Depends on your development environment - change a fundamental header file in C, C++, even Java, and you can waste several builds watching the dependencies come out of the woodwork... clean rebuilds are the best way to handle large change packages during development, IMHO.
And missing an error in the plane is a tad bit more important than in a program.
WRONG!
You have no way of knowing what other people are programming, but buddy, if you miss some types of errors in some types of programs, Bad Things Happen. Many of the moving things in the world are fly-(or drive-) by-wire - what do you think is figuring out "for this input, do that output"? A program.
Quick example - the first Ariane 5 blew up due to untested software and hardware reuse - they took a piece of equipment that worked fine on the Ariane 4 series, stuck it on the 5, and didn't test it throught the full flight envelope - but a software design decision that worked on the A4 (not to bounds-check certain values) was invalid on the A5 (you COULD get that many bits of the values on that rocket) and when it happened, the software said "It's gotta be bad hardware! Turn it off!" and swapped to the redundant module - which was busy locking up with the same error. Boom.
Now look at the millions of lines of software in the Boeing 777. The microcontroller in your grandmother's pacemaker. The computer-controlled milling machine that shapes the turbine blades of jet engines.
And it's not like if the copilot could go fly -another- plane instead of co-piloting that the airline would get any benefit. They don't get more money for being able to fly more planes.
No, he'd get laid off. That's economics. Airlines don't want spare bodies getting paid salaries - if they could train chimps to fly, and the FAA would approve, they'd do that.
Lines of code may be a horrible way to measure productivity, but it's very hard to get accountants to believe in "function points".
I agree that XP isn't as efficient as "normal" programming when both programmers are experienced - what I've seen work in practice is the use of XP (adapted slightly) as a training ground, where inexperienced personnel are combined with experienced personnel to cross-train. It works both ways - you can have a wily old Unix coder sentenced to work on MFC sitting next to a kid who's only done Visual Basic, and both will learn a lot, and be more productive as a team than either or both separately. It's a matter of chemistry as much as management.
But then, that's just an opinion. It matters little.
At one job, we'd spend six hours working the bug fixes due for first shift (there weren't enough servers to go around, so we subs got to do shift work, too - I was on second - 12 hour days, 10 on the clock), fired up the Compile-From-Hell (Ada - sssllllooooowwww compiler, lots of code), and checked out, headed for the casinos (Mississippi Gulf Coast, the first had just gotten installed). Came back a couple of hours later (or on a page from the Puritan who volunteered to watch the builds scroll past) and started testing to finish out our shift.
Later on that same job, they sent us to the lead corp's site to do final integration (since their on-location management had screwed up, we all got drawn back to the mothership). We had to do the same stuff, but with more code - and even back at Home Base, they only gave us 2 servers for our builds, so they took hours. One night approaching midnight, the VP in charge of the project walks through - only 2 of us are there, both subs, and we're both playing Minesweeper! He asks what we're up to, looking pointedly at the game screens, and I pop up the build windows behind it (X on Windows) and show him the lines scrolling past, 1 every 45 seconds, and guestimate that we have another 45 minutes before the builds finish, but can't leave in case they break. He nods, we chat nicely for a bit... and three days later, we have 2 more build servers.
Grad School - in something Not Computer Science. Look at the world, pick something else that you're interested in (that computers are also used for), and go get a grad degree in it. Accounting. Astronomy. Biology. Chemistry. Various types of Engineering. Law. Medicine. Anything that processes data needs computer people, or needs people doing it who can also program, or (even better) lead computer people.
If you can get a job while in grad school (or through grad school), fine - otherwise, live on scholarships/loans/handouts from parents.
It's almost not enought to be able to do one thing, and do it well, unless that one thing is so rare it's almost unrecognizable - and there are way too many folks out there who designed VBScript web pages competing for the jobs with "programmer" in the title.
We never hear about the VCs who stay out of companies. "Due diligence" was something the lawyers and accountants argued about during mergers, but there were some select VC groups doing the same thing with their investments.
