Hear, hear! The LEAF project has made things so much easier since the dark old LRP days. Back then, I posted a huge piece of documentation, just to help people set up a NAT firewall. Now, everyone has made it so incredibly automatic that it's nearly automatic.
Those of you wanting to check it out may wonder which flavor to start with. I suggest Bering (though I use Bering-ucLibC).
So you're either revved up about building one of these for yourself or you're just reading on because you've nothing better to do. I can understand that.
It's not that I have nothing better to do. It's that I do have something better to do. Two kids, two jobs, a bunch of homework to grade, and a bicycle that isn't fixing itself.
This allows me to fantasize about having the time to make one of those beauties!
Matlab is a Black Box mostly because it has huge numbers of built in functions. In order to make sure that all of these functions work properly (you would be amazed by how many commercial functions don't actually do what is advertised) by building our own and then compairing the two
Indeed. Also, they charge thousands of dollars for add-on packages that are mostly, or sometimes entirely, comprised of trivial functions. I think the basic tool is awesome for undirected number crunching, but I am less impressed with the toolboxes.
And once you've figured out exactly how you want to crunch those numbers, naturally it is time to go to a compiled language. Mathworks tries to make that easy, but frankly I have always just started from scratch.
(1) Mathematica is not so hot for processing data sets. Since it works so symbolically, you can try some transformation and end up waiting forever for an endless symbolic expression to appear on the screen, all because of a typo. This can be mitigated using $PrePrint=Short[#,7]&, but that has its own problems.
(2) Anything using Numerical Recipes is immediately suspect (not to mention the license issues). There used to be a document at JPL about this. For a critique, see this compilation.
NAG generally rocks, but I have to say I've been disappointed of late with their nonlinear optimizers. Case in point: the SQP optimizer allows minor iterations to wander outside the parameter space boundaries. OK, fine, you say. I'll just return NaN for some objective function components. Whoops! That makes your program crash. Not just return an error code, but crash, because they try taking a square root.
If they're going to wander like that, they should check return values, or at least the sum of components, for NaN.
Another case in point -- just last week we fed the same 5 dimensional nonlinear problem to both the MS Excel optimizer (MS calls it the "Solver" and it has historically been famously bad bad bad), and to every one of the NAG optimizers, even slow ones like simplex. Result: Excel beat them all. I was shocked.
In NAG's favor, we have another optimization problem we ask it to do in about 400 dimensions. Granted, that objective function is closer to quadratic, but even so I was impressed with NAG's performance -- always finding the unique minimum, even when it had to run for days.
Oh, and BTW, if anyone ever tells you computers are "fast enough" now, they're not.
Well, I thought that because I haven't used it for a while, having switched to LEAF. And I never even heard of MicroBSD. So when I parsed BSD, I immediately thought it was my former darling.
I thought others may have had the same problem. It's not exactly like there's a lot of discussion of either Pico- or Micro-, you know.
I know this is somewhat OT, but how goes current integration of Palm and GPS? That is, I know you can buy an amazing little GPS tab for a Palm that plugs into the CF slot, and that you can buy some map software that looks to have better European street info than Garmin's own maps.
But do they work well together? If I use the CF slot for the GPS can I no longer add memory those maps will want so much?
Should I just hope to get the Garmin Palm device (which is not out yet) before I travel this spring, and hope it lets me use third party maps?
And what about those GPS tabs? It looks like at least some of them need to be factory serviced in order to switch between WAAS and Europe's experimental equivalent of EGNOS.
Around the office we would use this feature to vamp off whoever happened to have the fastest machine at the time. Woe betide the lucky bastard with the 200MHz PPro! He would have 5 copies of EOModeler.app running.
Whew -- for a minute I thought this was an article referring to a bizarre fallout with PicoBSD. PicoBSD is a neat little FreeBSD-on-a-single-floppy distro. Kind of an equivalent to Linux's admirable Leaf Project.
