That doesn't do your company any good when you leave them.
Sorry, explain the part where that matters to me?
It's called taking pride in your work. I'm paid to develop software.That doesn't just mean it compiles and runs, but that it's readable and (hopefully) extensible by the people to come later. I try to code well enough that I could be hit by a truck tomorrow and the next guy could easily pick it up. Has the side effect of me being able to read code I wrote 2 years ago, which is great as I'm on a very long term project.
The common thread I'm seeing here is that we probably need more "black boxing" of small bits of reusable code. That really means 1) being able to trust the originator at least as much as you trust your compiler writer and 2) Well defined and documented interfaces.
What's wrong with this picture? Well, you're not likely to find these two things together unless the coder was doing it for financial gain. There's tons of good code out there, but it's poorly documented. Likewise, there's code that's easy to use, but riddled with bugs. Commercial code is no guarantee of quality, but such code was probably designed from the beginning to be reusable (if it's being sold as a component), so it's much more likely to plug into your app easily than Joe Coder's complex matrix math package that he designed for his own use, no matter how good it might otherwise be.
On any given day, there's probably code that I would happily pay a small sum for to not have to develop myself as long as I could be fairly sure it worked. Example: I had a bunch of data from a sensor I wanted to display in my app. I'd have happily paid $40 or so to get a simple 2D/3D charting object I could plug into C++ and keep going. Even if there was a small license fee attached, because the purchase cost is much less than the cost of my time coding and debugging this thing unless we're going to sell tens of thousands of copies of my application -- which we're not. Instead I had to spend half the day learning enough Microsoft Foundation Classes (MFC) to build the graph myself along with miscellaneous features. My expertise is not MFC and I don't want it to be! I have other much more interesting things to code than charting routines. I could probably find something for sale that cost $299 and had tons of bells and whistles, or I could have embedded an Excel charting object, but I just need a very simple task done.
I'm sure many of the posters here would agree that pride has little to do with it. There are parts of our jobs where our individual skills are required, and some aspects so far removed, that learning that particular skillset is simply too expensive for the current project. This is when you need to be able to buy software components.
I think here on/., we are all denying it like crazy and ludicrously thinking that SCO is suing for *no* reason. Why would a company sue unless it has at least something to base its case on? I mean, i think it is more likely than not, that SCO did find some code which infringes on its
SCO is suing for a reason, but that reason is not related to software or a desire to get infringing code out of public distribution. It's called "Shake 'em and see what falls out." All they need is sufficient cause of action so it doesn't look like a frivolous suit to the court and it works to their advantage. Look at the results: their stock price has increased astromically in such a short time. They have a huge amount of public visibility, and maybe, just maybe, IBM will buy them out. Sure, software types may look down on them but who cares? If you only care about stock price, you need to impress the analysts. Say what you will, but buried in all the stupidity, SCO is doing something right: the price is rising steadily. I'm beginning to think I should have picked up some shares back when it was just around $6 or so.
You forget one obvious exception: those of us who got first degrees when computers were scarce, and returned to school when they were much more prevalent. Before my senior year of college I had never even heard of the internet. By the time I returned for my MS, it was everywhere. No more long lines for registration or payment issues, etc. Access to Lexis-Nexis and many other expensive databases for free (now that I completed my degree, I miss this a lot). Best of all: being able to talk directly to experts in the field: I would email professors across the country who were doing relevant research to ask questions that my own searching couldn't answer. Usenet groups like comp.arch.embedded, etc. The only lines I stood in were at the bookstore and even then I could have ordered most books online, but chose not to.
Dating? The last 10 or so women I've gone out with, including the one I spend my life with, were found online.
The difference is night & day. Night & supernova even.
Radio works by modulating information onto a higher-frequency carrier wave. Let's look at the simplest example: a CW (Morse Code) signal on say, a 14MHz carrier. At a rate of 15wpm, the information is modulated onto the carrier by switching the carrier on and off at just over 1Hz. The receiver, by mixing the signal down to the audio passband (around 1000Hz or so) and allowing extraction of that 1Hz signal, is behaving as a low pass filter.
