The ppc backend is more computationally expensive than the x86 backend, in my experience. gcc runs fast on athlons because the x86 codegen doesn't take as much time as ppc codegen. (I'm 90% sure about this, since I didn't check it carefully or anything, and it was a while ago. I used gcc 2.94 when I tried a bit of stuff.) #define X(x,y) x##y
So good an idea in fact that it's been had before:)
I used to be subscribed to the list, which was win32@lists.debian.org, but I can't find an archive of it, and i don't know if it still exists.
If you search around, you may be able to find some posts describing an package repository of.deb packages built for cygwin or mingw. #define X(x,y) x##y
> So get UltraSparc's score, double it, and you
> get... 876, which is more than the shoddy 573 the P4 gets
And you double it how? I'd be impressed if you could show me where you could buy an UltraSPARC that ran at 1.8GHz. I'm assuming the fastest Ultra you can buy is the 900MHz one, and the P4 kicks its butt. The key is that the P4 does run that fast, but nobody knows how to make an UltraSPARC run that fast, even if they wanted to.
Also note that you'd need to double the speed of the memory bus, and halve all latencies in everything, especially the RAM, to get double the SPECcpu score.
Sun's *SPARC cpus are way behind everyone else, but they keep selling them because they put them in very reliable machines and sell good support. Their good name helps, too. High processing speed is not their forte.
Read Silicon Insider: http://www.realworldtech.com/page.cfm?Section=Colu mns&Subject=Insider, he has some comments on SPARC. #define X(x,y) x##y
The major problem would be IO bandwidth, so you'd have to use something other than multiple cards in a 32bit 33MHz PCI bus. That has a theoretical limit of 132 MB/s, which is too low a bisection bandwidth for any respectable gigE switch. NVidia's nforce chipset has two PCI busses, just like my old Mac clone from Daystar, so that would help some. Still, with a server mobo and some 64bit 66MHz PCI busses, you'd have 528MB/s per bus. That's decent, but still nowhere near what you want for 24 GE ports. You'd also need to buy several 4-port NICs, and find slots to put them in. Switches really are hard to do with software on commodity machines. Routing is another matter, unless you have a lot of subnets and need a lot of bisection bandwidth, I guess. (If you were going to be limited by the fact that all the traffic was going up or down through one of the interfaces, and not between two internal ones, then a computer with NICs could probably do the job reasonably well. #define X(x,y) x##y
Legal measures, like requiring the opt-out to actually work, might get somewhere. It would be good if spammers had to be cautious to avoid fines because they sent you spam after you opted out. (and yes, I think just a fine, not the death penalty or anything would be appropriate, since honest mistakes could happen. If the penalty is too harsh, spammers might not get fined if they can show that there was some kind of honest mistake. If there's a 100 $ fine that they have to pay, no court wrangling or bullshit like that, then they'll behave.) #define X(x,y) x##y
That wasn't a troll (err, not necessarily), since it was a quote from Mad TV. Man, I haven't watched that show for years. (not that I've missed it...)
#define X(x,y) x##y
Re:partially correct (HP-UX) -you need a K&R compi
on
GCC 3.0 Released
·
· Score: 1
> Nothing to do with it. Everything after stage1 is a consistency check.
Not quite. Assuming GCC generates faster/better code than the native compiler, installing GCC compiled by GCC will make compiling other program happen faster.
Thanks for pointing out the consistency check going on with stage 3. I hadn't known why it was useful to build gcc again with gcc.
good point about not letting Weird Shit take too much time away from the good stuff. Keeping the computer in a family room should be enough to stop kids from actively seeking porn or blood & gore, or any stuff they think will make them look bad in their parents eyes. It would be embarrasing to look at most Bad Stuff in front of your parents, or even for them to know that you had been looking at it. Using censorware to block stuff is not necessary, since I think fear of embarrasment will do the trick. #define X(x,y) x##y
logging URLs with stuff after '?' stripped off (to clean off GET form submission data, to avoid being to invasive) is way different from logging chat sessions. I don't know how a 10 year old would feel about having their every move online logged, but I know I don't like the idea. Parents don't need to read everything their kids type. #define X(x,y) x##y
I think most of the stuff that's not appropriate for kids is not very entertaining for them either, and they probably won't find nasty stuff unless they are trying to.
