I am sorry to say this but some of these complaints are valid or are believed to be valid.
A little background. I live in Denmark (Nothern EU) where windmills are used a lot. In fact we have the largest windmill producer in the world.
In Denmark we have laws stating how close you can put windmills to people. Studys have confirmed that the noise from the windmills do stress people close to the windmills. They produce a low humming noice that some people will get stress from other will get headaches and again some will not be bothered at all.
It is also a fact that the wings produce reflections from sunlight in a manner similar to stroboskobe when turning at certain speeds. Some people get ill from this - even people who are not epileptic!
I have never heard about women getting problems with their period. I guess that may be hysteria.
In Denmark the windmill producers try to reduce the reflections by coating the wings in a different ways. But most importantly we are moving the windmills to sea.
At sea there are now people living so it is less of a issue as long as they are far away from land. At sea the wind conditions are usually better resulting in more power per hour hence a better economy.
My mom has a share in a collection of windmills just outside of Copenhagen. You can see pictures here from a introductory trip this summer http://pix.xbsd.net/middelgrund-molle/.
My suggestion to New Yorkers and others would be to contact the Danish government and ask for advice and perhaps Vestas (http://www.vestas.com/) if you want to by windmills.
Quote: "In that case, no local exploit is needed--the attacker either uses sudo, or just sniffs the password the next time the user uses su..."
Unless the user has . in his/her PATH the attacker will not be able to "replace" any programs like the sh. The attacker cannot log keyboards strokes or snif the tty without root privileges. Hence if your system is set up in a good manner the attacker cannot just get the root password. Unfortunately most users can be tricked into writing their own password and most systems like ubuntu and MacOS has sudo ALL as default.
The post is somewhat misleading. Allthough Debitel is a German company the sentence relates to the Danish daughter company which is also named Debitel.
The Danish daughter comapany had a net result of 6,000,000 kr eq. 115,000 $ in 2004.
Thus a fine of 2,000,000 kr is a lot. They even got a 'rebate'. The previous fines was 100 kr pr. email which would have resulted in 4,800,000 kr.
One thing most people tend to forget is that when you choose propriarity software you let the provider decide upon things as hardware and other software you can use.
With Linux/BSD it is possible to buy allmost any type of computer and run the OS. On the other hand Mac OS X only runs on MacPPC and MS Windows for the most part on x86.
Take a small thing like flash. Yes it's possible to use flash on linux x86 dsitributions but what if you decide to use a Power or Sparc based computer? Then you are out of luck.
A lot of people ask me: Why would I want to use a non-x86 based computer? Perhaps because it makes your programs run faster. Perhaps it provides superior functionality that helps the OS writers prevent security problems (NX). There are lots of other examples.
The fact is that your choice between non-free software and free software inflicts upon the list of choices you get on other things (ie. hardware) and not just for you but for other people as well.
The dll's are for 32bit x86 only which means that all those who run on PPC, UltraSparc etc. are out of luck. This actually helps the x86 monopoly.
With access to the source it's possible to at least port the codec to new architectures (in most cases). That's why opensource matters more than most care to think.
The ULE scheduler was supposed to be the default scheduler. It was placed as the default in 5.2. Between 5.2 and 5.3 kernel preemption was enabled. Some serious flaws in ULE popped up leading to crashes and other instability. Due to the complexity of ULE and preemption together only a handful of core people are able to fix this. Since 5.x has been hold back long enough it was decided to drop ULE as the default scheduler for 5.3 and concentrate on releasing 5.3. This doesn't mean that nobody is working on fixing the problem nor that ULE will not be the default scheduler. It is just going to take a while before it happens.
The reason for totally disabling ULE in 5.3 was to focus on other bugs in 5.3 and fix ULE on current (6.0) and then backport this to a later 5.x release.
I suggest you read cvs-src summaries at http://www.xl0.org/FreeBSD/ which gives a view on what is happening on current.
How come this has not turned into a Slashdot poll? I believe a lot of the Slashdot reades run Linux/BSD or something like it and would benefit from a graphics card that 'Just Works (TM)'.
