Back when I was in college (three guesses where...), there was (and still is) a building on campus with five floors and four elevators. It was a mess. If you were unfortunate enough to have a class on the fifth floor, you had to endure sharing it for more than a minute with dozens of sweating and smelly riders (it wasn't air conditioned)... roughly 1/4 of whom got off on each floor (slightly weighted towards higher floors, but not as much as you'd think). Going down after class was even worse... the cab would be COMPLETELY filled on the fifth floor, but stop at every single floor on the way down to take on additional riders who wouldn't have fit inside. During the span of a single class change, maybe a dozen people would actually go from one floor of the building to another (rarely, professors... they just took the stairs because it was less frustrating).
What I suggested was simple... between classes, each elevator should serve only two stops: the ground floor, and ONE other floor (#1 goes to 2, #2 goes to 3, #3 goes to 4, and #4 goes to 5). It would mean that someone wanting to go from the fifth floor to the second would have to go down to ground, switch elevators, then go up to the second... but the other 99.9% of the riders would have an express ride directly to where they wanted to go -- increasing capacity by eliminating time-consuming futile stops (especially on the "down" trip). The doors would open, the elevator would completely empty out, the cab would completely fill, and go straight to the other floor it served, where the same process would be repeated. No more stopping at intermediate floors where "nobody" (statistically) ever got out, and nobody could have boarded anyway.
Of course, the suggestion was completely ignored. But I still think it made complete sense...
When it comes to saving the day at a low price, the US Postal Service's Express Mail takes the grand prize. They deliver on Saturdays, for no additional cost (unlike FedEX, UPS, and DHL). They accept packages on Saturdays until early afternoon (in big cities, at least), and actually deliver them on SUNDAY -- for the SAME COST as weekday service. I don't think FedEx, UPS, and DHL even OFFER Sunday delivery as an option.
With Express Mail, you can literally ship something in the morning on Christmas Eve, knowing that one of Santa Claus' blue-clad delegates will be ringing their doorbell on Christmas morning to deliver it (and probably say , "Ho, Ho, Ho!" while he's at it).
For hobbyists who work on things over the weekend, Express Mail is a godsend. Find out that you need some part for your robot on Friday night after work, and you can have it shipped Saturday morning and arrive on Sunday.
Priority Mail is a close second, though. Faster, cheaper, AND more reliable than FedEx Ground (they really, REALLY suck... I've caught them literally lying about making delivery attempts when they were running late; once, when I was having my house worked on and had more than a dozen people mulling around the house, they claimed that "nobody was home". Bastards! They're RUINING FedEx's good name...) Best of all, with Priority Mail, if you miss the delivery on Friday... there's always Saturday. Unlike FedEx/DHL/UPS, who won't even let you go pick it up until the next business day...
It appears that Microsoft is at least responsible enough to be a patent whore (like IBM) rather than a patent troll (like SCO). The difference between the two? Patent whores have open, transparent licensing terms and use them to earn royalties, not put others out of business or drive them to drop the product in disgust. Twenty-five cents (apparently, capped off at some quantity that most flash manufacturers will exceed anyway) isn't going to break the bank on a flash drive. Even if the manufacturer passes along 100% of the cost and every distributor along the way marks it up 100%, you've added maybe a buck or two to a device selling for more than $25 anyway (not counting drives sold at a loss the day after Thanksgiving).
Actually, the annual cap was a brilliant move by Microsoft. By ensuring that the largest vendors (like Camera makers themselves) pay only a pittance, they're minimizing the risk that someone like Canon might try to roll out its own filesystem to avoid paying royalties. At 2-5 cents per card, it just wouldn't be worth the support headaches.
Plus, as a few have pointed out, a manufacturer could sidestep the royalties by simply not formatting the drive. What's likely to happen is that mass-market flash (bought by Joe Clueless at Wal Mart) will be sold formatted, but made by companies that make enough to easily exceed the capped royalties and end up paying 10 cents per unit or less, and bulk eBay flash (made by smaller companies that would have to pay the full amount) will be sold unformatted, with the expectation that anyone buying a crate of flash drives [or buying on eBay, or buying at a flea market] can be subjected to a little more inconvenience since they can't easily return it the moment they discover it's not formatted and will have to either throw it in a drawer or actually read the instructions and figure out how to format it themselves.
I agree... DigiKey rocks! I've placed orders that were submitted less than a minute before the same-day shipping deadline, and every last one went out right on schedule. Reasonable shipping charges, too.
My only real complaint is the occasional weird gaps in their inventory. They might stock 5 thousand different male DB-9 sockets of every type, but don't have shrouded 6-pin headers (used to connect Atmel AVR programmer to circuit for in-system programming), black ribbon cable, or the cheap & handy (for robot use) Dinsmore 1490 compass module. Devantech ultrasonic sensors, Fuji 401-series PIR sensors (half the cost of the Panasonic ones they sell), and board-to-board connectors like Samtec's elevated socket strip (handy for making robot daughterboards that plug into each other vertically).
I'd also like to take this opportunity to promote USPS Express Mail. It's cheap, they deliver next-day on Saturday for no extra charge, and you can even send something on Saturday and it'll be delivered first thing on Sunday morning FOR THE SAME PRICE AS WEEKDAY SERVICE. Just TRY that with FedEx or UPS!!! I placed an order with DigiKey the day before Hurricane Katrina hit Miami (wisely choosing Express Mail over FedEx), and got my order first thing on Sunday morning while 70% of Miami was still without power, and the mailman had to climb over the fallen tree in the front yard.
Eeeek. Accidental submission after being interrupted by someone and continuing to write without remembering the exact original topic. (puts paper bag over head).
From what I remember, a given hunk of flash is almost guaranteed to have at least one corrupt bit after 10 years (probably more, possibly sooner), and be nearly worthless in the context of financial data/programs/things for which tiny errors are catastrophic after about 20 years. HOWEVER, things for which errors are acceptable (like digital images and audio) are likely to be recoverable by mere mortals at home after even 30 years, and substantially recoverable by professionals after 50 or more. Put another way, a run of the mill DVD-ROM drive preserved in a cool, dry location for 25 years will probably be able to read images from a DVD+R burned 25 years earlier, but might choke on a file or two due to corruption in filesystem-related bits, and might have tiny errors in files read from it. Years later, the filesystem itself will have degraded to the point where something expecting pristine iso-9660 will choke on it, but someone with custom firmware that can rip the raw data straight from the disk for analysis and reassembly elsewhere will probably be able to pull it off.
The biggest problem, 15-25 years from now, is likely to be either filesystem degradation (so a clueless nontechnical home user won't be able to do filesystem copies, even if the future drives are backwards-compatible with cd/dvd, though a savvy s'ware engineer might be able to grab something from sourceforge or somewhere to do the job), or the fact that nontechnical users might very well lack the ability to read old media (read: Orrin Hatch and his Disney Comrades successfully imposing laws that assume anything not signed and encrypted by hollywood is pirated and thus must not be allowed to be readable by legal drives).
Fast forward two or three years, and buy a new Chinese-made ${optical media} player from Wal-Mart.
It supports all 4 optical standards: CD(-ROM), DVD(-ROM), HD(-ROM), and BD(-ROM).
It supports every meaningful video and audio codec associated with those optical-disc standards:
It supports MPEG-1. It supports MPEG-2 -- both MP@ML (the original "DVD" resolutions) *and* MP@HL (high-def extensions). It supports MPEG-4, including the subset commonly known as "DivX". It supports WMV/HD.
It also supports just about every reasonable permutation of the above codecs and their related technologies, including MP3 audio.