The point he's making is that EE design for the most part has to work using standard, off the shelf parts. Only in exceptional cases should you use custom parts... like a 0.0001 ohm 200W resistor with a.00001 deg/C temp coefficient. Or if the designer doesn't know that 4.7kohm is a standard value, but designs in 4.8kohm +/-.001%, it's probably going to be expensive to produce the design.
The "optimum" depends on what you're designing for. Manufacturability has to be high on the list of priority for 99% of circuits. Remember the old saying "Any idiot can design the expensive version; it takes experience to make it cheap!"
I know I would have been far more interested in EE if we were building a transister radio or something useful rather than just tinkering with simple low/high/band pass filters and verifying Ohm's Law. Granted these are worthwhile skills, but you don't
Actually, if you think about it, a radio is just a low-pass filter!
What exactly did they say when you asked? Have you made sure that they understand what you want to do? (Create a driver that makes the card work on linux, that anyone can get, potentially increasing the sales for the card). The key is to present the request not as "we need this" but as "you will get this if we can get that".
Speaking as someone who spent a few years working for a manufacturer of data acquisition/control interface hardware, this is not very likely to work. The mfr has a much better idea than the average customer of what the market is like. He has to in order to stay in business. That customer only knows his own needs; the manufacturer hears from many customers across a broad spectrum each day. Now, an intelligent company will try to give you all the information you need to write your own driver so you're appreciative and more willing to deal with them again, but they're not likely to take the time to write and support one just because you say their market share might rise.
I started my career as an EE working in customer support before I was allowed to design anything; I've heard all these issues before many times. In fact, I was surprised at National Instrument's attitude: about 10 years ago I was designing a GPIB interface card and NI very kindly lent us (lent, not leased. Free!) a GPIB bus analyzer for about 6 weeks or so for debugging. Other than a few possible customer referrals, there really wasn't anything in it for them.
Finally, if you're coming out of tech school with an engineering degree or something of the sort, (ie without a significant liberal arts background) now might be a good time to round off your educations with some books about religion, philosophy, economics, politics and business (to name a few). While the subjects might sound drab, you might just find your calling (econ for me).
Second that! One of the best things my undergraduate education did for us engineers was require much more English/Lit classes than just Freshman Comp 101. I just finished The Virgin's Knot last night. It inspired me to Google around to learn enough about weaving to try making a tiny piece of fabric with found cotton twine. Beautiful story. Every human on the planet should read Gabriel Garcia-Marquez's One Hundred Years of Solitude at least once. Someday I will read it in its original Spanish. It's that good! Love in the Time of Cholera is an easier and quicker read, but not quite as good. I also have started The Portable MBA cause I want a better understanding of the aspects of business I don't know about. Cost me $1 at Half-Price books. Marcia Stigum's The Money Market might be a bit dated now (I read it back in '88) but it clarified a lot about how money makes more money. Anyone know an easy introduction to corporate finance for an EE/SW geek?
Recently enjoyed- The Door Into Summer - Heinlein Blue Champagne - Varley Effective Java - good. Although I only use Java for my home projects. Work is C++ Real Time Design Patterns Erotic Nudes (actually this is my gf's book, but the photography is spectacular)
On my radar, or in progress: The Hagakure - Book of the Samurai The Sheltering Sky
More generally: find a library or a good used bookstore and just browse. When you're paying $0.50 for a paperback, you can afford to make a few mistakes in your choices. I also have found a lot of great books -- and great OLD books (I have a few novels from the mid 19th century I paid no more than $1 or so for) at library sales.
Don't be ridiculous. I presently develop software for a medical device. At last check we had around 500kloc running on custom hardware that costs the customer about $80,000 per instrument. What possible benefit could there be to Joe Random Developer seeing the code without having intimate knowledge of the design requirements and the hardware? For that matter, how could he change anything without having the hadware to test it on?
If you want to improve safety, insist on better scrutiny by outside agencies a la CE, UL, CSA, (or in our case, all of the above as well as the US FDA) to certify the software, but saying that Open Source is a solution is just silly.