I work in company where it has taken a while to get the CEO and others to understand the benefits of not having extended rights. If you want to make them understand let them manage their own PC. They will get infested and crash a lot (usually). When they ask for help install a fresh version and run a virus check on their files. Do not waste time on restoring there program settings. Instead tell them it's the best way to deal with the problem at hand (it is!). After losing time on this the CEO will listen to arguments like: "We/you are wasting time and time is money." He will ask you what can be done. Tell him he will lose his admin rights and you will manage his PC (add more arguments). When he agrees make sure his PC runs smooth for a long time and when there is a problem you fix it quickly. After a while he will appriciate that he get's his job done and the admin waste less time on reinstalling his PC. When the CEO (replace with some head guy) understands why normal users shouldn't have extended rights you can tell him that you would like his backing to take away extended rights from the normal users. This is a very short explanation on what to do. The point is to explain to the management why it's a benefit to give up their rights - time/money!
It is not allways easy to convince the CEO but it's worth the time. You will need the managements blessing to deploy tighter security. Most people don't get it the first time you explain why it's necessary and it will take forever to explain it to a 1000 users. That's why you need the CEO to tell them.
You're absolututely right about the missing LVM. Having seen veritas, solaris lvm, vinum (freebsd), lvm and evms (linux) I have to say Linux should not take the same road. It must be done better than those LVM's. A good LVM should make it possible to change an existing disk into an lvm disk without destroying data (even on/). Also the LVM should be capable of creating raid systems transpaerently as well as resizing partitions on the fly. Generally I feel that those who impement LVM's do not understand the word Logical. The LVM should just provide the admin with 'a disk' which he/she can configure in any way (almost). I have been told the AIX LVM comes close to this.
A thing which sucks in Linux is networking. Here I'm talking about drivers and the configuration. I still have to take the time to make shure all interfaces use the right speed and duplex mode. To do this I have to use mii-tool which does't work well with all drivers. My complaint is the lack of conformity among the drivers. Look at the BSD drivers and see how it's done. Second provide a man page for every driver with good documentation especially about bugs and flaws.
The last thing I want to bring forward is the missing mpath daemon. Solaris has a very clever daemon which can group interfaces together to let the OS us more than one NIC on the same network. Thus if a wire/switch/interface fails the daemon automatically moves routes and IP's from one interfcae to the other. NB I'm not talking just about link failure but also the event where packets just gets dropped on the way through the network. I know I can by my way out of the problem but I think it should be a part of Linux.
Generally I believe that Linux needs more consistency. It still feels like a bunch of tools smacked together with a kernel created by thousands of people with different oppinions. It needs to feel like one OS from the bottom and up.
Also OpenFirmware is supposed to live in ROM or in flash. The removes the dependence in the harddrive. Those who have tried a Compaq machine will know what a hazzle it is to have the tools on the harddrive.
My experience with firmware comes form sun mostly. It works really well and is easy to understand and very configurable.
I'd love to see the BIOS go away but not with EFI or the Phoenix stuff. I don't understand why they wont support OpenFirmware instaead.
I bought a Rio 800 in LA last year and brought it back to Denmark where I live.
It has 128 MB ram which is fine for my purpose. The finish is nice and it came with a nice beltpouch.
What I don't like is the battery. Diamond claim 11 hours of playback. I have never had it play more than 2 hours. Worse is the recharging and the software that tells the battery status.
Some days ago I connected it to the recharger. After 24 hours it was up to 22%. After another 4 hours of recharging it was down to 8%. It's totally f*cked up.
I have send two emails to support but I have never gotten a reply. They seem to ignore me. I'm sending another email to try one more time. After that I will take the thing apart and try to get another rechargeble battery on my own.
We have these at my university and they suck ass. They are oftentimes not working. And I don't think its a maintainence issue, becuase we have a ton of stand alone Suns are they are rock solid, never having any troubles.
We use them at work where I'm the sysadmin and they are used at my Univesity. It's my my experience that they work fine. Not perfect but fine.
However if one is not carefull about configuring the Ray Server software or tries to tweak it beyond knowledge it will not work very well. So you have probably seen the work of a poorly skilled sysadm.
Why update to 2.4-series when it's not stable yet? Perhaps it will be now that Linus has left for 2.5?