To the entire (non-Chinese) industry's chagrin, it doesn't just support the "officially-blessed" combos... throw it a BD-ROM authored like a "DVD", and it's perfectly happy to treat it like a 10-hour long DVD. Shove in a DVD9 authored with the "HD-DVD" filesystem, and it's perfectly content to treat it like a 30-minute long HD-DVD. Or, potentially bypassing some royalty payments at the disc end (if royalties get charged to disc manufacturers, as opposed to player manufacturers), a normal 9-gig DVD-ROM encoded like a "normal" DVD, but using the MP@HL format at 720p24 or 480p60 to encode an hour or two of HD-quality video onto an old-fashioned 9-gig dual layer red-laser DVD and save a ton of manufacturing costs and royalty payments.
See, unlike American and Japanese companies (increasingly, the dinosaurs of the electronics industry who live in a DRM-laden universe where everyone is perpetually afraid of getting sued by Hollywood for allowing consumers to do something Not Authorized), the Chinese companies see no reason NOT to just treat everything like a "bucket" of capabilities, and let the player's firmware sort out any reasonable permutation.
But wait! How can something become a "standard" if it's not officially blessed by some official body, like the DVD Forum or Sony? Well, look at it this way... if every $79 optical video player sold at Wal-Mart can play some disc, but a $129 Sony from Best Buy can't, word will quickly get around, and people will just buy the cheaper player from Wal-Mart instead because it "just works". Sony can grouse and bitch about "illegitimate formats" until it's bankrupt. Consumers won't care. They'll buy the cheaper drives that play everything painlessly.
If you're American, ask yourself... "can my DVD player handle discs encoded with mp3-encoded audio?" In all likelihood, unless it's an ultra high-end player made by Sony or Denon... "Yes, it can." Officially, American DVD players don't have to support it (only AC3 and PWM)... but every last cheap Chinese-made player at Wal-Mart and Best Buy does, and for all intents and purposes, it IS a de-facto standard that any player sold in America had BETTER support unless its manufacturer wants to see a staggering number of "defective" players returned by consumers who discovered that it couldn't play some movie encoded with MP3 audio instead of AC3 or PWM. If some capability becomes nearly ubiquitous by virtue of being universally supported by "low-end" players, high-end players eventually WILL have to support it, too, regardless of whether anyone has ever formally blessed it as "official".
Unless, of course, the American and Japanese manufacturers and media conglomerates convince Congress to pass some nazi-ish law, like, "The Home Media Player Standards-Compatibility Improvement Act" that purports to weed out "incompatible" players, but really acts to forbid extensions as well as deficient implementations...
If I were in college now, rather than 10 years ago, THIS would likely become my "real" system... the one that mostly lived in my dorm room, chained to the desk, tethered to a nice Lexmark buckling-spring keyboard w/trackpoint and second TFT display, and removed only for trips home and to lan parties... augmented by a thin & light ultraportable that I actually carried around to my classes and used at the library and viewed as my "toy" computer, even if it were actually the one I ended up using for 90% of my classwork.
Of course, that's not a whole lot different from how I lived in college... a laptop as a "real" system was utterly out of the question back then, of course, so I had a fully-loaded desktop system to die for, but I still have fond memories of my old DEC HiNote CT475 (horrifyingly expensive and underpowered, but small enough to fit in a manila envelope) and even my older Compaq Aero (though its 7.5" screen was DEFINITELY too small for comfort by any objective standard and left me cross-eyed more than a few times).
No, actually it's perfectly capable of replicating fine wine... except for its compliance with the Fine Wine Freedom, Liberty and Justice Act of 2317 (and its European companion, the "Rule on High-quality Spirits-reproduction"), which requires that all replicators be equipped to recognize the molecular structure of a fine wine and produce malt liquor instead when asked to violate the intellectual property of hardworking vineyard owners...
As it so happens, I'm taking a break right this moment from the project that's consumed most of my free time for the past month, and likely to consume substantially all of it for the next week and a half: a string of 50 serial-networked Atmel ATtiny2313 microcontrollers, each juggling 4 independent RGB LEDs (total cost, not including support and development hardware, ~$250). As soon as I'm convinced that the bootloader I'm (hopefully) finishing this weekend can, in fact, successfully re-flash the chips once I've soldered them together and can't easily get at the SPI pins, the first 10-controller/40-LED segment will get built.
What will it (hopefully) be able to do? At the minimum, I'm hoping to program the following effects:
* each LED individually flickering like a candle flame, randomly and independently changing colors
* whole tree changing color in sync: red->yellow->green->cyan->blue->violet->red. This was actually a hard mode to program, because I had to re-do the entire LED-flashing algorithm to support temporarily combining the pair of 16-byte color buffers used by the two ports (each port controls 2 LEDs) into one big 32-byte buffer and increase the color resolution from 2 bits to 3 bits for slow whole-tree fades. I also had to change the pin assignments, to make sure that each port bit controlled an element of the same color on both ports.
* "Stupid Mode", where the controller sends 3-bytes in a row specifying the 2-bit RGB values for each of the four LEDs. I don't really plan to do much with this mode, but it's there as a backup plan if the bootloader scheme fails and I can't reflash the firmware at some future point. By cranking the baudrate up to 38.4k and speeding the controller up to 20MHz, I could probably pull off some decent effects. But I really want to make each controller smart and autonomous, so the controller just has to say, "group 3, slowly fade from red to violet, then hold using flicker effect #2 for 8 seconds before proceeding to the next instruction".
* of course, there's also "zombie" mode, activated by powering up the string without the controller. It's rather stupid and simple, but knows how to randomly change each LED's color every second or so. It's there to make sure that if I royally screw up the firmware on Christmas Eve, I'll still have an ok-looking tree to show off...
There will ultimately be 5 segments, connected together with a 5-pin DIN female socket on one end, and similar male plug on the other (3 wires total... Vcc, Gnd, and Serial data). One end will plug into the power supply and controller, which has a beefier microcontroller (ATMega162) to act like an orchestra's conductor and coordinate their activities when appropriate (and reflash the downstream microcontrollers when I update their firmware). Normally, all 50 chips will just watch the Serial line for 9600-baud serial broadcasts from the controller, but when given a reason and an opportunity to do so, they can individually pull the bus low to send crude signals back to the controller.
I'm still undecided about making them individually-addressable. My current plan is to make them group-addressable, with software-selectable group sizes of 3, 4, and 6. I can easily accomplish this by grounding none or one of the two "A" pins (3 possible groups: for 3-LED groups, they'd be 1-2-3-1, 2-3-1-2, and 3-1-2-3; for 4-LED groups, it would be irrelevant (each controller has 4 LEDs); for 6-LED groups, they'd be 1-2-3-4, 5-6-1-2, and 3-4-5-6. If I make them individually-addressable, I could do it the "easy" way and hardcode the base address of the first LED into each microcontroller's (basically-)immutable bootloader... but then I'll have to connect the strings in the right order. Alternatively, I can daisy-chain a fourth wire from the second "A" pin of each to chip to the first "A" pin of the next and use the neighbor-to-neighbor signaling to enumerate the string at reset. But really, as ideologically appealing as it seems to make each LED individually-addressable, I
Is there actually some real, (presumably orange) honest-to-god CHEESE that's actually called "American Cheese"? Because I've never seen the word "cheese" on so-called "American Cheese" that wasn't immediately followed by the qualifying word, "food". As in, "cheese food". If all American Cheese is, by definition, "Cheese Food" and NOT real cheese, how can there be such a thing as "American Cheese" in the first place? Or was there at one time such a thing as real American Cheese, but nobody ate it because it didn't melt nicely until some guy whose last name was "Kraft" invented "cheese food", figured out how to wrap the slices in heatshrink plastic to keep them from drying out and going bad, and turned (so-called) "American Cheese" into a viable product?
possession of the source != permission to disclose it.
