I found that redhat's 2.4.7-10 is a sweetspot. Any later, and IO performance kills interactivity.
Really embarassing, actually. Apparently none of the kernel developers use it as an interactive desktop, else I don't see how this could come to pass. The newer kernels DO work fine as serveres, but that's not the end-all-be-all of linux's tasks...
yes, but that type can be polymorphic. So while not parametric polymorphism (what we normally call gerneric), or 2nd order $\Lambda$ calc (erm. Is this equivalent to system-F? I can never recall), they can be applied to arguments of differening type.
move/home and/usr/local to the same partition. If you now make 3 ish 3-4MB distro partitions, its actually quite easy to keep one stable partition and two testers: allowing you to play around, yet easily mount your imporant files to actually evaluate the testing partition.
I can't decide whether I'm just being pedantic, or disagreable, but:
silence is not an absolute. Unless you are in a vacuum. At some level, random vibrations -- hell, even thermal noise -- can be thought of as sound.(*) The point is that you don't need a moving part to generate sound.
Silence can be defined as a relative mesurement, tho. You define a threshold floor, and everything below it is silent. In a typical room -- say my kitchen, when the damn fridge compressor isn't making a racket -- you're looking at 26 dB ish.
There are CPU fans out there that are rated at 27 dB. Add a case, and that fan is silent, relative to the background noise of my kitchen. In summary: a noise is silent relative to the enviroment if it can't be detected (huge hole in reasoning, here: detected by what?) in that environment.
(*) there was also the case of the video card or system bus that made a sound while a window was being dragged. The conclusion was that there either was a loose wire that was deforming under induction, or else the random ions in the air itself were being affected by the bits flying down the bus.
amusingly, I just installed the ACME clone wily, and have been trying to teach myself mouse chording. I do this every other year or so, before I give up and go back to emacs.
I have a friend who swears by the real acme, so perhaps it is a whole-os-experience thing.
Sorry, I was being cute. What I meant was that as soon as you charge for a presumed infraction, you legitimize it. Perhaps not in the eye of the law, but definitely in the eye of the consumer.
So what is the point of a levy? A bribe to stop the RIAA suing today, without any guarantee about tomorrow? Seems shortsighted. But at least with a levy you can choose to buy CD-Rs or not.
How is this mandatory license going to work? Do I get to choose between the (non-gratis) RIAA sponsored free (speech) internet where I can send whatever bits I want, or the (gratis) open internet which censors what I can say? Pay for free speech? That seems like a horrible world -- we are now thinking of making reality WORSE than orwell's imagination. Great:-|
Look: no business model deserves protection by mandatory licensing. The net was not invented to only be accessible to people who want to pay a music tax. Who next? Movie studios too? Up-and-coming Sci-fi authors who publish their work online? I want my share of the pie as well.
Not that I'd mind communism... I like Iain M Banks. I just never thought I would get it via capitalism. Perhaps that was what Marx was on about all this time.
Seriously, I pay for things all the time: I don't shoplift even when I could get away with it, I tip well for service already recieved. Thus is a PR (primarily) and marketing (secondarily) problem. That has to be the fix. Legislating morality (c.f. drug war) won't work. Making it more attractive to be a legal downloader than illegal will.
This is not hard: a voluntary, non ursurous licence, would suffice. They could subcontract (civil contracts, not law!) to ISPs to track mp3 transfers. Eventually, they build up a case that they present to the court, and the downloader has to produce a license or pay a HEFTY fine. Just like busses work in europe.
If you finess license and tracking just a little, they could also get free grass-roots marketing, and fantastic marketing data. All you need to do is work with the consumer, not against them.
Oh wait, my counting version only works with atomic numbers. A statement like "I'd like quote-unquote-one thousand items..." leaves it ambigious whether you mean to quote the next 1000 words, starting with "items" or whether you want to quote "thousand".
Oh bother. I propose we move to a fully parenthesized prefix syntax.
shouldn't that be "Take an action quote-item-unquote, Suzy"?
Otherwise you're quoting the space in front of the word. Alternatively, you'd need to provide a counter, so you could disambiguate "Take an action quote-unquote-one item, suzy" which quotes only item as opposed to quoting item susy (quote-unquote-two)? Since you're now using quote-unquote as one terminal, you might as well drop one of them, and just use quote-two.
