Entering text into a workstation is a long way from being "yesterday's problem". Any proposed replacement for that role would need to at minimum deliver equal speed at equal cost. Dvorak delivers this whereas your proposals seem unlikely to do so.
I think Opera is the greatest thing since sliced bread and I will be ecstatic if a good port materializes on Unix. If I'm not overloaded with work I sometimes spend 20-30 hours a week in front of Opera. Here's why it is so great:
Very well designed keyboard interface including many single-key hotkeys. You never have to leave home row to go forward, back, navigate over the links on a page, move from one open document to another, follow a link, close an open document, and many other functions. If you like a vi style interface you will love the speed and control of navigating with Opera. I like its keyboard navigation better than Lynx.
Capable of keeping many windows open without performance problems. I very often have 30 windows open at once. MDI helps out here. For those with limited bandwidth this is a key feature, since it lets a big page trickle in while you're reading an already-loaded page.
Capable of launching a new window in the background, with just a single ctrl-shift-left click. Way quicker than Netscape or IE.
Very configurable in terms of customizing page rendering. I normally prefer to turn off all color and fonts, so that I can easily see what's text, what's a new link, what's a followed link. I haven't gotten it to 100% perfect but it is very good, much better than Netscape.
Very stable, except for some javascript. If javascript is off, it never dies. Nor have I seen any resource leaks. Explorer and Netscape are too unstable for me to enjoy.
Excellent hot list (bookmarks) access and configuration. No need to open a separate editing window to edit your hot list, and new hot list entries are automatically placed in the currently active hot list folder.
Good plugin compatibility (including Sun JVM) for those occasions when it's necessary.
Opera is not just an alternative browser, it's a more industrial-strength browser for people who need a highly productive interface. I think it will carve out a niche among power users on Unix systems.
Text based log files are not remotely administrable, so you must establish some form of secure local administration practice such as ssh to manage them. That doesn't scale if you have to manage (say) 1,000 servers.
What do you mean here? How are text logs more or less administrable in a secure/remote/automated way? Many people do manage 1,000 Unix systems without personally logging into them to perform maintenence.
The new Quake 3 effectively uses multiprocessing, for the first time in a computer action game AFAIK. Multiprocessing is so cheap and availiable now that I think it will be a checklist item for computer action gamers shortly. With a motherboard like the Abit BP-6 (dual socketed Celeron, $140) it is cheaper to get a lower clocked dual system than a slightly faster clocked uniprocessor system.
The software industry's response to this crisis is to concede defeat-to shoot for software that's "good enough." What "good enough" means is, of course, a matter of some debate, but critics say it is quickly becoming a euphemism for "riddled with errors"-particularly in the overheated, rush-to-market realm of net apps.
Software has been rushed to market ever since there has been software. Net distribution has reduced the friction of putting out a new version and made it practical to have a massive public beta or to cheaply distribute a patch to new customers. But good enough software goes back decades earlier. Unix was developed as "good enough" for cryin out loud.
Thompson seems to think that released bug fixes are a measure of bugginess.
Windows NT, hyped endlessly as a way to revitalize computing, was shipped in such a state that this year Microsoft released its fifth-fifth-"service pack," designed partly to patch the latest swarm of bugs.
That's just stupid. True, if a package were bug-free, no bug fixes would ever need to go out. But everyone who uses software seriously has understood for a long time that large software products all have lots of bugs. The fact that MS sends out bug fixes only means that they have the same feet of clay as everyone else.
Thompson uses some phrases without identifying where they come from.
..."good enough"......"death march"...
AFAIK those phrases were defined in the context of software development by James Bach and Ed Yourdon, respectively. They should be credited.
All in all it was a pretty fluffy piece of work I thought.
...keep in mind that evolution took 3 billion years to evolve human "intelligence". I don't see why the machines should be able to do it faster. Evolution is a horribly slow designer. Human intelligence works much faster than evolution. But we are limited by the fact that we haven't been able to design smarter people. If we manage to design a machine that is smarter than a person, and the machine is in turn able to design a smarter machine than anything we could come up with, then away we go! It doesn't particularly matter how much smarter each generation is; after five or twenty or a hundred generations, machine intelligence will dwarf human intelligence. And that will be nifty! ..brian "the optimist"
Fine, you didn't make it into the real tree. Publish a patch on your website. Let the public decide how useful the thing is.
IIRC, this was how Minix was supposed to work, and was a failure. You could publish patches all you wanted but you couldn't redistribute the modified system. The result was overly cumbersome for people who wanted to use the patches, because of dependence of one patch on another, patch application order, etc. Linux's GPL made it easy to use patches that were not in the tree.
