Is it because its slow? X runs just fine on low end hardware. Pentium era machines have no problem with it. Perhaps you are confusing X with the window manager/desktop environment.
But which efficient environments do modern distributions have? On Fedora Core 1 I see four options GNOME, KDE, TWM, and Failsafe Terminal. That is two times bloatware, and two times too primitive for the average user. Where is the nice stable configurable KDE or GNOME like environment, that will not use a lot of RAM and CPU power?
But if you want the control (or don't trust it) then you can switch it off.
That is not a good analogy. Sure you can choose not to use HT, it will give you the same control over the system as you would have on a computer without HT. But there is no way you could utilize the full power of the CPU without HT.
feeling that hyper threading really is just
to make life easier for novice/beginner programmer
to write programs in "high" level languages
That is completely wrong. HT has nothing to do with programming language. Good compilers for high level languages will do a best effort to arange instructions such that pipelining will be as efficient as possible. But it simply isn't possible to fully utilize the execution units with a single thread of execution. HT is a good idea no matter which language you are using. But of course it does have a lot of pitfalls to avoid, which completely contradicts your claim about it being intended for novices. Can the kernel itself ensure all potential performance problems of HT are avoided, without any help from user mode applications?
it is acceptable to have source available on request only.
Yes, but then you must accompany the binary distribution with a written offer to supply the source to anybody who make that request within the next three years at no more than the cost of physically distributing the source. It would be so much easier to just accompany the binary distribution with source codes. Otherwise they have to keep the source around for three years, and deal with requests for source, which they cannot make any money of, otherwise they would be violating the GPL.
Spam with forged sender addresses causes bounces. There is no way the server generating the bounce can know if the return address is real or forged. Consequently, there is no way it can prevent bouncing to forged addresses without at the same time making email unreliable by messages silently disapearing. If you are annoyed by bounces to forged addresses, there is a fairly simple solution.
Your mail server must first of all recognize bounces. It is easy to tell a bounce from a legitimate email. Sometimes faked bounces are sent, but those faked bounces are either spam or virus, so dropping some of those by mistake would be a good thing. When you have grouped messages into bounces and other messages, you filter bounces against a list of outgoing messages. If the bounce does not contain the message ID of any outgoing message, it is not a legitimate bounce and can safely be discarded. Obviously this means you need to save the message ID of all outgoing messages from at least a week back, preferably longer timee.
When actually deciding not to deliver a message, do so as gracefully as possible. And this applies to the original message as well as bounces. You can prevent having to generate the bounce yourself by refusing the mail as early as possible. At the very latest at the end of DATA you give an error stating that the mail was not received. As soon as you have send a positive acknowledgement of the DATA you have responsibility for the message, and must send a bounce message if you cannot or willnot deliver it. By refusing to receive the message you make it somebody elses responsibility. The end of DATA is the very latest possibility to refuse receiving a message, but if possible it should be rejected earlier.
Forged email addresses are out of spec. They shouldn't be bouncing to them.
There is no way to know if the address was forged. So actually they must bounce them. Since it is the original spam with a forged address, that is violating the spec, those are the one that must not be sent. Good luck with getting the spammers to comply.
After a drive (and company) failing with the SyQuest Sparq drive, I swore off all proprietary removable storage devices. CDROMs all the way, baby.
I couldn't agree more. At the time the SparQ drive did offer a nice removable media at a good price. But shortly after I bought my drive, the company was gone. I still have my drive and five disks, and it works some of the time. Yet I cannot trust it with any of my data, because if the drive stops working completely, I have no other drive to read my disks. But what the heck, I can buy a 30GB harddisk at the same price as two 1GB SparQ disks.
It crashes Safari. Now, admittedly, I don't know whether this is a Safari bug, a Java bug, a bug in the applet, or some combination thereof
It cannot be a bug in the applet. An applet is not allowed to do anything, that could crash your browser. If an applet is actually able to crash the browser it is by definition a bug in the browser. (Here I consider the JVM to be a part of the browser, so the bug could be in either the browser or the JVM). Of course it might very well be a combination of bugs in browser and applet. But the most critical bug then is the browsers missing ability to gracefully handle bugs in the applet.