The simple truth is that management will decide what type of product is shipped. Great engineers with shitty management still equals trouble,
And vice versa. An excellent mananger (yes,/. such things exist on this planet!) won't know the product is bad unless the engineers say so. Although in this particular case management was at fault, it's really dumb to assume all management is stupid. Most mgrs are actually fairly bright and can make intelligent decisions when engineers properly communicate product issues to them. Ultimately the product is the responsibility of the corporation not the individual. If you decide to not say something because historically no one paid any attention to you, then you have just become the weak link in the quality chain.
The solution probably lies along the way of admitting that computers, like all other physical systems, will fail. Now decide what the failure should look like. Sudden catastrophic failure without exceptional input is rare in the real world. e.g., my car is unlikely to explode as I'm driving to work; it's more likely that a small puff of steam out the exhaust will signal a slowly failing head gasket and give me time to get it fixed if I notice the problem. Likewise, computers should fail slowly. Got a divide-by-zero exception? Return to an earlier saved, safe state with some data loss since then. Or perhaps turn that divide by zero into a divide by a really small number and signal an error. These changes have to happen not just in software, but will require new fault-tolerant hardware architectures to implement them.
Other mechanisms that come to mind are parallel execution paths where if one path fails, it alerts the other one not to take that particular branch, or to look ahead for a "safe" path.
What I'm getting at is that in order to write failure-proof software, we'll need entire new methodologies of development, and perhaps new hardware. Right now our recovery mechanisms are limited to such things as watchdog timers or rollback/redo logs. Primitive, but it's a start.
I don't know about some of these development teams they are talking about but around here, you don't just implement "ideas" you might have had over the weekend. "Hey! Wouldn't it be cool if it did this... !" If it's not a requirement, it doesn't go in.
That's what I used to think. Then I began to see that my work environment is much more disciplined than most. I've been asked to remove "cool ideas" even though they worked, because they were not in the requirements document and therefore the testers would not write test cases for them. The more I learn about how business software is written, the more I wonder that it works at all. I mean, we're in a development phase (just before release) where changing a single line of code has to first be justified before the change and then peer reviewed after check in and paperwork filed, etc.
After a long day I sometimes wish we could shortcut the process, but then I remember what defects in the end product could mean and just sigh and suck it up.
Good points. But some of us (the cheaper ones, anyway:-) just buy art because we like it. The most expensive artwork I own is a handpainted dish I paid $100 for a few years ago. I do have a framed litho from an artist whose work generally costs more than I would pay, but that was free, as we took a class together and one requirement was that students distribute their prints to each other (I got the better part of that deal by far). If you want good art, but don't want to pay high prices, go to art school sales or local art fairs. You can pick up amazing stuff at bargain basement prices. That's where most of my other stuff comes from... that, and making friends with budding artists who like to give stuff away.
Taking a class at a local art school is an excellent experience for those of us in very technical professions. Being surrounded by artists is a great place to be, creativity for creativity's sake. If I could figure out how to get my employer to pay for it, I'd jump at the chance to get an MFA in a second! I have tons of ideas that use technology -- mostly kinetic sculpture -- as artistic expression but can't find time to complete.
I would guess that the people wise enough to use your service will be the ones that can do all of this themselves.
Not always. Consider this: I can build simple dynamic websites, and I understand the need for proper testing. But not having any background in the subject, unless I wanted to spend a lot of time learning about common problems/failure modes, I wouldn't have much of a clue where to start/how to write test cases, etc. And I probably wouldn't be motivated to spend the time learning about those types of quality issues as its time I could best spend elsewhere. You could argue that in this case I'd be better served by farming the entire job out to someone who could do it all themselves, but I'm considering the case where a small operator knows enough to build a usable site, but can't afford to hire a developer to do all the work.
Frex: a few years ago I had a small company I ran out of a spare bedroom. I wrote my own ad copy/submitted product to trade rags hoping for reviews, cause I could do a good enough job writing copy (had experience doing it at an earlier job), but couldn't justify an advertising consultant. But I wasn't a professional photographer, so I would hire one to take photos of the products I wanted to feature. $100 and I get a MUCH better job than I could have done myself.
Also, clicking the "Checkout" button gives warnings about the secutity certificate not matching the host name which might put some people off.
