For corporate needs, $1000 is nothing, especially if it saves time
Every time I see something aimed at business that I think is overpriced, I remember this dialog that occurred at the job I had out of school as a newly minted engineer:
guy-on-phone: "Hi, I'm the shop supervisor at $LargeMidwestManufacturer. We just broke this box that plugs into our SPC computer and all it has on it is your company name and phone number. I need to know the part number so I can write a purchase order for a new one." Me (after asking a few more questions to identify device): "OK, that's the SPxxx and the price without cables is $400" guy-on-phone: "!choke*splutter!!. That little box is 400 bucks? Jeez. Oh, whatever, can we get two shipped out overnight, 8am delivery?" Me: "Sure, as long as we get the P.O by 4pm. Just put my name on the PO and I'll make sure they rush it out."
Now, the reason I always think of this is because my reaction was the same as the shop foreman's, and I knew that it cost us all of $20 to make "that little black box." I also knew that our competitor's price was 2x ours (but theirs had more features)
The point is twofold: it's easier to spend money when it's not coming out of your own pocket and (2) if not having that $400 item is slowing down your production line, you want it there yesterday, damn the cost.
OK, somebody explain to me what the big deal with a GMail account is. As far as I can tell it's just free webmail with 1Gb storage. Yet people are so excited about it. Why?
I mean, free webmail is everywhere. I have untold gigs of storage on my HD and unused old HDs in the closet and I have never come close to more than a meg or so of saved email (if people send me pictures I want to keep, I just detach the file and save it somewhere else).
There's not a lot of profit in 20 cent digital prints.
I don't know how much profit there actually is, but remember that "not a lot of" does not necessarily equal zero. In this particular case, if there is 1 cent profit, that's 5%, 2 cents, that's 10%. In both cases, greater profit than a lot of consumer businesses enjoy. IOW, it looks doable.
OT: 'Jane Boxwine' is interesting, and [to me] connotes a different sort of person than 'Jane Sixpack'.
Yeah, it reminds me of a woman I knew who could not stop working out. Great abs, but totally obsessed with looking good.
Anyway... the fundamental problem with the car/computer analogy that I haven't seen mentioned yet is that cars have been around long enough and have become so basic to American society that *most* owners know there are some basic things that need to be done. Gotta get gas, change the oil, check coolant before winter, etc. Now, there are some people who don't have their oil changed more frequently than once every 10,000 miles or so, but they're in the minority.
But basic computer "maintenance" is a recent issue for the general population. Most people don't know they should have a firewall or antivirus software, or if they do, that they need to keep the virus definitions up to date. Most people aren't stupid. This a training issue, but how do you reach the person who bought a computer 5 years ago and just decided to get broadband? Well, the broadband provider could offer tutorials on security and related issues. At the very least, a booklet or CDROM with a number to call if you want (at extra cost) a field tech to checkout your system. This probably won't work because the provider will most likely look at it as a way to make more money, and the service will be grossly overpriced. So maybe local PC tech services should start targetting home users and offering "computer tuneups" (I have seen a few advertised) that do it at a more reasonable price.
And it is getting worse. At home, my machines are behind two routers (one is the crappy Qwest DSL box, the other is wireless: cheaper than paying for Actiontec's upgrade), both with NAT and one with a firewall. After months of no activity, I'm beginning to see a lot of failed SSH attempts to my primary Linux machine. Enough so I finally shut down port forwarding of SSH as I don't really need outside access to it. Our lone WinXP box has up to date virus subscription and is regularly updated and appears to be OK, but it's scary to see how many attacks are being made on my systems, and those are only the ones I know about.
Sounds like a good program. The only problem I would have with it is that I never use my Blockbuster card. As in I haven't entered one of their stores in months simply because I can never find anything there I like.
I'm not sure what this means, but the only movies I haven't been able to get are those labeled 'Not yet released' and we've been Netflix subscribers since February. Then again, we do prefer non-mainstream movies, so I guess they tend to be more available.
have it back the next night by midnight for a dollar credit
So could I, in theory. Problem is when I get home and I decide that (a) I'm really not in the mood for a movie after all, (b) wife decides to "distract" me from movie:-), or (c) I fall asleep and don't see all of it. Stuff like this is not an issue with Netflix.
