The biggest user-visible difference between Debian and Ubuntu kernels is that Debian kernels are always "-server" style kernels, that is, no preemption and low clock rate. This lowers responsiveness but can vastly improve throughput for server tasks. So yes, Debian is still a server distribution, although on modern hardware the difference is more subtle.
Try Debian Lenny with xfce. Mine boots to under 30 MB RAM for console login, 160 MB RAM for GUI, which is pretty heavy but almost all of that is just the X server which is getting extremely large. Console login is still rathger small and that's including a couple of services in the background, so it's not like the kernel and shell are bloated.
He might mean from bootloader to KDM login, which is still absolutely absurd because just loading that much executable data off the D420's disk takes more than 10 seconds. I run a tight Debian Lenny on my D420 and it takes longer than that just for *console* login.
LAME also has significant IO speed and memory speed components, which seem to be slowed down in general in the test environment. Besides that, the new scheduler gives generally lower priority to batch tasks. And GCC might be worse. Compiz is running. So there are a few slowdowns that really stack up. I'm not sure if they account for the whole 50%, however.
That's just retarded. KDE and GNOME libraries comprise similar functionality to the Windows foundation, in different ways with different strengths, and it just happens you are able to run both at once.
There are plenty of "small" media players you can install that depend on as few libs as possible, including some that run via a command line interface. You have the full spectrum of features and requirements available to you. Try saying that about Windows, where a multi-megabyte beast like foobar2000 is considered "lightweight".
Funny you should mention that, since "premature optimisation" is so common it's now an industry term.
A good commercial example is TES4: Oblivion, which runs in a few threads by default, but performs much better when forced to 1 major thread. The split to threads happened before the code itself was optimised (horizontal optimisation before vertical, a huge mistake), and the end result was that the synchronization design actually reduced throughput and increased latency.
I had NetBSD 3.0 i386 on a Pentium 3 laptop with an old hard disk, that would boot from stage 1 bootloader to console login in about 13 seconds. The complete POST before the bootloader was about 2-3 seconds as well. Granted, NetBSD is about as lightweight as you can get these days, but think of just how old and weak that hardware is.
Well, that's the idea. The proliferation of the GPL encourages a more open software market on the whole, by forcing companies to either compete (by reproducing code under a different license) or cooperate (by using GPL code and releasing their improvements). Either way, everyone benefits to some degree. That's why we have Linux.
BSD, MIT, etc. licenses exist for those who want to improve the market by lowering production costs rather than encouraging competition. That's why we have OSX.
Maybe it's not the interface you want, but you can get this very, very easily in Linux (or in fact, any Unix-like, even Cygwin) without any third party software.
find/media/disk1/media/disk2/home -type f -exec grep -Hn "monkey" {} \;
You can wrap that in a convenient shell script or even a GUI if you want. You can narrow down to just.txt files by adding -iname "*.txt" as well, and so on. find is ridiculously useful and it's a shame so many people never harness it.
It's even simpler than that. The vast majority of, say, Red Hat's income is through support contracts, which will be maintained even through an economic crisis. Even when they are set to be renewed, what is the client going to do, refuse support and risk even higher costs, switch to another vendor with even more expensive support and migration costs, or just renew the contract for the usual, well-known cost? Companies based on support fees will stay very strong.
That's certainly true for stable distributions, but seeing how quickly distributions like Ubuntu jump on new features, it's possible that btrfs will be an option for / within 2-3 years. Certainly not default, of course.
Well, reiserfs shot to default status for some distributions almost immediately after its release, and long before it was stable.
A number of enterprise-grade open source projects, such as most of the free J2EE stack and Linux, have attractive books available. That's well handled.
Commercial training is another matter though. I completed highly recommended week-long Oracle training and discovered it was not far removed from an online tutorial, yet took much longer. I guess that's just how some people prefer to receive knowledge.
Purchasing Windows doesn't give you an "assured" version either. The industry has learned that hard lesson over and over. You're much better off just licensing an open distribution like Red Hat, because you get the corporate support side as well as the community audit side.
The fact is that even if you don't have time to read the source, other people do, and a complete distribution has the unique level of multi-party quality assurance money can't buy.
Microsoft is probably the worst possible example anyway. They regularly put in their own malware. There's no audit required to know that WGA is pure and simple malware. It's absolutely moronic to name them as an example of an "assured" solution vendor.
All quantum crypto does is make it impossible to eavesdrop without being detected. It does not secure the data itself. You still use symmetric ciphers to protect the data, and those are theoretically demolished by quantum computing too.
Basically if we ever get practical quantum computing, ANY "search for solution in large space" problem is deflated, and we may as well give up on crypto entirely.
Without a test harness in place, you wouldn't find that bug anyway. Now that it's been found, I'm certain tests are in place. The reason software is getting so much more stable in recent years is that bugs are becoming more familiar and better understood, and processes are evolving to handle bugs better.
Slashdot Mirror
Thanks, and my apologies for tl;dr behavior.
The biggest user-visible difference between Debian and Ubuntu kernels is that Debian kernels are always "-server" style kernels, that is, no preemption and low clock rate. This lowers responsiveness but can vastly improve throughput for server tasks. So yes, Debian is still a server distribution, although on modern hardware the difference is more subtle.
That bug reports seems to only apply to upgraded installs, not fresh installs. Perhaps it doesn't address the issues in this benchmark.