Doesn't the efficient 3D graphics implementations for XFree86 more or less already bypass X? I mean of course X is involved during initial setup, but you will map part of the graphics hardware memory directly into the clients address space. (Obviously that only works locally, but I don't care that much about performance of remote applications. In most cases network will be the bottleneck for remote applications. As long as it works I don't care if remote X displays require more CPU time).
I'm not saying that you shouldn't reconsider your layers. I don't know exactly how much OpenGL will allow you to do, but if it is a decent interface, implementing an X server on top of it should be possible. Now if that X server could even implement OpenGL and take advantage of the underlying OpenGL, we would have some really interesting stuff. And yes in that case it might even make sense to implement OpenGL on top of the hardware rather than X. However X is more than just putting 2D and 3D graphics on the screen. It also handles multiple clients and windows, which is something I don't think you could do with OpenGL. So while your idea might be interesting, I would like to see a proof of concept before accepting it as a good idea.
Multiple X implementations for Linux is not a problem. In fact it could be an advantage. The claim that it becomes harder for vendors like ATI and NVidia to target the platform is of course wrong. They only have to support one of them. So what if we end up in a situation where ATI users will need one X server and NVidia users will need another X server? As long as both X servers implement the correct protocol. Different features of course could be a problem to some extent, but since we are talking about different hardware it's obvious, that there will be differences. If it is going to happen, I hope both forks stay open (is that required by the license used by XFree86?), at least if it is open a new branch can try to merge the two if divergence become too much.
Gee, that doesn't sound at all like a security hole, no.
It is only a security hole if not done correctly. There already exist diskeless machines, that will boot from an image on a server on the network. You of course only want to do such a thing on a trusted localnet. I'd like my computers to be completely controlled through a network interface, but please give us some extra ethernet interfaces and clearly mark one of them the console interface. Of course I wouldn't want the console interface connected to the internet. Using it for the inside interface on a NAT box would be acceptable, but I would prefer having an extra interface.
Of course people not answering the poll will affect the result. Any professional should know, that the larger percentage of people not wanting to answer, the less you can trust the result. But how about people deliberately giving incorrect answers? How do you find out how large a percentage they are? (Idea for the next slashdot poll perhaps?)
That definition of randomness does make sense. Unfortunately it is undecidable, so you can never prove something is random according to the definition. You can prove something is not random, if you can find a program generating it. But if you cannot find such a program, you don't know if it is because it doesn't exist, or if you just didn't look on the right one.
As for finding a language given the string, it isn't hard to find a regular language containing the string, the hard part is to find the right language. It is trivial to define a regular language that contains all strings. But in this case it probably isn't the right one.
Any process can send any message to any other process. Talk about insecure.
Accourding to http://security.tombom.co.uk/shatter.html
it is much worse than just that. Not only can anyone send such a message, but the messages can even force the receiver to execute arbitrary code.
If they can make money from selling a poor game for Linux, at least I hope that would open some game developers eyes. Imagine what money could then be made from a good game.
he probably was driven over the edge because he had to recompile the kernel to use his video card.
I wouldn't go that far, just because I had to recompile the kernel. OTOH if it still didn't work after recompiling my kernel, because the documentation the driver developers got from the hardware manufactures was incorrect or inadequate, then I might feel like killing somebody.
I would like a C language variation that let me specify how many bits i would like to use for a variable
In C we already have types like uint8_t and int16_t. In addition to that you can use bitfields for fields in your structs. What more would you ask for? Remember, that often extra space is actually added between your variables, for performance reasons. Alligned access is faster than unalligned access.
No, they cannot blame slashdot for this one. I ask for a gif file, it gives me an html file and claims it is text/plain. That is not caused by high load on the server. That is probably caused by an admin who doesn't know how to properly configure a webserver. Too many webservers violating the standards is a major problem to the world wide web.