I know this is somewhat OT, but how goes current integration of Palm and GPS? That is, I know you can buy an amazing little GPS tab for a Palm that plugs into the CF slot, and that you can buy some map software that looks to have good European street info.
But do they work well together? If I use the CF slot for the GPS can I no longer add memory those maps will want so much?
Should I just hope to get the Garmin Palm device (which is not out yet) before I travel this spring, and hope it lets me use third party maps?
And what about those GPS tabs? It looks like at least some of them need to be factory serviced in order to switch between WAAS and Europe's experimental equivalent of EGNOS.
When the Challenger exploded, the only commission member to actually give a damn about figuring out what happened was the late physicist Richard Feynman. He was hardly a space shuttle expert, but he had the right kind of skeptical mind.
Without his intrepid investigations, we probably still wouldn't know what happened (though some NASA engineers might). His investigation was thorough enough to find myriad safe (software) and unsafe (mission cancellation policies) aspects of the shuttle program.
Thanks for the offer...I am curious! (Probably won't make it to lovely Seattle again for some time though)
I find your experience believable, if not entirely germane. I've gone hundreds of miles on rollerblades with no problem either, but that doesn't mean they stop as well as a bike, let alone a car.
Essentially, if your upcoming halt in a Segway is not a sudden surprise, it makes sense it would stop very well...the mechanism will be able to read your body's cues, speed up the wheels to get them in front of your center of mass, and thereby eliminate the torque during your stop.
However, if you run the thing into, say, an unseen curb, I think you will endo. Or whatever you call it on a Segway. I've done just that twice on a bicycle with one endo (as a kid) and one non-endo (as an adult). So statistically that's what?....maybe a 50/50 endo probability with a 5% confidence level;-)
Rollerblades, BTW, are weird to hit unseen obstacles in. The tiny wheels mean of course that they stop short on relatively small irregularities (let alone curbs). On the other hand, having two legs means that hitting something usually results in nothing more than an awkward stagger as you quickly lift the other leg and thrust it in front of you. I haven't fallen while skating forward in years (don't ask about backward, ow!).
Of course, none of this debate has much to do with whether they belong on sidewalks. Personally, I think they should be treated like bikes -- legally expected to ride the streets, but unoficially ignored if they take sidewalks at reasonably low speeds. (I always ride my bike on the street).
OK, this is apropos of none of the previous posts (at least at my typical cruising altitude of +3), but here I go anyway...
Have you ever noticed how Segway boosters cite its stability and concomitant safety? Now, I'm willing to believe it's pretty stable, but it's time to point out the fallacy in assuming that implies safety, either for the rider or other users of public thoroughfares.
Unlike the skateboards in Snow Crash (which show N. Stephenson understood this issue far better than these fools) the Segway has NO RADAR. It can't tell if the rider is about to collide with something, like a person, pothole, or curb.
Why is that important? Well, for the other people on the pavement, it's mostly an issue of getting run into. For a Segway rider, though, I have yet to see anyone else point out the pothole and curb issue. Look at what you've got:
1) Much smaller wheels than a bicycle
2) A center of mass about as high, and not nearly as far behind the point of ground contact as a bicycle has.
Together, these mean that
A) Potholes, bump and curbs will stop a Segway much more forcefully than a bicycle
B) You can't stop a Segway nearly as quickly when you find yourself approaching one.
It's basic physics:
I) The smaller wheels, on encountering an object, have a steeper effective inclined plane
II) The torque about the point of ground contact when stopping is greater in the Segway, since the angle of force is almost perpindicular to it.
Given that stopping fast is already a problem with bicycles (here's how to stop a bicycle as fast as possible) since they can endo instead of skidding or ABSing like cars, it's going to be even worse for a Segway. I estimate that a Segway doesn't stop any better than rollerblades, and that's not too good.
I also think it likely a Segway can't turn away from trouble too easily either -- at least if it needs to keep both wheels on the ground.
They are amazing fridge magnets. And the platters, BTW, make superb coasters for your coffee table.
I am a little worried that there may be something toxic on the platters...it's not like they intend for disk platters to be food service approved.