The only physical video rental place I have ever seen that has a lot of the kind of movies I like is Discount Video on Hennepin Ave in Minneapolis (speaking of which, according to your website, I think I used to live a block or so away from you... "Dacotah View apartments" sound familiar?)
And now here's a plug for Netflix:-) Three movies out at any time for $20/month. No late fees: I never knew how important this was until I noticed that I was no longer rushing to view a movie just so I could return it on time. We probably go through 6-7 movies a month on Netflix and they have a selection no local video store can match. Next best thing to Video On Demand.
Well gee, I guess it's just a fluke that only large corporations are doing it
Are you really that dense or are you just not very good at trolling?
There are many small companies doing it; it simply doesn't make the nightly news. There are small manufacturers all over the net whose blogs claim that their PC boards are made by Olimex in Bulgaria (whose prices are so cheap it is not worth my time to make them in my basement any more). I've done this: not with Olimex, but from other foreign vendors. It's no more difficult than dealing with the guy in the next town. I know of a small business of about 6 employees that outsources programming to India. I could go on, but since I can't prove any of my anecdotes, you'll probably just claim I'm making it up.
If you honestly think this is only the province of large corporations, you're sadly mistaken.
It is also very bad for the small "mom and pop" companies. Only large corporations can afford to do out sourcing
NO. Where do you people come up with this stuff!?!
I can hop on RentACoder and get someone in Eastern Europe to do code that I couldn't hope to compete with on price alone. Hell, a friend and I considered a business plan based on bidding cheap for local (US) contracts then farming them out on eLance or RentACoder to see if it would work.
I can use foreign PC board manufacturers and assemblers at fractions of the US price. And that's just off the top of my head without doing any additional research.
With the net and FedEx, it's getting to the point that *anyone* can outsource. I'm waiting for the day I see housewives in the Ukraine advertising online that they will sew the latest Donna Karan (or somesuch) knockoff dresses custom fitted to your measurements for $30! I suspect it's not far off.
I gave it 30 minutes. After that I was so sick of it that I walked out of the theater knowing that it was pouring cats & dogs outside and I was soaked through by the time I got home.
will look like unmaintainable trash to most other programmers
Unmaintainable trash is *not* good code. Not in a team environment anyway. Hell, even when I'm doing stuff only I will see, I write it in a maintainable fashion because I already know that I'm going to revisit it in a year (whether or not I plan to!) and I really won't want to spend half a day trying to figure out what I was thinking.
I agree with you on option 1 except as stated above and I'd loosen up the code ownership a bit.
It's good to get this perspective sometimes. It's been so long since I didn't have a car I forget how inconvenient some things I take for granted can be. I like my Thinkpad T40 cause I can walk around the building with it and if I take it home, I just put it in a Targus bag and toss it in the back of the car. But if I had to take it on my bike or lug it around on the multiple busses it would take me to get home, it would be a pain: the combo of T40, external power supply, extra battery and bag gets uncomfortable real fast.
I don't think that's really an issue; in 6 years of S/W development at this job, I have *never* had a computer upgraded piecemeal: IT just rolls out completely new PCs every 2 years (or earlier if we complain about something). Right now we have 1.6GHz Thinkpads with 512M ram running Win2k Pro and that'll be just fine for a few more years development.
But I don't see the need for any more portability in the office environment. If I have to take my notepad to a meeting, I just eject it from a docking station and go to the meeting where it will automatically connect to the wireless network if I don't plug it into a network outlet. If I had a smaller device, now I'd need to also take along a full size keyboard to use it efficiently and that would defeat the whole portability issue. Where I do see these things taking off is in lab and service use. There have been many times I've wished for a tiny PC that I could use to snoop on a serial line as I'm integrating a new device our EE dept just built, instead of having to balance a full size PC and monitor on a lab cart. Service people could connect to a machine with a tiny handheld device that could have service manuals and extensive diagnostics. This is where it would be really useful in my world.