Nevertheless, I think it's a straw man. Most likely, Diebold doesn't want to release THEIR source, and they're using Windows as the excuse. If Diebold wants to take the high road, let them release the source to THEIR code, including a diff file showing only their changes to the source received from Microsoft, and let some idiot bureaucrat in the state capitol make a fool of himself going on CNN and demanding the source code to Windows itself while everyone laughs at his pettiness.
Re:It'll be a sad state of affairs when this happe
on
A Flu Pandemic?
·
· Score: 2, Interesting
> As one of those researchers, I've got to ask -- given that Taiwan is already breaking > the Tamiflu patent, what makes you think my bosses are insane enough to invest that > research money when the product is going to be confiscated?
Because even if they are, it's not causing Roche and everyone down the royalty chain to lose any actual money. Roche is already producing at 100% capacity, selling 100% of its output at its chosen price point, and refuses to license manufacturing rights to anyone else. As long as Roche is still selling 100% of its chosen capacity at its chosen price point, it has no right to bitch.
Personally, I take immense comfort from the high likelihood that right now, multiple factories in mainland China are almost certainly working around the clock, secretly cranking out pirated Tamiflu as fast as they possibly can, quietly filling warehouses with it, under the direction of one or more government officials motivated mainly by the thought of making billions of dollars and euros selling it to Americans and Europeans on eBay (or direct) for, say, a thousand dollars/euros for a 3-month twice-daily supply, when/if a real, honest to god pandemic strikes and the rights of Tamiflu's IP owners fall off the bottom of their list of concerns.
Think about it... the Chinese have little regard for IP anyway, they're faced with a potential future domestic crisis whose sole possible cure -- Tamiflu -- is already being rationed and suffering from limited availability. Does ANYONE *seriously* think they're going to sit back with their hands neatly folded, obediently refraining from violating Roche's IP and settling for the crumbs Roche might allocate to them at some outrageous, inflated price? And of course, if they DO make lots and lots of it, and demand far outstrips supply worldwide, the fact that they'll have to build a few new skyscrapers just to warehouse the money they'll make selling it abroad just seals the deal.
The genuine danger is that if no pandemic emerges within 5 years or so, some "bright" government official in China will decide to keep it from going to waste (since they won't be able to sell it, or even admit it exists, under any conditions besides an outright pandemic) and order it ground up and added to chicken feed. THAT would be a Very, Very Bad Thing(TM).
1080i60 = 1920 x 1080 @ 60 interlaced fields/second (540 scanlines redrawn every 1/60th sec)
720p60 = 1280 x 720 @ 60 progressive frames/second (all 720 scanlines redrawn every 1/60th sec)
480p60 is kind of messy.
There's the square-pixel variant, whose resolution is 640 x 480 @ 60 progressive frames/second.
There's also the rectangular-pixel 16:9 variant, whose resolution is (704 | 720) x 480 @ 60 progressive frames/sec. Offhand, I don't remember which one is the official one.
I think 480i60 has the same square-pixel (4:3) and rectangular-pixel (16:9) variants as 480p60.
Bandwidth-wise, 1080i60 and 540p60 are indistinguishable. 540p60 requires ~33.75KHz, while 480p60 requires the same as plain VGA: ~31.5KHz. In a CRT, this basically means the manufacturer can push a few components to the bounds of their tolerance and get away with using a single flyback transformer to handle both 480p60 and 1080i60. Since 720p60 requires ~45KHz, there's no way to fake it... you need an upgrade to handle at least two radically different scanrates. Which leads us to the second problem...
What do multisync monitors do when you radically switch display modes (say, between full-screen "DOS" and whatever resolution you normally run Windows at)? That's right... the screen turns black, a relay thunks, and the display comes back a second or two later. Unfortunately, Joe Sixpack isn't sophisticated enough to understand the engineering rationale behind the thunk-blink, and views it as a drawback. Especially when he's surfing between WWF Wrestling (in glorious 480i60) and ABC (in 720p60).
So, the way HDTV manufacturers look at it, they can make a TV that natively supports 480i60, 480p60, 720p60, and 1080i60, charge $20 more to cover the $2 extra it cost them to natively support 720p60 in addition to the rest, and then deal with endless bitching from unsophisticated customers who are mad because their TV goes "thunk" and blacks out when they're channel-surfing. Or, they can omit 720p60, pocket the savings, and be mostly left alone because average ignorant consumers wouldn't know the difference between 720p60 and 1080i60 if their lives depended on it:-(
The ONLY company I know of that makes TVs that NATIVELY support 480i60, 480p60, 720p60, AND 1080i60 is Monivision. Their CRT-based HDTVs are basically big multisync monitors with increased dot pitch (it's not practical to make a 35" CRT with.20mm dot pitch because it would be unacceptably dark due to the holes in the mask being so small, and would be gross overkill at that size and resolution anyway.) If I were going to buy a CRT-based HDTV, it would almost definitely be one of theirs for that precise reason.
Of course, there's a good solution to the "thunk" problem when channel-surfing: leave the TV's display mode alone and hack-convert everything to that mode while surfing, THEN (optionally) pick the best native display mode once the user hits another button on the remote. That way, Joe Sixpack can enjoy thunk-free surfing, and WE can enjoy HDTV programs at their native scanrates:-)
The good news is that natively-progressive TVs are slowly edging out natively-interlaced HDTVs. The arrival of 1080p HD-DVDs will abolish any remaining demand for 1080i60 TVs among high-end users, and the flood of affordable 720p60 DLP, LCoS, and Plasma sets will edge them out among mid-range users. The remaining ones, who'll end up buying sub-32" CRT sets for years to come, will basically have TVs that display 480p60, even if they're technically doing 1080i60. With screens that small, the added horizontal detail of anything higher than 480p is lost and smeared anyway, and the scanlines are close enough together that the difference between 1080 sharp interlaced scanlines and 540 blurred progressive scanlines is academic (it's one of the reasons why it's practically impossible to look at a 27" "HDTV" set and figure out whether it's displaying something as 480p or 1080i... with low-end CRT-based HDTVs, the shadow mask is coarser than the nominal vertical resolution, and adjacent scanlines between fields end up lighting up the same phosphor triads anyway).
As lots of people tend to forget (even people WITH HDTVs):
* the difference between broadcast NTSC or composite NTSC and studio/dvd-quality NTSC (via s-video) is dramatic and noticeable.
* the difference between studio/dvd quality NTSC via s-video ("480i60") and 480p60 is night and day. Someone who just had PRK/LASEK the previous day could still tell the difference between the two on a 27" TV twenty feet away in a smoke-filled room.
* On a natively 720p60 set, 720p60 looks noticeably better than 480p60. On a natively 1080i60 set, 720p60 is almost indistinguishable from 480p60 because the TV downsamples to 540 scanlines, then kell-filters them to prevent flicker. The net result is almost the same amount of vertical detail as 480p60, with only slightly more horizontal detail. Meanwhile, the GPU and CPU are working almost twice as hard.
* On a natively 720p60 set, 1080i60 is nearly indistinguishable from 480p60, because the TV just throws away half of the scanlines, resamples the 540 that remain up to 720, then resamples the alleged 1920 horizontal pixels to 1280. On a natively 1080i60 set, 1080i60 COULD exhibit greater detail than 480p60... except for the tiny problems of interline twitter (requiring kell filtering), inadequate GPU/CPU power to really pull off their best work at 1920x1080, and the fact that programmers fundamentally don't understand the realities of computer-generated interlaced video and inevitably produce games that look great on the progressive-scan monitors connected to their dev boxen, but have glaring artifacts and deficiencies when viewed on a real interlaced display.