In the late days of the boom, I got a phone call from a VC who wanted to talk with me about one of our tool vendors; I'd used the tools, referred other people to the company, and reported quite a few bugs, so apparently I showed up in their databases. I ended up spending 3 hours one night talking with a senior guy at the VC firm, who actually knew and understood the business area, had done his homework about the competitors, and wanted honest opinions about the company - it's products, it's staff, it's support, what I thought they'd gotten right, what I thought they were doing wrong... all in all, a very enjoyable experience. I got the impression that these guys were being very careful where they put their multi-million dollar investments.
I hope that's one of the firms that's still out there, keeping the markets afloat (and trolling for bargains, I'm sure). They're the kind of exception that makes the "invest-in-anything-with-a-dot-in-it" VCs look like such fools.
Another reason (in addition to the excellent ones other folks have listed) - cost. Power4 chips are over-engineered, compared with "consumer" CPUs like the G4, P4, and 970. Hannibal's article mentions that at the same clock speed, some instructions execute faster on the 970 simple because of the thickness of the oxide layers used in the transistor gates. It's a different emphasis - high reliability and expense versus "less" (still acceptable to 80% of the world) reliability and acceptable mass-production cost-per-chip.
Plus, the Power4 is really designed as a server/Big Iron chip - it's really 2 CPUs on 1 die - and that's just not what an iMac needs.
The article points out that a comparison with the AMD chip would be appropriate, but it's not "out there" right now as a basis for comparison. Hannibal says he'll probably use the 970 as a reference when he gets hold of the Opteron and does his down-in-the-registers review of it.
IMHO, one of the best things they could do with this is start a "smart radio" configuration standard, where the digital radio components could be discovered or described, and the "software radio" would use the appropriate interfaces.
Of course, concepts like this usually fall within the bounds of the IEEE, or worse, the ITU. Still, this could be a good tool to drive such technologies.
My understanding is that the classical commercial radio modulation schemes are a price/performance tradeoff, with most of the price part based on the technologies available before 1970. It all has to be modulated and demodulated with "dumb" hardware, and the receivers have to be commodity items.
That said, we could probably start stuffing multi-channel digital signals into the spaces between the existing analog channels, or do some frequency hopping spread spectrum into those gaps, and get some decent performance in existing bandwidth.
If you look, you'll see that this legislation takes existing law and tunes it - they're not adding pages to the US Code, they're revising it - the same thing you do to your code base.
Don't you wish you could make (l)users go to jail for some of the dumb stuff they've pulled?;)
As for the need for anti-"cyber" stuff - people spend a lot of time and energy on this forum complaining about how obsolete the law is compared with technology. Well, the Congresscritters have just taken a step forward. Maybe not in the direction we the "techno-elite" would like, but they moved. You don't like it? Get out and vote!
Finally - someone who actually READ some (or all)of it!
Talk about knee-jerk reactions - they want the sentencing bureau to review sentencing guidelines! What a horrible danger to our civil liberties! I know if I found out my wife/child/significant other's plane had crashed because some yahoo hacked into ATC or killed the power at the field as they were on approach, I'd want 'em facing life.
IF you read the bill, you see that there is language in there for intent, and premeditation, and malice... I'm far more concerned about the "quick'n'easy" wiretap and trace authority, although something like it's been needed since Stoll was working with his Cuckoo's Egg.
I figured this would be a hot-button topic on/. - I hoped there would be actual thought involved before most people pushed those buttons.
There were two lines of growth there. Motorola thought they could assembly-line satellites and bring the per-unit cost down dramatically; almost all satellites today are custom-made, even the big "families" like the Boeing 601 and 702. Motorola was going to mass-produce identical Iridium satellites and get their per-uit cost down.
There was also going to be a boom in cheap launch technology. Companies like Rotary Rocket, Beale Aerospace, and Kistler were growing to compete for the business of keeping Iridium, ICO, Globalstar, and especially Teledesic in space. Since all four have scaled back or disappeared, the funding for the new launch technology development disappeared too. If they'd succeeded, the cost per pound to low orbit would have been 20% of the current price - making a large constellation affordable.
Hello, Iridium customer service? My phone was stolen two hours ago in the Hong Kong airport... what's that? My customer ID? Phone number? Sure... it's where? Shanghai? Yes, I authorize anti-theft deterrence...
and somewhere in downtown Shanghai, a head explodes....