I'm having a hard time thinking of things that a kid could see on the 'net that you couldn't talk with them about.
I'll admit that I haven't had to be in the position of supervising kids browsing the 'net, so I'm probably thinking in too abstract terms. I've taught Sunday school for a few years now, so I have some experience with kids around age 10, though. Maybe I'm overly optimistic in my idea that kids can understand the fact that some of the stuff on the 'net is _not_ good wisdom or words to live by (e.g. racist material). I do think that knowing something about what's out there won't hurt anyone over age 10. I would guess that the Internet helps introduce difficult topics for conversation.
One thing I just thought of is that this only works if you have quite tolerant views, like I do, and don't think it's a problem for kids to know what sex is about, even before they're old enough to want to do it. The English language is so full of phrases with double entendres because it has a long history of being spoken by prudes and other wimpy people who would die if they said penis (insert Monty Python song here:). (e.g. There's a big difference between two people having sex and two people sleeping in the same bed because there was a limited number of beds. However, saying "they slept together" means they had sex, so you have to waste a lot of words explaining that they slept in the same bed without having sex. English is dumb because people in the past (and wimps in the present) like to pretend they aren't talking about things even when they are.) Anyway, now you might have some idea why I think it's more or less ok for kids to look at web sites once they've reached a certain level of maturity,
and can take it. Obviously a 5 year old isn't going to understand some of the stuff you'd need to explain after they tripped across certain web sites, so actually being there _while_ they browse is important, because they will be highly unlikely to read some stuff with you right there. This is the same reason teenagers hide their porn from their parents. (OTOH, 5 year olds probably don't read well enough to get too far on the web, but a 7 or 8 year old who reads well could give parents something to worry about.) #define X(x,y) x##y
> Passing through each one is a super-cooled liquid nitrogen that pushed
> the temperature of the tube down towards 0 degrees kelvin.
But not very close to 0 there, sissy boy. Real men use liquid helium, which doesn't even liquify until a couple degrees K. There's a good reason why it costs a lot more. Cooling LN2 down to those temperatures could be done, but it just isn't the same. It's texture is all wrong. Just last week, I dropped the temp. a few more degrees, down to where helium becomes superfluid. Now we're talking smooth sound... (err, I mean "I definitely sounded better; It was worth it". Almost missed my audiophile-speak...)
I also use lead cables, which become superconducting at the temperatures I'm talking about. Good old fashioned type I superconduction, because the newfangled high temp. type II superconductors just don't sound as vibrant.
(In real life, silver, gold, and copper don't superconduct, even though they are the best conductors at room temp (in that order, BTW).) #define X(x,y) x##y
You don't need accurate clocks in the mechanism that spins the CD. All you need is to read it into a memory buffer, and feed the data from the buffer to a DAC at a fixed rate. Of course, that _does_ have to happen at a controlled rate.
CD audio is random access (AFAIK), but due to the brilliant design, there are no position markers, so the drive can't tell where it is when it tries to seek somewhere. It can only get to approximately where you asked it to. The reason like CD paranoia (see xiph.org) exists is to match up the data from the end of one read to the start of the next. I seem to remember something else about anti-jitter, too, but I can't remember why. I actually don't remember a whole bunch about this right now, so anyone who cares should go and read the docs that explain it all really well, on the cdparanoia site at xiph.org. #define X(x,y) x##y
LOL! How are they for sniffing? Sounds like you've been doing some of that...
(In case anyone hasn't figured out that he's joking, remember that you can read data from a CD with a _very_ low bit-error rate. My Samsung CDRW drive's manual says it has an error rate of "mode 1: 1 block/10^12, Mode 2: 1 block/10^9".) #define X(x,y) x##y
Well, obviously the speaker itself has to be analog. I doubt that they'd be content to accept whatever digital->analog converter the speaker manufacturer decided to use, either... (No, I don't think a DAC would colour the sound either, but it might add some slight noise if it wasn't great.) Audiophiles will probably also want to choose their own amplifier as well.