In short, 1080i60 doesn't have a whole lot to offer more than 480p60 for action-related games due to all the filtering necessary to prevent interline twitter, and inadequate raw GPU/CPU horsepower to really handle 1920x1080 properly. 720p60 has definite potential to offer better-looking games because 1280x720 is still a reasonable resolution as far as the GPU/CPU is concerned... but at the moment, natively 720p60 TVs only represent about 1/3 of the total in America (unfortunately), and 720p60 looks like $#!+ on most natively 1080i60 TVs.
At the implementation level, upgrading chips capable of 480p60 to 1080i60 is a comparatively small tweak, because most of the increased bandwidth goes into permitting faster pixel-to-pixel color changes. The actual scanrate (~33.75KHz) isn't a whole lot higher than 480p60's (~31.5kHz). Unfortunately, you can't fool Mother Nature... making the leap to 720p60 requires ~45KHz, because the real or metaphorical electron beam has a LOT more ground to cover in the same amount of time. Put another way, you can do some nasty hacks and claim that a given circuit is technically capable of "1080i", even if its REAL capabilities aren't much better than 480p because the horizontal detail will be all smeared and blurred due to inadequate bandwidth... but making the leap to 720p requires real upgrades that cost real money. And ultimately, the 2/3 majority of American HDTV owners whose sets are inherently 1080i won't see much of an improvement anyway, and will probably bitch about games that only support 480p and 720p.
The REAL surge in "HDTV" console gaming will come in another 2-4 years, once natively-720p TVs have displaced enough older 1080i TVs (at least among gamers who know the difference and care), and Nintendo's NEXT console WILL support 720p (but probably won't bother with 1080i, leading to more waves of grousing and complaining about its lack of "true" support for HDTV).
Even at 16 bytes, it's STILL too long to easily remember, though;-)
Adding a bit to my earlier comment, it's IMPERATIVE that there be a standard way to express IPv6 addresses in a compressed form that's human-memory-friendly. If there isn't, I have no doubt that several dozen parties will come up with slightly different compression + expression schemes of their own, and at least a half-dozen will likely become co-standards... and anyone unfortunate enough to have to convert a human-format address into its binary form will have lots of fun trying to figure out by context what form it's in. Say...
"Cisco 1.0" form -- $8b 79 03 3e 2a 83 57 5c af 09 63$
Let's all be honest... the biggest reason we fear IPV6 is the fact that we'll never, ever be able to actually remember IP addresses by memory anymore. "aaa.bb.ccc.ddd" fits nicely within the human short term memory stack. "aaa.bb.ccc.ddd.ee.f.ggg.hh.iii.j.kk.lll.mm.nn.ooo .p.qqq.rr.sss.tt.uuu.vvv.ww.x.yyy.zzz.AA.BBB.CCC.D .EEE.FF" overflows it, by several orders of magnitude. AT BEST, it might be possible to keep a tiny handfull of IP addresses straight, IF they're mostly zeroes AND have those zeroes in the same locations.
Instead of using all 128 bits for address space, let's set aside the first bit for human-friendly data compression: if the first bit is "1", we assume that 127 significant bits of address data follow. If it's "0", we'd transparently insert 64 "0" digits, then continue with another 63 significant digits of address info.
Yeah, it would cut the potential address space in half... but as more than a few have already pointed out, IPv6 isn't exactly hurting for address space anyway. The big benefit is that for the first few years/decades/centuries of use, nobody would ever actually HAVE to deal with a full-blown 128-bit address... IPV6 would have a de-facto 64-bit address space, ready to become 127 bits when the need finally arises. IP addresses would double in length, but 8 bytes are still within most people's capabilities. For the foreseeable future, nobody outside of Cisco or academia would actually have to bother with 32-byte IP addresses.
Short of that, the only way ANYONE will be able to deal with 128-bit addresses is if someone comes up with a scheme for constructing plausible nonsense words that unambiguously map to real 128-bit addresses, using letter combos to encode 2 to 4 bits per letter. Say, compressing big repeating chunks of zeroes and ones with escape patterns down to 40 or 50 bits, then using those bits to specify grapheme (morpheme? "nonsense word"?) consonant-vowel combos vaguely resembling romanicized Japanese, like "fa muka gu zade no kwatule yo". It would look weird, but by re-establishing some kind of contextual link to plausible language, people would at least have SOME chance of (mostly) getting it right from memory.
The whole idea of using normal power lines for data transmission is just plain silly, at least in urban areas. When you factor in the costs of all the extra hardware needed to inject and extract the data while blocking the electricity itself, then add the problems inherent in trying to send data down what's literally an unshielded bare wire, then compare the cost of doing that to, say, hanging a fibre bundle on the pole, the whole rationale for using the power lines goes down the toilet.
The ONLY reason American power companies care about broadband-over-power lines is because they want to be the third data carrier in their market, standing alongside the phone and cable companies. The catch is, due to government regulation, getting permission to hang fibre bundles next to power lines (with zero mutual interference) is a major problem, vulnerable to challenges by the incumbent phone and cable companies. But if they can somehow squeeze data down the power lines, it's easier to slide it past the regulators.
I guarantee -- the moment the FCC officially codifies the right of power companies to offer broadband and makes it an unalienable right not subject to refusal or regulation by states, every last power company in America will instantly abandon broadband over power lines and run fibre instead unless they're explicitly prohibited from doing so.
Yeah, but having your airbags triggered by a thief is several orders of magnitude worse than having a window broken. By law, the car commits electronic suicide the moment the airbags deploy and can't be driven, period, until a few thousand dollars worth of repair work has been done. Not just the ECU and airbags... the wires and sensors need to be replaced too. And then there's the matter of legal liability... any repair shop has to buy an insurance policy against someone who tries to sue in the future claiming the recertified airbag system deployed unnecessarily or failed to deploy when necessary due to some fault of theirs.
For all intents and purposes (in America, at least), a car whose airbags have deployed is effectively "totaled" because, from the insurance company's perspective, it's cheaper to pay the claim as a total loss and sell the car to a broker for export to some third-world country where the car can legally be repaired without the airbags and recertification than it is to pay to have it repaired, recertified, and liability-insured for use in the US.
EXACTLY what happened in Miami after Andrew. People who otherwise wouldn't have been particularly affected by Andrew turned on CNN, saw reports that people were paying $10 for a bag of ice in Miami, and promptly ran out, rented a U-Haul truck, filled it with bagged ice, and headed down to Miami.
When they arrived, they quickly discovered that they weren't alone. In fact, *SO MANY* people had the exact same idea that the price of ice fell to $5/bag within a day, then $3, then $2, then $1. People can bitch about how terrible it is to exploit disaster victims, but the fact is, due to greed, a HELL of a lot of bagged ice became available in Miami... ice that wouldn't otherwise have been available. While government agencies and relief organizations were holding meetings to plan their strategic response, millions of bags of ice were already on their way.
The truth is, guys were selling bagged ice for $2-3 out of trucks RIGHT ACROSS THE STREET from government aid caravans giving it away for free. The difference was, getting free ice meant waiting in line for an hour or two. Buying it from the guy with a truck might have cost ten bucks, but it meant you could be heading home in 30 seconds. Later, as competition increased, the ice guys started driving up and down streets like ice cream trucks and offered to-the-door delivery.
