in the entire state.
Americans hate lawyers, as do I. In my view, the person who, after multiple counts and recounts is resorting to using lawyers for the sole purpose of getting a judge to appoint him President.
For all the blame throwing, Bush has contested the vote in more counties than Gore. But ask yourself one question before assessing Gore's actions. Suppose on election night Gore has won by 1900 votes, not Bush. Do you really think that Bush would not have gained the lead by now ?? Do you think he would have used any less tools to challenge the election than Gore ? I suspect he would have gained far more votes than Gore has (were the tabled turned), and it would have been done much more smoothly.
BTW, Gore's lawyer, Boyd, is the lead government lawyer in the Microsoft case, don't know if anyone's mentioned that yet. This shakes my faith in the Reno case against them, IMO, he has damaged his credibility severely by arguing specious cases on Gore's behalf.
Gore's lawyer is David Boies, not Boyd. He was IBM's lead attorney when they had their antitrust case dropped. His strategy then was to stall and delay - a very successful strategy. He was the Attorney General's offices lead attorney against Microsoft. By most accounts he was stellar against Microsoft. He also represented Napster against the record companies. This is clearly a very efficient lawyer who enjoys taking on cases of national importance that work at the edges of the interpretation of the law. Don't forget, the value of a lawyer is based on how he does with what he's got. So far Boies has beaten Microsoft for the government, beaten the government for IBM, and negotiated a settlement for Napster. I don't think he is worried about the phone ringing for more cases.
Let's all hope that the case in Florida moves in ways that allow us to maximally trust the accuracy of the process. I doubt that will happen - I don't think either candidate really wants it.
Of course a majority of Debian users are the sort of people who would oppose any kind of commercial involvement in their distribution, because they're all hardcore Stallmanists who believe that communism onlt failed because of the oppression of evil American capitalism. But, luckily enough, all of these people earn enough $$$ to be able to spout such rhetoric from the comfort of their luxury homes.
I think you are either alienated by Debian fanatics or smoking crack.
I view the most wonderful thing about Debian as its attention to even very small packaging details. These are the sort of things that really add up in the long run, and get forgotten about in the push to get a new commercial distribution out the door.
The Debian packagers are NOT packaging for commercial reasons. In fact, the vast vast majority of them are packaging primarily for themselves, and because they benefit from others doing the same. Because of their motivation, the packages come out later, cleaner, and in general make administering linux boxes easier on the admin. That is what I appreciate about Debian, and I view it as a direct result of the lack of commercial push. In the end, I use Debian because my boxes run better, nothing more, and nothing less.
Communism is collective ownership and labor organization to the benefit of all. If that makes Debian communism, so be it. It is fairly obvious to me that aspects of commercial packaging work against the packaging being clean. It focuses more on feature completeness (a big selling point) and timely releases (another big selling point).
At least Dabian makes my boxes run more smoothly. It is going to take any commercial packager that is not Debian based at least two years to catch up to their thoroughness and care in packaging. And if that is the case I would be happy to use some other distribution. But for now I find Debian cleaner and more thorough than at least Mandrake and Redhat (and probably more than that too).
Gore ass-lickers like to act like Gore is intelligent...however when one considers the secret arm's deal to the Russians to allow weapons to be sold to Iran; selling access to the Chinese; and being such a habitual liar that it isn't possible to know until the next if what Al Gore said was true or not...really makes one wonder how intelligent this man-woman is!
Gore did graduate college Cum Laude from Harvard. He did flunk out of Divinity School at Vanderbilt after returning from his enlisted service in Nam. He then DROPPED OUT of Vanderbilt law in order to serve as a congressman for Tennessee. Gore has served in national office ever since.
Nader graduated from Princeton Magna Cum Laude.
He went on to Ivy League law school, and has INVENTED the role of consumer advocate in the US, saving consumers lives and millions of their dollars in the process.
Bush graduated Yale with a 77 average. He went to the National Guard to not show up. He went to business school. Tell me how a Harvard MBA can fail miserably, losing millions of dollars in the oil business in West Texas. Oh, yeah. He was a drunk. Then, Harken bought his oil company primarily for his name - a buy that cost them about $5 million dollars. That is a lot to pay for a company losing money. Then Dubya was bailed out several times by friends of the then Vice President, Dad. After becoming a millionaire through these handouts, Dubya went into politics.