Don't even get me started on making little Rabbit-Ears with your fingers when you talk. Makes me violent just thinking about it.
Just because you are running a VM doesn't mean too much - in fact, it may be a bad thing, as security becomes less of a concern (no need to upgrade, doesn't matter if I am hacked).
Imagine a Java sendmail installation running on a JVM. I may be able to compromise jsendmail, and while this won't give me local root, I may be able to use the compromised sendmailer to implement social engineering hacks.
Also, if the VM is designed to run "safe" languages that rely on static type checking for safety (such as java) a non-too-impossible attach on the machine it is running to randomly corrupt memory could be used to compromise the underlying system. See the recent paper by Appel and Govindavajhala.
Which isn't to say I don't think this is a great idea. Writing your servers in a safe language is the easiest way to avoid rampant overflow errors, making jsendmail that much harder to compromise.
Also, don't forget that all those malformed-request attacks against poorly written PERL CGIs that fail to validate strings passed to a subshell: perl runs in a VM, but does no good when the attack slips by the VM to the base hardware.
While it is weird that you can't buy a free standing 16" 1600x1200 LCD monitor, if you could, I would expect it to cost about the same as a 17" 1280x1024 LCD: in the order of $500.
The biggest problem with LCDs, IIRC, is yeild yeild yeild, and that is more dependent on size than resolution.
However, as you point out, I'm sure the auxilliary circuitry that a separate display would need adds $200 to the price compared to how the costs break down for the display integrated in a laptop.
I'm running phoebe (bought a Radeon 9000, had to upgrade X, thought I'd just let RH take care of it) and have seen HUGE decreases in interactive performance.
Any sort of IO will make video playback stutter, and vmem page in/out will bring the machine to a non-interactive state for 20 odd seconds. Folks, the phoebe kernel's interactivity sucks under load. (FWIW, I have an 800Mhz Duron and 768MB ram)
I was hoping that the new O(1) scheduler and latency would fix this.
I had the same problem with stock 2.4.18 kernels, but switching back to redhat's patched-to-the-gills 7.3 kernel made ALL the problems go away. So this is clearly a kernel tuning issue.
I've tried to report these (and similar) bugs to redhat, but for X related issues at least, they only allow mailing list subscribers to post, so there's my feedback in the bit bucket.
Re:Why isn't a 2D OS enough?
on
Opencroquet
·
· Score: 1
Actually, I wanted to write a WM that shrank unfocused windows and moved them to the side of the screen. The relative sizes of the applications reflects the natural size of the window and how long since it was last active.
Whatever you are working on is in the middle of the screen, and when you switch apps, it is initially just in the background. The longer you spend with the new app, the previous app shrinks and moves towards the edge. No icons, just enlightenment style snapshots of the window.
yes, that is true: on paper one dvi-d link caps out at 165Mhz, which apparently specs out to 1600x1200. However, it _is_ possible to get a single DVI-D link to drive 19x12: Witness that people have gotten both the sony 23", Apple 23", and Samsung 24" driven digitally buy off the shelf cards that only supply single-link DVI (as you say, no card on the market provides dual-link DVI-D).
To whit, the ATI 9700 Pro has been verified to work even though it shouldn't. I'm currently installing Xfree 4.3.0 to try to drive an ATI 9000 Pro at that resolution. If it doesn't work, Dell get the p232w back. The Apple 23 inch is driven over single link DVI (well, ADC, which is DVI-D +other cables), and that is the offical way to do it.
This is contrary to what both ATI's marketing AND specs say. The common claim is that the blanking delay, which is there as a holdover from analog days, can be be used to push the additional 3600+ pixels per redraw to the monitor. Alternately, you can feed it less than 60hz.
Reminds you of the old joke: the american asks the frenchman about his new theory "Yes, but does it work in practice?", while the frenchman asks about the american's prototype "well, does yours work in theory?"
I thought the same. If gigabit ethernet were faster than scsi or ide, then why wouldn't ne have NAS drives inside the computer.
Mind you, some work has been done on switched busses for motherboard use, and perhaps, maybe, you could emulate something like that, but once again, going to the NIC+NET would seem to be a big bottleneck.