Personally, I wish stores WOULD jack up the cost of things like D cells and plywood before the storm... it would stop a$$#0735 from trying to buy a six-week supply at the last minute, wiping out the stores and ensuring that most people can't buy them at ANY price. If people KNEW that D cells would cost four times as much the day before a storm hits, and twice as much for a day or two afterward, they'd be more likely to stock up on them ahead of time... or at least, buy just enough to get through the next few days.
They were/are called "Field Emission Displays" (FED), if I recall correctly. The idea was to take a flat plate of glass (or equivalent) with normal CRT phosphors, but behind each phosphor put multiple tiny solid-state electron emitters. In other words, instead of taking a single electron beam and sweeping it repeatedly across the screen, put multiple weak electron beams behind each subpixel. The rationale for using multiple beams was avoidance of dead pixels. If there were, say, six emitters behind each subpixel, one or two of them could fail outright or progressively, and the viewer would likely never notice the difference... and the few that WERE bad enough to be noticeable could be programmed around by simply increasing the intensity of that specific subpixel.
I first found out about FED displays (yeah, I know "FED Display" is redundant) when I went to CES in Orlando in 1996(?). I thought for sure FED would rule the day (not for laptops, of course, because they used too much power... but for home TVs). I'm not really sure whether the concept proved to be unfeasible, or whether FED actually mutated something else. My guess is that cheap, hi-res DLP, far bigger and cheaper TFT panels than anyone dared to predict a decade ago, and cheaper & less-fragile plasma all chipped away at the manufacturers' motives for funding it, and they all went after one or more of the other technologies instead.
IMHO, it's a pity. DLP has good intensity, is indestructable, and high resolution, but suffers from either rainbows or high cost & convergence problems. TFT has high resolution, but is limited in brightness when used with projectors, and both size & pixel-perfection in direct displays. Brand new plasma sets look impressive, but they're the lowest-rez sets of all, and I'd personally rather watch a 34" CRT than endure ugly stretched images necessitated by plasma's vulnerability to burn-in due to pillarboxing instead. CRT... well, CRT has probably the best overall color saturation, but is size-limited by the tube itself (unless, of course, your living room is big enough to drive a forklift in).
Sigh. I really, REALLY like the concept of FED displays... they have/had the potential to give us the best features from all the other alternatives (though possibly at a staggering cost).
Fine... earn more revenue by putting power outlets *everywhere* with coin slots that give 10 minutes of power for a quarter instead of hiding and disabling every visible outlet in the terminal...
The perverse irony of power outlets in terminals is that the newest terminals have the fewest and most carefully-hidden outlets of all. Old terminals from the 1960s have outlets everywhere, because they wanted to make life easy for the janitorial staff.
I hope to god PalmSource ends up acquiring Tapwave's IP, makes the DRM to use its features optional (enabling free open-source software to take advantage of it), and formally adopts it as the reference hardware platform for the next generation of PalmOS devices (particularly phones). With BOTH free AND commercial apps available, and Palm's official blessing as an optional standard (but for all intents and purposes mandatory for new PalmOS hardware going forward), it would be hard for them to fail. Everyone knows PocketPC hardware thrashes most Palm hardware... but people still buy Palm devices because it's a friendlier, more hackable platform that occasionally enables someone to do something cool and creative that Palm themselves never envisioned (can anyone even FATHOM Microsoft officially blessing a PocketPC equivalent of Hackmaster, let alone mentioning its existence and documenting its API?)
Slashdot Mirror
Back when I was in college (three guesses where...), there was (and still is) a building on campus with five floors and four elevators. It was a mess. If you were unfortunate enough to have a class on the fifth floor, you had to endure sharing it for more than a minute with dozens of sweating and smelly riders (it wasn't air conditioned)... roughly 1/4 of whom got off on each floor (slightly weighted towards higher floors, but not as much as you'd think). Going down after class was even worse... the cab would be COMPLETELY filled on the fifth floor, but stop at every single floor on the way down to take on additional riders who wouldn't have fit inside. During the span of a single class change, maybe a dozen people would actually go from one floor of the building to another (rarely, professors... they just took the stairs because it was less frustrating).
What I suggested was simple... between classes, each elevator should serve only two stops: the ground floor, and ONE other floor (#1 goes to 2, #2 goes to 3, #3 goes to 4, and #4 goes to 5). It would mean that someone wanting to go from the fifth floor to the second would have to go down to ground, switch elevators, then go up to the second... but the other 99.9% of the riders would have an express ride directly to where they wanted to go -- increasing capacity by eliminating time-consuming futile stops (especially on the "down" trip). The doors would open, the elevator would completely empty out, the cab would completely fill, and go straight to the other floor it served, where the same process would be repeated. No more stopping at intermediate floors where "nobody" (statistically) ever got out, and nobody could have boarded anyway.
Of course, the suggestion was completely ignored. But I still think it made complete sense...
When it comes to saving the day at a low price, the US Postal Service's Express Mail takes the grand prize. They deliver on Saturdays, for no additional cost (unlike FedEX, UPS, and DHL). They accept packages on Saturdays until early afternoon (in big cities, at least), and actually deliver them on SUNDAY -- for the SAME COST as weekday service. I don't think FedEx, UPS, and DHL even OFFER Sunday delivery as an option.
With Express Mail, you can literally ship something in the morning on Christmas Eve, knowing that one of Santa Claus' blue-clad delegates will be ringing their doorbell on Christmas morning to deliver it (and probably say , "Ho, Ho, Ho!" while he's at it).
For hobbyists who work on things over the weekend, Express Mail is a godsend. Find out that you need some part for your robot on Friday night after work, and you can have it shipped Saturday morning and arrive on Sunday.
Priority Mail is a close second, though. Faster, cheaper, AND more reliable than FedEx Ground (they really, REALLY suck... I've caught them literally lying about making delivery attempts when they were running late; once, when I was having my house worked on and had more than a dozen people mulling around the house, they claimed that "nobody was home". Bastards! They're RUINING FedEx's good name...) Best of all, with Priority Mail, if you miss the delivery on Friday... there's always Saturday. Unlike FedEx/DHL/UPS, who won't even let you go pick it up until the next business day...
It appears that Microsoft is at least responsible enough to be a patent whore (like IBM) rather than a patent troll (like SCO). The difference between the two? Patent whores have open, transparent licensing terms and use them to earn royalties, not put others out of business or drive them to drop the product in disgust. Twenty-five cents (apparently, capped off at some quantity that most flash manufacturers will exceed anyway) isn't going to break the bank on a flash drive. Even if the manufacturer passes along 100% of the cost and every distributor along the way marks it up 100%, you've added maybe a buck or two to a device selling for more than $25 anyway (not counting drives sold at a loss the day after Thanksgiving).
Actually, the annual cap was a brilliant move by Microsoft. By ensuring that the largest vendors (like Camera makers themselves) pay only a pittance, they're minimizing the risk that someone like Canon might try to roll out its own filesystem to avoid paying royalties. At 2-5 cents per card, it just wouldn't be worth the support headaches.
Plus, as a few have pointed out, a manufacturer could sidestep the royalties by simply not formatting the drive. What's likely to happen is that mass-market flash (bought by Joe Clueless at Wal Mart) will be sold formatted, but made by companies that make enough to easily exceed the capped royalties and end up paying 10 cents per unit or less, and bulk eBay flash (made by smaller companies that would have to pay the full amount) will be sold unformatted, with the expectation that anyone buying a crate of flash drives [or buying on eBay, or buying at a flea market] can be subjected to a little more inconvenience since they can't easily return it the moment they discover it's not formatted and will have to either throw it in a drawer or actually read the instructions and figure out how to format it themselves.