Slashdot Mirror
Apt-get.
Which is an update program, and not a package program. Mandrake has MandrakeUpdate (along with stable and cooker sources), and I'm sure most other rpm based distros have similar tools.
Ever tried to use those tools ??
You are right though - this is not strictly a deb vs rpm issue.
More packages in the standard tree - fewer compiled by other people who are not part of the distribution.
Again, squat to do with deb versus rpm. And for that matter, Mandrake's Cooker has most of what I want, being an open submission of Mandrake rpms.
Actually this has a ton to do with reliability. The entire point of having a distribution is that they will ensure that packaging is reliable and consistent. When third party vendors start offering Redhat RPMs, or Mandrake RPMs, or you start using contrib RPMs, package dependencies get less consistent, and the machine is just not as clean.
I get everything I need at debian.org in woody non-us and non-free, all packaged by debian maintainers.
Debconf.
Okay now THATs more like it
Debconf is a little scripting conf that allows packages to pop up questions in console windows during the package installation that ask you if you would like to clobber your existing configuration files with new ones or not. The default is always do not clobber. Or, a package can pop up a reminder about a configuration issue associated with a package, and either show or do not show the reminder in the future. Lastly, they have help hints if you cannot resolve a dependency that tell you what to do !! These have helped me a few times.
How is .deb better then .rpm?
Apt-get.
Cleaner dependencies.
More package maintainers.
More packages in the standard tree - fewer compiled by other people who are not part of the distribution.
More testing.
Debconf.
No backward compatibility break between rpm-3 and rpm-4 for debs.
Nuff said.
This is a dramatic change in computer using from Windows/Mac to linux.
First, recognize that linux can do just about any network function possible on a computer. Most of these require some sys admin skills to accomplish, and have HOWTOs included with the distribution.
Other. simpler functions, have a variety of ways to receive help. Usenet is a great resource for linux. So are the help documents for KDE/GNOME. So are man pages. So are web sites.
In general, for linux, you NEVER call a help line, or consult a hard copy. The references are on the computer or on the net. Just search for what you want with Google, check the HOWTOs, acclimatize to using man -k and locate, and enjoy the ride.
You will eventually come to realize that help resources for linux dwarf those available for vendor operating systems, since the help inevitably comes from the user base.
You are missing the point.
The software installs as encrypted software plus key in your CPRM sector.
Any use will require encryption keys plus encrypted media.
No way in he!! open source will ever be allowed to read CPRM sectors, as that would make the encryption keys copyable.
Totally illegal to defeat this measure under DMCA.
Copying original enrypted media is accepted as commonplace - the protectiion resides in the keys that you cannot even legally see unless you are software provided by the media copyright owners.
That is right - the installation program will give more rights to YOUR hard drive to copyright owners than to you.
That is another thing I like about Debian. Although I can understand why a beginner would dislike it.
The packaging is done with a finer granularity. They often choose a minimal number of packages to, say, have a working emacs. Then you can apt-get all the elisp you need to increase its function.
The same with tetex. There are, I think, three packages that are necessary. Then you can add docs, fonts, and postprocessors to make it tetex on 'roids.
But it is really about choice and flexibility - the linux way. And I think this applies especially to Debian. Of course, flexibility also means more knowledge is required to achieve the functionality you desire.
trying to associate single neurons with particular brain processes is like trying to find a needle in a haystack.
No one was suggesting assigning functions to single neurons - but rather studying the nervous system on the basis of the signals of single neurons. I would suggest that there is an enormous difference. Researchers have used signal detection theory in several well known examples in the labs of Ranulfo Romo and Bill Newsome to demonstrate that signals of some neurons, when averaged together in groups of around 100, can predict the decision of the animal in discrimination trials. These studies on relatively simple percepts can and will be elaborated into more complex behaviors as time progresses.
Using microelectrode technology we have been able to verify that sensory disturbances exist in focal hand dystonia, a sensory-motor disorder that can be understood as a complication of repetitive stress injury. Such detailed observations as we made cannot be made with EEG. However, MEG studies were started after the animal studies to demonstrate recordings that were abnormal and predicted by our hypothesis. (EEG was inappropriate because the area in question is orthogonal to the radial vectors picked up in EEG).