The only data point I have is that there are no particular EPA requirements for disposal of the platters. So they probably don't contain anything too dangerous.
These countless manufacturers making home audio MP3 devices with integrated hard drives are missing the point: you want a single point of storage in your house for MP3's, or you are going to have chaos.
Networks are not just for web-browsing. They're also for sharing files, like those MP3's. And if you're going to have them somewhere, you might as well have them on your PC where: (1) You can rip with that expensive Athlon or PPC processor using LAME (2) There exist myriad MP3 control tools like id3tool and mp3gain (3) Your portable MP3 player hooks up (4) You may be downloading files from mp3.com or other places (5) [Important] You have a chance of keeping the hard drive and fan whines away from your audio equipment
Thus, you want the home audio component you buy to simply get its music off the network. Audiotron and SLiMP3 do just that, which makes them the only reasonable such devices.
Between them, I prefer the Audiotron because it has digital output, which means I can use the expensive DAC in my receiver rather than the cheap one in the unit. The SLiMP3 is impressively flexible, though!
Looking at the public comments Jay links to, it appears this campaign has been successful in inducing many people to write in favor of changing those terms in the standards.
It seems to me, however, that Bruce Perens, in the many posts you see above, is right. That is, these standards are as good as it gets.
So -- write saying you support them, to counterbalance all the previous trigger-happy or (IMO) zealous individuals.
Slashdot Mirror
Hear, hear! The LEAF project has made things so much easier since the dark old LRP days. Back then, I posted a huge piece of documentation, just to help people set up a NAT firewall. Now, everyone has made it so incredibly automatic that it's nearly automatic.
Those of you wanting to check it out may wonder which flavor to start with. I suggest Bering (though I use Bering-ucLibC).
From the TapLight website:
So you're either revved up about building one of these for yourself or you're just reading on because you've nothing better to do. I can understand that.
It's not that I have nothing better to do. It's that I do have something better to do. Two kids, two jobs, a bunch of homework to grade, and a bicycle that isn't fixing itself.
This allows me to fantasize about having the time to make one of those beauties!
I meant, of course, that this ain't the racism.
>>So "french" is now a race?
> Yes.
Uh, better work on your English, lga.
> The people of France are a race.
I wonder if the millions of "Berbers" living in ghettos outside the French cities would agree with that.
America has racism, and nationalism. But this ain't it.
We do tax it, though.
Iraqi army -- CHARGE!
(ducks)
Matlab is a Black Box mostly because it has huge numbers of built in functions. In order to make sure that all of these functions work properly (you would be amazed by how many commercial functions don't actually do what is advertised) by building our own and then compairing the two
Indeed. Also, they charge thousands of dollars for add-on packages that are mostly, or sometimes entirely, comprised of trivial functions. I think the basic tool is awesome for undirected number crunching, but I am less impressed with the toolboxes.
And once you've figured out exactly how you want to crunch those numbers, naturally it is time to go to a compiled language. Mathworks tries to make that easy, but frankly I have always just started from scratch.
Although Octave is great to mimic Matlab 5.2, some neat features in Matlab 6.0 are not there yet.
Three-dimensional arrays come to mind. Grrr.
Two notes:
(1) Mathematica is not so hot for processing data sets. Since it works so symbolically, you can try some transformation and end up waiting forever for an endless symbolic expression to appear on the screen, all because of a typo. This can be mitigated using $PrePrint=Short[#,7]&, but that has its own problems.
(2) Anything using Numerical Recipes is immediately suspect (not to mention the license issues). There used to be a document at JPL about this. For a critique, see this compilation.
Mathematics is not a plural. Thus, you don't need the "s".
NAG generally rocks, but I have to say I've been disappointed of late with their nonlinear optimizers. Case in point: the SQP optimizer allows minor iterations to wander outside the parameter space boundaries. OK, fine, you say. I'll just return NaN for some objective function components. Whoops! That makes your program crash. Not just return an error code, but crash, because they try taking a square root.