Also, you seem to have forgotten that this has already been done, with USB instead of ethernet. You can get USB speakers that (obviously) work without a sound card. #define X(x,y) x##y
gzip doesn't really help much on audio. I wouldn't bother, since I'd only use this on a local network anyway. I guess if you're encrypting, then gzip doesn't add much overhead, though.
I actually have used something like this, but with netcat execing an mp3 decoder with stdout going to a socket. The other end of the socket is on a computer too slow to decode MP3s in real time, running rawplay with stdin coming from a socket.
The slow machine (a P75) has bad memory bandwidth too, so the 1.5Mb/s data stream still slows it down noticeably. Fortunately, I'm shuffling around my computers, and I'm retiring the P75. #define X(x,y) x##y
You might want to use Vorbis. (see www.xiph.org) The encode/decode functions are in a library, so it should be easy to use. You could look at the source code for oggenc to see what library calls it makes. Check out the docs at http://www.xiph.org/ogg/vorbis/doc/vorbisenc/ (some of the links in the encoding library API ref are 404, it seems...).
Another reason for using vorbis is that it will keep Fraunhoffer from suing you for using an MP3 encoder. #define X(x,y) x##y
Your previous posts have made some good points, but you struck out on this one, AFAICT.
The ration we're talking about is mass to surface area = mass / area.
Mass is proportional to volume, which is proportional to distance cubed.
Area is proportional to distance squared.
m:s ratio \propto d^3 / d^2 = d
As d (the scale of the aircraft) increases, so does the m:s ratio. (I have omitted constants, since they don't affect how the ratio scales with d. This is like O() notation in algorithm analysis.)
> twiddled for drag coefficient and so on - if I'm not mistaken the square-cube law would only
> apply to the mass (not weight) of the helium
enclosed.
No, that's totally bogus. Earth has gravity. Therefore weight is proportional to mass. Helium is lighter than air, so there is more buoyant force than there is weight. If you want to talk about buoyancy - weight as the "effective weight", acting in the upward direction, then it increases when you scale up the craft. Making it bigger increases the volume of air displaced by helium, and also increases the weight of the material the envelope is made of. The mass of air displaced increases faster than the mass of the envelope when you make the balloon bigger, so we have the obvious conclusion that big balloons can lift more than scaled-down versions of the same design.
However, "effective weight" is not what matters when the wind is pushing on something to make it go faster. Imagine trying to push something heavy on the the space station. You don't have to do anything to keep it in the air, but it takes a while to get it going fast. For a closer analogy, think of the classic fly-pushing-cruise-ship thing. The ship will eventually go somewhere, but it will take a while to accelerate:)
Now, to consider the issue of what will happen to a big ass airship, I imagine that forces due to the wind could be proportional to the area of the ship (when the ship is anchored, for example). This would require the strength of the ship to go up as d^2, which could be hard to do. When drifting, if forces (due to shearing forces, maybe) are proportional to d, then the absolute strength of a piece of the stuff used to make it has increase as you make the ship bigger. A rope that has to hold more tension has to hold it along every point of its length. Being big wouldn't help an airship withstand large wind forces.
(I'm probably wrong about something in that last paragraph, since I've never done any fluid dynamics except for some simple stuff in first year.)
Whoever sends or points at a.doc file more than likely has the capability to create a more useful version of the document. Simply state that the document should be published in a file format that is actually standardized upon.
This is all well and good (and I used to do it myself at school, where I am more or less justified in demanding that stuff be available to Unix users). However, using linux at work is very different. If everyone you work with is using some MS crap, the only way you can reasonably expect people to let you keep using linux is if you don't cause any hassle for anybody else. This means converting the docs yourself, either using your owm copy of MS Word (which was installed on the partition of your disk that you shrank to make room for a linux or BSD system:).
I was in that position last summer, and I had to use Wine (for Lotus Notes, which is actually an interesting program), and boot into windoze every now and then to use excel. Lotus Notes mostly works with wine, and has an excel and word viewer, so that saved some rebooting. Converters like antiword are also useful. I sent stuff to other people in HTML format or just ASCII email, since the stuff I had to write was only stuff like short reports on technical stuff. I would have pulled out LaTeX and made a PDF if necessary.