The contract is with the satellite builder and launcher - the FCC is being picky, all right. This is SOP for geosync comsats, but Iridium's a different model - the FCC's having to stretch to accomodate the changes. Unfortunately, stretching isn't something they do particularly well.
Re:It's a matter of finding things again...
on
Unlimited Airwaves
·
· Score: 1
It's not a routing question - these aren't necessarily point-to-point links - if you want to implement those, you'll have to look higher in the protocol stack:). The technology here is down in the physical layer. IPv6 is WAY above this.
Spread-spectrum is based on codes and pseudo-random numbers and fast clocks. The codes determine the uniqueness of the chip sequence for each user; the clocks determine when the chips occur; the pseudo-random numbers determine the chip sequence based on the code and the clock tick. At least, in some versions;) It's not simple, and not very cheap - it's taken years for the cellular industry to get CDMA to affordable levels, and that's one of the fundamental technologies needed for spread-spectrum applications.
You can do optical networking without fiber - and in about 6 more weeks of grad school, I'll be able to tell you all about it (ugh).
You still have to "bound" the area of the overall spectrum that you look for signal in. The size of the slots is a function of the speed of your clocks, size of your code, and a lot of other technology-specific issues.
Multiplexing over channels doesn't harm the signal provided the code/time space is large enough for all the users - it is still possible to get interference, but it should look like white noise to the receiver, which is why the "noise floor" keeps getting referenced. The more users, and the more power each emits, the higher the noise floor - and beyond a certain level, you can't find the signal in all the noise.
Re:It's a matter of finding things again...
on
Unlimited Airwaves
·
· Score: 2, Informative
Good synopsis.
One of the problems with his proposal is that there's so much old equipment out there that doesn't tolerate interference - so you have to retire a "use" for a band before you can free it up for spread-spectrum, or the SS signals generate noise for the existing users, who're using 20 year old technology. That's been done successfully once, to my knowledge - when the FCC made the wireless carriers pay to relocate and reequip point-to-point microwave relays in the A band. $Billions spent.
Of course, if you want to go spread spectrum, you have to have codes - and the more users you have, the more complex the codes need to be - there is a limit to the number of users you can have, based on the uniqueness of any given code set.
I spent some time looking at various web sites about this yesterday - seems the original Flyer would Dutch roll from take-off to landing, and was very unfriendly in ground-effect. This made landing - interesting - until they finally cracked it up. Good thing it went so slowly that it didn't hurt so much when they hit.
By today's standards, the thing's unflyable - horrible control authority, CG all wrong, underpowered... Orville and Wilbur had to be talented in the first place to fly it. Of course, this is the basic device that we started from to derive "today's standards". I hope none of the replica teams crack up... there's enough aviation hysteria these days, without a "reenactment" generating more bad press.
Must be fun, inventing a whole science, and a set of industries.
The folks doing this kind of work usually ARE utilities, or in joint ventures with them. They have their buildings built for these purposes.
That said, a fiber bundle or a set of satellite dishes gets you lots of bandwidth these days. "Trunking" gains new meaning when you can have 5 OC-3s in a 1-inch conduit.
So they're saying we have to look at this HTML message and click on something to receive verification - when the whole anti-virus world would shout "No! Stop! Don't do that!"
My crystal ball says... within 180 days of this going into effect, a virus family appears that looks like spam and tries to hijack Outlook based email clients when you click on that seal. Anyone care to venture a wager?
Not to mention that every time you "verify", you provide feedback to Trusted Sender that their email has been received and read. Oh, yeah - don't worry about your privacy - we can do that FOR you.
Nobody has to crash a Linux distro to prove that it passes Federal standards for front, side, and rear collisions (although the SSCA would be similar on the software side). Very few software companies will be threatened with class-action lawsuits because their software design exploded in a rear impact (the Pinto distro, anyone?). There's no Corporate Average Software Efficiency requirement that the operating system be able to perform 25.5 MFLOPs across a variety of loads, applications, and system configurations.