I've come up with a similar but not identical explanation: I've heard this on a laptop without speakers. I hypothesised that the currents when moving windows created a regular enough change in charge on some circuit traces that were close to the case, and that the inductive change was causing the AIR around the circuit board to react. All you need is some ions and they should react to the electric field.
My guess is that the net fluctuation of a computer is pretty random, and hence isn't percieved as noise by humans, but when moving a large block of bits around, there's enough repetition or constructive interference, or some sort of harmonic interaction *waves hands furiously* that allows the net interaction with air to be heard as sound waves.
you bought a $3K machine so you could spend more money buying an editor that has regexps AS A FEATURE! Something I take for granted is touted as a major checkmark. That is a warning sign, or at least a sign of incompatible world-view.
But seriously, if you like the mouse, then yes, emacs is not your friend.
As for applescript, I'm suprised that OS-X ports of emacs don't have some foreign-function call interface that lets them script and be scripted by applescript.
ah. You have taken the "hell is endothermic" physics test (google that for a laugh: any test whose answers include "take into account the fact that I still have not succeeded in having sexual relations with her" has got to be good).
On a more serious note, look into laser evaporation. It turns out that if you have a laser and an atom, you can tune laser so that only in the presence (sp?) of positive dopler shift (ie, atom moving towards the laser source) will the atom be able to absorb a photon. If you gradually tune the laser to a smaller and smaller band, and you have such a laser pointing from every which way, you have effectively used a laser to cool the atom.
Think of it as shooting ball bearings to stop a bowling ball.
Slashdot Mirror
I found that redhat's 2.4.7-10 is a sweetspot. Any later, and IO performance kills interactivity.
Really embarassing, actually. Apparently none of the kernel developers use it as an interactive desktop, else I don't see how this could come to pass. The newer kernels DO work fine as serveres, but that's not the end-all-be-all of linux's tasks...
resurrecting duckling
the problem there is that you're allowing assignment in an expression context. ASSSERT wants an expression that evaluates to a boolean.
So is eiffel statically type safe yet? Didn't it use to have some funky covariance that was thought safe for a long time, but turned out not to be?
Actually that sounds exactly like the behavior you describe. The whole "mytype" business -- erm, perhaps that was LOOM -- same idea anyways.
yes, but that type can be polymorphic. So while not parametric polymorphism (what we normally call gerneric), or 2nd order $\Lambda$ calc (erm. Is this equivalent to system-F? I can never recall), they can be applied to arguments of differening type.
Hence they deserveto be called generic.
(you forgot to close your tag)
/home and /usr/local to the same partition. If you now make 3 ish 3-4MB distro partitions, its actually quite easy to keep one stable partition and two testers: allowing you to play around, yet easily mount your imporant files to actually evaluate the testing partition.
<pedantic>
move
</pedantic>
I can't decide whether I'm just being pedantic, or disagreable, but:
silence is not an absolute. Unless you are in a vacuum. At some level, random vibrations -- hell, even thermal noise -- can be thought of as sound.(*) The point is that you don't need a moving part to generate sound.
Silence can be defined as a relative mesurement, tho. You define a threshold floor, and everything below it is silent. In a typical room -- say my kitchen, when the damn fridge compressor isn't making a racket -- you're looking at 26 dB ish.
There are CPU fans out there that are rated at 27 dB. Add a case, and that fan is silent, relative to the background noise of my kitchen. In summary: a noise is silent relative to the enviroment if it can't be detected (huge hole in reasoning, here: detected by what?) in that environment.
(*) there was also the case of the video card or system bus that made a sound while a window was being dragged. The conclusion was that there either was a loose wire that was deforming under induction, or else the random ions in the air itself were being affected by the bits flying down the bus.
amusingly, I just installed the ACME clone wily, and have been trying to teach myself mouse chording. I do this every other year or so, before I give up and go back to emacs.
I have a friend who swears by the real acme, so perhaps it is a whole-os-experience thing.
perhaps it's time to build that inside-out house with the padded walls... on the outside?
So is The Star making it up as it goes along, or does it know something google doesn't?
Sorry, I was being cute. What I meant was that as soon as you charge for a presumed infraction, you legitimize it. Perhaps not in the eye of the law, but definitely in the eye of the consumer.