These two posts make me wonder: is there a market for "homebrewed website" quality checking? Say, a person with a lot of web development experience (definitely not me: I'm an embedded head:-) who accepts a small fee for testing your site just before you publish it to the world and finds bugs/makes usability suggestions/stress tests it. Anyone want to take this idea and run with it?
..since I realize how quickly fortunes change, I have been making "what if" plans. Financially, I can survive an extended period of unemployment. I was always big on saving & investing and don't spend much of what I make. But I have been considering many business ideas. I spend some time each day thinking and researching niches that a small 1-2 person company can enter and prosper in. Both for side projects I can do now, and things that could support me full time should I become unemployed. We are all going to be on the unemployment line at some time in our careers (I've been there once), why wait for it to happen? Plan now!
What does a store gain from having a small web site? I think that a web site for a small shop will not do any good unless the costumers can find it in google when they are searching for the products directly, and the site has, at least, descriptions, photos and prices of the items to be sold.
OK, here's one. There's a pasta restaurant/store I pass quite frequently. I've heard they're good and they do takeout, but I either never have time to stop in and get a menu (we tend to eat late and it takes forever to decide what to get), or I don't see convenient parking, so I keep going. One day it occurs to me to lookup their name in Google, and lo & behold, they have a website listing their menus, daily specials, hours, etc. Suddenly I have a menu I can browse at my leisure and the restaurant now has a new customer who knows how good their food is. At a bare minimum, I think all restaurants could benefit with a static page that had their hours of operation and a menu. If they added a dynamic update of the waiting time, so much the better.
That's cool. I pretty much dropped all interest in the Motorola MCUs that I used before once I started with the Atmel ones. They are amazing little processors.
I agree with the guy who said that equating hardware hacking with case modding is like equating spoilers on a Civic with real car rebuilding.
I haven't been up all night with my AVRs cause I need the sleep:-) But I'm always interested when someone mentions solenoid valves. What's the project?
If humans like freedom, then why is Microsoft so popular? I think humans, in general, prefer convenience over freedom.
Microsoft is popular because it gives most users freedom. But their definition of "freedom" is obviously different from yours. The average (home) user will communicate with email, browse the Web or buy stuff online, write letters, do taxes and manage finances and play games. That's pretty much it, and Windows gives sufficient options in those areas that its users are satisfied. You are right: humans generally prefer convenience over freedom (until the freedom is taken away:-), but in this case I don't think they have to make a choice.
Hackers are a different breed. I became interested in Linux back in 95-96 or so when I heard about Slackware. I remember waiting til 1-2am to download the disks at 14.4kbps because there was less traffic at that time. I could only do one a night cause I had to go to work in the morning. Then the struggles to get my modem working (took a while to figure out that it didn't like my WinModem:-)I liked Linux because I could get development tools easily and things just seemed to "fit together" better than the Microsoft stuff I was using at the time (WFWG). As a user, I was quite happy with Microsoft & Quicken stuff. As a developer, I wanted to use Linux more and more to the point now where my primary OS for the last 3 years has been some RedHat flavor of Linux.
I'm planning to "give something back" by building a "Linux hardware hacker" site describing some stuff I've done and how to get started doing real-world data acquisition and control under Linux using homebuilt hardware (reading temperatures, controlling lights & motors, etc). Having a day job where I get paid to do similar things with Windows, I know that Linux is much simpler for these purposes. But more to the point, if I were doing my experimental stuff under Windows, I'd have never thought of making it public so others could build on what I've done. There is an entirely different mindset and I have seen the change in myself as I moved from using one OS to the other.
Don't be too creeped out. I work with someone who built a computer-controlled exercise bike about 15 years ago (actually a trainer for regular bike!). It's an idea that crops up from time to time among hardware hackers.
No software will be perfect, therefore, no software will sell itself. Unless your university plans on making the financial risk of targeted advertising, the money just won't come. WTF? Lots of software sells itself simply by filling a void
I think what he means is that if you don't tell other people that the software exists (i.e., marketing), no one can purchase it.
My approach to starting a business has always been: 1. Find a group of potential customers I can reach. 2. Decide what product I can sell them. 3. (hopefully) Profit!!!