So the relative scale of procuding a single additional instance of a car may be comparable to the scale of producing a single additional copy of a Linux distribution, or other free software - but the cost basis is very, very different.
Slashdot Mirror
"... and very hot during re-entry."
Depends on your development environment - change a fundamental header file in C, C++, even Java, and you can waste several builds watching the dependencies come out of the woodwork... clean rebuilds are the best way to handle large change packages during development, IMHO.
And missing an error in the plane is a tad bit more important than in a program.
WRONG!
You have no way of knowing what other people are programming, but buddy, if you miss some types of errors in some types of programs, Bad Things Happen. Many of the moving things in the world are fly-(or drive-) by-wire - what do you think is figuring out "for this input, do that output"? A program.
Quick example - the first Ariane 5 blew up due to untested software and hardware reuse - they took a piece of equipment that worked fine on the Ariane 4 series, stuck it on the 5, and didn't test it throught the full flight envelope - but a software design decision that worked on the A4 (not to bounds-check certain values) was invalid on the A5 (you COULD get that many bits of the values on that rocket) and when it happened, the software said "It's gotta be bad hardware! Turn it off!" and swapped to the redundant module - which was busy locking up with the same error. Boom.
Now look at the millions of lines of software in the Boeing 777. The microcontroller in your grandmother's pacemaker. The computer-controlled milling machine that shapes the turbine blades of jet engines.
And it's not like if the copilot could go fly -another- plane instead of co-piloting that the airline would get any benefit. They don't get more money for being able to fly more planes.
No, he'd get laid off. That's economics. Airlines don't want spare bodies getting paid salaries - if they could train chimps to fly, and the FAA would approve, they'd do that.
Lines of code may be a horrible way to measure productivity, but it's very hard to get accountants to believe in "function points".
I agree that XP isn't as efficient as "normal" programming when both programmers are experienced - what I've seen work in practice is the use of XP (adapted slightly) as a training ground, where inexperienced personnel are combined with experienced personnel to cross-train. It works both ways - you can have a wily old Unix coder sentenced to work on MFC sitting next to a kid who's only done Visual Basic, and both will learn a lot, and be more productive as a team than either or both separately. It's a matter of chemistry as much as management.
But then, that's just an opinion. It matters little.
At one job, we'd spend six hours working the bug fixes due for first shift (there weren't enough servers to go around, so we subs got to do shift work, too - I was on second - 12 hour days, 10 on the clock), fired up the Compile-From-Hell (Ada - sssllllooooowwww compiler, lots of code), and checked out, headed for the casinos (Mississippi Gulf Coast, the first had just gotten installed). Came back a couple of hours later (or on a page from the Puritan who volunteered to watch the builds scroll past) and started testing to finish out our shift.
Later on that same job, they sent us to the lead corp's site to do final integration (since their on-location management had screwed up, we all got drawn back to the mothership). We had to do the same stuff, but with more code - and even back at Home Base, they only gave us 2 servers for our builds, so they took hours. One night approaching midnight, the VP in charge of the project walks through - only 2 of us are there, both subs, and we're both playing Minesweeper! He asks what we're up to, looking pointedly at the game screens, and I pop up the build windows behind it (X on Windows) and show him the lines scrolling past, 1 every 45 seconds, and guestimate that we have another 45 minutes before the builds finish, but can't leave in case they break. He nods, we chat nicely for a bit... and three days later, we have 2 more build servers.
Grad School - in something Not Computer Science. Look at the world, pick something else that you're interested in (that computers are also used for), and go get a grad degree in it. Accounting. Astronomy. Biology. Chemistry. Various types of Engineering. Law. Medicine. Anything that processes data needs computer people, or needs people doing it who can also program, or (even better) lead computer people.
If you can get a job while in grad school (or through grad school), fine - otherwise, live on scholarships/loans/handouts from parents.
It's almost not enought to be able to do one thing, and do it well, unless that one thing is so rare it's almost unrecognizable - and there are way too many folks out there who designed VBScript web pages competing for the jobs with "programmer" in the title.
Good luck.
