This is in the FAQ: I suspect OpenMOSIX will not speed up compilation very much, because small bursty tasks, like compilers, probably perform poorly on MOSIX's process-migration architecture. As I understand it, it should work much better for interactive use or scientific computing where the tasks can usefully stay on each machine for a long time.
But if you want to post some comaritive performance measurements I'm sure everybody would find it interesting.
You could look at HP's iPAQ music center. I haven't seen one in the flesh.
It's bigger and probably louder than the Slim Devices unit, but it doesn't require a PC to be running. Unsurprisingly I hear it runs Linux/Samba/etc inside too.
They used to have one under the HP brand with a CD-R, but perhaps Carly downsized it.
judge-for-yourself is amusingly lame. If you feel like you're not getting enough Weight Loss, Make Money Fast, and Net Detective offers in email, jfy is the perfect place to find them.
Even if Windows did double drive letters to give you AA:, it would break practically every program that tried to use them assuming they were single. I hate to think how many libraries have something like
You can already do software upgrades for modern performance cars and motorbikes. Just insert a new EPROM, or on more modern machines, reflash the existing chip. This can control fuel injection, timing, response to throttle, etc.
If you're really keen, you can get software to design new maps yourself, though without a dynometer, fuel/air sensor and other tools you're probably just making things worse. In addition, the programming can be more tightly tuned to the setup and application of your own machine.
Yes, you could presumably break the engine if you get it wrong, but I guess most people are not that dumb -- they get a good chip from a reputable dealer and enjoy a nice 5% boost.
It's funny, when I first read this, I thought it was a spoof of Larry McVoy defending the bitkeeper licence and whining about opensource people being thieves. Scary thought.
Much as I would like to see encryption be commonplace, I think there are harder problems than just getting it built into many mail clients.
First of all, most deployed software is insecure, and most machines are configured badly. If everybody used OpenPGP, then there would be key-stealing Outlook/Word/IE worms.
It is not completely clear that PGP's web of trust system can scale up to a system where most users are naive and many keys are compromised. Will it really cope with being flooded with key signatures that are not properly validated (against photo id, etc), or that were made with compromised keys.
I think these can be overcome, but it requires more than just shelling out to PGP. It will need some really serious thought about how to write a user interface that clearly explains security actions without overwhelming the user. It needs better investment in infrastructure to keep keyservers and revocation lists up to date. Possibly it needs smarter trust metrics that can cope with Joe AOLer's tendency to sign anyone's key when he's asked.
The merger numbers showed that HP was actually losing money on every MS-loaded PC sold. In any case, at least a few percent of HP's sales now run Linux. (Like you, I am too lazy to check the numbers.) So "probably makes 1000x times more selling MS-loaded computers" is, primae facie, crap.
It looks like a really interesting technique. However, the value of the second page (and your linux credibility:-) is somewhat eroded by the fact that the images don't show up. Looking at the HTML shows lines like this:
Rather than downloading the source and recompiling, a few of Slashdot's Finest have tried to download all the images by brute force and ignorance. It's kind of funny watching their half-arsed attempts to write a recursive download script.
Take for example the shining wit from sentry.seaspace.com (192.150.113.5). First, with charming naivete, he tries to give a wildcard to an HTTP request:
GET/tiles/zoom064/x014400_y019200/* HTTP/1.0
D'oh! But our hero is not foiled so easily. Next, he tries to write a little script to download the images using wget, but unfortunately a 1-line shell script seems to be too hard. The server sees about a hundred requests like this, for nonexistent files.
GET/tiles/zoom064/x014400_y019200/x.big.png
If something doens't work, just try harder! Another attempt at this lame little script misinterprets the filename coding and produces a few thousand requests for other nonexistent files.
You might think that somebody who's written a script like this would keep an eye on it to see if it's working properly, rather than allowing it to run for an hour getting one error message after another. But apparently not for Mr Seaspace. Has he gone off to apply his l33t leeching skills to some other site? Is he tending to his goats? We'll never know.