:-|
So what is the point of a levy? A bribe to stop the RIAA suing today, without any guarantee about tomorrow? Seems shortsighted. But at least with a levy you can choose to buy CD-Rs or not.
How is this mandatory license going to work? Do I get to choose between the (non-gratis) RIAA sponsored free (speech) internet where I can send whatever bits I want, or the (gratis) open internet which censors what I can say? Pay for free speech? That seems like a horrible world -- we are now thinking of making reality WORSE than orwell's imagination. Great
Look: no business model deserves protection by mandatory licensing. The net was not invented to only be accessible to people who want to pay a music tax. Who next? Movie studios too? Up-and-coming Sci-fi authors who publish their work online? I want my share of the pie as well.
Not that I'd mind communism... I like Iain M Banks. I just never thought I would get it via capitalism. Perhaps that was what Marx was on about all this time.
Seriously, I pay for things all the time: I don't shoplift even when I could get away with it, I tip well for service already recieved. Thus is a PR (primarily) and marketing (secondarily) problem. That has to be the fix. Legislating morality (c.f. drug war) won't work. Making it more attractive to be a legal downloader than illegal will.
This is not hard: a voluntary, non ursurous licence, would suffice. They could subcontract (civil contracts, not law!) to ISPs to track mp3 transfers. Eventually, they build up a case that they present to the court, and the downloader has to produce a license or pay a HEFTY fine. Just like busses work in europe.
If you finess license and tracking just a little, they could also get free grass-roots marketing, and fantastic marketing data. All you need to do is work with the consumer, not against them.
but... but...
we already have this sort of tax: In canadia and also (I think) in germania there is a tax on all CD-R media that goes directly to the music industry.
It has always seemed to me a bad idea: since I've already paid the fine, I can commit the crime with a clear conscience.
Oh wait, my counting version only works with atomic numbers. A statement like "I'd like quote-unquote-one thousand items..." leaves it ambigious whether you mean to quote the next 1000 words, starting with "items" or whether you want to quote "thousand".
Oh bother. I propose we move to a fully parenthesized prefix syntax.
shouldn't that be "Take an action quote-item-unquote, Suzy"?
Otherwise you're quoting the space in front of the word. Alternatively, you'd need to provide a counter, so you could disambiguate "Take an action quote-unquote-one item, suzy" which quotes only item as opposed to quoting item susy (quote-unquote-two)? Since you're now using quote-unquote as one terminal, you might as well drop one of them, and just use quote-two.
Don't even get me started on making little Rabbit-Ears with your fingers when you talk. Makes me violent just thinking about it.
Just because you are running a VM doesn't mean too much - in fact, it may be a bad thing, as security becomes less of a concern (no need to upgrade, doesn't matter if I am hacked).
Imagine a Java sendmail installation running on a JVM. I may be able to compromise jsendmail, and while this won't give me local root, I may be able to use the compromised sendmailer to implement social engineering hacks.
Also, if the VM is designed to run "safe" languages that rely on static type checking for safety (such as java) a non-too-impossible attach on the machine it is running to randomly corrupt memory could be used to compromise the underlying system. See the recent paper by Appel and Govindavajhala.
Which isn't to say I don't think this is a great idea. Writing your servers in a safe language is the easiest way to avoid rampant overflow errors, making jsendmail that much harder to compromise.
Also, don't forget that all those malformed-request attacks against poorly written PERL CGIs that fail to validate strings passed to a subshell: perl runs in a VM, but does no good when the attack slips by the VM to the base hardware.
Can anyone comment on the difference compared to the Phoebe beta? Specifically, the beta is unresponsive and laggy as hell under any sort of IO load.
A better IO reponsive kernel might make me bother.
While it is weird that you can't buy a free standing 16" 1600x1200 LCD monitor, if you could, I would expect it to cost about the same as a 17" 1280x1024 LCD: in the order of $500.
The biggest problem with LCDs, IIRC, is yeild yeild yeild, and that is more dependent on size than resolution.
However, as you point out, I'm sure the auxilliary circuitry that a separate display would need adds $200 to the price compared to how the costs break down for the display integrated in a laptop.
Oh shit, that sucks!