I agree... DigiKey rocks! I've placed orders that were submitted less than a minute before the same-day shipping deadline, and every last one went out right on schedule. Reasonable shipping charges, too.
My only real complaint is the occasional weird gaps in their inventory. They might stock 5 thousand different male DB-9 sockets of every type, but don't have shrouded 6-pin headers (used to connect Atmel AVR programmer to circuit for in-system programming), black ribbon cable, or the cheap & handy (for robot use) Dinsmore 1490 compass module. Devantech ultrasonic sensors, Fuji 401-series PIR sensors (half the cost of the Panasonic ones they sell), and board-to-board connectors like Samtec's elevated socket strip (handy for making robot daughterboards that plug into each other vertically).
I'd also like to take this opportunity to promote USPS Express Mail. It's cheap, they deliver next-day on Saturday for no extra charge, and you can even send something on Saturday and it'll be delivered first thing on Sunday morning FOR THE SAME PRICE AS WEEKDAY SERVICE. Just TRY that with FedEx or UPS!!! I placed an order with DigiKey the day before Hurricane Katrina hit Miami (wisely choosing Express Mail over FedEx), and got my order first thing on Sunday morning while 70% of Miami was still without power, and the mailman had to climb over the fallen tree in the front yard.
Eeeek. Accidental submission after being interrupted by someone and continuing to write without remembering the exact original topic. (puts paper bag over head).
From what I remember, a given hunk of flash is almost guaranteed to have at least one corrupt bit after 10 years (probably more, possibly sooner), and be nearly worthless in the context of financial data/programs/things for which tiny errors are catastrophic after about 20 years. HOWEVER, things for which errors are acceptable (like digital images and audio) are likely to be recoverable by mere mortals at home after even 30 years, and substantially recoverable by professionals after 50 or more. Put another way, a run of the mill DVD-ROM drive preserved in a cool, dry location for 25 years will probably be able to read images from a DVD+R burned 25 years earlier, but might choke on a file or two due to corruption in filesystem-related bits, and might have tiny errors in files read from it. Years later, the filesystem itself will have degraded to the point where something expecting pristine iso-9660 will choke on it, but someone with custom firmware that can rip the raw data straight from the disk for analysis and reassembly elsewhere will probably be able to pull it off.
The biggest problem, 15-25 years from now, is likely to be either filesystem degradation (so a clueless nontechnical home user won't be able to do filesystem copies, even if the future drives are backwards-compatible with cd/dvd, though a savvy s'ware engineer might be able to grab something from sourceforge or somewhere to do the job), or the fact that nontechnical users might very well lack the ability to read old media (read: Orrin Hatch and his Disney Comrades successfully imposing laws that assume anything not signed and encrypted by hollywood is pirated and thus must not be allowed to be readable by legal drives).
Fast forward two or three years, and buy a new Chinese-made ${optical media} player from Wal-Mart.
It supports all 4 optical standards: CD(-ROM), DVD(-ROM), HD(-ROM), and BD(-ROM).
It supports every meaningful video and audio codec associated with those optical-disc standards:
It supports MPEG-1.
It supports MPEG-2 -- both MP@ML (the original "DVD" resolutions) *and* MP@HL (high-def extensions).
It supports MPEG-4, including the subset commonly known as "DivX".
It supports WMV/HD.
It also supports just about every reasonable permutation of the above codecs and their related technologies, including MP3 audio.
To the entire (non-Chinese) industry's chagrin, it doesn't just support the "officially-blessed" combos... throw it a BD-ROM authored like a "DVD", and it's perfectly happy to treat it like a 10-hour long DVD. Shove in a DVD9 authored with the "HD-DVD" filesystem, and it's perfectly content to treat it like a 30-minute long HD-DVD. Or, potentially bypassing some royalty payments at the disc end (if royalties get charged to disc manufacturers, as opposed to player manufacturers), a normal 9-gig DVD-ROM encoded like a "normal" DVD, but using the MP@HL format at 720p24 or 480p60 to encode an hour or two of HD-quality video onto an old-fashioned 9-gig dual layer red-laser DVD and save a ton of manufacturing costs and royalty payments.
See, unlike American and Japanese companies (increasingly, the dinosaurs of the electronics industry who live in a DRM-laden universe where everyone is perpetually afraid of getting sued by Hollywood for allowing consumers to do something Not Authorized), the Chinese companies see no reason NOT to just treat everything like a "bucket" of capabilities, and let the player's firmware sort out any reasonable permutation.
But wait! How can something become a "standard" if it's not officially blessed by some official body, like the DVD Forum or Sony? Well, look at it this way... if every $79 optical video player sold at Wal-Mart can play some disc, but a $129 Sony from Best Buy can't, word will quickly get around, and people will just buy the cheaper player from Wal-Mart instead because it "just works". Sony can grouse and bitch about "illegitimate formats" until it's bankrupt. Consumers won't care. They'll buy the cheaper drives that play everything painlessly.
If you're American, ask yourself... "can my DVD player handle discs encoded with mp3-encoded audio?" In all likelihood, unless it's an ultra high-end player made by Sony or Denon... "Yes, it can." Officially, American DVD players don't have to support it (only AC3 and PWM)... but every last cheap Chinese-made player at Wal-Mart and Best Buy does, and for all intents and purposes, it IS a de-facto standard that any player sold in America had BETTER support unless its manufacturer wants to see a staggering number of "defective" players returned by consumers who discovered that it couldn't play some movie encoded with MP3 audio instead of AC3 or PWM. If some capability becomes nearly ubiquitous by virtue of being universally supported by "low-end" players, high-end players eventually WILL have to support it, too, regardless of whether anyone has ever formally blessed it as "official".
Unless, of course, the American and Japanese manufacturers and media conglomerates convince Congress to pass some nazi-ish law, like, "The Home Media Player Standards-Compatibility Improvement Act" that purports to weed out "incompatible" players, but really acts to forbid extensions as well as deficient implementations...
If I were in college now, rather than 10 years ago, THIS would likely become my "real" system... the one that mostly lived in my dorm room, chained to the desk, tethered to a nice Lexmark buckling-spring keyboard w/trackpoint and second TFT display, and removed only for trips home and to lan parties... augmented by a thin & light ultraportable that I actually carried around to my classes and used at the library and viewed as my "toy" computer, even if it were actually the one I ended up using for 90% of my classwork.
Of course, that's not a whole lot different from how I lived in college... a laptop as a "real" system was utterly out of the question back then, of course, so I had a fully-loaded desktop system to die for, but I still have fond memories of my old DEC HiNote CT475 (horrifyingly expensive and underpowered, but small enough to fit in a manila envelope) and even my older Compaq Aero (though its 7.5" screen was DEFINITELY too small for comfort by any objective standard and left me cross-eyed more than a few times).
No, actually it's perfectly capable of replicating fine wine... except for its compliance with the Fine Wine Freedom, Liberty and Justice Act of 2317 (and its European companion, the "Rule on High-quality Spirits-reproduction"), which requires that all replicators be equipped to recognize the molecular structure of a fine wine and produce malt liquor instead when asked to violate the intellectual property of hardworking vineyard owners...
As it so happens, I'm taking a break right this moment from the project that's consumed most of my free time for the past month, and likely to consume substantially all of it for the next week and a half: a string of 50 serial-networked Atmel ATtiny2313 microcontrollers, each juggling 4 independent RGB LEDs (total cost, not including support and development hardware, ~$250). As soon as I'm convinced that the bootloader I'm (hopefully) finishing this weekend can, in fact, successfully re-flash the chips once I've soldered them together and can't easily get at the SPI pins, the first 10-controller/40-LED segment will get built.
