You know, it was that logic that made me go for the degree in the first place. It was arguably the biggest mistake of my life. If I had gotten a job back in the.com boom (and I had no shortage of offers), then I would now be financially secure, and when the government said that I would have to pay my own medical treatment, I wouldn't have had to scramble to find the first job I could without being able to finish the degree. It probably would have even given me the reserves to go back to get a degree again later, without piling up over $20,000 in debt.
So now, thanks to that logic, I'm worse than broke, barely making my medical bills and the student loan people assure me that no, even if I go bankrupt (which is looking more and more likely at this point, regardless of my feelings), I'll have to pay back that $20,000 anyway, and I won't ever get a chance to get back on my feet.
Thankfully, "most of" *DOES* count. So does my track record. My abilities speak for themselves, as does my portfolio (which includes things like compilers, 3D renderers, and the like), and now I'm starting to get real job offers. Sure, I don't expect to find a job in hardware design (although I've had a pleasant surprise lately in that area...), but in the software world, degrees only matter if you are applying through the HR department. Applying through people who understand the realities of the industry actually works out. It's now only a matter of improving my health and waiting.
Moncton is just where I got stuck when I took ill. As soon as I'm in good health again, I would like to move, but getting something that will pay relocation isn't as easy as it sounds (and after all the medical bills and missed time, I have no savings to use, and what I make will take a long time to move on).
Well, actually, I have way more than 5 years even of paid experience, and a total of nearly 17 when including non-paid. If there were real jobs here, I would have one.
There are a few major obstacles for me and my degree though. I'm in a really bad financial situation, thanks in part to some really hefty medical expenses. More than this, I had a lot of trouble in the classroom environment taking more than a few courses, thanks to being autistic (it doesn't really affect me much in the work world thankfully).
As for the CS/CompEng seperation, there is now a faculty of Computer Science that offers the CS degree, the ECE (Electrical and Computer Engineering) department offers Computer Engineering, and I seem to recall there is something closer to MIS as well, although I never really looked into it.
Actually, a few things, "for the record". It's a degree in Computer ENGINEERING, not Computer Science. The two curriculums are quite different. I don't imagine the average CS degree teaches the finer points of analog filter design, state machine theory or DSP.
Second, and perhaps more importantly... the curriculum is also in parallel threads. I managed to complete the software and VLSI threads. I can back that up easily enough. And I have a high GPA behind it as well.
I don't see a degree being much of an asset beyond a certain point, though. Once I'm properly established, I'll just start my business. *I* certainly don't care whether or not I have a degree.
Fine... this isn't a job posting, this is an actual description of my job duties, but I'll make it look like one:
Senior SQL Administrator/Programmer (PostgreSQL)
Senior Web Applications Developer (PHP). E-Commerce experience an asset.
Toolchain/Plugin Developer (C/C++). Develop PostgreSQL plugins and in-house applications.
Senior Network Administrator on a heterogeneous FreeBSD/Linux/WindowsXP environment. Must possess strong skills in server application deployment and windows interoperability. Must possess a background in firewall and network design.
BCS/BEE and/or 10+ years of proven software design experience.
Background in cryptography is an asset.
Must be willing to work overtime when necessary at 1x pay.
Benefits Package: none, contract basis, terminatable at any time without severance package
Pay: $14 Canadian/hour
Wait a minute... what am I doing? Is anyone looking for a developer (or hardware engineer for that matter)?
Most of what I mentioned is barely CPU bound, although even my home machine is multiprocessor. It's not so much a matter of budget as it is of flexibility. Doing development with non-standard environments (say, developing for a system with 3 different types of CPU's and an FPGA (which probably contains one of the CPU's)) requires a fair bit of scripting.
As for context-sensitive autocomplete, there is a script or two do do just that in Vim (notice I didn't say vi, but Vim). Vim has plenty of other nice things, like syntax coloring, folding, autoindenting, and all the things you would expect in a programmer's editor, with all the speed of the vi interface.
Notice that I said that Linux is a workstation OS, not a desktop OS. Your list of tasks is far shorter than mine. I do heavy development on everything from kernel-level and embedded systems to web applications to hardware designs. I also run quite a bit of server software on the other machines around here.