#define X(x,y) x##y
I think the main advantage of Mac hardware is in the market that Motorola targetted the G4: embeded systems. A laptop is pretty close to an embeded system, or at least has similar requirements: low power and small space. A G4 uses a lot less electricity to do the same amount of number crunching as most other processors that are at least as fast, x86 or otherwise. (i.e. not counting StrongARM/XScale or stuff like that.)
Running without a fan is nice. My 486 laptop does that. When I feel like getting a new laptop, I'll probably try to get one based on the powerpc (or StrongARM, if anyone makes them!).
As for absolute performance, Mac fans should really stop deluding themselves. Everyone has their reasons for buying what they buy, so of course one is not automatically an idiot for buying a Mac that is more expensive and slower for general purpose computing than a new PC. Anyway, please peruse Silicon Insider, especially Apple's Power Failure
One thing you'll notice if you have a PPC is that compiling stuff takes longer than on x86. gcc just plain does more work when generating ppc code than x86 code. (i.e. generating ppc code takes longer than generating x86 code, using a cross compiler.). Presumably, this is because it has a lot more registers, among other differences, so there is a lot more stuff for gcc to consider while optimizing. Another thing is that, especially on my quad PPC604 mac clone from Daystar, compiling doesn't scale very well. I think the fact that the 512kB L2 cache is shared among the 4 CPUs, and has to be accessed over a relatively slow bus, really kills things. CPU-bound tasks that don't need to touch a lot of memory do a lot better. This has improved a great deal with recent Macs, especially the G4 which has great SMP support. (the G3 and G4 have per-processor L2 cache, and have the tag-check hardware on-die, even though the main SRAMs are external.)
Well, that was one of my more disorganized and wandering posts...:) #define X(x,y) x##y
woody's not released yet. I guess lyberth was only considering released distros. (I run woody on my x86 machines, and unstable on my ppc.) #define X(x,y) x##y
I like to bind the key that says "caps lock" to be another Control key. That's where control is supposed to be (it's there on VT100 keyboards, and on Sun's keyboards for their SPARCStations). I don't use caps lock, so I got rid of it from my keyboard. As an emacs user, having a conveniently placed control key is really nice. (It's nice for all the zillions of other programs that use emacs-like editting functions, too.)
BTW, the best program for messing with your X keymap is xkeycaps. It gives you a handy GUI for doing stuff so you don't have to waste your time learning xmodmap syntax until you learn it without trying from seeing it often enough. #define X(x,y) x##y
Linux strace doesn't trace fork(2)/clone(2) by default. You have to use -f for that. Read that man page, and practice by tracing a simple shell script to see what it does and what you can make of it. Use -s to show longer strings if you want. (The default is to show... after 32 character strings, except for file names, which are always shown in full.)
I recomment you strace all system calls from the installer (i.e. don't use -e), and filter later with grep | less, so you make sure you don't miss any interesting data. #define X(x,y) x##y
Re:sounds good n ecological to me
on
Hi-Tech Repo Man
·
· Score: 1
who runs the busses in your city? I've never ridden a bus that stops at every stop even if nobody is getting on or off. Maybe you don't know this because you're too special to take the bus.
If Americans (and, er, us Canadians too) weren't so pig-headed, they would figure out that instead of trying to check their email and browse the web and talk on a cell phone while attempting to control a ton of steel moving down a road fast enough to kill them if they screw up, they could do all that stuff while sitting on a bus, leaving the driving to someone who isn't distracted.
Public transportation has great potential, especially if somebody gets smart and hooks up a bus with ethernet and phone jacks so people can make good use of their time during their commute. #define X(x,y) x##y
Thank you! I totally agree that Americans (and Canadians) are way too in love with their cars, and this results in a lot of stuff being fucked up.
Think about this: Lots of companies are spending time developing stuff that lets you browse the web, check your email, or do whatever you want, while you're driving. This is ridiculous. To drive safely, you need to pay attention to the road. Lots of people try to do other stuff during their commute, because it's so long that it would be a huge waste of time otherwise. If they weren't so stubborn about using cars, they could just take public transportation, and do all that stuff they want to get done, and leave the driving to someone who will do a good job. Obviously, a bus crowded to standing-room-only capacity is not great for getting work done with your laptop, but if people designed public transportation so that useful work could be done while travelling, things would be a whole lot better.