Yep, I agree that would be more interesting. It's kind of a hard problem though, especially for software as large as the kernel. I worked on this a bit, and I think that it's probably impossible to generate it automatically: you need some human intervention to say which couplings are really meaningful and which ones are accidental. For example, having a line from every file in the kernel to printk would not really help anyone.
The kernel map is more about being pretty and giving a very-high-level feel for the complexity and structure of the kernel, and not so much a practical tool for programmers.
Serving any page request is a constant-time operation, regardless of zoom level. (I wrote it, but if you don't believe me look at the source.)
The server is using Apache's mod_throttle to try to prevent overexcited slashdot denizens from trying to recursively fetch the whole thing. You weren't trying that by any chance?:-)
Possibly the network is slow somewhere but I doubt the machine is overloaded.
If you really want to mirror it then download the source, which is much smaller (about 100kB?) Run the image generator on your own machine and set it up according to the instructions.
16 If a man has an emission of semen, he shall bathe his whole body in water, and be unclean until the evening. 17Everything made of cloth or of skin on which the semen falls shall be washed with water, and be unclean until the evening. 18If a man lies with a woman and has an emission of semen, both of them shall bathe in water, and be unclean until the evening.
19 When a woman has a discharge of blood that is her regular discharge from her body, she shall be in her impurity for seven days, and whoever touches her shall be unclean until the evening. 20Everything upon which she lies during her impurity shall be unclean; everything also upon which she sits shall be unclean. 21Whoever touches her bed shall wash his clothes, and bathe in water, and be unclean until the evening. 22Whoever touches anything upon which she sits shall wash his clothes, and bathe in water, and be unclean until the evening; 23whether it is the bed or anything upon which she sits, when he touches it he shall be unclean until the evening. 24If any man lies with her, and her impurity falls on him, he shall be unclean seven days; and every bed on which he lies shall be unclean.
I hope you abide by these rules just as strictly. If you hold hands with your wife or friend when she has her period, or allow her to sit on your sofa or in your car, you're going straight to hell! And don't forget, she has to make a burnt offering of turtledoves or pigeons if she has a heavy period.
This is what Whitfield Diffie refers to as the "marijuanaization of crypto": a majority of interests think the law is dumb and it is widely flouted, but for various reasons it stays in place, partly because many police would need to find useful jobs if it were removed.
It's a conference speech, not a serious essay. Coherence is optional. Having an appropriate mix of flattery, cute ear-candy analogies, and provocative trolling is much more important.
Everybody's talking about Bruce for fifteen minutes, so his mission is accomplished. Presumably his ego is stroked, recognition and purchases of his books will go up marginally, and he's more likely to be invited next time.
Richard Gabriel did a better job of describing the futility of doing software these days:
Over the years I've despaired that the ways we've created to build software matches less and less well the ways that people work effectively. More so, I've grown saddened that we're not building the range of software that we could be, that the full expanse of what computing could do--to enhance human life, to foster our creativity and mental and physical comfort, to liberate us from isolation from knowledge, art, literature, and human contact--is left out of our vision. It seems that high-octane capitalism has acted like an acid or a high heat to curdle and coagulate our ways of building software into islands that limit us.
But like survivors, we've managed to make these islands homes. We've found the succulent but bitter fruit that can sustain us, the small encrusted or overfurred creatures we can eat to survive, the slow-moving and muddy streams from which we drink against the urge to spit it out. Being survivors, we can make do with little. But how little like life is such an existence.
Keynes said something like "economics is ideal for somebody with the temperament of an artist but not the talent." (I don't have the exact quote here.)
Describing the parties as media companies, tech companies, and consumers is part of the problem!
We are: people, citizens, individuals, readers, and so on. We are not merely "consumers". The word has several unjustified associations that help media companies define the terms of the debate.
"Consumer" implies that people passively and mindlessly eat what they're given. Even in the status quo this is not true: people criticize, filter, and converse with media; in the internet age it is even less true.
"Consumer" implies that people only consume and do not produce; in fact there is no sharp boundary, and the web blurs it even more.
"Consumer" implies that the party has no rights or
responsibilities other than to consume.
You should reject description of yourself as a "consumer", and use "citizen", "person", "individual" or something similar instead.