This is just one example of a case in which recordings made in lab animals transferred nearly completely to humans in the wild.
EEG-type monitoring also does not measure the level of anything - it is only sensitive to changes in vector averaged voltages. Whereas signals carried by neurons have bandwidths of around 1-4 kHz, EEG hardly detects any energy above 100 Hz. All you get is a grossly averaged signal envelope that represents some aspects of action potentials and synaptic currents - in hypothesis only. If all the neurons in a cortical area had very high firing rates, but were largely asynchronized, EEG would measure nothing.
Even knowing the neural signals exactly it is very difficult to predict the EEG with a high precision.
Even so, we have studied field potentials - a kind of greatly amplified EEG - in depth in very simple cerebral cortical systems. It is far from obvious, even in such a simple case, how one can go from field potentials to neural signals.
The signal processing of the brain represents such a challenge because the purest signals can be measured only by microelectrodes with tip exposures of 5 microns or less - as originally described by Galambos in 1943. That presents an enormous technical difficulty that challenges investigators still.
If you've done EEG you know the limitations, and the variance, and the problems of uneven electrical conductance interfaces, and the studies to hypothesize what synaptic and action potential derived currents contribute to signals. I speak with EEG researchers regularly, and respect their work immensely. I don't think they would suggest a helmet such as the one described in the article would be adequate for much.
And the gamma band (25-40 Hz)?? What about the gamma band ??
The one really invoked in attention ??
Truly, claims that anyone understands if or how a toy like this might work are mere pie in the sky. We'd need to understand how attention worked before we could train it using EEG waves that demonstrate at best a very weak noisy reflection of SOME but certainly not MOST brain events,
In animal experiments people recorded EEG in the 1930s and 1940s as soon as amplifier technology began being applied to Neuroscience. Then in 1943 a researcher named Galambos (famous for co-discovering echo-location in bats) saw that using very small electrode tip exposures allowed recording from single neurons.
This breakthrough led the animal researchers to all but throw away EEG as a useful tool, although there are tons of human data still. But the problems with the EEG are several. First, it only reflects a vector averaging of many million neurons and synaptic currents. And it only reflects the dot product of that average with a radial vector.
Researchers estimate about 100 neurons in one area of the brain would be sufficient to carry the information in a percept, with perhaps 5, possilby as many as 10, brain areas involved. Such things are immeasurable by EEG. The things that are measurable are the oscillating features of brain processing, which could for all we know be epiphenomenonological. Or not. We really don't know yet. But the studies of the signals carried by single neurons clearly bear close relation to brain processing, and have been very difficult to relate to EEG-type monitoring.
I think its great what they're doing. But, Isn't debian mostly alpha stuff. I mean it's always in the testing phases so what's the real point here?
...
Try running potato - the stable distro.
Woody is unstable. If you apt-get upgrade your woody regularly you will occasionally have some fun manual work to do. Like that upgrade last week that broke perl and thus debconf.
But the debian releases are in general more stable and MUCH MUCH easier to maintain than any RPM based distro. The packaging is just clean.
The reason the packaging is clean is simple. There are 644 Debian packagers, most of them system administrators running Debian for a living. It is human nature to be lazy, so these administrators each do their job on Debian so that their day jobs are easy and they can play quake while RedHat, Mandrake, and SuSE admins are resolving package dependencies and compiling unsupported packages from tarballs.
Now where is that Dubya patch for Quake again
As Ralph Nader would say, the federal government consists of people handing out more and more to the major corporations, and stomping all over consumers' rights. It is really a shame that we have evolved into a government of the corporations, and not a government of the people.
Major corporate mergers like this one should be seriously challenged. Instead, once a REPUBLICAN goes into office, the merger goes through unanimously. I don't think it would have taken too much longer had Gore been president-elect, though.
Another Naderism "The only difference between Gore and Bush is how fast their knees hit the floor when big business walks in the room."
It's time for RMS & ESR to behave in the same fashion, IMO.
RMS lives in a world of free software. He dislikes M$, but largely he preaches free software to the masses. He practices what he preaches, and is not involved in commercialism.