If they're going to wander like that, they should check return values, or at least the sum of components, for NaN.
Another case in point -- just last week we fed the same 5 dimensional nonlinear problem to both the MS Excel optimizer (MS calls it the "Solver" and it has historically been famously bad bad bad), and to every one of the NAG optimizers, even slow ones like simplex. Result: Excel beat them all. I was shocked.
In NAG's favor, we have another optimization problem we ask it to do in about 400 dimensions. Granted, that objective function is closer to quadratic, but even so I was impressed with NAG's performance -- always finding the unique minimum, even when it had to run for days.
Oh, and BTW, if anyone ever tells you computers are "fast enough" now, they're not.
Well, I thought that because I haven't used it for a while, having switched to LEAF. And I never even heard of MicroBSD. So when I parsed BSD, I immediately thought it was my former darling.
I thought others may have had the same problem. It's not exactly like there's a lot of discussion of either Pico- or Micro-, you know.
Satisfied?
I know this is somewhat OT, but how goes current integration of Palm and GPS? That is, I know you can buy an amazing little GPS tab for a Palm that plugs into the CF slot, and that you can buy some map software that looks to have better European street info than Garmin's own maps.
.
But do they work well together? If I use the CF slot for the GPS can I no longer add memory those maps will want so much?
Should I just hope to get the Garmin Palm device (which is not out yet) before I travel this spring, and hope it lets me use third party maps?
And what about those GPS tabs? It looks like at least some of them need to be factory serviced in order to switch between WAAS and Europe's experimental equivalent of EGNOS
Help a fellow electronics fetishist!
A shame? Well, yes and no.
Around the office we would use this feature to vamp off whoever happened to have the fastest machine at the time. Woe betide the lucky bastard with the 200MHz PPro! He would have 5 copies of EOModeler.app running.
Whew -- for a minute I thought this was an article referring to a bizarre fallout with PicoBSD. PicoBSD is a neat little FreeBSD-on-a-single-floppy distro. Kind of an equivalent to Linux's admirable Leaf Project.
I know this is somewhat OT, but how goes current integration of Palm and GPS? That is, I know you can buy an amazing little GPS tab for a Palm that plugs into the CF slot, and that you can buy some map software that looks to have good European street info.
But do they work well together? If I use the CF slot for the GPS can I no longer add memory those maps will want so much?
Should I just hope to get the Garmin Palm device (which is not out yet) before I travel this spring, and hope it lets me use third party maps?
And what about those GPS tabs? It looks like at least some of them need to be factory serviced in order to switch between WAAS and Europe's experimental equivalent of EGNOS.
Help a fellow electronics fetishist!
Oh heck, no. It will sound every iota as good as it would if it has gone through a month of production.
Will the recording process suffer due to the hurry?
What kind of silly question is that?
Will the recording process suffer due to the hurry?
No, the recording process will use the same microphones and digital tapes it always does, I would think. The sound engineering may suffer however.
Will the recording process suffer due to the hurry?
A process is not sentient. Only sentient beings can experience suffering.
Without his intrepid investigations, we probably still wouldn't know what happened (though some NASA engineers might). His investigation was thorough enough to find myriad safe (software) and unsafe (mission cancellation policies) aspects of the shuttle program.
Who will be our Feynman now?
I agree that the sentence frament you quote is untrue. But for this particular application, one might legitimately ask:
How do you plan to back up that one-time pad?
Thanks for the offer...I am curious! (Probably won't make it to lovely Seattle again for some time though)
;-)
I find your experience believable, if not entirely germane. I've gone hundreds of miles on rollerblades with no problem either, but that doesn't mean they stop as well as a bike, let alone a car.
Essentially, if your upcoming halt in a Segway is not a sudden surprise, it makes sense it would stop very well...the mechanism will be able to read your body's cues, speed up the wheels to get them in front of your center of mass, and thereby eliminate the torque during your stop.