I think reality is more or less like that, except that there is another amount, the loss of goodwill or future business caused by bad publicity (D). X = A * B * C + D.
You can reduce D by either having good publicity, or manipulating the media or covering things up. I suspect in some countries where lawsuits are less popular/successful, it dominates ABC.
People in a business who want to "do the right thing" even when it's not strictly necessary can try to justify it by pointing to D, which is kind of hard to predict or quantify.
Cannabis making people peaceful is probably at least as much due to the cultural context of it in the west as to anything inherent in the drug. Consider the use of hashish by Assassins to sooth fears before battle, or of marijuana by present-day African militias.
Right, and that's why it was originally a good idea in C -- because in pre-ANSI C variables were very weakly typed, and you could easily use a (float *) when you wanted an (int *), with semicatastrophic consequences. If the compiler doesn't warn you, then it can be hard to keep them straight, so using Hungarian is an adequate workaround. Even in those days using a good Lint might have been better though.
Of course modern compilers do keep types (more or less) straight, and so perpetuating this is pretty dumb.
OS/2 also gave a priority boost to the GUI foreground application. (Should I be using past tense? Probably;-)
Anyhow, it would be an interesting project to try to implement this on Unix. It probably doesn't strictly require the kernel to know about GUIs. It might be enough for the window manager to just advise the kernel to boost a process's priority when it's got focus.
One impediment is that of course making a process less nice is a privileged operation. You might fix this by having a setuid helper, or perhaps by creative use of process capabilities. Arguably user processes *should* be able to make themselves more nice, and then recover, but that would be a more controversial and intrusive change...
Some distributions run X at a negative niceness, to make it more responsive. If we implemented priority inversion then this could work even better: applications would get a boost when X was waiting for them to respond to input or redraw themselves.
Here's a Doxygenated version of the Samba source code. You can see that not much of the source has real documentation yet, so only a few functions have really nice documentation. This actually makes it a nice example for a project thinking about changing across, because even without much markup Doxygen still gives you useful cross-referencing a source browsing.
On the whole, the cross-referencing is less useful than Exuberant-etags and id-utils under emacs, but it's still pretty cool.
Doxygen can produce TeX output, however it doesn't look great, and for a project as big as Samba it can overflow internal limits in the default build of LaTeX. I'm sure you could patch it to look better.
Another weakness is that it does not know about CVS. It's nice to see the history of the code integrated with the current state. LXR will do that for you, but at the cost of a more complicated server-side installation.
Slashdot Mirror
GNU Make already does that. RTFM.
This is in the FAQ: I suspect OpenMOSIX will not speed up compilation very much, because small bursty tasks, like compilers, probably perform poorly on MOSIX's process-migration architecture. As I understand it, it should work much better for interactive use or scientific computing where the tasks can usefully stay on each machine for a long time.
But if you want to post some comaritive performance measurements I'm sure everybody would find it interesting.
You could look at HP's iPAQ music center. I haven't seen one in the flesh.
It's bigger and probably louder than the Slim Devices unit, but it doesn't require a PC to be running. Unsurprisingly I hear it runs Linux/Samba/etc inside too.
They used to have one under the HP brand with a CD-R, but perhaps Carly downsized it.
judge-for-yourself is amusingly lame. If you feel like you're not getting enough Weight Loss, Make Money Fast, and Net Detective offers in email, jfy is the perfect place to find them.
Even if Windows did double drive letters to give you AA:, it would break practically every program that tried to use them assuming they were single. I hate to think how many libraries have something like
..
if (isletter(p[0])) && (p[1]==':'))
You can already do software upgrades for modern performance cars and motorbikes. Just insert a new EPROM, or on more modern machines, reflash the existing chip. This can control fuel injection, timing, response to throttle, etc.
If you're really keen, you can get software to design new maps yourself, though without a dynometer, fuel/air sensor and other tools you're probably just making things worse. In addition, the programming can be more tightly tuned to the setup and application of your own machine.
Yes, you could presumably break the engine if you get it wrong, but I guess most people are not that dumb -- they get a good chip from a reputable dealer and enjoy a nice 5% boost.