I will post my answers here instead. Not because I expect Microsoft to read them here, but rather because I want to make a few points about how I filled in this survey. First of all my answers are honest and mostly realistic (except I don't think Microsoft is willing to take the steps needed to satisify me. I also don't think I would trust them if they aparently did take those steps). The list of improvements Microsoft could do is short. Some people have said they should make their software GPL. I didn't go that far. Compliance with open standards and source availability is important to me, in addition if Microsoft followed those two requests, it would limit the amount of damage they could do to the computer and software industry. The list of possible improvements for Linux is two minor details, but those really are the most important places I could find any need for improvements. It is really Linux specific, so I don't see how Microsoft could ever use it to improve Windows. But I did honestly answer the question. A lot of people might have other suggestions about how to improve Linux. But most of them would probably really be about improving software and hardware to work with Linux rather than improvements of Linux which is what was asked about. And I don't see any information in my answers, which Microsoft could use to attack Linux more than they already do. And I don't see how they could use my suggestions for improvements without making life easier for the free software community. (Sorry about the bad formatting of the rest of this comment, but I had to remove some linebreaks to satisfy slashdot. Did you know, there actually is a minimum required average number of chars per line?)
1. What best describes your Linux knowledge level? Expert 2. How long have you used Linux? More than 2 years 3. Do people ask you for Linux advice? Yes 4. What best describes the kind of Linux user you are? Informal team computer expert 5. Please describe your primary home Linux machine. monitor type: 17" memory (RAM): 320MB processor type: Intel processor speed: 568MHz number of CPUs: 1 hard drive space: 111GB length of your ownership: 1-2 years 6. What OS came with your primary Linux system when you got it? No OS installed 7. What kind of network connectivity do you use for your home Linux machine? connecting to internet: Cable modem connecting to home network: Ethernet switch 8. What is the primary Linux distribution you use? at home: Red Hat at work: Red Hat 9. What region are you from? Europe 10. Who would you recommend use Linux at home? Everyone 11. Rank the reasons you like Linux for the home. not important = 0 very important = 4 1 Its cheaper 2 There is more free add-on software 0 Better gaming experience 2 Ability to run on old hardware 3 More secure than Windows 1 I want to get more Linux experience to help my career 2 No need to constantly keep installing updates and fixes 1 Integration with consumer electronics devices 3 Better performance than Windows 3 Easier to use than Windows 1 Easier to install than Windows 3 Better install and uninstall of additional software 4 No enforced license registration 4 Better scripting 3 Better command line 1 Better device and peripheral support 4 Access to source code 4 Easier to customize exactly how I want it 3 More intuitive, simpler to understand 2 Linux community support 3 Better reliability 1 The satisfaction of not giving Microsoft more money. 3 I don't trust Microsoft 3 I don't want to use proprietary software 1 I don't want to use commercial software 12. Rank the importance of various tasks you use your home Linux system for. not important = 0 very important = 4 0 Playing games 3 Hosting games 3 Internet gateway 3 Firewall 3 A home file server 3 A home web server 3 Word processing N/A Spreadsheets N/A Databases 3 E-mail 2 Web browsing 1 Home finances 3 Programming
isnt reverse engineering any M$ product against the EULA?
Maybe, but then the EULA would be against the law in some countries. Where I live the law explicitly allows you to do certain things, which include making backup copies, fixing bugs, and reverse engineering in some cases. And the law says those rights cannot be given up by agreement. So even if you had put your signature on a contract saying you had no rights to perform those actions, the law would still give you those rights.
that sort of thing has been going on since the mainframe days.
What happened on those mainframes was AFAIK not an emulation like VirtualPC and others. It was more like the way multiple user space programs run on top of the same kernel. So it was really a small kernel on top of which you could run kernels as user space programs. The most important (but not the only) difference is being able to run natively on the CPU.
Slashdot Mirror
Is it because its slow? X runs just fine on low end hardware. Pentium era machines have no problem with it. Perhaps you are confusing X with the window manager/desktop environment.
But which efficient environments do modern distributions have? On Fedora Core 1 I see four options GNOME, KDE, TWM, and Failsafe Terminal. That is two times bloatware, and two times too primitive for the average user. Where is the nice stable configurable KDE or GNOME like environment, that will not use a lot of RAM and CPU power?
But if you want the control (or don't trust it) then you can switch it off.