I'm a big fan of bicycles myself, BTW. You can't exactly do a lot of work while on a bike, but it's fun and it keeps me in shape:) #define X(x,y) x##y
Slashdot Mirror
The ppc backend is more computationally expensive than the x86 backend, in my experience. gcc runs fast on athlons because the x86 codegen doesn't take as much time as ppc codegen. (I'm 90% sure about this, since I didn't check it carefully or anything, and it was a while ago. I used gcc 2.94 when I tried a bit of stuff.)
#define X(x,y) x##y
So good an idea in fact that it's been had before :)
I used to be subscribed to the list, which was win32@lists.debian.org, but I can't find an archive of it, and i don't know if it still exists.
If you search around, you may be able to find some posts describing an package repository of .deb packages built for cygwin or mingw.
#define X(x,y) x##y
That's "the true north strong and free" to you, eh. Learn the frickin national anthem before you bother with the map. (j/k:)
#define X(x,y) x##y
> So get UltraSparc's score, double it, and you
> get
And you double it how? I'd be impressed if you could show me where you could buy an UltraSPARC that ran at 1.8GHz. I'm assuming the fastest Ultra you can buy is the 900MHz one, and the P4 kicks its butt. The key is that the P4 does run that fast, but nobody knows how to make an UltraSPARC run that fast, even if they wanted to.
Also note that you'd need to double the speed of the memory bus, and halve all latencies in everything, especially the RAM, to get double the SPECcpu score.
Sun's *SPARC cpus are way behind everyone else, but they keep selling them because they put them in very reliable machines and sell good support. Their good name helps, too. High processing speed is not their forte.
Read Silicon Insider: http://www.realworldtech.com/page.cfm?Section=Col
#define X(x,y) x##y
The major problem would be IO bandwidth, so you'd have to use something other than multiple cards in a 32bit 33MHz PCI bus. That has a theoretical limit of 132 MB/s, which is too low a bisection bandwidth for any respectable gigE switch. NVidia's nforce chipset has two PCI busses, just like my old Mac clone from Daystar, so that would help some. Still, with a server mobo and some 64bit 66MHz PCI busses, you'd have 528MB/s per bus. That's decent, but still nowhere near what you want for 24 GE ports. You'd also need to buy several 4-port NICs, and find slots to put them in. Switches really are hard to do with software on commodity machines. Routing is another matter, unless you have a lot of subnets and need a lot of bisection bandwidth, I guess. (If you were going to be limited by the fact that all the traffic was going up or down through one of the interfaces, and not between two internal ones, then a computer with NICs could probably do the job reasonably well.
#define X(x,y) x##y
Legal measures, like requiring the opt-out to actually work, might get somewhere. It would be good if spammers had to be cautious to avoid fines because they sent you spam after you opted out. (and yes, I think just a fine, not the death penalty or anything would be appropriate, since honest mistakes could happen. If the penalty is too harsh, spammers might not get fined if they can show that there was some kind of honest mistake. If there's a 100 $ fine that they have to pay, no court wrangling or bullshit like that, then they'll behave.)
#define X(x,y) x##y
That wasn't a troll (err, not necessarily), since it was a quote from Mad TV. Man, I haven't watched that show for years. (not that I've missed it...)
#define X(x,y) x##y
> Nothing to do with it. Everything after stage1 is a consistency check.
Not quite. Assuming GCC generates faster/better code than the native compiler, installing GCC compiled by GCC will make compiling other program happen faster.
Thanks for pointing out the consistency check going on with stage 3. I hadn't known why it was useful to build gcc again with gcc.
#define X(x,y) x##y
good point about not letting Weird Shit take too much time away from the good stuff. Keeping the computer in a family room should be enough to stop kids from actively seeking porn or blood & gore, or any stuff they think will make them look bad in their parents eyes. It would be embarrasing to look at most Bad Stuff in front of your parents, or even for them to know that you had been looking at it. Using censorware to block stuff is not necessary, since I think fear of embarrasment will do the trick.