Reverse the order of 2&3 and you can find you've wasted a lot of time on something you can't sell because the cost of reaching the market is greater than the production cost of the product.
Because there's a point for many people (not all) where paying a reasonable fee for a 'legit', reliably-good datafile, is much more convenient than spending the time and effort to sift through multiple p2p networks full of unknowns.
That is something that many people forget or don't realize. This is exactly what I would want from the music companies. I'm not a starving student anymore: I have plenty of money to spend on CDs, but I want to be able to conveniently preview what I hear so the money isn't wasted. My time is valuable, so I don't want to spend it looking for poorly recorded crap on Kazaa. I've noticed that most of the CDs I purchased recently were because I heard songs on the soundtrack of one movie or another. I rarely listen to the radio other than NPR (that's how I discovered India Arie before she was popular), and MTV etc is just a waste of time.
The RIAA and their ilk should just forget about the people who can't afford to buy CDs. Trying to stop piracy from that quarter is a waste of their time and we all know it's a losing battle. When I was a student with no money, it would have been no big deal to spend my free time amassing thousands of tracks online. Now I just want to plunk down my cash and play a CD with no effort. I'm in the market they should be trying to serve. Instead, the more I find out about their tactics, the more I want to just buy CDs directly from the artists and bypass them altogether.
We are unfortunate enough to have a phone number 1 digit out from a major TV rental shop. On a regular basis we get the following call:
So use it to your advantage. Do you know any competing rental shops that would give you a referral bonus for sending potential customers their way? Do you tell the people calling what services your company offers? Some of them might turn into customers for you. Think like a businessperson: use every contact to your advantage.
Slashdot Mirror
It's called taking pride in your work. I'm paid to develop software.That doesn't just mean it compiles and runs, but that it's readable and (hopefully) extensible by the people to come later.
I try to code well enough that I could be hit by a truck tomorrow and the next guy could easily pick it up. Has the side effect of me being able to read code I wrote 2 years ago, which is great as I'm on a very long term project.
The common thread I'm seeing here is that we probably need more "black boxing" of small bits of reusable code. That really means 1) being able to trust the originator at least as much as you trust your compiler writer and
2) Well defined and documented interfaces.
What's wrong with this picture? Well, you're not likely to find these two things together unless the coder was doing it for financial gain. There's tons of good code out there, but it's poorly documented. Likewise, there's code that's easy to use, but riddled with bugs. Commercial code is no guarantee of quality, but such code was probably designed from the beginning to be reusable (if it's being sold as a component), so it's much more likely to plug into your app easily than Joe Coder's complex matrix math package that he designed for his own use, no matter how good it might otherwise be.
On any given day, there's probably code that I would happily pay a small sum for to not have to develop myself as long as I could be fairly sure it worked.
Example: I had a bunch of data from a sensor I wanted to display in my app. I'd have happily paid $40 or so to get a simple 2D/3D charting object I could plug into C++ and keep going. Even if there was a small license fee attached, because the purchase cost is much less than the cost of my time coding and debugging this thing unless we're going to sell tens of thousands of copies of my application -- which we're not. Instead I had to spend half the day learning enough Microsoft Foundation Classes (MFC) to build the graph myself along with miscellaneous features. My expertise is not MFC and I don't want it to be! I have other much more interesting things to code than charting routines. I could probably find something for sale that cost $299 and had tons of bells and whistles, or I could have embedded an Excel charting object, but I just need a very simple task done.
I'm sure many of the posters here would agree that pride has little to do with it. There are parts of our jobs where our individual skills are required, and some aspects so far removed, that learning that particular skillset is simply too expensive for the current project. This is when you need to be able to buy software components.
SCO is suing for a reason, but that reason is not related to software or a desire to get infringing code out of public distribution. It's called "Shake 'em and see what falls out." All they need is sufficient cause of action so it doesn't look like a frivolous suit to the court and it works to their advantage. Look at the results: their stock price has increased astromically in such a short time. They have a huge amount of public visibility, and maybe, just maybe, IBM will buy them out. Sure, software types may look down on them but who cares? If you only care about stock price, you need to impress the analysts. Say what you will, but buried in all the stupidity, SCO is doing something right: the price is rising steadily.