What will it (hopefully) be able to do? At the minimum, I'm hoping to program the following effects:
* each LED individually flickering like a candle flame, randomly and independently changing colors
* whole tree changing color in sync: red->yellow->green->cyan->blue->violet->red. This was actually a hard mode to program, because I had to re-do the entire LED-flashing algorithm to support temporarily combining the pair of 16-byte color buffers used by the two ports (each port controls 2 LEDs) into one big 32-byte buffer and increase the color resolution from 2 bits to 3 bits for slow whole-tree fades. I also had to change the pin assignments, to make sure that each port bit controlled an element of the same color on both ports.
* "Stupid Mode", where the controller sends 3-bytes in a row specifying the 2-bit RGB values for each of the four LEDs. I don't really plan to do much with this mode, but it's there as a backup plan if the bootloader scheme fails and I can't reflash the firmware at some future point. By cranking the baudrate up to 38.4k and speeding the controller up to 20MHz, I could probably pull off some decent effects. But I really want to make each controller smart and autonomous, so the controller just has to say, "group 3, slowly fade from red to violet, then hold using flicker effect #2 for 8 seconds before proceeding to the next instruction".
* of course, there's also "zombie" mode, activated by powering up the string without the controller. It's rather stupid and simple, but knows how to randomly change each LED's color every second or so. It's there to make sure that if I royally screw up the firmware on Christmas Eve, I'll still have an ok-looking tree to show off...
There will ultimately be 5 segments, connected together with a 5-pin DIN female socket on one end, and similar male plug on the other (3 wires total... Vcc, Gnd, and Serial data). One end will plug into the power supply and controller, which has a beefier microcontroller (ATMega162) to act like an orchestra's conductor and coordinate their activities when appropriate (and reflash the downstream microcontrollers when I update their firmware). Normally, all 50 chips will just watch the Serial line for 9600-baud serial broadcasts from the controller, but when given a reason and an opportunity to do so, they can individually pull the bus low to send crude signals back to the controller.
I'm still undecided about making them individually-addressable. My current plan is to make them group-addressable, with software-selectable group sizes of 3, 4, and 6. I can easily accomplish this by grounding none or one of the two "A" pins (3 possible groups: for 3-LED groups, they'd be 1-2-3-1, 2-3-1-2, and 3-1-2-3; for 4-LED groups, it would be irrelevant (each controller has 4 LEDs); for 6-LED groups, they'd be 1-2-3-4, 5-6-1-2, and 3-4-5-6. If I make them individually-addressable, I could do it the "easy" way and hardcode the base address of the first LED into each microcontroller's (basically-)immutable bootloader... but then I'll have to connect the strings in the right order. Alternatively, I can daisy-chain a fourth wire from the second "A" pin of each to chip to the first "A" pin of the next and use the neighbor-to-neighbor signaling to enumerate the string at reset. But really, as ideologically appealing as it seems to make each LED individually-addressable, I
Is there actually some real, (presumably orange) honest-to-god CHEESE that's actually called "American Cheese"? Because I've never seen the word "cheese" on so-called "American Cheese" that wasn't immediately followed by the qualifying word, "food". As in, "cheese food". If all American Cheese is, by definition, "Cheese Food" and NOT real cheese, how can there be such a thing as "American Cheese" in the first place? Or was there at one time such a thing as real American Cheese, but nobody ate it because it didn't melt nicely until some guy whose last name was "Kraft" invented "cheese food", figured out how to wrap the slices in heatshrink plastic to keep them from drying out and going bad, and turned (so-called) "American Cheese" into a viable product?
possession of the source != permission to disclose it.
Nevertheless, I think it's a straw man. Most likely, Diebold doesn't want to release THEIR source, and they're using Windows as the excuse. If Diebold wants to take the high road, let them release the source to THEIR code, including a diff file showing only their changes to the source received from Microsoft, and let some idiot bureaucrat in the state capitol make a fool of himself going on CNN and demanding the source code to Windows itself while everyone laughs at his pettiness.
> As one of those researchers, I've got to ask -- given that Taiwan is already breaking
> the Tamiflu patent, what makes you think my bosses are insane enough to invest that
> research money when the product is going to be confiscated?
Because even if they are, it's not causing Roche and everyone down the royalty chain to lose any actual money. Roche is already producing at 100% capacity, selling 100% of its output at its chosen price point, and refuses to license manufacturing rights to anyone else. As long as Roche is still selling 100% of its chosen capacity at its chosen price point, it has no right to bitch.
Personally, I take immense comfort from the high likelihood that right now, multiple factories in mainland China are almost certainly working around the clock, secretly cranking out pirated Tamiflu as fast as they possibly can, quietly filling warehouses with it, under the direction of one or more government officials motivated mainly by the thought of making billions of dollars and euros selling it to Americans and Europeans on eBay (or direct) for, say, a thousand dollars/euros for a 3-month twice-daily supply, when/if a real, honest to god pandemic strikes and the rights of Tamiflu's IP owners fall off the bottom of their list of concerns.
Think about it... the Chinese have little regard for IP anyway, they're faced with a potential future domestic crisis whose sole possible cure -- Tamiflu -- is already being rationed and suffering from limited availability. Does ANYONE *seriously* think they're going to sit back with their hands neatly folded, obediently refraining from violating Roche's IP and settling for the crumbs Roche might allocate to them at some outrageous, inflated price? And of course, if they DO make lots and lots of it, and demand far outstrips supply worldwide, the fact that they'll have to build a few new skyscrapers just to warehouse the money they'll make selling it abroad just seals the deal.
The genuine danger is that if no pandemic emerges within 5 years or so, some "bright" government official in China will decide to keep it from going to waste (since they won't be able to sell it, or even admit it exists, under any conditions besides an outright pandemic) and order it ground up and added to chicken feed. THAT would be a Very, Very Bad Thing(TM).
1080i60 = 1920 x 1080 @ 60 interlaced fields/second (540 scanlines redrawn every 1/60th sec)
720p60 = 1280 x 720 @ 60 progressive frames/second (all 720 scanlines redrawn every 1/60th sec)
480p60 is kind of messy.
There's the square-pixel variant, whose resolution is 640 x 480 @ 60 progressive frames/second.
There's also the rectangular-pixel 16:9 variant, whose resolution is (704 | 720) x 480 @ 60 progressive frames/sec. Offhand, I don't remember which one is the official one.
I think 480i60 has the same square-pixel (4:3) and rectangular-pixel (16:9) variants as 480p60.
Cost, and Joe Sixpack's stupidity.
:-(
.20mm dot pitch because it would be unacceptably dark due to the holes in the mask being so small, and would be gross overkill at that size and resolution anyway.) If I were going to buy a CRT-based HDTV, it would almost definitely be one of theirs for that precise reason.
:-)
Bandwidth-wise, 1080i60 and 540p60 are indistinguishable. 540p60 requires ~33.75KHz, while 480p60 requires the same as plain VGA: ~31.5KHz. In a CRT, this basically means the manufacturer can push a few components to the bounds of their tolerance and get away with using a single flyback transformer to handle both 480p60 and 1080i60. Since 720p60 requires ~45KHz, there's no way to fake it... you need an upgrade to handle at least two radically different scanrates. Which leads us to the second problem...
What do multisync monitors do when you radically switch display modes (say, between full-screen "DOS" and whatever resolution you normally run Windows at)? That's right... the screen turns black, a relay thunks, and the display comes back a second or two later. Unfortunately, Joe Sixpack isn't sophisticated enough to understand the engineering rationale behind the thunk-blink, and views it as a drawback. Especially when he's surfing between WWF Wrestling (in glorious 480i60) and ABC (in 720p60).