It's funny, when I first read this, I thought it was a spoof of Larry McVoy defending the bitkeeper licence and whining about opensource people being thieves. Scary thought.
Much as I would like to see encryption be commonplace, I think there are harder problems than just getting it built into many mail clients.
First of all, most deployed software is insecure, and most machines are configured badly. If everybody used OpenPGP, then there would be key-stealing Outlook/Word/IE worms.
It is not completely clear that PGP's web of trust system can scale up to a system where most users are naive and many keys are compromised. Will it really cope with being flooded with key signatures that are not properly validated (against photo id, etc), or that were made with compromised keys.
I think these can be overcome, but it requires more than just shelling out to PGP. It will need some really serious thought about how to write a user interface that clearly explains security actions without overwhelming the user. It needs better investment in infrastructure to keep keyservers and revocation lists up to date. Possibly it needs smarter trust metrics that can cope with Joe AOLer's tendency to sign anyone's key when he's asked.
The merger numbers showed that HP was actually losing money on every MS-loaded PC sold. In any case, at least a few percent of HP's sales now run Linux. (Like you, I am too lazy to check the numbers.) So "probably makes 1000x times more selling MS-loaded computers" is, primae facie, crap.
... the angry monkey dance, on the other hand, is the acme of professionalism, reasoned argument, taste and decorum.
It looks like a really interesting technique. However, the value of the second page (and your linux credibility :-) is somewhat eroded by the fact that the images don't show up. Looking at the HTML shows lines like this:
<img src="file:///F:/kernel-2.5.33-drivers.png" alt="">
Oops.
Take for example the shining wit from sentry.seaspace.com (192.150.113.5). First, with charming naivete, he tries to give a wildcard to an HTTP request:
GET /tiles/zoom064/x014400_y019200/* HTTP/1.0
D'oh! But our hero is not foiled so easily. Next, he tries to write a little script to download the images using wget, but unfortunately a 1-line shell script seems to be too hard. The server sees about a hundred requests like this, for nonexistent files.
GET /tiles/zoom064/x014400_y019200/x.big.png
If something doens't work, just try harder! Another attempt at this lame little script misinterprets the filename coding and produces a few thousand requests for other nonexistent files.
You might think that somebody who's written a script like this would keep an eye on it to see if it's working properly, rather than allowing it to run for an hour getting one error message after another. But apparently not for Mr Seaspace. Has he gone off to apply his l33t leeching skills to some other site? Is he tending to his goats? We'll never know.
Ka-plunk.
Yep, I agree that would be more interesting. It's kind of a hard problem though, especially for software as large as the kernel. I worked on this a bit, and I think that it's probably impossible to generate it automatically: you need some human intervention to say which couplings are really meaningful and which ones are accidental. For example, having a line from every file in the kernel to printk would not really help anyone.
The kernel map is more about being pretty and giving a very-high-level feel for the complexity and structure of the kernel, and not so much a practical tool for programmers.
The server is using Apache's mod_throttle to try to prevent overexcited slashdot denizens from trying to recursively fetch the whole thing. You weren't trying that by any chance? :-)
Possibly the network is slow somewhere but I doubt the machine is overloaded.
If you really want to mirror it then download the source, which is much smaller (about 100kB?) Run the image generator on your own machine and set it up according to the instructions.
16 If a man has an emission of semen, he shall bathe his whole body in water, and be unclean until the evening. 17Everything made of cloth or of skin on which the semen falls shall be washed with water, and be unclean until the evening. 18If a man lies with a woman and has an emission of semen, both of them shall bathe in water, and be unclean until the evening.
19 When a woman has a discharge of blood that is her regular discharge from her body, she shall be in her impurity for seven days, and whoever touches her shall be unclean until the evening. 20Everything upon which she lies during her impurity shall be unclean; everything also upon which she sits shall be unclean. 21Whoever touches her bed shall wash his clothes, and bathe in water, and be unclean until the evening. 22Whoever touches anything upon which she sits shall wash his clothes, and bathe in water, and be unclean until the evening; 23whether it is the bed or anything upon which she sits, when he touches it he shall be unclean until the evening. 24If any man lies with her, and her impurity falls on him, he shall be unclean seven days; and every bed on which he lies shall be unclean.