In your case, as a desktop user, or perhaps even a desktop poweruser, Windows is a pretty logical choice. Engineers like myself need something more. It really is that simple.
It's like notepad versus Vim. Notepad is way easier to use, you can pretty much just start it up and start typing. It takes minutes to learn. Vim takes *YEARS* of practice to fully master the subtleties of it, but no notepad user can edit text as rapidly as I can. If you don't do much more than write an occasional post-it to yourself, or a quick letter or something, notepad or its ilk is a better fit for what you do. If you do heavy editing of syntactic text files (like sourcecode or HDL), where saving a couple keystrokes can really matter, Vim is well worth the time required to learn.
I respectfully disagree. Windows's interface is not the slightest bit scriptable, and while easy to use if you don't know it, is *really* difficult to use for those of us who do. It took me nearly 20 minutes to get Windows XP to connect to my network (which has a slightly unusual configuration that the "wizards" didn't expect), and 6 lines (20 seconds) at a command-line to get Linux on. Convincing XP to let me alter my own IP address was a chore in and of itself, while Linux merrily lets me make virtual network devices, bridge segments, set up firewalling the way I want it, etc.
I don't argue that Windows is a better desktop OS. It is. Windows thinks it's smarter than you. When this is true, it's a good tradeoff. But I'm not a desktop user. In the workstation (as opposed to desktop) environment, Linux wins out with room to spare.
Not too late to short some stock. You know their stock will take a major nosedive before too long, and it's near peak now. The more it goes down, the more money you make.
If I were involved in stock trading, I'd do this myself. Good way to make a pile of money.
Photovoltaic solar panels are not now and never will be viable as a source of energy, but this doesn't rule out solar power entirely. A heliostat (an array of mirrors computer-aimed at a central target) is much more efficient in terms of the ROI energy produced, and usually in general as well. A well-tuned heliostat produces nice, clean energy all day long (not just when the sun is up... it melts sodium and uses this to fuel turbines even at night). The total cost per kilowatt-hour is much, much lower than any other solar method.
Actually, you missed something. The chipset they are using is designed specifically for low-speed applications (like attaching to 8-bit microcontrollers), and does enough of the link-level work by itself that the main CPU doesn't have to. The card will buffer (in internal memory) the incoming packets, screen out any incorrect MAC addresses (unless you are in promisc mode), and even allow the system to access the packets byte by byte, rather than doing a bulk copy.
While the effective bulk transfer rate would be rather limited, it is quite possible that this system could survive on a saturated network. 500k/s of total network traffic isn't much, but 500k/s of traffic devoted to that machine is quite substantial. It might even be able to provide services that have very low bandwidth requirements.
I have a box of arcnet hardware, and an arcnet card is still available in my firewall machine should I ever need to connect something arcnet-enabled. It's 1MBit, but it has the advantage that you can run a cable up to a kilometer long between cards (not exactly a supported configuration, but it can be done) over regular TV coax. If you ever need to solve the "last-mile" problem, arcnet is a cheap way to do it.
You probably mean the 3c509 (which was a really nice 16-bit ISA card), not the 3c905 (which is a PCI 10/100 card). I have a couple 3c509's in my router box, they are very nice cards.
All very good reasons. Those of us who do a lot of embedded development also do it for learning purposes. An old 8-bit machine from the '80s is a lot easier to debug than a modern 8-bit microcontroller (more interactivity, easier to place probes in the hardware, etc), while simultaneously having the same limitations.
Chances are, if you can cram contiki onto an Atari or C64, you can get a TCP/IP stack into an embedded controller of some sort.
Yes, you are absolutely right, uploading a bad firmware to a drive would quite likely cause damage. The point, though, is that no such firmware upload is being performed.
Unlike things like wireless networking cards, the firmware of a drive is flashed into the device itself, not stored in volatile memory. Unless you deliberately set out to overwrite it, it should remain untouched and gracefully accept whatever commands you send it (anything short of a "write firmware" command, which typically would require some sort of confirmation string ('AA 55' is a common string in the industry because it's so hard to accidentally do).
This firmware still has to interpret commands from the computer, *INCLUDING* the write firmware command. Usually, the firmware will have a dispatch table written into non-writable memory which will handle updating. Some firmwares will even have multiple slots with cryptographic checksums to ensure validity, reverting to a known-good firmware upon failure (most good motherboard BIOS will have this feature).