I'm serious. I have had a lot of issues with 2.4. Ram filesystem, RAID, NFS, eepro-driver, loopback, ppp and just plain scheduler panics.
So for those people that doesn't have to use 2.4 because of some hardware related issues (SCSI controllers and whatnot) sticking with an older and better tested kernel is a good idea.
Just think about all the people that upgraded to kernel 2.4.15 and got a nasty surprise when they rebooted...
In my experience most accounting software is based on the client/server model. Still most of the software is written entirely for Windows. The accounting personel at my company has nothing to do with servers - at least as little as I have anything to do with accounting.
I would LOVE to see the serverside software written for ex. Solaris or BSD/Linux. That would make it an easier job for me whenever we upgrade our license (3-4 times a year) and also provide with a more stable platform than Win2K!
I don't care kind of software the accounting personel runs. If they are used to windows and there is a windows client - fine with me.
I can only recommend www.noise-control.de together with Fujitsu silentdrives. Don't go for Athlon but something less powerconsuming and more heat tolerant.
- You can build some very silent PC's with Cyrix processors.
In one link: href="http://www.gnu.org/philosophy/why-free.html
It's not allways about having the most features. Sometimes you you should consider whether you are doing the right thing - like not not using propriatary software (IE).
Hmm.. to start this I should say which Athlon systems I know by hand. 1 Kryotech 700@900MHz on a Microstar MB with 3/4gB ECC RAM. 6 clusternodes running 600MHz on Asus K7M with normal RAM (1 part). 1 600MHz on a Microstar MB with 384mB MCT RAM and 1 600MHz with a MB I don't know. Almost all of the machines there has had problems with RAM which we've had to change. All the PSU's in the cluster-nodes has been changed due to lack of quality. They have passed memtest-2.2 and still they crash with VM-kernel errors. The Kryotech machine is the worst of them all. In the half year we've had it has not had an uptime of more than a week. RAM has been changed several times and the BIOS has been flashed as well. Several kernels has been tried but nothing works. Worst of all it seems like Kryotech couldn't care less... Only one of the computeres (the one with a MB I don't know) has been stable. All of the others crash every once in a while.
Whether it's hardware related or it's due to some bad code in the kernel I don't know. One thing is for sure. I would not bet on an Athlon for production use.
Slashdot Mirror
I am sorry to say this but some of these complaints are valid or are believed to be valid.
A little background. I live in Denmark (Nothern EU) where windmills are used a lot. In fact we have the largest windmill producer in the world.
In Denmark we have laws stating how close you can put windmills to people. Studys have confirmed that the noise from the windmills do stress people close to the windmills. They produce a low humming noice that some people will get stress from other will get headaches and again some will not be bothered at all.
It is also a fact that the wings produce reflections from sunlight in a manner similar to stroboskobe when turning at certain speeds. Some people get ill from this - even people who are not epileptic!
I have never heard about women getting problems with their period. I guess that may be hysteria.
In Denmark the windmill producers try to reduce the reflections by coating the wings in a different ways. But most importantly we are moving the windmills to sea.
At sea there are now people living so it is less of a issue as long as they are far away from land. At sea the wind conditions are usually better resulting in more power per hour hence a better economy.
My mom has a share in a collection of windmills just outside of Copenhagen. You can see pictures here from a introductory trip this summer http://pix.xbsd.net/middelgrund-molle/.
My suggestion to New Yorkers and others would be to contact the Danish government and ask for advice and perhaps Vestas (http://www.vestas.com/) if you want to by windmills.
I can only agree. I use it to manage software installation and upgrade for all MS machines at the company where I work.
With unattended one can install and update not only MS products but all win32 software that can be installed/patched unattended.
Quote:
"In that case, no local exploit is needed--the attacker either uses sudo, or just sniffs the password the next time the user uses su..."
Unless the user has . in his/her PATH the attacker will not be able to "replace" any programs like the sh. The attacker cannot log keyboards strokes or snif the tty without root privileges. Hence if your system is set up in a good manner the attacker cannot just get the root password.
Unfortunately most users can be tricked into writing their own password and most systems like ubuntu and MacOS has sudo ALL as default.
The post is somewhat misleading. Allthough Debitel is a German company the sentence relates to the Danish daughter company which is also named Debitel.