ESR is a poster boy for Open Source commercial movement.
Lumping them together is doing a large disservice to both of them.
This is a really notable advance. It takes linux nfs from junky v2 userspace to fast state-of-the-art v3 kernel support.
Big businesses like accountability, someone they can point a finger at and say 'Make it work'.
IBM, or RedHat, or VA Linux, or dozens of other companies will support linux and fix problems for you, allowing you to hold them accountable.
I still think this is a straw argument - You can CHOOSE anyone in the world, and they would have source code access that allows the fixing of ANY bug. That will NEVER work with a closed source vendor.
Microsoft wants to drive more traffic to its web site. Its security postings are one mechanism to do so. That takes precendent over things like full disclosure, or serving the security community.
Web traffic is $$.
Don't even think Microsoft cares about security - they don't except for its ability to make them look bad. If they can market something as secure, it really doesn't matter whether it is or not.
And this is a direct attempt to hit BugTraq squarely in the wallet by taking most of their web traffic, and having them click through to Microsoft.
Hey - I did the same thing.
:)
Really nice to see KDE in Debian.
And really nice to have a working apt.
Apt draws volunteers to Debian because it makes Debian better than most of the other distros. But if the value of Debian is replicated somewhere else, w/out all of the inconveniences of Debian, then your volunteer force diminishes.
Here begins some serious speculation.
1) Apt on other distros will not work. Apt depends on package dependencies being done very cleanly, and this is simply not true in any other distro to the same extent that it is in Debian. The other distros need not only apt, but they also need a packaging policy.
2) Debian is self-supporting. People who find Debian and enjoy it because it is done for the benefit of its volunteers generically enjoy the distro. This is not going away any time soon. One might argue that Debian is competitive with develops with other distros, but I don't think that is true. Other distros pay their supporters, and Debian is still a distro of volunteers.
3) Debian developers are among the most stringent Free Software supporters. They are in it to create the best Free Software distribution possible. Many people think they already have it. There are literally no challengers - distributions with strict Free Software guidelines.
4) Debian is an active development environment. Apt is just one example - it is not a killer app in and of itself. Debian initscripts are better (IMHO) than those of the other distros I have checked. Debian security is up there as well.
5) Debian has more packages than Redhat or Mandrake or SuSe. They have these packages in their 'official' distribution - not available packaged by someone else at rpmfind.
Basically, I think the care in packaging of Debian is about 18+ months ahead of anyone else. And for that reason I can see no reason to even consider another distribution for my boxes.
We use Redhat at work. I get called a weenie for supporting Debian. But administering my Debian boxes takes 1/10th the time it takes me to administer my Redhat boxes. And that is the only reason I need to stay with Debian. YMMV.
Don't get me wrong, I like Debian for a lot of reasons, but apt is the absolute killer app that keeps me on Debian. I now find myself wondering, if RedHat adopts apt, what would happen to Debian.
The beauty of Debian is how cleanly it is packaged - not just apt. It is a distribution done by system administrators largely for themselves. It is not going anywhere. As long as it still benefits the package maintainers, Debian will thrive.
Other distributions are packaged for profit. That brings in other motives. Distributions will sell themselves as packaging things more rapidly or packaging things for 586 instead of 386 (like Mandrake). Distributions will fund a lot of development and sell themselves as containing the latest compiler (Redhat). Debian is packaged because it works the way a bunch of system administrators think Free Software should work.
And there will always be plenty of room for that. If you and/or others go running to Redhat once apt is working well with rpm, so be it. Debian will continue to thrive because it cannot die. It can't lose its funding because it hardly has any. Debian works because volunteers make it work (and a very small but growing number of paid developers).
This is a great next step for rpm based distributions. However, the cleanliness of debian packaging is only part apt.
Most of it comes from thorough policy and packaging guidelines from The Debian Documentation Project. Until other distributions develop such comprehensize packaging policies, the package will not interrelate as well (read - dependency problems will screw up apt). Debian maintainers spent a lot of time thinking out these compehnsive guidelines.
I can rarely upgrade Redhat distributions cleanly without tons of rpm commands ignoring dependencies - however, I find this trivially simple with debian. And the capabilities of dpkg and rpm offer no advantages. But the packaging policies do.