By what logic? Do you really think companies will pay you more than you cost them? Look, imagine you need a car to get to work. You can get the same car for 10,000 or 15,000 or 20,000 depending on what dealership in town you go to. Why would you pay 20,000 for the same product (including warranty, etc) when you could get it for 10,000? You wouldn't. Neither would a business. If they're paying someone X, it's because it's worth X to them to keep that person on staff. It's that simple.
I'm not arguing against distance learning per se, only against the type of people who so often seem to think it's a good idea, and the type of schools that seem to cater to them.
Really? What exactly would you say is wrong with UIUC or Stanford ? I considered taking either an online course in patterns, taught by Ralph Johnson this summer, or one in Bioinformatics at Stanford. I'd like to know what you thing is wrong with these "types" of schools.
I have no first-hand experience with any MBA program, so take this with a large flake of kosher salt. I have, however, thought about and discussed it with others, including MBAs.
I think a large part of the MBA curriculum is discussion with your peers and face to face meeting. An online-only MBA is probably going to be seriously lacking in these aspects so a better choice may be taking the more rote courses: accounting, finance, etc online and the people-interface stuff: management, marketing in an actual classroom. That, and while tech-centered business is very accepting of online degrees, the management field is still very much focused on face-to-face interaction, so you may find that a UP MBA is not as valued as one from a local university.
Me, I started out by reading The Portable MBA while exercising to get an overview of what I would be in for and more background to decide if I do want to go back to school to do it. Like you, I have a full time job as an S/W developer (and my employer will cover tuition 100% -- my favorite benefit after the paycheck:-) and a family, so I would have to take a lot of classes online; this is why I've been considering the downside to doing it all online.
But, in light of the fact that they approved Windows for use in medical equipment of any kind, I doubt they're doing their stated role here
They don't. FDA will not tell you what operating system you may/may not use on your device any more than they will dictate what manufacturer makes the motor in your circulation pump. They are concerned with the system as a whole. Therefore if the FDA auditor asks you what happens if the system goes BSOD just after it began to apply that therapeutic dose of xrays, your response should be to point to the hardware circuit that overrides the software dosage control and shuts the system down gracefully without harm to the patient.
The problem with looking at only a single component (even one as central as the OS) and screaming "that can fail, don't use it" is that it ignores the requirement for complete system design that should be done in any properly engineered product. Not surprisingly, most software outside regulated industries seems to do just that!
it is irresponsible in the extreme to rely on somebody else's box 'o' bugs as part of your life-safety system. period
The microprocessor/microcontroller and peripheral ICs are also "somebody else's box 'o' bugs" Are you planning on individually validating all of them as well?
The fact is that reality intrudes. Code anything but very very simple systems in machine/assembly and watch your defect rates climb and your time to market extend to infinity. Modern medical systems are very complex, and even with five year development cycles, you're not going to be coding more than a few 1000 lines in assembly.
I'm no microsoft fan either, but even in medical applications it makes sense to use their products sometimes: 1) You can find qualified developers without taking a year to replace the guy who just left. 2) Need to store patient/instrument performance data? There are more database solutions available than for the popular embedded OSes. 3) The graphic designer you need for the GUI for that complex interface is going to need off the shelf tools or you have to spawn another project to build them yourself 4) Really safety critical stuff can be isolated to custom hardware running only your own code (in most cases I'm aware of, this is what happens anyway -- the GUI stuff is just for user interface and data storage), or executes in hardware with no software component: don't have code that monitors if the user opens the cover so you can shut off the X-Ray; use a simple interlock switch. 5) (and this is the/. heresy) Winders ain't that bad if you can a) control the hardware platform it runs on, and in an embedded system you'd better be able to; b) your software is the ONLY thing running on top of the OS (again, fairly easy to do) and c) you limit the device drivers you didn't write yourself to only those approved by Microsoft.
As embedded Linux gets more into the mainstream it will displace Microsoft for the bigger systems, but that takes time.