I hope you abide by these rules just as strictly. If you hold hands with your wife or friend when she has her period, or allow her to sit on your sofa or in your car, you're going straight to hell! And don't forget, she has to make a burnt offering of turtledoves or pigeons if she has a heavy period.
Ah, there's a good god for the 21st century.
This is what Whitfield Diffie refers to as the "marijuanaization of crypto": a majority of interests think the law is dumb and it is widely flouted, but for various reasons it stays in place, partly because many police would need to find useful jobs if it were removed.
--
0x60a15ec5
The cup holders in my Alfa Romeo aren't big enough to hold American-brand drink bottles. I have to buy Evian :-)
Everybody's talking about Bruce for fifteen minutes, so his mission is accomplished. Presumably his ego is stroked, recognition and purchases of his books will go up marginally, and he's more likely to be invited next time.
Richard Gabriel did a better job of describing the futility of doing software these days:
Keynes said something like "economics is ideal for somebody with the temperament of an artist but not the talent." (I don't have the exact quote here.)
Visual C++ likewise.
Describing the parties as media companies, tech companies, and consumers is part of the problem!
We are: people, citizens, individuals, readers, and so on. We are not merely "consumers". The word has several unjustified associations that help media companies define the terms of the debate.
"Consumer" implies that people passively and mindlessly eat what they're given. Even in the status quo this is not true: people criticize, filter, and converse with media; in the internet age it is even less true.
"Consumer" implies that people only consume and do not produce; in fact there is no sharp boundary, and the web blurs it even more.
"Consumer" implies that the party has no rights or
responsibilities other than to consume.
You should reject description of yourself as a "consumer", and use "citizen", "person", "individual" or something similar instead.
I think reality is more or less like that, except that there is another amount, the loss of goodwill or future business caused by bad publicity (D). X = A * B * C + D.
You can reduce D by either having good publicity, or manipulating the media or covering things up. I suspect in some countries where lawsuits are less popular/successful, it dominates ABC.
People in a business who want to "do the right thing" even when it's not strictly necessary can try to justify it by pointing to D, which is kind of hard to predict or quantify.
Cannabis making people peaceful is probably at least as much due to the cultural context of it in the west as to anything inherent in the drug. Consider the use of hashish by Assassins to sooth fears before battle, or of marijuana by present-day African militias.
Right, and that's why it was originally a good idea in C -- because in pre-ANSI C variables were very weakly typed, and you could easily use a (float *) when you wanted an (int *), with semicatastrophic consequences. If the compiler doesn't warn you, then it can be hard to keep them straight, so using Hungarian is an adequate workaround. Even in those days using a good Lint might have been better though.
Of course modern compilers do keep types (more or less) straight, and so perpetuating this is pretty dumb.
OS/2 also gave a priority boost to the GUI foreground application. (Should I be using past tense? Probably ;-)
Anyhow, it would be an interesting project to try to implement this on Unix. It probably doesn't strictly require the kernel to know about GUIs. It might be enough for the window manager to just advise the kernel to boost a process's priority when it's got focus.
One impediment is that of course making a process less nice is a privileged operation. You might fix this by having a setuid helper, or perhaps by creative use of process capabilities. Arguably user processes *should* be able to make themselves more nice, and then recover, but that would be a more controversial and intrusive change...
Some distributions run X at a negative niceness, to make it more responsive. If we implemented priority inversion then this could work even better: applications would get a boost when X was waiting for them to respond to input or redraw themselves.
On the whole, the cross-referencing is less useful than Exuberant-etags and id-utils under emacs, but it's still pretty cool.
Doxygen can produce TeX output, however it doesn't look great, and for a project as big as Samba it can overflow internal limits in the default build of LaTeX. I'm sure you could patch it to look better.
Another weakness is that it does not know about CVS. It's nice to see the history of the code integrated with the current state. LXR will do that for you, but at the cost of a more complicated server-side installation.
--
Martin