Linux doesn't inherit the 25 years of UNIX history. Other UNIXes are either SVR chidlren or BSD children. Linux "does its own thing" in a couple of key areas.
The linux IP stack was home grown. Note the plethora of of DOS attacks on linux's stack. These basically did not affect any other UNIX more often than not. Traditionally there have also been scalability/reliability concerns amongst ircd operators for instance, where last I heard freeBSD was the platform of choice.
The VFS/vnode layer in linux is quite different from the 4.4BSD implementation (or the SVR4 one for that matter). For more information, consider reading the websites/papers on GFS (the Global Filesystem at University of Minnesota). To paraphrase, the linux VFS layer is very local-file-system oriented, thus making it tricky to implement distributed/remote file-systems.
UNIX has been a moving target for years and years. Most modern unix deriviates are just that -- branches of the original. Linux is different. It does not contain any code retaining any of the original UNIX licenses. It was developed to work similarly to UNIX, and in many ways it does and sometimes can even work better. But it is not derived from the original codebase, nor from any of the subsequent branches. This has its good and bad points. Two "bad points" are mentioned above. The good points of starting with a clean slate (as far as implementation goes - functionality must be mimicked) are obvious to anyone doing software development.
Lots of stuff in the PeeCee world is still CPU bound - Nearly any game for instance
Disk scheduling. Operating systems have the ability to bundle and reorder disk ops to make them more efficient. The CPU has historically been so much faster that it can be worth CPU cycles to actually figure out how to pre-arrange data / access patterns
Your notion that an OS waits around doing nothing while a disk that runs on the order of 1000 times slower is pretty much wrong. Typically when there's page fault, you suspend the current pid, start the page service routine, and start running some other job off the run and/or ready queue. When the PF has been serviced, you pick up where you left off. This scheme is only detrimental when 1) you only care about single task performance -- in that case use DOS. 2) when you have thrashing
There is of course a finite limit to the amount of usefullness a 234ghz chip would do in a system that was otherwise held at todays levels -- Amdahl's law guarantees this. But one reason that it's often not worth upgrading PCs anymore as opposed to buying new ones is that all of the surrounding technologies are advancing right along with the processors -- perhaps at a slower pace, but still making forward progress.
just posting all of the personal information on every employee of the RIAA ? If any of them happen to get egged, defecated on, or receive bodily harm, who's really going to be that upset ? Is this moral? Is this legal ? There is afterall, a societal precedent for being mad at someone over who they work for. The educated reader will of course remember the fine film "Clerks", in which there is a discussion about the legitimacy of killing all of the contractors working on the Death Star. These employees of the RIAA --- like the contractors, know who they're working for. They're fully embracing the evil of their corrupt organization by choosing their own employment. They're just as guilty as Ms. Rosen. Not that I'd ever suggest anything illegal. This is, afterall, a question about what the moral duty of someone would be who happened to have this information available to them. Certainly not a suggestion. Certainly not foreshadowing of a sinister plan. Nope.
Slashdot Mirror
Linux doesn't inherit the 25 years of UNIX history. Other UNIXes are either SVR chidlren or BSD children. Linux "does its own thing" in a couple of key areas.
The linux IP stack was home grown. Note the plethora of of DOS attacks on linux's stack. These basically did not affect any other UNIX more often than not. Traditionally there have also been scalability/reliability concerns amongst ircd operators for instance, where last I heard freeBSD was the platform of choice.
The VFS/vnode layer in linux is quite different from the 4.4BSD implementation (or the SVR4 one for that matter). For more information, consider reading the websites/papers on GFS (the Global Filesystem at University of Minnesota). To paraphrase, the linux VFS layer is very local-file-system oriented, thus making it tricky to implement distributed/remote file-systems.
UNIX has been a moving target for years and years. Most modern unix deriviates are just that -- branches of the original. Linux is different. It does not contain any code retaining any of the original UNIX licenses. It was developed to work similarly to UNIX, and in many ways it does and sometimes can even work better. But it is not derived from the original codebase, nor from any of the subsequent branches. This has its good and bad points. Two "bad points" are mentioned above. The good points of starting with a clean slate (as far as implementation goes - functionality must be mimicked) are obvious to anyone doing software development.
- Lots of stuff in the PeeCee world is still CPU bound - Nearly any game for instance
- Disk scheduling. Operating systems have the ability to bundle and reorder disk ops to make them more efficient. The CPU has historically been so much faster that it can be worth CPU cycles to actually figure out how to pre-arrange data / access patterns
- Your notion that an OS waits around doing nothing while a disk that runs on the order of 1000 times slower is pretty much wrong. Typically when there's page fault, you suspend the current pid, start the page service routine, and start running some other job off the run and/or ready queue. When the PF has been serviced, you pick up where you left off. This scheme is only detrimental when 1) you only care about single task performance -- in that case use DOS. 2) when you have thrashing
There is of course a finite limit to the amount of usefullness a 234ghz chip would do in a system that was otherwise held at todays levels -- Amdahl's law guarantees this. But one reason that it's often not worth upgrading PCs anymore as opposed to buying new ones is that all of the surrounding technologies are advancing right along with the processors -- perhaps at a slower pace, but still making forward progress.just posting all of the personal information on every employee of the RIAA ? If any of them happen to get egged, defecated on, or receive bodily harm, who's really going to be that upset ? Is this moral? Is this legal ? There is afterall, a societal precedent for being mad at someone over who they work for. The educated reader will of course remember the fine film "Clerks", in which there is a discussion about the legitimacy of killing all of the contractors working on the Death Star. These employees of the RIAA --- like the contractors, know who they're working for. They're fully embracing the evil of their corrupt organization by choosing their own employment. They're just as guilty as Ms. Rosen.
Not that I'd ever suggest anything illegal. This is, afterall, a question about what the moral duty of someone would be who happened to have this information available to them. Certainly not a suggestion. Certainly not foreshadowing of a sinister plan. Nope.