I'm beginning to think I should have picked up some shares back when it was just around $6 or so.
You forget one obvious exception: those of us who got first degrees when computers were scarce, and returned to school when they were much more prevalent.
Before my senior year of college I had never even heard of the internet. By the time I returned for my MS, it was everywhere. No more long lines for registration or payment issues, etc. Access to Lexis-Nexis and many other expensive databases for free (now that I completed my degree, I miss this a lot). Best of all: being able to talk directly to experts in the field: I would email professors across the country who were doing relevant research to ask questions that my own searching couldn't answer. Usenet groups like comp.arch.embedded, etc.
The only lines I stood in were at the bookstore and even then I could have ordered most books online, but chose not to.
Dating? The last 10 or so women I've gone out with, including the one I spend my life with, were found online.
The difference is night & day. Night & supernova even.
Radio works by modulating information onto a higher-frequency carrier wave. Let's look at the simplest example: a CW (Morse Code) signal on say, a 14MHz carrier. At a rate of 15wpm, the information is modulated onto the carrier by switching the carrier on and off at just over 1Hz. The receiver, by mixing the signal down to the audio passband (around 1000Hz or so) and allowing extraction of that 1Hz signal, is behaving as a low pass filter.
The point he's making is that EE design for the most part has to work using standard, off the shelf parts. Only in exceptional cases should you use custom parts... like a 0.0001 ohm 200W resistor with a .00001 deg/C temp coefficient. Or if the designer doesn't know that 4.7kohm is a standard value, but designs in 4.8kohm +/- .001%, it's probably going to be expensive to produce the design.
The "optimum" depends on what you're designing for. Manufacturability has to be high on the list of priority for 99% of circuits. Remember the old saying "Any idiot can design the expensive version; it takes experience to make it cheap!"
Actually, if you think about it, a radio is just a low-pass filter!
Speaking as someone who spent a few years working for a manufacturer of data acquisition/control interface hardware, this is not very likely to work. The mfr has a much better idea than the average customer of what the market is like. He has to in order to stay in business. That customer only knows his own needs; the manufacturer hears from many customers across a broad spectrum each day. Now, an intelligent company will try to give you all the information you need to write your own driver so you're appreciative and more willing to deal with them again, but they're not likely to take the time to write and support one just because you say their market share might rise.
I started my career as an EE working in customer support before I was allowed to design anything; I've heard all these issues before many times. In fact, I was surprised at National Instrument's attitude: about 10 years ago I was designing a GPIB interface card and NI very kindly lent us (lent, not leased. Free!) a GPIB bus analyzer for about 6 weeks or so for debugging. Other than a few possible customer referrals, there really wasn't anything in it for them.
Second that! One of the best things my undergraduate education did for us engineers was require much more English/Lit classes than just Freshman Comp 101.
I just finished The Virgin's Knot last night. It inspired me to Google around to learn enough about weaving to try making a tiny piece of fabric with found cotton twine. Beautiful story.
Every human on the planet should read Gabriel Garcia-Marquez's One Hundred Years of Solitude at least once. Someday I will read it in its original Spanish. It's that good! Love in the Time of Cholera is an easier and quicker read, but not quite as good.
I also have started The Portable MBA cause I want a better understanding of the aspects of business I don't know about. Cost me $1 at Half-Price books.
Marcia Stigum's The Money Market might be a bit dated now (I read it back in '88) but it clarified a lot about how money makes more money. Anyone know an easy introduction to corporate finance for an EE/SW geek?
Recently enjoyed-
The Door Into Summer - Heinlein
Blue Champagne - Varley
Effective Java - good. Although I only use Java for my home projects. Work is C++
Real Time Design Patterns
Erotic Nudes (actually this is my gf's book, but the photography is spectacular)
On my radar, or in progress:
The Hagakure - Book of the Samurai
The Sheltering Sky
More generally: find a library or a good used bookstore and just browse. When you're paying $0.50 for a paperback, you can afford to make a few mistakes in your choices. I also have found a lot of great books -- and great OLD books (I have a few novels from the mid 19th century I paid no more than $1 or so for) at library sales.