The only time that we'll ever have to be upset is if we build an AI machine that is capable of building AI machines that are smarter than the AI machines we are capable of building
That wouldn't upset me--it would please and excite me to no end. That event will be the beginning of post-evolutionary development. I can hardly wait.
Sun NC's aren't necessarily more difficult to maintain, but the fact that people are more familiar with M$ products that I just can envision this being cost effective.
Every year, millions more people have to learn all kinds of obscure system administration tools on MS OSes when all they want to do is get their work done. There is an enormous productivity cost to making people their own sysadmins. It would be a humanitarian action to make people not need to get so familiar with the administration of MS OSes.
Sure, you find it interesting to administer your own systems. I do too. But most people don't. In a heterogenous, patched-together academic lab, most people are just frustrated by all the administrative maintenence users have to perform just to make things work. The promise of a uniform and reliable environment is appealing to the nontechnical 90% of users, not the 10% who happen to enjoy exploring the intricacies of technology. Those kids will hopefully discover the Unix command line! The server is still plenty interesting, complex and exposed to users. The only thing technical kids would give up is the hardware and low level software access.
I think you need a lesson in resisting the hypnotic gaze of Steve Jobs. Here's what's wrong with a network of dickless iMacs booting off a MacOSX server:
The iMacs are going to have a shelf life of three or four years, just like every PC and Mac at every school. Then they are going to remain in service for another three or four years, stinking up the place, just like every PC and Mac at every school. By centralizing everything except the video framebuffer and mouse/kb, Sunray should make upgrades easier.
AFAIK OSX is not available for large systems. It is desirable to centralize the server to reduce administrative load. Sun sells arbitrarily large boxes.
I think geeks overestimate the degree to which OS's influence the way people learn about computers.
This geek thinks that the affordances provided by an OS have tons to do with the direction people go with computers. If a system has only a command line, like my beloved Commodore, the user has a sink or swim experience. Those who swim quickly learn a whole lot about computers. Those who sink are in trouble. For best results there should be a bundled scripting/shell language and a programming language, ideally interpreted and with graphics.
If there is a GUI and a command line, like current Windows and Unix systems, then those who have the hacker nature will get into the CLI before long and the learning process starts. Non hackers can get by just fine.
If there is really just a GUI or menu driven interface, like MacOS or green screens, no user ever learns that much about computers from the OS, although they may have all kinds of other good learning experiences on the system. The growth of a potential hacker may be stunted.
I do agree that schools are not a useful place to have multiple platforms. Learning the details of one OS
Sun is afraid that they will wake up one day to find that someone's gone and written them out of the loop with a clustering technology that makes fast, effective use of all those MIPS going unused on folks' desks.
MIPS are about the least scarce thing I can think of for a network administrator at a facility like the one described in the article. (Disk space is a close second.) Every new PC has enough MIPS to choke a horse, way more than is required by the applications people want to run. And yet the average school or university computer lab is a mess due to unauthorized changes made to the systems by users, and differences between different generations of systems.
A more centralized computing environment is about delivering consistent, uniform, controlled, reliable user services. Very few people need more MIPS, but everyone except a bithead needs a consistent experience from all systems, with upgrades also happening system-wide. A centralized server delivers on these requirements. Users won't miss the MIPS.
These Sunrays are not NCs that execute code locally, whether that be Java or anything else. They are not even as smart as XTerms. They are graphical dumb terminals.
It is mostly young people who want the graphics goodies. Old people want to play the games they played as kids, like board games, card games, war games, old arcade games, etc.
It's also the case that young people are willing to put up with the flakiness and learning curve associated with high tech games. Most older people are unable or unwilling to deal with crap like drivers, determining hardware requirements, and learning complex controllers.
I think there is a lot of gaming revenue potential in zero-user-administration type devices, but in the retro and lower-tech area, not the games on the covers of PC gaming mags.
Your logic would mean that if by some act of genious you were able to make it rain chocolate syrup every wednesday, then if you want to drink the syrup you should pay. Otherwise, just put out an umbrella and be very careful not to let any fall into your mouth. How can you not think that is absurd?
Right! It's about time we exposed these antipiracy people for what they are: greedy chocolate syrup rain hoarders.
Your suspicions are all the more reason why you should join the fsf cadres. Code can only free so much, you need small arms to put the teeth behind the GPL.
Eat me, you moron! BNN is strident conservative bitching with a thin veneer of humor. Whereas the Onion rocks. See "Twelve More Pie-Fucking Movies In The Works". A witty comment, and one that would have been weakened if it were bowlderized.
"equipment that easily could break 40 bit encryption in real time on every link being monitored"
Isn't this an undefined capability? Finding keys isn't the same as signal processing where you can define a bandwidth and say you can process it in real time with a given piece of equipment. Depending on how many unique keys are used per megabyte of data, the job of finding those keys could be easier or harder.