We never hear about the VCs who stay out of companies. "Due diligence" was something the lawyers and accountants argued about during mergers, but there were some select VC groups doing the same thing with their investments.
In the late days of the boom, I got a phone call from a VC who wanted to talk with me about one of our tool vendors; I'd used the tools, referred other people to the company, and reported quite a few bugs, so apparently I showed up in their databases. I ended up spending 3 hours one night talking with a senior guy at the VC firm, who actually knew and understood the business area, had done his homework about the competitors, and wanted honest opinions about the company - it's products, it's staff, it's support, what I thought they'd gotten right, what I thought they were doing wrong... all in all, a very enjoyable experience. I got the impression that these guys were being very careful where they put their multi-million dollar investments.
I hope that's one of the firms that's still out there, keeping the markets afloat (and trolling for bargains, I'm sure). They're the kind of exception that makes the "invest-in-anything-with-a-dot-in-it" VCs look like such fools.
Plus, the Power4 is really designed as a server/Big Iron chip - it's really 2 CPUs on 1 die - and that's just not what an iMac needs.
The article points out that a comparison with the AMD chip would be appropriate, but it's not "out there" right now as a basis for comparison. Hannibal says he'll probably use the 970 as a reference when he gets hold of the Opteron and does his down-in-the-registers review of it.
IMHO, one of the best things they could do with this is start a "smart radio" configuration standard, where the digital radio components could be discovered or described, and the "software radio" would use the appropriate interfaces.
Of course, concepts like this usually fall within the bounds of the IEEE, or worse, the ITU. Still, this could be a good tool to drive such technologies.
My understanding is that the classical commercial radio modulation schemes are a price/performance tradeoff, with most of the price part based on the technologies available before 1970. It all has to be modulated and demodulated with "dumb" hardware, and the receivers have to be commodity items.
That said, we could probably start stuffing multi-channel digital signals into the spaces between the existing analog channels, or do some frequency hopping spread spectrum into those gaps, and get some decent performance in existing bandwidth.
Don't you wish you could make (l)users go to jail for some of the dumb stuff they've pulled? ;)
As for the need for anti-"cyber" stuff - people spend a lot of time and energy on this forum complaining about how obsolete the law is compared with technology. Well, the Congresscritters have just taken a step forward. Maybe not in the direction we the "techno-elite" would like, but they moved. You don't like it? Get out and vote!
Talk about knee-jerk reactions - they want the sentencing bureau to review sentencing guidelines! What a horrible danger to our civil liberties! I know if I found out my wife/child/significant other's plane had crashed because some yahoo hacked into ATC or killed the power at the field as they were on approach, I'd want 'em facing life.
IF you read the bill, you see that there is language in there for intent, and premeditation, and malice... I'm far more concerned about the "quick'n'easy" wiretap and trace authority, although something like it's been needed since Stoll was working with his Cuckoo's Egg.
I figured this would be a hot-button topic on /. - I hoped there would be actual thought involved before most people pushed those buttons.
There were two lines of growth there. Motorola thought they could assembly-line satellites and bring the per-unit cost down dramatically; almost all satellites today are custom-made, even the big "families" like the Boeing 601 and 702. Motorola was going to mass-produce identical Iridium satellites and get their per-uit cost down.
There was also going to be a boom in cheap launch technology. Companies like Rotary Rocket, Beale Aerospace, and Kistler were growing to compete for the business of keeping Iridium, ICO, Globalstar, and especially Teledesic in space. Since all four have scaled back or disappeared, the funding for the new launch technology development disappeared too. If they'd succeeded, the cost per pound to low orbit would have been 20% of the current price - making a large constellation affordable.
It would also be a good anti-theft mechanism:
Hello, Iridium customer service? My phone was stolen two hours ago in the Hong Kong airport... what's that? My customer ID? Phone number? Sure... it's where? Shanghai? Yes, I authorize anti-theft deterrence...
and somewhere in downtown Shanghai, a head explodes....
The contract is with the satellite builder and launcher - the FCC is being picky, all right. This is SOP for geosync comsats, but Iridium's a different model - the FCC's having to stretch to accomodate the changes. Unfortunately, stretching isn't something they do particularly well.