#define X(x,y) x##y
logging URLs with stuff after '?' stripped off (to clean off GET form submission data, to avoid being to invasive) is way different from logging chat sessions. I don't know how a 10 year old would feel about having their every move online logged, but I know I don't like the idea. Parents don't need to read everything their kids type.
#define X(x,y) x##y
I think most of the stuff that's not appropriate for kids is not very entertaining for them either, and they probably won't find nasty stuff unless they are trying to.
:). (e.g. There's a big difference between two people having sex and two people sleeping in the same bed because there was a limited number of beds. However, saying "they slept together" means they had sex, so you have to waste a lot of words explaining that they slept in the same bed without having sex. English is dumb because people in the past (and wimps in the present) like to pretend they aren't talking about things even when they are.) Anyway, now you might have some idea why I think it's more or less ok for kids to look at web sites once they've reached a certain level of maturity,
I'm having a hard time thinking of things that a kid could see on the 'net that you couldn't talk with them about.
I'll admit that I haven't had to be in the position of supervising kids browsing the 'net, so I'm probably thinking in too abstract terms. I've taught Sunday school for a few years now, so I have some experience with kids around age 10, though. Maybe I'm overly optimistic in my idea that kids can understand the fact that some of the stuff on the 'net is _not_ good wisdom or words to live by (e.g. racist material). I do think that knowing something about what's out there won't hurt anyone over age 10. I would guess that the Internet helps introduce difficult topics for conversation.
One thing I just thought of is that this only works if you have quite tolerant views, like I do, and don't think it's a problem for kids to know what sex is about, even before they're old enough to want to do it. The English language is so full of phrases with double entendres because it has a long history of being spoken by prudes and other wimpy people who would die if they said penis (insert Monty Python song here
and can take it. Obviously a 5 year old isn't going to understand some of the stuff you'd need to explain after they tripped across certain web sites, so actually being there _while_ they browse is important, because they will be highly unlikely to read some stuff with you right there. This is the same reason teenagers hide their porn from their parents. (OTOH, 5 year olds probably don't read well enough to get too far on the web, but a 7 or 8 year old who reads well could give parents something to worry about.)
#define X(x,y) x##y
> Passing through each one is a super-cooled liquid nitrogen that pushed
> the temperature of the tube down towards 0 degrees kelvin.
But not very close to 0 there, sissy boy. Real men use liquid helium, which doesn't even liquify until a couple degrees K. There's a good reason why it costs a lot more. Cooling LN2 down to those temperatures could be done, but it just isn't the same. It's texture is all wrong. Just last week, I dropped the temp. a few more degrees, down to where helium becomes superfluid. Now we're talking smooth sound... (err, I mean "I definitely sounded better; It was worth it". Almost missed my audiophile-speak...)
I also use lead cables, which become superconducting at the temperatures I'm talking about. Good old fashioned type I superconduction, because the newfangled high temp. type II superconductors just don't sound as vibrant.
(In real life, silver, gold, and copper don't superconduct, even though they are the best conductors at room temp (in that order, BTW).)
#define X(x,y) x##y
You don't need accurate clocks in the mechanism that spins the CD. All you need is to read it into a memory buffer, and feed the data from the buffer to a DAC at a fixed rate. Of course, that _does_ have to happen at a controlled rate.
CD audio is random access (AFAIK), but due to the brilliant design, there are no position markers, so the drive can't tell where it is when it tries to seek somewhere. It can only get to approximately where you asked it to. The reason like CD paranoia (see xiph.org) exists is to match up the data from the end of one read to the start of the next. I seem to remember something else about anti-jitter, too, but I can't remember why. I actually don't remember a whole bunch about this right now, so anyone who cares should go and read the docs that explain it all really well, on the cdparanoia site at xiph.org.
#define X(x,y) x##y
LOL! How are they for sniffing? Sounds like you've been doing some of that...
(In case anyone hasn't figured out that he's joking, remember that you can read data from a CD with a _very_ low bit-error rate. My Samsung CDRW drive's manual says it has an error rate of "mode 1: 1 block/10^12, Mode 2: 1 block/10^9".)