The Danish daughter comapany had a net result of 6,000,000 kr eq. 115,000 $ in 2004.
Thus a fine of 2,000,000 kr is a lot. They even got a 'rebate'. The previous fines was 100 kr pr. email which would have resulted in 4,800,000 kr.
With Linux/BSD it is possible to buy allmost any type of computer and run the OS. On the other hand Mac OS X only runs on MacPPC and MS Windows for the most part on x86.
Take a small thing like flash. Yes it's possible to use flash on linux x86 dsitributions but what if you decide to use a Power or Sparc based computer? Then you are out of luck.
A lot of people ask me: Why would I want to use a non-x86 based computer? Perhaps because it makes your programs run faster. Perhaps it provides superior functionality that helps the OS writers prevent security problems (NX). There are lots of other examples.
The fact is that your choice between non-free software and free software inflicts upon the list of choices you get on other things (ie. hardware) and not just for you but for other people as well.
The dll's are for 32bit x86 only which means that all those who run on PPC, UltraSparc etc. are out of luck. This actually helps the x86 monopoly.
With access to the source it's possible to at least port the codec to new architectures (in most cases).
That's why opensource matters more than most care to think.
The ULE scheduler was supposed to be the default scheduler. It was placed as the default in 5.2. Between 5.2 and 5.3 kernel preemption was enabled. Some serious flaws in ULE popped up leading to crashes and other instability. Due to the complexity of ULE and preemption together only a handful of core people are able to fix this.
Since 5.x has been hold back long enough it was decided to drop ULE as the default scheduler for 5.3 and concentrate on releasing 5.3.
This doesn't mean that nobody is working on fixing the problem nor that ULE will not be the default scheduler. It is just going to take a while before it happens.
The reason for totally disabling ULE in 5.3 was to focus on other bugs in 5.3 and fix ULE on current (6.0) and then backport this to a later 5.x release.
I suggest you read cvs-src summaries at http://www.xl0.org/FreeBSD/ which gives a view on what is happening on current.
How come this has not turned into a Slashdot poll?
I believe a lot of the Slashdot reades run Linux/BSD or something like it and would benefit from a graphics card that 'Just Works (TM)'.
I work in company where it has taken a while to get the CEO and others to understand the benefits of not having extended rights.
If you want to make them understand let them manage their own PC. They will get infested and crash a lot (usually). When they ask for help install a fresh version and run a virus check on their files. Do not waste time on restoring there program settings.
Instead tell them it's the best way to deal with the problem at hand (it is!).
After losing time on this the CEO will listen to arguments like: "We/you are wasting time and time is money."
He will ask you what can be done. Tell him he will lose his admin rights and you will manage his PC (add more arguments). When he agrees make sure his PC runs smooth for a long time and when there is a problem you fix it quickly. After a while he will appriciate that he get's his job done and the admin waste less time on reinstalling his PC.
When the CEO (replace with some head guy) understands why normal users shouldn't have extended rights you can tell him that you would like his backing to take away extended rights from the normal users.
This is a very short explanation on what to do. The point is to explain to the management why it's a benefit to give up their rights - time/money!
It is not allways easy to convince the CEO but it's worth the time. You will need the managements blessing to deploy tighter security. Most people don't get it the first time you explain why it's necessary and it will take forever to explain it to a 1000 users. That's why you need the CEO to tell them.
It's very simple: choice
;)
Linux/BSD gives me the choice to do what I want with my computer. Nobody can EOL my software and nobody decides how things should work on my computer.
In other words: I'm free to paint the bikeshed whatever color I want.
Though I do tend to like the color on the FreeBSD bikeshed
A thing which sucks in Linux is networking. Here I'm talking about drivers and the configuration. I still have to take the time to make shure all interfaces use the right speed and duplex mode. To do this I have to use mii-tool which does't work well with all drivers. My complaint is the lack of conformity among the drivers. Look at the BSD drivers and see how it's done. Second provide a man page for every driver with good documentation especially about bugs and flaws.