It's not a routing question - these aren't necessarily point-to-point links - if you want to implement those, you'll have to look higher in the protocol stack :). The technology here is down in the physical layer. IPv6 is WAY above this.
;) It's not simple, and not very cheap - it's taken years for the cellular industry to get CDMA to affordable levels, and that's one of the fundamental technologies needed for spread-spectrum applications.
Spread-spectrum is based on codes and pseudo-random numbers and fast clocks. The codes determine the uniqueness of the chip sequence for each user; the clocks determine when the chips occur; the pseudo-random numbers determine the chip sequence based on the code and the clock tick. At least, in some versions
You can do optical networking without fiber - and in about 6 more weeks of grad school, I'll be able to tell you all about it (ugh).
Take a look at how many people are buying satellite radios - maybe we can find some actual, real-world numbers and not just /. guesses.
You still have to "bound" the area of the overall spectrum that you look for signal in. The size of the slots is a function of the speed of your clocks, size of your code, and a lot of other technology-specific issues.
Multiplexing over channels doesn't harm the signal provided the code/time space is large enough for all the users - it is still possible to get interference, but it should look like white noise to the receiver, which is why the "noise floor" keeps getting referenced. The more users, and the more power each emits, the higher the noise floor - and beyond a certain level, you can't find the signal in all the noise.
Good synopsis.
One of the problems with his proposal is that there's so much old equipment out there that doesn't tolerate interference - so you have to retire a "use" for a band before you can free it up for spread-spectrum, or the SS signals generate noise for the existing users, who're using 20 year old technology. That's been done successfully once, to my knowledge - when the FCC made the wireless carriers pay to relocate and reequip point-to-point microwave relays in the A band. $Billions spent.
Of course, if you want to go spread spectrum, you have to have codes - and the more users you have, the more complex the codes need to be - there is a limit to the number of users you can have, based on the uniqueness of any given code set.
Gawd, I'm having a grad school flashback!
They actually controlled it to some degree through body lean - their hips were cradled, and one of the control sets was wired up to that cradle.
By all accounts I've seen, it was a full-body experience.
I spent some time looking at various web sites about this yesterday - seems the original Flyer would Dutch roll from take-off to landing, and was very unfriendly in ground-effect. This made landing - interesting - until they finally cracked it up. Good thing it went so slowly that it didn't hurt so much when they hit.
By today's standards, the thing's unflyable - horrible control authority, CG all wrong, underpowered... Orville and Wilbur had to be talented in the first place to fly it. Of course, this is the basic device that we started from to derive "today's standards". I hope none of the replica teams crack up... there's enough aviation hysteria these days, without a "reenactment" generating more bad press.
Must be fun, inventing a whole science, and a set of industries.
The folks doing this kind of work usually ARE utilities, or in joint ventures with them. They have their buildings built for these purposes.
That said, a fiber bundle or a set of satellite dishes gets you lots of bandwidth these days. "Trunking" gains new meaning when you can have 5 OC-3s in a 1-inch conduit.
So they're saying we have to look at this HTML message and click on something to receive verification - when the whole anti-virus world would shout "No! Stop! Don't do that!"
My crystal ball says... within 180 days of this going into effect, a virus family appears that looks like spam and tries to hijack Outlook based email clients when you click on that seal. Anyone care to venture a wager?
Not to mention that every time you "verify", you provide feedback to Trusted Sender that their email has been received and read. Oh, yeah - don't worry about your privacy - we can do that FOR you.
Do you really want Your Tax Dollars sending you spam?
A difference in the analogy - regulation.
Nobody has to crash a Linux distro to prove that it passes Federal standards for front, side, and rear collisions (although the SSCA would be similar on the software side). Very few software companies will be threatened with class-action lawsuits because their software design exploded in a rear impact (the Pinto distro, anyone?). There's no Corporate Average Software Efficiency requirement that the operating system be able to perform 25.5 MFLOPs across a variety of loads, applications, and system configurations.
So the relative scale of procuding a single additional instance of a car may be comparable to the scale of producing a single additional copy of a Linux distribution, or other free software - but the cost basis is very, very different.