Don't be ridiculous.
I presently develop software for a medical device. At last check we had around 500kloc running on custom hardware that costs the customer about $80,000 per instrument. What possible benefit could there be to Joe Random Developer seeing the code without having intimate knowledge of the design requirements and the hardware? For that matter, how could he change anything without having the hadware to test it on?
If you want to improve safety, insist on better scrutiny by outside agencies a la CE, UL, CSA, (or in our case, all of the above as well as the US FDA) to certify the software, but saying that Open Source is a solution is just silly.
And vice versa. An excellent mananger (yes,
The solution probably lies along the way of admitting that computers, like all other physical systems, will fail. Now decide what the failure should look like. Sudden catastrophic failure without exceptional input is rare in the real world. e.g., my car is unlikely to explode as I'm driving to work; it's more likely that a small puff of steam out the exhaust will signal a slowly failing head gasket and give me time to get it fixed if I notice the problem. Likewise, computers should fail slowly. Got a divide-by-zero exception? Return to an earlier saved, safe state with some data loss since then. Or perhaps turn that divide by zero into a divide by a really small number and signal an error.
These changes have to happen not just in software, but will require new fault-tolerant hardware architectures to implement them.
Other mechanisms that come to mind are parallel execution paths where if one path fails, it alerts the other one not to take that particular branch, or to look ahead for a "safe" path.
What I'm getting at is that in order to write failure-proof software, we'll need entire new methodologies of development, and perhaps new hardware. Right now our recovery mechanisms are limited to such things as watchdog timers or rollback/redo logs. Primitive, but it's a start.
That's what I used to think. Then I began to see that my work environment is much more disciplined than most. I've been asked to remove "cool ideas" even though they worked, because they were not in the requirements document and therefore the testers would not write test cases for them. The more I learn about how business software is written, the more I wonder that it works at all.
I mean, we're in a development phase (just before release) where changing a single line of code has to first be justified before the change and then peer reviewed after check in and paperwork filed, etc.
After a long day I sometimes wish we could shortcut the process, but then I remember what defects in the end product could mean and just sigh and suck it up.
Good points. But some of us (the cheaper ones, anyway :-) just buy art because we like it. The most expensive artwork I own is a handpainted dish I paid $100 for a few years ago. I do have a framed litho from an artist whose work generally costs more than I would pay, but that was free, as we took a class together and one requirement was that students distribute their prints to each other (I got the better part of that deal by far).
If you want good art, but don't want to pay high prices, go to art school sales or local art fairs. You can pick up amazing stuff at bargain basement prices. That's where most of my other stuff comes from... that, and making friends with budding artists who like to give stuff away.
Taking a class at a local art school is an excellent experience for those of us in very technical professions. Being surrounded by artists is a great place to be, creativity for creativity's sake. If I could figure out how to get my employer to pay for it, I'd jump at the chance to get an MFA in a second! I have tons of ideas that use technology -- mostly kinetic sculpture -- as artistic expression but can't find time to complete.
Not always. Consider this: I can build simple dynamic websites, and I understand the need for proper testing. But not having any background in the subject, unless I wanted to spend a lot of time learning about common problems/failure modes, I wouldn't have much of a clue where to start/how to write test cases, etc. And I probably wouldn't be motivated to spend the time learning about those types of quality issues as its time I could best spend elsewhere.
You could argue that in this case I'd be better served by farming the entire job out to someone who could do it all themselves, but I'm considering the case where a small operator knows enough to build a usable site, but can't afford to hire a developer to do all the work.
Frex: a few years ago I had a small company I ran out of a spare bedroom. I wrote my own ad copy/submitted product to trade rags hoping for reviews, cause I could do a good enough job writing copy (had experience doing it at an earlier job), but couldn't justify an advertising consultant. But I wasn't a professional photographer, so I would hire one to take photos of the products I wanted to feature. $100 and I get a MUCH better job than I could have done myself.
That's the kind of thing I'm thinking of.