That is not a good analogy. Sure you can choose not to use HT, it will give you the same control over the system as you would have on a computer without HT. But there is no way you could utilize the full power of the CPU without HT.
feeling that hyper threading really is just to make life easier for novice/beginner programmer to write programs in "high" level languages
That is completely wrong. HT has nothing to do with programming language. Good compilers for high level languages will do a best effort to arange instructions such that pipelining will be as efficient as possible. But it simply isn't possible to fully utilize the execution units with a single thread of execution. HT is a good idea no matter which language you are using. But of course it does have a lot of pitfalls to avoid, which completely contradicts your claim about it being intended for novices. Can the kernel itself ensure all potential performance problems of HT are avoided, without any help from user mode applications?
it is acceptable to have source available on request only.
Yes, but then you must accompany the binary distribution with a written offer to supply the source to anybody who make that request within the next three years at no more than the cost of physically distributing the source. It would be so much easier to just accompany the binary distribution with source codes. Otherwise they have to keep the source around for three years, and deal with requests for source, which they cannot make any money of, otherwise they would be violating the GPL.
Spam with forged sender addresses causes bounces. There is no way the server generating the bounce can know if the return address is real or forged. Consequently, there is no way it can prevent bouncing to forged addresses without at the same time making email unreliable by messages silently disapearing. If you are annoyed by bounces to forged addresses, there is a fairly simple solution.
Your mail server must first of all recognize bounces. It is easy to tell a bounce from a legitimate email. Sometimes faked bounces are sent, but those faked bounces are either spam or virus, so dropping some of those by mistake would be a good thing. When you have grouped messages into bounces and other messages, you filter bounces against a list of outgoing messages. If the bounce does not contain the message ID of any outgoing message, it is not a legitimate bounce and can safely be discarded. Obviously this means you need to save the message ID of all outgoing messages from at least a week back, preferably longer timee.
When actually deciding not to deliver a message, do so as gracefully as possible. And this applies to the original message as well as bounces. You can prevent having to generate the bounce yourself by refusing the mail as early as possible. At the very latest at the end of DATA you give an error stating that the mail was not received. As soon as you have send a positive acknowledgement of the DATA you have responsibility for the message, and must send a bounce message if you cannot or willnot deliver it. By refusing to receive the message you make it somebody elses responsibility. The end of DATA is the very latest possibility to refuse receiving a message, but if possible it should be rejected earlier.
Forged email addresses are out of spec. They shouldn't be bouncing to them.
There is no way to know if the address was forged. So actually they must bounce them. Since it is the original spam with a forged address, that is violating the spec, those are the one that must not be sent. Good luck with getting the spammers to comply.
After a drive (and company) failing with the SyQuest Sparq drive, I swore off all proprietary removable storage devices. CDROMs all the way, baby.
I couldn't agree more. At the time the SparQ drive did offer a nice removable media at a good price. But shortly after I bought my drive, the company was gone. I still have my drive and five disks, and it works some of the time. Yet I cannot trust it with any of my data, because if the drive stops working completely, I have no other drive to read my disks. But what the heck, I can buy a 30GB harddisk at the same price as two 1GB SparQ disks.
It crashes Safari. Now, admittedly, I don't know whether this is a Safari bug, a Java bug, a bug in the applet, or some combination thereof
It cannot be a bug in the applet. An applet is not allowed to do anything, that could crash your browser. If an applet is actually able to crash the browser it is by definition a bug in the browser. (Here I consider the JVM to be a part of the browser, so the bug could be in either the browser or the JVM). Of course it might very well be a combination of bugs in browser and applet. But the most critical bug then is the browsers missing ability to gracefully handle bugs in the applet.
Doesn't the efficient 3D graphics implementations for XFree86 more or less already bypass X? I mean of course X is involved during initial setup, but you will map part of the graphics hardware memory directly into the clients address space. (Obviously that only works locally, but I don't care that much about performance of remote applications. In most cases network will be the bottleneck for remote applications. As long as it works I don't care if remote X displays require more CPU time).
I'm not saying that you shouldn't reconsider your layers. I don't know exactly how much OpenGL will allow you to do, but if it is a decent interface, implementing an X server on top of it should be possible. Now if that X server could even implement OpenGL and take advantage of the underlying OpenGL, we would have some really interesting stuff. And yes in that case it might even make sense to implement OpenGL on top of the hardware rather than X. However X is more than just putting 2D and 3D graphics on the screen. It also handles multiple clients and windows, which is something I don't think you could do with OpenGL. So while your idea might be interesting, I would like to see a proof of concept before accepting it as a good idea.