Apt will help a lot though.
This also shows that competition in Free Software is good. If debian innovates, the innovations can be copied to rpm based distros. And everyone wins.
I don't know anything about soft updates.
FFS is basically a log-structured file system (like ReiserFS) that also uses soft updates. This is a way that the file system can determine if the files are intact or not at every point in time.
For the development TUX2 (which I've read more about), the file system writes a tree of inodes to disk. They are organized, and have a sort of 'completeness' bit. The last thing in the tree written is the completeness bit, which then allows the file system to be updated atomically without a journal. Basically, you write a bunch of files to disk, which are recognizable as being grouped. Then, the last thing you do is write a bit on disk that allows you to see that the group has been written intact. So in a recovery, you merely check groups one by one to verify their completeness bits are written. In more advanced schemes the file system can also tell you which of several groups are potentially not intact to begin with. Recovery from crashes are as fast or faster than journalled file system crashes. But file systems still need fscks. You cannot guarantee the integrity of data simply because the file system didn't crash.
I've seen numerous tests of FFS in massive compiles in which the compilation is just allowed to end when the power goes off. FFS does extremely well.
However, for the name server, an administrator wants uniformity. I recently had a situation where one group had name servers on Bind 4 OpenBSD, and everyone else was on Bind 8
Compiling BIND 8 is not exactly rocket science. It is what I do for my name servers anyway. But stop hyping on name servers - they are a little task suitable to a 386 in a closet running linux kernel 2.0. Setting them up on a large number of computers in a small area is just opening potential doors. Named is the number one mechanism of computer breakins in the last 5 years.
You will also note that Red Hat installs Bind 8 as a caching server only, and I've seen networks with hundreds of UNIX workstations, all running such caching servers - named can be considered client software, although I understand that nscd can now serve a similar purpose.
I find this to be quite silly. A reasonable named can serve a few hundred very active computers without issue. Matter of fact, our UCSF campus uses two on the entire campus serving tens of thousands of machines. Redhat has no business setting up named by default in any capacity - they caused hundreds if not thousands of break-ins by doing this in Redhat 5.x releases. If you want named to run, you ought to be able to figure it out. Named is not a garden variety server task. Any server task should be considered seriously as a potential breakin mechanism. If it is not really necessary, it should be closed.
Unless, of course, you relish reinstalling the OS.
As far as mature memory management goes (if I understand your point correctly), I think an administrator will try to avoid swap/paging on a mission-critical system. And how many high-capacity, heavy-load system run with a single processor?
This is much more pervasive. A single large compile will push almost any machine to the limits of its memory management. You ought to be able to retain use of the mouse at the same time you compile. This is MUCH more robust on the BSDs. This is something I would notice on any machine I use on my desktop - and I am a very conservative users of resources.
Things have got to scale. Right now, both Linux and BSD fall short on this point. I look forward to the day when this is no longer the case.
Linux outscales almost all commercial Unices using the 2.4 kernels with respect to load scaling and maximal TCP/IP throughput. That was entirely the point of the development series. So it is coming.
Actually, ReiserFS is a good deal faster than ext2,
This is sort of a loaded statement, as it depends on the nature of the data. In particular, if the blocksize is small or comparable to the average file sizes, then it is pretty much a wash. In cases with lots of files smaller than the ext2 blocksize, Reiserfs will have an edge.
and FreeBSD's FS is inbetween ext2 and ReiserFS in speed. I don't know why people say journeling FSs are slow
Maybe they benchmarked them ?? Seriously, FFS is a really solid file system, different from ext2 or ReiserfFS.
the issue with journalling is that you need to keep your journal synchronous on disk. This creates a bit of inflexibility in the VFS used by the kernel. Soft updates allow a lot more flexibility, as does tux2, another soft update system using atomic updates of phase trees. The issue is whether restructuring the file system to allow atomic updates causes you more speed loss than adding a journal. Generally journalling will lose, although ReiserFS has other reasons that make it fast. Those will largely be copied into soft updates FSs like Tux2, and you will be able to choose a faster soft update file system or a slower journalled system.