I get the feeling that you don't realize the complexity of some of these devices. Many modern embedded control systems have the same issues as distributed server-based architectures because, well, that's what they *are*
The best products seem to come from places that have unlimited time
Actually the best products come from places that know how to make good use of time. Make a reasonable development schedule. Marketing wants a new feature? OK, either push the schedule out or drop some less important feature that will take an equivalent amount of time. Plan design and code reviews up front and build them into the schedule. Allow sufficient minor integrations along the way so the customer can see how the project is progressing and give feedback before it gets too far off course. Allocate enough time for QA to create test cases, execute the tests and make sure there's enough time for developers to fix the inevitable problems. Monitor the metrics coming out of the schedule: number of defects found this week; number fixed; number waiting to be fixed or fixed and waiting to be tested, number of features still not yet implemented; adjust schedule as necessary.
In short, approach software development like you would an engineering project. It's done every day and many organizations churn out quality product.
I don't really understand the rush to make linux as small as possible.
Here's a contrived (but realistic) example: a small manufacturer makes powered dollies that can move loads of up to 1000 lbs by the operator pushing the handle on the dolly. Now they identify a market opportunity for a smarter dolly that has variable maximum speed depending on how much is loaded onto it, won't move if the weight is over the limit, has steerable front wheels that will turn at a rate that won't send the load flying off, and is controllable from a handheld computer with 20' range.
Now, of course, all of what I described could be done with a small microcontroller and some programming, but there are many aspects that having a real OS (no, I didn't say real *time*) would make this device much easier to program and the cost of the final product means that the slight extra cost of a Linux-capable SBC (single board computer) versus say an AVR based SBC is negligible. e.g., maybe they want a touch-screen interface, want to do nightly runtime data offload to a central server or other tasks that are so much easier as part of the OS.
The Linux based solution may take 2 weeks to complete and get on the market; the non-OS based solution could take much longer. So factor in that this thing (the handheld control) needs to be super-rugged (perhaps enclosed in a NEMA-4 washdown-rated enclosure so it can be used in the rain or hosed down at the end of the day after being used in a meat-packing plant) to withstand being banged around all day, very small and low power and you can see why a tiny Linux based solution looks very attractive.
That, and it's cool:-) This is why I love developing embedded systems!
Re:Embed Linux? It's not an ideal choice.
on
How to Embed Linux
·
· Score: 1
I think what's happening is that the term embedded is being watered down.
This bothered me for quite a while, then I realized it doesn't really matter. People with "real embedded" (which in my book means having knowledge of low-level control) experience will know what I mean when I use the term.
The embedded system I work on in my day job runs on a 1.5GHz Pentium, has over 1MLOC, costs around $800,000, uses close to 2kW of power, and at around 900lbs, is definitely not portable. And I still see posts here from people who think that embedded means the system has to fit in 16k of ROM and sip from batteries!
Not really. Engine control software "decides what the engine will do" based on what the operator desires, not blindly following what the operator is doing. The operator isn't sending 1000 FIRE signals to sparkplugs every second; the operator may want to rev to 8000 RPM, but the ECM will cut fuel at 6000, etc.
The ECM translates operator requests and adjusts for programmed limitations, current sensor states and engine control parameters to try to fulfill those requests, the UAV controls are doing essentially the same thing, but the operator requests (go from point A to point B) have been programmed in before flight started.
Slashdot Mirror
Every time I see something aimed at business that I think is overpriced, I remember this dialog that occurred at the job I had out of school as a newly minted engineer:
Now, the reason I always think of this is because my reaction was the same as the shop foreman's, and I knew that it cost us all of $20 to make "that little black box." I also knew that our competitor's price was 2x ours (but theirs had more features)
The point is twofold: it's easier to spend money when it's not coming out of your own pocket and (2) if not having that $400 item is slowing down your production line, you want it there yesterday, damn the cost.
OK, somebody explain to me what the big deal with a GMail account is. As far as I can tell it's just free webmail with 1Gb storage. Yet people are so excited about it. Why?
I mean, free webmail is everywhere. I have untold gigs of storage on my HD and unused old HDs in the closet and I have never come close to more than a meg or so of saved email (if people send me pictures I want to keep, I just detach the file and save it somewhere else).
So am I missing something here?