You don't need much CPU to compute a simple checksum (like MD5 or the like). Any 8-bit micro could do the job. The only requirement would be to make a page or two of the firmware non-writable (which is supported in a lot of modern flash-rams and flash-programmable microcontrollers) to do the actual checksumming algorithm.
I've never heard of an open-source CDROM drive, so the complaints about this algorithm being proprietary are mostly irrelevant. The only issue is hostile OEM firmware (DRM or other misfeatures), which is a case for market pressure (if I found that my drive had been crippled by an OEM firmware, I'd demand a refund).
I think it's much more likely that they could sue LG. Mandrake was merely distributing an enhancement for their CDROM support that would work on virtually any CDROM drive outside of this range. This would classify that use as "normal use".
A product that dies during normal use is a problem for the manufacturer. LG should just recall them before this becomes a fiasco.
IDE = Intelligent Drive Electronics. ATAPI = ATA Packet Interface (basically SCSI transport over IDE). The point here is that control of something like the physical head position is entirely in the domain of the onboard microcontroller and not the host computer.
So if the host machine asks for sector X, but sector X doesn't exist, the drive electronics should simply report "sector not found" and return an error, not crash the head.
Digital cryptographic signatures work pretty well. If the hardware will only function with a signed firmware, then it's unlikely you will do much besides temporarily disable the drive, requiring a reflash.
All the parts you mentioned used to be sold at Radio Shack, and I even have a few on hand.
They used to sell keylock switches for security systems. I suspect they still carry them. This would just be a straight replacement of the power switch with another switch. Simple enough. You could also pilfer these out of old machines that used to have a keyboard lock.
As for the little sound device, the chip you are looking for is the ISD1000A (or, since it's now out of production, one of the newer versions of the same line). Radio Shack used to sell this chip a while back. It's a 20-second audio record/playback IC. Just flash in your 20 second clip of a car starting, and wire it up to trigger when you boot (tie the trigger line to the/reset line on the ISA bus and you will be fine).
A half-dozen stepper motors, a couple rollers, and a microcontroller to keep it all in sync. Rather like the paper feeder mechanism on any modern printer (which you can easily cannibalize for this, if you wish, although you will probably have to rebuild the logic board). Stack in some bills into a hidden compartment, and voila!
Then you could always replace the light-gap cards with a real magstripe, and add a reader head and a couple op-amps. Tie it to the microcontroller, and away you go!
Punchcards for authentication... why didn't I think of that? I've been spending all my time tinkering with microcontrollers and iButtons when I have a perfectly good RS232C-interface punch card reader sitting here gathering dust. I mean, a punchcard stores a whopping 80 bytes, which is quite sizable for a password or encryption key!
A cooler way would be to use one of Dallas Semiconductor's iButton devices as a key for your computer. An iButton is a little ROM chip in a sealed metal can that looks rather like a watch battery. You snap it into one of the holders, and it'll easily fit on your keyring (it's comfortably small too). Some iButtons even have secure facilities like SHA1 hash algorithms, onboard realtime clocks, and the like.
I'm planning to make a device based on a PIC12F675 8-pin microcontroller to do just this. I'm even considering adding a hardware watchdog timer, for good measure. For the less hardware-inclined, they sell serial and parallel-port cables that allow software to do the work instead.
The embedded readers are small enough that they could easily be mounted on a peripheral (a monitor or keyboard), or possibly even fitted into the case of a laptop. The embedded readers require dedicated hardware, but this can be a single 12-pin SSOP, taking a negligable amount of space.
The best part is that even the bare ROM version of the iButton has a 64-bit, factory-lasered, guaranteed-unique key.
Granted, a lot of these projects are designed by amateurs and are drastically unprofessional (this one, for instance, could use a lot of help). But sometimes these "amateurs" also happen to be professionals as well, in which case the plans tend to be solid. Often, the "cheap knockoffs" end up with more features than the commercial products in such cases.
It's also false that it costs more to DIY. Sure, there is an initial investment in tools and equipment. The time required is heavily dependant on the skill of the builder and the quality of the tools. If you happen to have mass production equipment (like I once did), you can make prototypes relatively quickly, and even do small production runs. And so long as you already have the tools required and a few good suppliers, it's often much, much cheaper to make your own.