I don't think we are likely to see a large scale move toward commercial games on any Unix including Linux. Here's why:
The platform with the most cutting-edge hardware is the most prominent gaming platform. Linux is nowhere near Windows on support for the most current graphics, sound and controller hardware. There is no comparison.
The growth areas in gaming are focused on people who bought their first computer recently and are not into reading manuals. I'm talking Deer Hunter and Myst territory. These are just the people who are never going to consider Linux.
The fractured nature of the Unix GUI scene does not help game designers who want as tightly defined a target platform as possible. A console is their dream, with Windows exhibiting way too much hardware diversity. Unix is worse--diverse software as well as hardware.
Linux's key strengths are wasted on games. No one wants to run new games on ancient hardware, or over networked X, or multiuser. Commercial games are not going to be GPL'd (at least not right away). And no one cares about uptime for gaming. But the windows strengths--consistent development platform, bleeding edge hardware support, market share, easy OS installs/upgrades--are exactly what game developers need.
Slashdot Mirror
I think we should be practical and go with the missiles.
Entering text into a workstation is a long way from being "yesterday's problem". Any proposed replacement for that role would need to at minimum deliver equal speed at equal cost. Dvorak delivers this whereas your proposals seem unlikely to do so.
- Very well designed keyboard interface including many single-key hotkeys. You never have to leave home row to go forward, back, navigate over the links on a page, move from one open document to another, follow a link, close an open document, and many other functions. If you like a vi style interface you will love the speed and control of navigating with Opera. I like its keyboard navigation better than Lynx.
- Capable of keeping many windows open without performance problems. I very often have 30 windows open at once. MDI helps out here. For those with limited bandwidth this is a key feature, since it lets a big page trickle in while you're reading an already-loaded page.
- Capable of launching a new window in the background, with just a single ctrl-shift-left click. Way quicker than Netscape or IE.
- Very configurable in terms of customizing page rendering. I normally prefer to turn off all color and fonts, so that I can easily see what's text, what's a new link, what's a followed link. I haven't gotten it to 100% perfect but it is very good, much better than Netscape.
- Very stable, except for some javascript. If javascript is off, it never dies. Nor have I seen any resource leaks. Explorer and Netscape are too unstable for me to enjoy.
- Excellent hot list (bookmarks) access and configuration. No need to open a separate editing window to edit your hot list, and new hot list entries are automatically placed in the currently active hot list folder.
- Good plugin compatibility (including Sun JVM) for those occasions when it's necessary.
Opera is not just an alternative browser, it's a more industrial-strength browser for people who need a highly productive interface. I think it will carve out a niche among power users on Unix systems.ergopro.com (mentioned elsewhere in these comments) looks to be cheaper.
If you have to ask "why vi over pico" you have a wonderful vi learning experience to look forward to.
What do you mean here? How are text logs more or less administrable in a secure/remote/automated way? Many people do manage 1,000 Unix systems without personally logging into them to perform maintenence.
The new Quake 3 effectively uses multiprocessing, for the first time in a computer action game AFAIK. Multiprocessing is so cheap and availiable now that I think it will be a checklist item for computer action gamers shortly. With a motherboard like the Abit BP-6 (dual socketed Celeron, $140) it is cheaper to get a lower clocked dual system than a slightly faster clocked uniprocessor system.
- Thompson seems to think bugginess is new.
- Thompson seems to think that released bug fixes are a measure of bugginess.
- Thompson uses some phrases without identifying where they come from.
..."good enough"... ..."death march"...
All in all it was a pretty fluffy piece of work I thought.The software industry's response to this crisis is to concede defeat-to shoot for software that's "good enough." What "good enough" means is, of course, a matter of some debate, but critics say it is quickly becoming a euphemism for "riddled with errors"-particularly in the overheated, rush-to-market realm of net apps.
Software has been rushed to market ever since there has been software. Net distribution has reduced the friction of putting out a new version and made it practical to have a massive public beta or to cheaply distribute a patch to new customers. But good enough software goes back decades earlier. Unix was developed as "good enough" for cryin out loud.
Windows NT, hyped endlessly as a way to revitalize computing, was shipped in such a state that this year Microsoft released its fifth-fifth-"service pack," designed partly to patch the latest swarm of bugs.
That's just stupid. True, if a package were bug-free, no bug fixes would ever need to go out. But everyone who uses software seriously has understood for a long time that large software products all have lots of bugs. The fact that MS sends out bug fixes only means that they have the same feet of clay as everyone else.