I don't know how much profit there actually is, but remember that "not a lot of" does not necessarily equal zero. In this particular case, if there is 1 cent profit, that's 5%, 2 cents, that's 10%. In both cases, greater profit than a lot of consumer businesses enjoy. IOW, it looks doable.
Yeah, it reminds me of a woman I knew who could not stop working out. Great abs, but totally obsessed with looking good.
Anyway... the fundamental problem with the car/computer analogy that I haven't seen mentioned yet is that cars have been around long enough and have become so basic to American society that *most* owners know there are some basic things that need to be done.
Gotta get gas, change the oil, check coolant before winter, etc. Now, there are some people who don't have their oil changed more frequently than once every 10,000 miles or so, but they're in the minority.
But basic computer "maintenance" is a recent issue for the general population. Most people don't know they should have a firewall or antivirus software, or if they do, that they need to keep the virus definitions up to date. Most people aren't stupid. This a training issue, but how do you reach the person who bought a computer 5 years ago and just decided to get broadband? Well, the broadband provider could offer tutorials on security and related issues. At the very least, a booklet or CDROM with a number to call if you want (at extra cost) a field tech to checkout your system. This probably won't work because the provider will most likely look at it as a way to make more money, and the service will be grossly overpriced.
So maybe local PC tech services should start targetting home users and offering "computer tuneups" (I have seen a few advertised) that do it at a more reasonable price.
And it is getting worse. At home, my machines are behind two routers (one is the crappy Qwest DSL box, the other is wireless: cheaper than paying for Actiontec's upgrade), both with NAT and one with a firewall. After months of no activity, I'm beginning to see a lot of failed SSH attempts to my primary Linux machine. Enough so I finally shut down port forwarding of SSH as I don't really need outside access to it. Our lone WinXP box has up to date virus subscription and is regularly updated and appears to be OK, but it's scary to see how many attacks are being made on my systems, and those are only the ones I know about.
Sounds like a good program. The only problem I would have with it is that I never use my Blockbuster card. As in I haven't entered one of their stores in months simply because I can never find anything there I like.
I'm not sure what this means, but the only movies I haven't been able to get are those labeled 'Not yet released' and we've been Netflix subscribers since February. Then again, we do prefer non-mainstream movies, so I guess they tend to be more available.
So could I, in theory. Problem is when I get home and I decide that (a) I'm really not in the mood for a movie after all, (b) wife decides to "distract" me from movie
The only physical video rental place I have ever seen that has a lot of the kind of movies I like is Discount Video on Hennepin Ave in Minneapolis (speaking of which, according to your website, I think I used to live a block or so away from you
And now here's a plug for Netflix
Are you really that dense or are you just not very good at trolling?
There are many small companies doing it; it simply doesn't make the nightly news. There are small manufacturers all over the net whose blogs claim that their PC boards are made by Olimex in Bulgaria (whose prices are so cheap it is not worth my time to make them in my basement any more). I've done this: not with Olimex, but from other foreign vendors. It's no more difficult than dealing with the guy in the next town. I know of a small business of about 6 employees that outsources programming to India.
I could go on, but since I can't prove any of my anecdotes, you'll probably just claim I'm making it up.
If you honestly think this is only the province of large corporations, you're sadly mistaken.
NO. Where do you people come up with this stuff!?!
I can hop on RentACoder and get someone in Eastern Europe to do code that I couldn't hope to compete with on price alone. Hell, a friend and I considered a business plan based on bidding cheap for local (US) contracts then farming them out on eLance or RentACoder to see if it would work.
I can use foreign PC board manufacturers and assemblers at fractions of the US price. And that's just off the top of my head without doing any additional research.
With the net and FedEx, it's getting to the point that *anyone* can outsource. I'm waiting for the day I see housewives in the Ukraine advertising online that they will sew the latest Donna Karan (or somesuch) knockoff dresses custom fitted to your measurements for $30! I suspect it's not far off.
I gave it 30 minutes. After that I was so sick of it that I walked out of the theater knowing that it was pouring cats & dogs outside and I was soaked through by the time I got home.
OMG! A voice of sanity. On Slashdot.
Repent ye sinners, for the end is nigh.
This would be a good time to announce my new Solar Hummer project on Sourceforge...