Try Debian Lenny with xfce. Mine boots to under 30 MB RAM for console login, 160 MB RAM for GUI, which is pretty heavy but almost all of that is just the X server which is getting extremely large. Console login is still rathger small and that's including a couple of services in the background, so it's not like the kernel and shell are bloated.
He might mean from bootloader to KDM login, which is still absolutely absurd because just loading that much executable data off the D420's disk takes more than 10 seconds. I run a tight Debian Lenny on my D420 and it takes longer than that just for *console* login.
LAME also has significant IO speed and memory speed components, which seem to be slowed down in general in the test environment. Besides that, the new scheduler gives generally lower priority to batch tasks. And GCC might be worse. Compiz is running. So there are a few slowdowns that really stack up. I'm not sure if they account for the whole 50%, however.
That's just retarded. KDE and GNOME libraries comprise similar functionality to the Windows foundation, in different ways with different strengths, and it just happens you are able to run both at once.
There are plenty of "small" media players you can install that depend on as few libs as possible, including some that run via a command line interface. You have the full spectrum of features and requirements available to you. Try saying that about Windows, where a multi-megabyte beast like foobar2000 is considered "lightweight".
That is the most I have ever learned from a single Slashdot post. Thank you very much :)
Great idea! We shall call it "hibernation"!
Eh, run Linux, where the entire network-enabled print system including all "drivers" is a couple of megabytes.
Funny you should mention that, since "premature optimisation" is so common it's now an industry term.
A good commercial example is TES4: Oblivion, which runs in a few threads by default, but performs much better when forced to 1 major thread. The split to threads happened before the code itself was optimised (horizontal optimisation before vertical, a huge mistake), and the end result was that the synchronization design actually reduced throughput and increased latency.
I had NetBSD 3.0 i386 on a Pentium 3 laptop with an old hard disk, that would boot from stage 1 bootloader to console login in about 13 seconds. The complete POST before the bootloader was about 2-3 seconds as well. Granted, NetBSD is about as lightweight as you can get these days, but think of just how old and weak that hardware is.
Well, that's the idea. The proliferation of the GPL encourages a more open software market on the whole, by forcing companies to either compete (by reproducing code under a different license) or cooperate (by using GPL code and releasing their improvements). Either way, everyone benefits to some degree. That's why we have Linux.
BSD, MIT, etc. licenses exist for those who want to improve the market by lowering production costs rather than encouraging competition. That's why we have OSX.
Maybe it's not the interface you want, but you can get this very, very easily in Linux (or in fact, any Unix-like, even Cygwin) without any third party software.
find /media/disk1 /media/disk2 /home -type f -exec grep -Hn "monkey" {} \;
You can wrap that in a convenient shell script or even a GUI if you want. You can narrow down to just .txt files by adding -iname "*.txt" as well, and so on. find is ridiculously useful and it's a shame so many people never harness it.
"It seems we're getting dupes from a parallel SlashDot."
It's called GoateeDot, and Cowboy Neal is a well respected editor who screens stories for accuracy, grammar, and spelling.
Fixed that for you.
What? I did.
It's even simpler than that. The vast majority of, say, Red Hat's income is through support contracts, which will be maintained even through an economic crisis. Even when they are set to be renewed, what is the client going to do, refuse support and risk even higher costs, switch to another vendor with even more expensive support and migration costs, or just renew the contract for the usual, well-known cost? Companies based on support fees will stay very strong.
You're right. In other news, fourier transform should be just as fast as random distribution.
Why not just use OpenOffice? Compatible with every modern OS and certainly more pleasant than Office 97.
That's certainly true for stable distributions, but seeing how quickly distributions like Ubuntu jump on new features, it's possible that btrfs will be an option for / within 2-3 years. Certainly not default, of course.
Well, reiserfs shot to default status for some distributions almost immediately after its release, and long before it was stable.
A FUSE ZFS guarantees it will never be the "default" filesystem anyway. BTRFS has a good shot at being your / in a couple of years.
A number of enterprise-grade open source projects, such as most of the free J2EE stack and Linux, have attractive books available. That's well handled.
Commercial training is another matter though. I completed highly recommended week-long Oracle training and discovered it was not far removed from an online tutorial, yet took much longer. I guess that's just how some people prefer to receive knowledge.
Purchasing Windows doesn't give you an "assured" version either. The industry has learned that hard lesson over and over. You're much better off just licensing an open distribution like Red Hat, because you get the corporate support side as well as the community audit side.
The fact is that even if you don't have time to read the source, other people do, and a complete distribution has the unique level of multi-party quality assurance money can't buy.
Microsoft is probably the worst possible example anyway. They regularly put in their own malware. There's no audit required to know that WGA is pure and simple malware. It's absolutely moronic to name them as an example of an "assured" solution vendor.
All quantum crypto does is make it impossible to eavesdrop without being detected. It does not secure the data itself. You still use symmetric ciphers to protect the data, and those are theoretically demolished by quantum computing too.
Basically if we ever get practical quantum computing, ANY "search for solution in large space" problem is deflated, and we may as well give up on crypto entirely.
Without a test harness in place, you wouldn't find that bug anyway. Now that it's been found, I'm certain tests are in place. The reason software is getting so much more stable in recent years is that bugs are becoming more familiar and better understood, and processes are evolving to handle bugs better.
Try 12-24 months. Debian make the best effort to "get it right" of any community-supported distro, and the results tend to be the best too.