An md5 sum 20 bytes long.
No it is not. It is only 16 bytes. A SHA1 hash OTOH is 20 bytes long.
fragmentation could be a real concern.
Multiple X implementations for Linux is not a problem. In fact it could be an advantage. The claim that it becomes harder for vendors like ATI and NVidia to target the platform is of course wrong. They only have to support one of them. So what if we end up in a situation where ATI users will need one X server and NVidia users will need another X server? As long as both X servers implement the correct protocol. Different features of course could be a problem to some extent, but since we are talking about different hardware it's obvious, that there will be differences. If it is going to happen, I hope both forks stay open (is that required by the license used by XFree86?), at least if it is open a new branch can try to merge the two if divergence become too much.
Gee, that doesn't sound at all like a security hole, no.
It is only a security hole if not done correctly. There already exist diskeless machines, that will boot from an image on a server on the network. You of course only want to do such a thing on a trusted localnet. I'd like my computers to be completely controlled through a network interface, but please give us some extra ethernet interfaces and clearly mark one of them the console interface. Of course I wouldn't want the console interface connected to the internet. Using it for the inside interface on a NAT box would be acceptable, but I would prefer having an extra interface.
Of course people not answering the poll will affect the result. Any professional should know, that the larger percentage of people not wanting to answer, the less you can trust the result. But how about people deliberately giving incorrect answers? How do you find out how large a percentage they are? (Idea for the next slashdot poll perhaps?)
That definition of randomness does make sense. Unfortunately it is undecidable, so you can never prove something is random according to the definition. You can prove something is not random, if you can find a program generating it. But if you cannot find such a program, you don't know if it is because it doesn't exist, or if you just didn't look on the right one.
As for finding a language given the string, it isn't hard to find a regular language containing the string, the hard part is to find the right language. It is trivial to define a regular language that contains all strings. But in this case it probably isn't the right one.
Any process can send any message to any other process. Talk about insecure.
Accourding to http://security.tombom.co.uk/shatter.html it is much worse than just that. Not only can anyone send such a message, but the messages can even force the receiver to execute arbitrary code.
If they can make money from selling a poor game for Linux, at least I hope that would open some game developers eyes. Imagine what money could then be made from a good game.
he probably was driven over the edge because he had to recompile the kernel to use his video card.
I wouldn't go that far, just because I had to recompile the kernel. OTOH if it still didn't work after recompiling my kernel, because the documentation the driver developers got from the hardware manufactures was incorrect or inadequate, then I might feel like killing somebody.
I would like a C language variation that let me specify how many bits i would like to use for a variable
In C we already have types like uint8_t and int16_t. In addition to that you can use bitfields for fields in your structs. What more would you ask for? Remember, that often extra space is actually added between your variables, for performance reasons. Alligned access is faster than unalligned access.
No, they cannot blame slashdot for this one. I ask for a gif file, it gives me an html file and claims it is text/plain. That is not caused by high load on the server. That is probably caused by an admin who doesn't know how to properly configure a webserver. Too many webservers violating the standards is a major problem to the world wide web.
isnt reverse engineering any M$ product against the EULA?
Maybe, but then the EULA would be against the law in some countries. Where I live the law explicitly allows you to do certain things, which include making backup copies, fixing bugs, and reverse engineering in some cases. And the law says those rights cannot be given up by agreement. So even if you had put your signature on a contract saying you had no rights to perform those actions, the law would still give you those rights.
Unless you're talking about their router's network traffic lights... which are probably blinken like crazy
Gotta be blinking so fast the individual blinks are not actually visible anymore.
Redhat/Fedora has announced that the next Fedora release will ship with 2.6.
There is alrady an rpm in their development directory. But why didn't they sign it?
Go read the doc.
If they want us to read it, use another format.
that sort of thing has been going on since the mainframe days.
What happened on those mainframes was AFAIK not an emulation like VirtualPC and others. It was more like the way multiple user space programs run on top of the same kernel. So it was really a small kernel on top of which you could run kernels as user space programs. The most important (but not the only) difference is being able to run natively on the CPU.