I've never seriously used it, but BSD: lacks a journaled file system
... but they've actually lost some utility in memory management that is more relevant for workstation users. So you could argue linux is now really good for big iron tasks, but the BSDs have really clean memory management that allows them to be heavily loaded and perform well.
But has soft updates, which are a substantial improvement over ext2 with respect to recovery from crashes. More seriously, what advantage does journalling have over soft (atomic) updates, and what advantages do soft updates have over journalling ?? I think reasonable arguments can be made that soft updates are faster, and at least as crash tolerant as journalling, and a heck of a lot easier to program and maintain wrt the VFS layer.
only recently migrated to the ELF format
So???
still uses Bind 4 (OpenBSD specifically)
And this one is really relevant for those 0.01% of machines begin used as name servers.
Does BSD have a multi-threaded IP stack? How does BSD perform on Mindcraft? Linux has been playing catch-up in this space for some time, and may have a big lead.
Actually, *BSD does much better with their IP stack than linux 2.2.* and 2.0.*. *BSD also does much better under heavy loads because it has a more mature memory management scheme. Linux is supposed to work on this in the next devel series now that more fine grained SMP locking is present.
But seriously, how much advantage do you reckon a multi-threaded IP stack makes on a single processor machine ??
The free BSDs are a very fine choice for a kernel and base utilities. There are some areas in which linux is better, and others in which the BSDs are better. Generally common server tasks work out better for *BSD than linux, and application availability and marketing are stronger for linux.
Linux's big recent push has largely oriented around big hardware - SMP, multiple NICs,
Well, linux has already outnumbered *BSDs and OSX combined.
But that is largely irrelevant. With the introduction of autoconf, the open source components of these technologies will complement each other. Openssh was taken from openbsd to *BSD and linux. GNOME and KDE are largely linux developed, but work fine on *BSD.
The largest linux companies look at big business UNIX and Microsoft as the competition for different markets. *BSD and linux will both continue to grow at the expense of Microsoft and mainframe Unices. The market dynamics may have a few people going from linux to *BSD (and fewer going the other direction), but the changes in user base for linux and *BSD are coming not from each other but from Microsoft and mainframe Unix.
The media loves to play up battles, like KDE/GNOME, Redhat/Mandrake... but the reality is that KDE and GNOME help each other more than they hurt each other through competition. The same is true of Mandrake and Redhat. Any improvements made by open source companies in software lead to strengthening of all open source companies' software.
Making a living summarizing /. posts for over a year now.
However, I clearly think that if Gore continues to go ahead with his lawyers in front of Democrat judges (who already have rewritten the law, in effect changing the rules of the game after the ball has been put in play), he's going to destroy his party.
The judges said the law was inconsistent. Therefore it needed interpretation. The interpretation was that the counties could recount if they so chose, and the certification deadline needed to be moved to accomodate. That is hardly changing the rules. So the counties that had large problems with ballots recounted manually. Or, rather, one of them did. Another was harassed by people flown in courtesy of the Republican party to harass vote recounters in Democratic counties. Another couldn't finish in time. None of the Republican counties chose to recount manually.
Those were choices made on a county by county basis. Really the secretary of state should've
1) set standards for manual recounting
2) had the entire state manually recount
No matter what happened in that case, I think it would have achieved maximal trust in the process in Florida and the US. Arguing that a piece of crap election process should be allowed to stand as is forms arguments of patent lunacy.
Let's not forget the accuracy of the count in Florida is at least 10 times worse than the vote tally difference. The entire Florida election is one big ugly mess. The voting and counting process is horribly inaccurate. In the case of a nearly tied vote, the only appropriate thing to do is to work as hard as possible given time constraints to improve the accuracy. Unfortunately Bush's Florida campaign manager is in charge of the process, the Florida legislature is Republican, and the State Supreme Court is stacked Democrat. Any moves made by either side would immediately be interpreted as partisan and destructive - yet improved voting accuracy is essential to our trust of the election process. Gore is fighting mainly for his votes, and Bush is fighting to force acceptance of the piece of crap. I'd rather see efforts made to achieve maximal accuracy in the entire state.
Americans hate lawyers, as do I. In my view, the person who, after multiple counts and recounts is resorting to using lawyers for the sole purpose of getting a judge to appoint him President.