I'm running phoebe (bought a Radeon 9000, had to upgrade X, thought I'd just let RH take care of it) and have seen HUGE decreases in interactive performance.
Any sort of IO will make video playback stutter, and vmem page in/out will bring the machine to a non-interactive state for 20 odd seconds. Folks, the phoebe kernel's interactivity sucks under load. (FWIW, I have an 800Mhz Duron and 768MB ram)
I was hoping that the new O(1) scheduler and latency would fix this.
I had the same problem with stock 2.4.18 kernels, but switching back to redhat's patched-to-the-gills 7.3 kernel made ALL the problems go away. So this is clearly a kernel tuning issue.
I've tried to report these (and similar) bugs to redhat, but for X related issues at least, they only allow mailing list subscribers to post, so there's my feedback in the bit bucket.
Actually, I wanted to write a WM that shrank unfocused windows and moved them to the side of the screen. The relative sizes of the applications reflects the natural size of the window and how long since it was last active.
Whatever you are working on is in the middle of the screen, and when you switch apps, it is initially just in the background. The longer you spend with the new app, the previous app shrinks and moves towards the edge. No icons, just enlightenment style snapshots of the window.
yes, that is true: on paper one dvi-d link caps out at 165Mhz, which apparently specs out to 1600x1200. However, it _is_ possible to get a single DVI-D link to drive 19x12: Witness that people have gotten both the sony 23", Apple 23", and Samsung 24" driven digitally buy off the shelf cards that only supply single-link DVI (as you say, no card on the market provides dual-link DVI-D).
To whit, the ATI 9700 Pro has been verified to work even though it shouldn't. I'm currently installing Xfree 4.3.0 to try to drive an ATI 9000 Pro at that resolution. If it doesn't work, Dell get the p232w back. The Apple 23 inch is driven over single link DVI (well, ADC, which is DVI-D +other cables), and that is the offical way to do it.
This is contrary to what both ATI's marketing AND specs say. The common claim is that the blanking delay, which is there as a holdover from analog days, can be be used to push the additional 3600+ pixels per redraw to the monitor. Alternately, you can feed it less than 60hz.
Reminds you of the old joke: the american asks the frenchman about his new theory "Yes, but does it work in practice?", while the frenchman asks about the american's prototype "well, does yours work in theory?"
well, then.
AOP is like metaprogramming, but more controlled and disciplined, so that it fits into the statically typed Class/Object/single inhertiance world.
I thought the same. If gigabit ethernet were faster than scsi or ide, then why wouldn't ne have NAS drives inside the computer.
Mind you, some work has been done on switched busses for motherboard use, and perhaps, maybe, you could emulate something like that, but once again, going to the NIC+NET would seem to be a big bottleneck.
I've come up with a similar but not identical explanation: I've heard this on a laptop without speakers. I hypothesised that the currents when moving windows created a regular enough change in charge on some circuit traces that were close to the case, and that the inductive change was causing the AIR around the circuit board to react. All you need is some ions and they should react to the electric field.
My guess is that the net fluctuation of a computer is pretty random, and hence isn't percieved as noise by humans, but when moving a large block of bits around, there's enough repetition or constructive interference, or some sort of harmonic interaction *waves hands furiously* that allows the net interaction with air to be heard as sound waves.
nah!
you bought a $3K machine so you could spend more money buying an editor that has regexps AS A FEATURE! Something I take for granted is touted as a major checkmark. That is a warning sign, or at least a sign of incompatible world-view.
But seriously, if you like the mouse, then yes, emacs is not your friend.
As for applescript, I'm suprised that OS-X ports of emacs don't have some foreign-function call interface that lets them script and be scripted by applescript.
Different [key]strokes for different folks.
ah. You have taken the "hell is endothermic" physics test (google that for a laugh: any test whose answers include "take into account the fact that I still have not succeeded in having sexual relations with her" has got to be good).
On a more serious note, look into laser evaporation. It turns out that if you have a laser and an atom, you can tune laser so that only in the presence (sp?) of positive dopler shift (ie, atom moving towards the laser source) will the atom be able to absorb a photon. If you gradually tune the laser to a smaller and smaller band, and you have such a laser pointing from every which way, you have effectively used a laser to cool the atom.
Think of it as shooting ball bearings to stop a bowling ball.