AFAIK those phrases were defined in the context of software development by James Bach and Ed Yourdon, respectively. They should be credited.
...keep in mind that evolution took 3 billion years to evolve human "intelligence". I don't see why the machines should be able to do it faster.
Evolution is a horribly slow designer. Human intelligence works much faster than evolution. But we are limited by the fact that we haven't been able to design smarter people. If we manage to design a machine that is smarter than a person, and the machine is in turn able to design a smarter machine than anything we could come up with, then away we go! It doesn't particularly matter how much smarter each generation is; after five or twenty or a hundred generations, machine intelligence will dwarf human intelligence. And that will be nifty!
..brian "the optimist"
Fine, you didn't make it into the real tree. Publish a patch on your website. Let the public decide how useful the thing is.
IIRC, this was how Minix was supposed to work, and was a failure. You could publish patches all you wanted but you couldn't redistribute the modified system. The result was overly cumbersome for people who wanted to use the patches, because of dependence of one patch on another, patch application order, etc. Linux's GPL made it easy to use patches that were not in the tree.
The only time that we'll ever have to be upset is if we build an AI machine that is capable of building AI machines that are smarter than the AI machines we are capable of building
That wouldn't upset me--it would please and excite me to no end. That event will be the beginning of post-evolutionary development. I can hardly wait.
Every year, millions more people have to learn all kinds of obscure system administration tools on MS OSes when all they want to do is get their work done. There is an enormous productivity cost to making people their own sysadmins. It would be a humanitarian action to make people not need to get so familiar with the administration of MS OSes.
Sure, you find it interesting to administer your own systems. I do too. But most people don't. In a heterogenous, patched-together academic lab, most people are just frustrated by all the administrative maintenence users have to perform just to make things work. The promise of a uniform and reliable environment is appealing to the nontechnical 90% of users, not the 10% who happen to enjoy exploring the intricacies of technology. Those kids will hopefully discover the Unix command line! The server is still plenty interesting, complex and exposed to users. The only thing technical kids would give up is the hardware and low level software access.
This geek thinks that the affordances provided by an OS have tons to do with the direction people go with computers. If a system has only a command line, like my beloved Commodore, the user has a sink or swim experience. Those who swim quickly learn a whole lot about computers. Those who sink are in trouble. For best results there should be a bundled scripting/shell language and a programming language, ideally interpreted and with graphics.
If there is a GUI and a command line, like current Windows and Unix systems, then those who have the hacker nature will get into the CLI before long and the learning process starts. Non hackers can get by just fine.
If there is really just a GUI or menu driven interface, like MacOS or green screens, no user ever learns that much about computers from the OS, although they may have all kinds of other good learning experiences on the system. The growth of a potential hacker may be stunted.
I do agree that schools are not a useful place to have multiple platforms. Learning the details of one OS
MIPS are about the least scarce thing I can think of for a network administrator at a facility like the one described in the article. (Disk space is a close second.) Every new PC has enough MIPS to choke a horse, way more than is required by the applications people want to run. And yet the average school or university computer lab is a mess due to unauthorized changes made to the systems by users, and differences between different generations of systems.
A more centralized computing environment is about delivering consistent, uniform, controlled, reliable user services. Very few people need more MIPS, but everyone except a bithead needs a consistent experience from all systems, with upgrades also happening system-wide. A centralized server delivers on these requirements. Users won't miss the MIPS.
These Sunrays are not NCs that execute code locally, whether that be Java or anything else. They are not even as smart as XTerms. They are graphical dumb terminals.
It's also the case that young people are willing to put up with the flakiness and learning curve associated with high tech games. Most older people are unable or unwilling to deal with crap like drivers, determining hardware requirements, and learning complex controllers.
I think there is a lot of gaming revenue potential in zero-user-administration type devices, but in the retro and lower-tech area, not the games on the covers of PC gaming mags.
How is this flame bait? Is the moderator who bumped this down just sarcasm-impaired?
Right! It's about time we exposed these antipiracy people for what they are: greedy chocolate syrup rain hoarders.
Free the chocolate rain!
Your suspicions are all the more reason why you should join the fsf cadres. Code can only free so much, you need small arms to put the teeth behind the GPL.
Eat me, you moron! BNN is strident conservative bitching with a thin veneer of humor.
Whereas the Onion rocks. See "Twelve More Pie-Fucking Movies In The Works". A witty comment, and one that would have been weakened if it were bowlderized.
"equipment that easily could break 40 bit encryption in real time on every link being monitored"
Isn't this an undefined capability? Finding keys isn't the same as signal processing where you can define a bandwidth and say you can process it in real time with a given piece of equipment. Depending on how many unique keys are used per megabyte of data, the job of finding those keys could be easier or harder.