The last thing I want to bring forward is the missing mpath daemon. Solaris has a very clever daemon which can group interfaces together to let the OS us more than one NIC on the same network. Thus if a wire/switch/interface fails the daemon automatically moves routes and IP's from one interfcae to the other. NB I'm not talking just about link failure but also the event where packets just gets dropped on the way through the network. I know I can by my way out of the problem but I think it should be a part of Linux.
Generally I believe that Linux needs more consistency. It still feels like a bunch of tools smacked together with a kernel created by thousands of people with different oppinions.
It needs to feel like one OS from the bottom and up.
Also OpenFirmware is supposed to live in ROM or in flash. The removes the dependence in the harddrive.
Those who have tried a Compaq machine will know what a hazzle it is to have the tools on the harddrive.
My experience with firmware comes form sun mostly. It works really well and is easy to understand and very configurable.
I'd love to see the BIOS go away but not with EFI or the Phoenix stuff. I don't understand why they wont support OpenFirmware instaead.
It has 128 MB ram which is fine for my purpose. The finish is nice and it came with a nice beltpouch.
What I don't like is the battery. Diamond claim 11 hours of playback. I have never had it play more than 2 hours. Worse is the recharging and the software that tells the battery status.
Some days ago I connected it to the recharger. After 24 hours it was up to 22%. After another 4 hours of recharging it was down to 8%. It's totally f*cked up.
I have send two emails to support but I have never gotten a reply. They seem to ignore me. I'm sending another email to try one more time. After that I will take the thing apart and try to get another rechargeble battery on my own.
Just my 2
We use them at work where I'm the sysadmin and they are used at my Univesity. It's my my experience that they work fine. Not perfect but fine.
However if one is not carefull about configuring the Ray Server software or tries to tweak it beyond knowledge it will not work very well. So you have probably seen the work of a poorly skilled sysadm.
Why update to 2.4-series when it's not stable yet? Perhaps it will be now that Linus has left for 2.5?
I'm serious. I have had a lot of issues with 2.4. Ram filesystem, RAID, NFS, eepro-driver, loopback, ppp and just plain scheduler panics.
So for those people that doesn't have to use 2.4 because of some hardware related issues (SCSI controllers and whatnot) sticking with an older and better tested kernel is a good idea.
Just think about all the people that upgraded to kernel 2.4.15 and got a nasty surprise when they rebooted...
2.4 shouldn't have been called 2.4 yet!
In my experience most accounting software is based on the client/server model. Still most of the software is written entirely for Windows. The accounting personel at my company has nothing to do with servers - at least as little as I have anything to do with accounting.
I would LOVE to see the serverside software written for ex. Solaris or BSD/Linux. That would make it an easier job for me whenever we upgrade our license (3-4 times a year) and also provide with a more stable platform than Win2K!
I don't care kind of software the accounting personel runs. If they are used to windows and there is a windows client - fine with me.
Just my thoughts.
I can only recommend www.noise-control.de together with Fujitsu silentdrives.
Don't go for Athlon but something less powerconsuming and more heat tolerant.
- You can build some very silent PC's with Cyrix processors.
Well take a look at the HTML-code...
In one link: href="http://www.gnu.org/philosophy/why-free.html
It's not allways about having the most features. Sometimes you you should consider whether you are doing the right thing - like not not using propriatary software (IE).
Hmm.. to start this I should say which Athlon systems I know by hand. 1 Kryotech 700@900MHz on a Microstar MB with 3/4gB ECC RAM. 6 clusternodes running 600MHz on Asus K7M with normal RAM (1 part). 1 600MHz on a Microstar MB with 384mB MCT RAM and 1 600MHz with a MB I don't know. Almost all of the machines there has had problems with RAM which we've had to change. All the PSU's in the cluster-nodes has been changed due to lack of quality. They have passed memtest-2.2 and still they crash with VM-kernel errors. The Kryotech machine is the worst of them all. In the half year we've had it has not had an uptime of more than a week. RAM has been changed several times and the BIOS has been flashed as well. Several kernels has been tried but nothing works. Worst of all it seems like Kryotech couldn't care less...
Only one of the computeres (the one with a MB I don't know) has been stable. All of the others crash every once in a while.
Whether it's hardware related or it's due to some bad code in the kernel I don't know. One thing is for sure. I would not bet on an Athlon for production use.