However, if you run the thing into, say, an unseen curb, I think you will endo. Or whatever you call it on a Segway. I've done just that twice on a bicycle with one endo (as a kid) and one non-endo (as an adult). So statistically that's what?....maybe a 50/50 endo probability with a 5% confidence level
Rollerblades, BTW, are weird to hit unseen obstacles in. The tiny wheels mean of course that they stop short on relatively small irregularities (let alone curbs). On the other hand, having two legs means that hitting something usually results in nothing more than an awkward stagger as you quickly lift the other leg and thrust it in front of you. I haven't fallen while skating forward in years (don't ask about backward, ow!).
Of course, none of this debate has much to do with whether they belong on sidewalks. Personally, I think they should be treated like bikes -- legally expected to ride the streets, but unoficially ignored if they take sidewalks at reasonably low speeds. (I always ride my bike on the street).
Have you ever noticed how Segway boosters cite its stability and concomitant safety? Now, I'm willing to believe it's pretty stable, but it's time to point out the fallacy in assuming that implies safety, either for the rider or other users of public thoroughfares.
Unlike the skateboards in Snow Crash (which show N. Stephenson understood this issue far better than these fools) the Segway has NO RADAR. It can't tell if the rider is about to collide with something, like a person, pothole, or curb.
Why is that important? Well, for the other people on the pavement, it's mostly an issue of getting run into. For a Segway rider, though, I have yet to see anyone else point out the pothole and curb issue. Look at what you've got:
1) Much smaller wheels than a bicycle
2) A center of mass about as high, and not nearly as far behind the point of ground contact as a bicycle has.
Together, these mean that
A) Potholes, bump and curbs will stop a Segway much more forcefully than a bicycle
B) You can't stop a Segway nearly as quickly when you find yourself approaching one.
It's basic physics:
I) The smaller wheels, on encountering an object, have a steeper effective inclined plane
II) The torque about the point of ground contact when stopping is greater in the Segway, since the angle of force is almost perpindicular to it.
Given that stopping fast is already a problem with bicycles (here's how to stop a bicycle as fast as possible) since they can endo instead of skidding or ABSing like cars, it's going to be even worse for a Segway. I estimate that a Segway doesn't stop any better than rollerblades, and that's not too good.
I also think it likely a Segway can't turn away from trouble too easily either -- at least if it needs to keep both wheels on the ground.
They are amazing fridge magnets. And the platters, BTW, make superb coasters for your coffee table.
I am a little worried that there may be something toxic on the platters...it's not like they intend for disk platters to be food service approved.
The only data point I have is that there are no particular EPA requirements for disposal of the platters. So they probably don't contain anything too dangerous.
These countless manufacturers making home audio MP3 devices with integrated hard drives are missing the point: you want a single point of storage in your house for MP3's, or you are going to have chaos.
Networks are not just for web-browsing. They're also for sharing files, like those MP3's. And if you're going to have them somewhere, you might as well have them on your PC where:
(1) You can rip with that expensive Athlon or PPC processor using LAME
(2) There exist myriad MP3 control tools like id3tool and mp3gain
(3) Your portable MP3 player hooks up
(4) You may be downloading files from mp3.com or other places
(5) [Important] You have a chance of keeping the hard drive and fan whines away from your audio equipment
Thus, you want the home audio component you buy to simply get its music off the network. Audiotron and SLiMP3 do just that, which makes them the only reasonable such devices.
Between them, I prefer the Audiotron because it has digital output, which means I can use the expensive DAC in my receiver rather than the cheap one in the unit. The SLiMP3 is impressively flexible, though!
Even more fun is planting one on somebody without them knowing it!
Go for a buttoned rear pants pocket, or equivalent location they'll never think to check.
Looking at the public comments Jay links to, it appears this campaign has been successful in inducing many people to write in favor of changing those terms in the standards.
It seems to me, however, that Bruce Perens, in the many posts you see above, is right. That is, these standards are as good as it gets.
So -- write saying you support them, to counterbalance all the previous trigger-happy or (IMO) zealous individuals.