These two posts make me wonder: is there a market for "homebrewed website" quality checking? Say, a person with a lot of web development experience (definitely not me: I'm an embedded head
..since I realize how quickly fortunes change, I have been making "what if" plans. Financially, I can survive an extended period of unemployment. I was always big on saving & investing and don't spend much of what I make. But I have been considering many business ideas. I spend some time each day thinking and researching niches that a small 1-2 person company can enter and prosper in. Both for side projects I can do now, and things that could support me full time should I become unemployed.
We are all going to be on the unemployment line at some time in our careers (I've been there once), why wait for it to happen? Plan now!
OK, here's one. There's a pasta restaurant/store I pass quite frequently. I've heard they're good and they do takeout, but I either never have time to stop in and get a menu (we tend to eat late and it takes forever to decide what to get), or I don't see convenient parking, so I keep going. One day it occurs to me to lookup their name in Google, and lo & behold, they have a website listing their menus, daily specials, hours, etc. Suddenly I have a menu I can browse at my leisure and the restaurant now has a new customer who knows how good their food is.
At a bare minimum, I think all restaurants could benefit with a static page that had their hours of operation and a menu. If they added a dynamic update of the waiting time, so much the better.
I haven't been up all night with my AVRs cause I need the sleep :-) But I'm always interested when someone mentions solenoid valves. What's the project?
Microsoft is popular because it gives most users freedom. But their definition of "freedom" is obviously different from yours. The average (home) user will communicate with email, browse the Web or buy stuff online, write letters, do taxes and manage finances and play games. That's pretty much it, and Windows gives sufficient options in those areas that its users are satisfied. You are right: humans generally prefer convenience over freedom (until the freedom is taken away
Hackers are a different breed. I became interested in Linux back in 95-96 or so when I heard about Slackware. I remember waiting til 1-2am to download the disks at 14.4kbps because there was less traffic at that time. I could only do one a night cause I had to go to work in the morning. Then the struggles to get my modem working (took a while to figure out that it didn't like my WinModem
I'm planning to "give something back" by building a "Linux hardware hacker" site describing some stuff I've done and how to get started doing real-world data acquisition and control under Linux using homebuilt hardware (reading temperatures, controlling lights & motors, etc). Having a day job where I get paid to do similar things with Windows, I know that Linux is much simpler for these purposes. But more to the point, if I were doing my experimental stuff under Windows, I'd have never thought of making it public so others could build on what I've done. There is an entirely different mindset and I have seen the change in myself as I moved from using one OS to the other.
Don't be too creeped out. I work with someone who built a computer-controlled exercise bike about 15 years ago (actually a trainer for regular bike!). It's an idea that crops up from time to time among hardware hackers.
I think what he means is that if you don't tell other people that the software exists (i.e., marketing), no one can purchase it.
My approach to starting a business has always been:
1. Find a group of potential customers I can reach.
2. Decide what product I can sell them.
3. (hopefully) Profit!!!
Reverse the order of 2&3 and you can find you've wasted a lot of time on something you can't sell because the cost of reaching the market is greater than the production cost of the product.
That is something that many people forget or don't realize. This is exactly what I would want from the music companies. I'm not a starving student anymore: I have plenty of money to spend on CDs, but I want to be able to conveniently preview what I hear so the money isn't wasted. My time is valuable, so I don't want to spend it looking for poorly recorded crap on Kazaa. I've noticed that most of the CDs I purchased recently were because I heard songs on the soundtrack of one movie or another. I rarely listen to the radio other than NPR (that's how I discovered India Arie before she was popular), and MTV etc is just a waste of time.
The RIAA and their ilk should just forget about the people who can't afford to buy CDs. Trying to stop piracy from that quarter is a waste of their time and we all know it's a losing battle. When I was a student with no money, it would have been no big deal to spend my free time amassing thousands of tracks online. Now I just want to plunk down my cash and play a CD with no effort. I'm in the market they should be trying to serve. Instead, the more I find out about their tactics, the more I want to just buy CDs directly from the artists and bypass them altogether.
So use it to your advantage. Do you know any competing rental shops that would give you a referral bonus for sending potential customers their way? Do you tell the people calling what services your company offers? Some of them might turn into customers for you.
Think like a businessperson: use every contact to your advantage.