For all the blame throwing, Bush has contested the vote in more counties than Gore. But ask yourself one question before assessing Gore's actions. Suppose on election night Gore has won by 1900 votes, not Bush. Do you really think that Bush would not have gained the lead by now ?? Do you think he would have used any less tools to challenge the election than Gore ? I suspect he would have gained far more votes than Gore has (were the tabled turned), and it would have been done much more smoothly.
BTW, Gore's lawyer, Boyd, is the lead government lawyer in the Microsoft case, don't know if anyone's mentioned that yet. This shakes my faith in the Reno case against them, IMO, he has damaged his credibility severely by arguing specious cases on Gore's behalf.
Gore's lawyer is David Boies, not Boyd. He was IBM's lead attorney when they had their antitrust case dropped. His strategy then was to stall and delay - a very successful strategy. He was the Attorney General's offices lead attorney against Microsoft. By most accounts he was stellar against Microsoft. He also represented Napster against the record companies. This is clearly a very efficient lawyer who enjoys taking on cases of national importance that work at the edges of the interpretation of the law. Don't forget, the value of a lawyer is based on how he does with what he's got. So far Boies has beaten Microsoft for the government, beaten the government for IBM, and negotiated a settlement for Napster. I don't think he is worried about the phone ringing for more cases.
Let's all hope that the case in Florida moves in ways that allow us to maximally trust the accuracy of the process. I doubt that will happen - I don't think either candidate really wants it.
Of course a majority of Debian users are the sort of people who would oppose any kind of commercial involvement in their distribution, because they're all hardcore Stallmanists who believe that communism onlt failed because of the oppression of evil American capitalism. But, luckily enough, all of these people earn enough $$$ to be able to spout such rhetoric from the comfort of their luxury homes.
I think you are either alienated by Debian fanatics or smoking crack.
I view the most wonderful thing about Debian as its attention to even very small packaging details. These are the sort of things that really add up in the long run, and get forgotten about in the push to get a new commercial distribution out the door.
The Debian packagers are NOT packaging for commercial reasons. In fact, the vast vast majority of them are packaging primarily for themselves, and because they benefit from others doing the same. Because of their motivation, the packages come out later, cleaner, and in general make administering linux boxes easier on the admin. That is what I appreciate about Debian, and I view it as a direct result of the lack of commercial push. In the end, I use Debian because my boxes run better, nothing more, and nothing less.
Communism is collective ownership and labor organization to the benefit of all. If that makes Debian communism, so be it. It is fairly obvious to me that aspects of commercial packaging work against the packaging being clean. It focuses more on feature completeness (a big selling point) and timely releases (another big selling point).
At least Dabian makes my boxes run more smoothly. It is going to take any commercial packager that is not Debian based at least two years to catch up to their thoroughness and care in packaging. And if that is the case I would be happy to use some other distribution. But for now I find Debian cleaner and more thorough than at least Mandrake and Redhat (and probably more than that too).
Gore ass-lickers like to act like Gore is intelligent...however when one considers the secret arm's deal to the Russians to allow weapons to be sold to Iran; selling access to the Chinese; and being such a habitual liar that it isn't possible to know until the next if what Al Gore said was true or not...really makes one wonder how intelligent this man-woman is!
Gore did graduate college Cum Laude from Harvard. He did flunk out of Divinity School at Vanderbilt after returning from his enlisted service in Nam. He then DROPPED OUT of Vanderbilt law in order to serve as a congressman for Tennessee. Gore has served in national office ever since.
Nader graduated from Princeton Magna Cum Laude.
He went on to Ivy League law school, and has INVENTED the role of consumer advocate in the US, saving consumers lives and millions of their dollars in the process.
Bush graduated Yale with a 77 average. He went to the National Guard to not show up. He went to business school. Tell me how a Harvard MBA can fail miserably, losing millions of dollars in the oil business in West Texas. Oh, yeah. He was a drunk. Then, Harken bought his oil company primarily for his name - a buy that cost them about $5 million dollars. That is a lot to pay for a company losing money. Then Dubya was bailed out several times by friends of the then Vice President, Dad. After becoming a millionaire through these handouts, Dubya went into politics.