And besides, you are missing how much *FUN* it is to build stuff. You know how Linux was written by a bunch of people for fun? Hardware is really no different, except that it has some sort of mystique that scares off a lot of people.
Slashdot Mirror
So now, thanks to that logic, I'm worse than broke, barely making my medical bills and the student loan people assure me that no, even if I go bankrupt (which is looking more and more likely at this point, regardless of my feelings), I'll have to pay back that $20,000 anyway, and I won't ever get a chance to get back on my feet.
Thankfully, "most of" *DOES* count. So does my track record. My abilities speak for themselves, as does my portfolio (which includes things like compilers, 3D renderers, and the like), and now I'm starting to get real job offers. Sure, I don't expect to find a job in hardware design (although I've had a pleasant surprise lately in that area...), but in the software world, degrees only matter if you are applying through the HR department. Applying through people who understand the realities of the industry actually works out. It's now only a matter of improving my health and waiting.
Moncton is just where I got stuck when I took ill. As soon as I'm in good health again, I would like to move, but getting something that will pay relocation isn't as easy as it sounds (and after all the medical bills and missed time, I have no savings to use, and what I make will take a long time to move on).
There are a few major obstacles for me and my degree though. I'm in a really bad financial situation, thanks in part to some really hefty medical expenses. More than this, I had a lot of trouble in the classroom environment taking more than a few courses, thanks to being autistic (it doesn't really affect me much in the work world thankfully).
As for the CS/CompEng seperation, there is now a faculty of Computer Science that offers the CS degree, the ECE (Electrical and Computer Engineering) department offers Computer Engineering, and I seem to recall there is something closer to MIS as well, although I never really looked into it.
Second, and perhaps more importantly... the curriculum is also in parallel threads. I managed to complete the software and VLSI threads. I can back that up easily enough. And I have a high GPA behind it as well.
I don't see a degree being much of an asset beyond a certain point, though. Once I'm properly established, I'll just start my business. *I* certainly don't care whether or not I have a degree.
No, I don't have both... I have most of a BScEE(CompEng) and well over 10 years programming experience. However, I live in Moncton. Need I say more?
Benefits Package: none, contract basis, terminatable at any time without severance package
Pay: $14 Canadian/hour
Wait a minute... what am I doing? Is anyone looking for a developer (or hardware engineer for that matter)?
As for context-sensitive autocomplete, there is a script or two do do just that in Vim (notice I didn't say vi, but Vim). Vim has plenty of other nice things, like syntax coloring, folding, autoindenting, and all the things you would expect in a programmer's editor, with all the speed of the vi interface.
In your case, as a desktop user, or perhaps even a desktop poweruser, Windows is a pretty logical choice. Engineers like myself need something more. It really is that simple.
It's like notepad versus Vim. Notepad is way easier to use, you can pretty much just start it up and start typing. It takes minutes to learn. Vim takes *YEARS* of practice to fully master the subtleties of it, but no notepad user can edit text as rapidly as I can. If you don't do much more than write an occasional post-it to yourself, or a quick letter or something, notepad or its ilk is a better fit for what you do. If you do heavy editing of syntactic text files (like sourcecode or HDL), where saving a couple keystrokes can really matter, Vim is well worth the time required to learn.
I don't argue that Windows is a better desktop OS. It is. Windows thinks it's smarter than you. When this is true, it's a good tradeoff. But I'm not a desktop user. In the workstation (as opposed to desktop) environment, Linux wins out with room to spare.
If I were involved in stock trading, I'd do this myself. Good way to make a pile of money.
Photovoltaic solar panels are not now and never will be viable as a source of energy, but this doesn't rule out solar power entirely. A heliostat (an array of mirrors computer-aimed at a central target) is much more efficient in terms of the ROI energy produced, and usually in general as well. A well-tuned heliostat produces nice, clean energy all day long (not just when the sun is up... it melts sodium and uses this to fuel turbines even at night). The total cost per kilowatt-hour is much, much lower than any other solar method.