Unmaintainable trash is *not* good code. Not in a team environment anyway. Hell, even when I'm doing stuff only I will see, I write it in a maintainable fashion because I already know that I'm going to revisit it in a year (whether or not I plan to!) and I really won't want to spend half a day trying to figure out what I was thinking.
I agree with you on option 1 except as stated above and I'd loosen up the code ownership a bit.
Coding standards exist for a reason.
This is not surprising: it's the way the world works. Think carefully about what't going on here on
It's good to get this perspective sometimes. It's been so long since I didn't have a car I forget how inconvenient some things I take for granted can be. I like my Thinkpad T40 cause I can walk around the building with it and if I take it home, I just put it in a Targus bag and toss it in the back of the car. But if I had to take it on my bike or lug it around on the multiple busses it would take me to get home, it would be a pain: the combo of T40, external power supply, extra battery and bag gets uncomfortable real fast.
I don't think that's really an issue; in 6 years of S/W development at this job, I have *never* had a computer upgraded piecemeal: IT just rolls out completely new PCs every 2 years (or earlier if we complain about something). Right now we have 1.6GHz Thinkpads with 512M ram running Win2k Pro and that'll be just fine for a few more years development.
But I don't see the need for any more portability in the office environment. If I have to take my notepad to a meeting, I just eject it from a docking station and go to the meeting where it will automatically connect to the wireless network if I don't plug it into a network outlet. If I had a smaller device, now I'd need to also take along a full size keyboard to use it efficiently and that would defeat the whole portability issue.
Where I do see these things taking off is in lab and service use. There have been many times I've wished for a tiny PC that I could use to snoop on a serial line as I'm integrating a new device our EE dept just built, instead of having to balance a full size PC and monitor on a lab cart. Service people could connect to a machine with a tiny handheld device that could have service manuals and extensive diagnostics. This is where it would be really useful in my world.
By what logic? Do you really think companies will pay you more than you cost them?
Look, imagine you need a car to get to work. You can get the same car for 10,000 or 15,000 or 20,000 depending on what dealership in town you go to. Why would you pay 20,000 for the same product (including warranty, etc) when you could get it for 10,000?
You wouldn't. Neither would a business. If they're paying someone X, it's because it's worth X to them to keep that person on staff. It's that simple.
Really? What exactly would you say is wrong with UIUC or Stanford ? I considered taking either an online course in patterns, taught by Ralph Johnson this summer, or one in Bioinformatics at Stanford. I'd like to know what you thing is wrong with these "types" of schools.
I have no first-hand experience with any MBA program, so take this with a large flake of kosher salt. I have, however, thought about and discussed it with others, including MBAs.
:-) and a family, so I would have to take a lot of classes online; this is why I've been considering the downside to doing it all online.
I think a large part of the MBA curriculum is discussion with your peers and face to face meeting. An online-only MBA is probably going to be seriously lacking in these aspects so a better choice may be taking the more rote courses: accounting, finance, etc online and the people-interface stuff: management, marketing in an actual classroom. That, and while tech-centered business is very accepting of online degrees, the management field is still very much focused on face-to-face interaction, so you may find that a UP MBA is not as valued as one from a local university.
Me, I started out by reading The Portable MBA while exercising to get an overview of what I would be in for and more background to decide if I do want to go back to school to do it. Like you, I have a full time job as an S/W developer (and my employer will cover tuition 100% -- my favorite benefit after the paycheck
They don't. FDA will not tell you what operating system you may/may not use on your device any more than they will dictate what manufacturer makes the motor in your circulation pump. They are concerned with the system as a whole. Therefore if the FDA auditor asks you what happens if the system goes BSOD just after it began to apply that therapeutic dose of xrays, your response should be to point to the hardware circuit that overrides the software dosage control and shuts the system down gracefully without harm to the patient.
The problem with looking at only a single component (even one as central as the OS) and screaming "that can fail, don't use it" is that it ignores the requirement for complete system design that should be done in any properly engineered product.
Not surprisingly, most software outside regulated industries seems to do just that!
The microprocessor/microcontroller and peripheral ICs are also "somebody else's box 'o' bugs" Are you planning on individually validating all of them as well?