#define X(x,y) x##y
Well, obviously the speaker itself has to be analog. I doubt that they'd be content to accept whatever digital->analog converter the speaker manufacturer decided to use, either... (No, I don't think a DAC would colour the sound either, but it might add some slight noise if it wasn't great.) Audiophiles will probably also want to choose their own amplifier as well.
Also, you seem to have forgotten that this has already been done, with USB instead of ethernet. You can get USB speakers that (obviously) work without a sound card.
#define X(x,y) x##y
gzip doesn't really help much on audio. I wouldn't bother, since I'd only use this on a local network anyway. I guess if you're encrypting, then gzip doesn't add much overhead, though.
I actually have used something like this, but with netcat execing an mp3 decoder with stdout going to a socket. The other end of the socket is on a computer too slow to decode MP3s in real time, running rawplay with stdin coming from a socket.
The slow machine (a P75) has bad memory bandwidth too, so the 1.5Mb/s data stream still slows it down noticeably. Fortunately, I'm shuffling around my computers, and I'm retiring the P75.
#define X(x,y) x##y
You might want to use Vorbis. (see www.xiph.org) The encode/decode functions are in a library, so it should be easy to use. You could look at the source code for oggenc to see what library calls it makes. Check out the docs at http://www.xiph.org/ogg/vorbis/doc/vorbisenc/ (some of the links in the encoding library API ref are 404, it seems...).
Another reason for using vorbis is that it will keep Fraunhoffer from suing you for using an MP3 encoder.
#define X(x,y) x##y
Your previous posts have made some good points, but you struck out on this one, AFAICT.
:)
The ration we're talking about is mass to surface area = mass / area.
Mass is proportional to volume, which is proportional to distance cubed.
Area is proportional to distance squared.
m:s ratio \propto d^3 / d^2 = d
As d (the scale of the aircraft) increases, so does the m:s ratio. (I have omitted constants, since they don't affect how the ratio scales with d. This is like O() notation in algorithm analysis.)
> twiddled for drag coefficient and so on - if I'm not mistaken the square-cube law would only
> apply to the mass (not weight) of the helium
enclosed.
No, that's totally bogus. Earth has gravity. Therefore weight is proportional to mass. Helium is lighter than air, so there is more buoyant force than there is weight. If you want to talk about buoyancy - weight as the "effective weight", acting in the upward direction, then it increases when you scale up the craft. Making it bigger increases the volume of air displaced by helium, and also increases the weight of the material the envelope is made of. The mass of air displaced increases faster than the mass of the envelope when you make the balloon bigger, so we have the obvious conclusion that big balloons can lift more than scaled-down versions of the same design.
However, "effective weight" is not what matters when the wind is pushing on something to make it go faster. Imagine trying to push something heavy on the the space station. You don't have to do anything to keep it in the air, but it takes a while to get it going fast. For a closer analogy, think of the classic fly-pushing-cruise-ship thing. The ship will eventually go somewhere, but it will take a while to accelerate
Now, to consider the issue of what will happen to a big ass airship, I imagine that forces due to the wind could be proportional to the area of the ship (when the ship is anchored, for example). This would require the strength of the ship to go up as d^2, which could be hard to do. When drifting, if forces (due to shearing forces, maybe) are proportional to d, then the absolute strength of a piece of the stuff used to make it has increase as you make the ship bigger. A rope that has to hold more tension has to hold it along every point of its length. Being big wouldn't help an airship withstand large wind forces.
(I'm probably wrong about something in that last paragraph, since I've never done any fluid dynamics except for some simple stuff in first year.)
#define X(x,y) x##y
This is all well and good (and I used to do it myself at school, where I am more or less justified in demanding that stuff be available to Unix users). However, using linux at work is very different. If everyone you work with is using some MS crap, the only way you can reasonably expect people to let you keep using linux is if you don't cause any hassle for anybody else. This means converting the docs yourself, either using your owm copy of MS Word (which was installed on the partition of your disk that you shrank to make room for a linux or BSD system :).