While the effective bulk transfer rate would be rather limited, it is quite possible that this system could survive on a saturated network. 500k/s of total network traffic isn't much, but 500k/s of traffic devoted to that machine is quite substantial. It might even be able to provide services that have very low bandwidth requirements.
I have a box of arcnet hardware, and an arcnet card is still available in my firewall machine should I ever need to connect something arcnet-enabled. It's 1MBit, but it has the advantage that you can run a cable up to a kilometer long between cards (not exactly a supported configuration, but it can be done) over regular TV coax. If you ever need to solve the "last-mile" problem, arcnet is a cheap way to do it.
You probably mean the 3c509 (which was a really nice 16-bit ISA card), not the 3c905 (which is a PCI 10/100 card). I have a couple 3c509's in my router box, they are very nice cards.
Chances are, if you can cram contiki onto an Atari or C64, you can get a TCP/IP stack into an embedded controller of some sort.
Unlike things like wireless networking cards, the firmware of a drive is flashed into the device itself, not stored in volatile memory. Unless you deliberately set out to overwrite it, it should remain untouched and gracefully accept whatever commands you send it (anything short of a "write firmware" command, which typically would require some sort of confirmation string ('AA 55' is a common string in the industry because it's so hard to accidentally do).
This firmware still has to interpret commands from the computer, *INCLUDING* the write firmware command. Usually, the firmware will have a dispatch table written into non-writable memory which will handle updating. Some firmwares will even have multiple slots with cryptographic checksums to ensure validity, reverting to a known-good firmware upon failure (most good motherboard BIOS will have this feature).
I've never heard of an open-source CDROM drive, so the complaints about this algorithm being proprietary are mostly irrelevant. The only issue is hostile OEM firmware (DRM or other misfeatures), which is a case for market pressure (if I found that my drive had been crippled by an OEM firmware, I'd demand a refund).
A product that dies during normal use is a problem for the manufacturer. LG should just recall them before this becomes a fiasco.
So if the host machine asks for sector X, but sector X doesn't exist, the drive electronics should simply report "sector not found" and return an error, not crash the head.
Digital cryptographic signatures work pretty well. If the hardware will only function with a signed firmware, then it's unlikely you will do much besides temporarily disable the drive, requiring a reflash.
They used to sell keylock switches for security systems. I suspect they still carry them. This would just be a straight replacement of the power switch with another switch. Simple enough. You could also pilfer these out of old machines that used to have a keyboard lock.
As for the little sound device, the chip you are looking for is the ISD1000A (or, since it's now out of production, one of the newer versions of the same line). Radio Shack used to sell this chip a while back. It's a 20-second audio record/playback IC. Just flash in your 20 second clip of a car starting, and wire it up to trigger when you boot (tie the trigger line to the /reset line on the ISA bus and you will be fine).
Then you could always replace the light-gap cards with a real magstripe, and add a reader head and a couple op-amps. Tie it to the microcontroller, and away you go!
Punchcards for authentication... why didn't I think of that? I've been spending all my time tinkering with microcontrollers and iButtons when I have a perfectly good RS232C-interface punch card reader sitting here gathering dust. I mean, a punchcard stores a whopping 80 bytes, which is quite sizable for a password or encryption key!
I'm planning to make a device based on a PIC12F675 8-pin microcontroller to do just this. I'm even considering adding a hardware watchdog timer, for good measure. For the less hardware-inclined, they sell serial and parallel-port cables that allow software to do the work instead.
The embedded readers are small enough that they could easily be mounted on a peripheral (a monitor or keyboard), or possibly even fitted into the case of a laptop. The embedded readers require dedicated hardware, but this can be a single 12-pin SSOP, taking a negligable amount of space.
The best part is that even the bare ROM version of the iButton has a 64-bit, factory-lasered, guaranteed-unique key.
It's also false that it costs more to DIY. Sure, there is an initial investment in tools and equipment. The time required is heavily dependant on the skill of the builder and the quality of the tools. If you happen to have mass production equipment (like I once did), you can make prototypes relatively quickly, and even do small production runs. And so long as you already have the tools required and a few good suppliers, it's often much, much cheaper to make your own.
And besides, you are missing how much *FUN* it is to build stuff. You know how Linux was written by a bunch of people for fun? Hardware is really no different, except that it has some sort of mystique that scares off a lot of people.