The fact is that reality intrudes. Code anything but very very simple systems in machine/assembly and watch your defect rates climb and your time to market extend to infinity. Modern medical systems are very complex, and even with five year development cycles, you're not going to be coding more than a few 1000 lines in assembly.
I'm no microsoft fan either, but even in medical applications it makes sense to use their products sometimes:
1) You can find qualified developers without taking a year to replace the guy who just left.
2) Need to store patient/instrument performance data? There are more database solutions available than for the popular embedded OSes.
3) The graphic designer you need for the GUI for that complex interface is going to need off the shelf tools or you have to spawn another project to build them yourself
4) Really safety critical stuff can be isolated to custom hardware running only your own code (in most cases I'm aware of, this is what happens anyway -- the GUI stuff is just for user interface and data storage), or executes in hardware with no software component: don't have code that monitors if the user opens the cover so you can shut off the X-Ray; use a simple interlock switch.
5) (and this is the
As embedded Linux gets more into the mainstream it will displace Microsoft for the bigger systems, but that takes time.
I get the feeling that you don't realize the complexity of some of these devices. Many modern embedded control systems have the same issues as distributed server-based architectures because, well, that's what they *are*
Actually the best products come from places that know how to make good use of time.
Make a reasonable development schedule. Marketing wants a new feature? OK, either push the schedule out or drop some less important feature that will take an equivalent amount of time. Plan design and code reviews up front and build them into the schedule. Allow sufficient minor integrations along the way so the customer can see how the project is progressing and give feedback before it gets too far off course. Allocate enough time for QA to create test cases, execute the tests and make sure there's enough time for developers to fix the inevitable problems. Monitor the metrics coming out of the schedule: number of defects found this week; number fixed; number waiting to be fixed or fixed and waiting to be tested, number of features still not yet implemented; adjust schedule as necessary.
In short, approach software development like you would an engineering project. It's done every day and many organizations churn out quality product.
Here's a contrived (but realistic) example: a small manufacturer makes powered dollies that can move loads of up to 1000 lbs by the operator pushing the handle on the dolly. Now they identify a market opportunity for a smarter dolly that has variable maximum speed depending on how much is loaded onto it, won't move if the weight is over the limit, has steerable front wheels that will turn at a rate that won't send the load flying off, and is controllable from a handheld computer with 20' range.
Now, of course, all of what I described could be done with a small microcontroller and some programming, but there are many aspects that having a real OS (no, I didn't say real *time*) would make this device much easier to program and the cost of the final product means that the slight extra cost of a Linux-capable SBC (single board computer) versus say an AVR based SBC is negligible. e.g., maybe they want a touch-screen interface, want to do nightly runtime data offload to a central server or other tasks that are so much easier as part of the OS.
The Linux based solution may take 2 weeks to complete and get on the market; the non-OS based solution could take much longer. So factor in that this thing (the handheld control) needs to be super-rugged (perhaps enclosed in a NEMA-4 washdown-rated enclosure so it can be used in the rain or hosed down at the end of the day after being used in a meat-packing plant) to withstand being banged around all day, very small and low power and you can see why a tiny Linux based solution looks very attractive.
That, and it's cool
This bothered me for quite a while, then I realized it doesn't really matter. People with "real embedded" (which in my book means having knowledge of low-level control) experience will know what I mean when I use the term.
The embedded system I work on in my day job runs on a 1.5GHz Pentium, has over 1MLOC, costs around $800,000, uses close to 2kW of power, and at around 900lbs, is definitely not portable. And I still see posts here from people who think that embedded means the system has to fit in 16k of ROM and sip from batteries!
Not really. Engine control software "decides what the engine will do" based on what the operator desires, not blindly following what the operator is doing. The operator isn't sending 1000 FIRE signals to sparkplugs every second; the operator may want to rev to 8000 RPM, but the ECM will cut fuel at 6000, etc.
The ECM translates operator requests and adjusts for programmed limitations, current sensor states and engine control parameters to try to fulfill those requests, the UAV controls are doing essentially the same thing, but the operator requests (go from point A to point B) have been programmed in before flight started.