I was in that position last summer, and I had to use Wine (for Lotus Notes, which is actually an interesting program), and boot into windoze every now and then to use excel. Lotus Notes mostly works with wine, and has an excel and word viewer, so that saved some rebooting. Converters like antiword are also useful. I sent stuff to other people in HTML format or just ASCII email, since the stuff I had to write was only stuff like short reports on technical stuff. I would have pulled out LaTeX and made a PDF if necessary.
#define X(x,y) x##y
As for absolute performance, Mac fans should really stop deluding themselves. Everyone has their reasons for buying what they buy, so of course one is not automatically an idiot for buying a Mac that is more expensive and slower for general purpose computing than a new PC. Anyway, please peruse Silicon Insider, especially Apple's Power Failure
One thing you'll notice if you have a PPC is that compiling stuff takes longer than on x86. gcc just plain does more work when generating ppc code than x86 code. (i.e. generating ppc code takes longer than generating x86 code, using a cross compiler.). Presumably, this is because it has a lot more registers, among other differences, so there is a lot more stuff for gcc to consider while optimizing. Another thing is that, especially on my quad PPC604 mac clone from Daystar, compiling doesn't scale very well. I think the fact that the 512kB L2 cache is shared among the 4 CPUs, and has to be accessed over a relatively slow bus, really kills things. CPU-bound tasks that don't need to touch a lot of memory do a lot better. This has improved a great deal with recent Macs, especially the G4 which has great SMP support. (the G3 and G4 have per-processor L2 cache, and have the tag-check hardware on-die, even though the main SRAMs are external.)
Well, that was one of my more disorganized and wandering posts... :)
#define X(x,y) x##y
woody's not released yet. I guess lyberth was only considering released distros. (I run woody on my x86 machines, and unstable on my ppc.)
#define X(x,y) x##y
I like to bind the key that says "caps lock" to be another Control key. That's where control is supposed to be (it's there on VT100 keyboards, and on Sun's keyboards for their SPARCStations). I don't use caps lock, so I got rid of it from my keyboard. As an emacs user, having a conveniently placed control key is really nice. (It's nice for all the zillions of other programs that use emacs-like editting functions, too.)
BTW, the best program for messing with your X keymap is xkeycaps. It gives you a handy GUI for doing stuff so you don't have to waste your time learning xmodmap syntax until you learn it without trying from seeing it often enough.
#define X(x,y) x##y
Linux strace doesn't trace fork(2)/clone(2) by default. You have to use -f for that. Read that man page, and practice by tracing a simple shell script to see what it does and what you can make of it. Use -s to show longer strings if you want. (The default is to show ... after 32 character strings, except for file names, which are always shown in full.)
I recomment you strace all system calls from the installer (i.e. don't use -e), and filter later with grep | less, so you make sure you don't miss any interesting data.
#define X(x,y) x##y
who runs the busses in your city? I've never ridden a bus that stops at every stop even if nobody is getting on or off. Maybe you don't know this because you're too special to take the bus.
If Americans (and, er, us Canadians too) weren't so pig-headed, they would figure out that instead of trying to check their email and browse the web and talk on a cell phone while attempting to control a ton of steel moving down a road fast enough to kill them if they screw up, they could do all that stuff while sitting on a bus, leaving the driving to someone who isn't distracted.
Public transportation has great potential, especially if somebody gets smart and hooks up a bus with ethernet and phone jacks so people can make good use of their time during their commute.
#define X(x,y) x##y
Thank you! I totally agree that Americans (and Canadians) are way too in love with their cars, and this results in a lot of stuff being fucked up.
:)
Think about this: Lots of companies are spending time developing stuff that lets you browse the web, check your email, or do whatever you want, while you're driving. This is ridiculous. To drive safely, you need to pay attention to the road. Lots of people try to do other stuff during their commute, because it's so long that it would be a huge waste of time otherwise. If they weren't so stubborn about using cars, they could just take public transportation, and do all that stuff they want to get done, and leave the driving to someone who will do a good job. Obviously, a bus crowded to standing-room-only capacity is not great for getting work done with your laptop, but if people designed public transportation so that useful work could be done while travelling, things would be a whole lot better.
I'm a big fan of bicycles myself, BTW. You can't exactly do a lot of work while on a bike, but it's fun and it keeps me in shape
#define X(x,y) x##y