So, the way HDTV manufacturers look at it, they can make a TV that natively supports 480i60, 480p60, 720p60, and 1080i60, charge $20 more to cover the $2 extra it cost them to natively support 720p60 in addition to the rest, and then deal with endless bitching from unsophisticated customers who are mad because their TV goes "thunk" and blacks out when they're channel-surfing. Or, they can omit 720p60, pocket the savings, and be mostly left alone because average ignorant consumers wouldn't know the difference between 720p60 and 1080i60 if their lives depended on it
The ONLY company I know of that makes TVs that NATIVELY support 480i60, 480p60, 720p60, AND 1080i60 is Monivision. Their CRT-based HDTVs are basically big multisync monitors with increased dot pitch (it's not practical to make a 35" CRT with
Of course, there's a good solution to the "thunk" problem when channel-surfing: leave the TV's display mode alone and hack-convert everything to that mode while surfing, THEN (optionally) pick the best native display mode once the user hits another button on the remote. That way, Joe Sixpack can enjoy thunk-free surfing, and WE can enjoy HDTV programs at their native scanrates
The good news is that natively-progressive TVs are slowly edging out natively-interlaced HDTVs. The arrival of 1080p HD-DVDs will abolish any remaining demand for 1080i60 TVs among high-end users, and the flood of affordable 720p60 DLP, LCoS, and Plasma sets will edge them out among mid-range users. The remaining ones, who'll end up buying sub-32" CRT sets for years to come, will basically have TVs that display 480p60, even if they're technically doing 1080i60. With screens that small, the added horizontal detail of anything higher than 480p is lost and smeared anyway, and the scanlines are close enough together that the difference between 1080 sharp interlaced scanlines and 540 blurred progressive scanlines is academic (it's one of the reasons why it's practically impossible to look at a 27" "HDTV" set and figure out whether it's displaying something as 480p or 1080i... with low-end CRT-based HDTVs, the shadow mask is coarser than the nominal vertical resolution, and adjacent scanlines between fields end up lighting up the same phosphor triads anyway).
As lots of people tend to forget (even people WITH HDTVs):
* the difference between broadcast NTSC or composite NTSC and studio/dvd-quality NTSC (via s-video) is dramatic and noticeable.
* the difference between studio/dvd quality NTSC via s-video ("480i60") and 480p60 is night and day. Someone who just had PRK/LASEK the previous day could still tell the difference between the two on a 27" TV twenty feet away in a smoke-filled room.
* On a natively 720p60 set, 720p60 looks noticeably better than 480p60. On a natively 1080i60 set, 720p60 is almost indistinguishable from 480p60 because the TV downsamples to 540 scanlines, then kell-filters them to prevent flicker. The net result is almost the same amount of vertical detail as 480p60, with only slightly more horizontal detail. Meanwhile, the GPU and CPU are working almost twice as hard.
* On a natively 720p60 set, 1080i60 is nearly indistinguishable from 480p60, because the TV just throws away half of the scanlines, resamples the 540 that remain up to 720, then resamples the alleged 1920 horizontal pixels to 1280. On a natively 1080i60 set, 1080i60 COULD exhibit greater detail than 480p60... except for the tiny problems of interline twitter (requiring kell filtering), inadequate GPU/CPU power to really pull off their best work at 1920x1080, and the fact that programmers fundamentally don't understand the realities of computer-generated interlaced video and inevitably produce games that look great on the progressive-scan monitors connected to their dev boxen, but have glaring artifacts and deficiencies when viewed on a real interlaced display.
In short, 1080i60 doesn't have a whole lot to offer more than 480p60 for action-related games due to all the filtering necessary to prevent interline twitter, and inadequate raw GPU/CPU horsepower to really handle 1920x1080 properly. 720p60 has definite potential to offer better-looking games because 1280x720 is still a reasonable resolution as far as the GPU/CPU is concerned... but at the moment, natively 720p60 TVs only represent about 1/3 of the total in America (unfortunately), and 720p60 looks like $#!+ on most natively 1080i60 TVs.
At the implementation level, upgrading chips capable of 480p60 to 1080i60 is a comparatively small tweak, because most of the increased bandwidth goes into permitting faster pixel-to-pixel color changes. The actual scanrate (~33.75KHz) isn't a whole lot higher than 480p60's (~31.5kHz). Unfortunately, you can't fool Mother Nature... making the leap to 720p60 requires ~45KHz, because the real or metaphorical electron beam has a LOT more ground to cover in the same amount of time. Put another way, you can do some nasty hacks and claim that a given circuit is technically capable of "1080i", even if its REAL capabilities aren't much better than 480p because the horizontal detail will be all smeared and blurred due to inadequate bandwidth... but making the leap to 720p requires real upgrades that cost real money. And ultimately, the 2/3 majority of American HDTV owners whose sets are inherently 1080i won't see much of an improvement anyway, and will probably bitch about games that only support 480p and 720p.
The REAL surge in "HDTV" console gaming will come in another 2-4 years, once natively-720p TVs have displaced enough older 1080i TVs (at least among gamers who know the difference and care), and Nintendo's NEXT console WILL support 720p (but probably won't bother with 1080i, leading to more waves of grousing and complaining about its lack of "true" support for HDTV).
Whoops... typo. The sample IPV6 address should have 16 bytes, not 32:
. p"
;-)
//"Popo the red bunny lives across the street from atoona the dancing hippo and plays foozball with Matt the green cat every day"//
"aaa.bbb.ccc.dd.ee.f.ggg.hh.iii.jj.k.lll.mmm.nn.o
Even at 16 bytes, it's STILL too long to easily remember, though
Adding a bit to my earlier comment, it's IMPERATIVE that there be a standard way to express IPv6 addresses in a compressed form that's human-memory-friendly. If there isn't, I have no doubt that several dozen parties will come up with slightly different compression + expression schemes of their own, and at least a half-dozen will likely become co-standards... and anyone unfortunate enough to have to convert a human-format address into its binary form will have lots of fun trying to figure out by context what form it's in. Say...
"Cisco 1.0" form -- $8b 79 03 3e 2a 83 57 5c af 09 63$
"Microsoft form" -- ~aj8 pde 87g 3af k8m 75j h83 b3k
"SensibleSix" form -- %@kato lawa natu fooza, framooto towe kimu%
"SensibleSix (OpenFoo, Variant 2)" form -- @@kato lawa natutu fooza, framooto toway keemoo%
"Apple Form" --
"Linksys form" -- $$a2 28 66 3c 2a 93 fb$
"Cisco 2.0" form -- $!8b,79,3d,98,ea,74$
Let's all be honest... the biggest reason we fear IPV6 is the fact that we'll never, ever be able to actually remember IP addresses by memory anymore. "aaa.bb.ccc.ddd" fits nicely within the human short term memory stack. "aaa.bb.ccc.ddd.ee.f.ggg.hh.iii.j.kk.lll.mm.nn.ooo .p.qqq.rr.sss.tt.uuu.vvv.ww.x.yyy.zzz.AA.BBB.CCC.D .EEE.FF" overflows it, by several orders of magnitude. AT BEST, it might be possible to keep a tiny handfull of IP addresses straight, IF they're mostly zeroes AND have those zeroes in the same locations.
Instead of using all 128 bits for address space, let's set aside the first bit for human-friendly data compression: if the first bit is "1", we assume that 127 significant bits of address data follow. If it's "0", we'd transparently insert 64 "0" digits, then continue with another 63 significant digits of address info.
Yeah, it would cut the potential address space in half... but as more than a few have already pointed out, IPv6 isn't exactly hurting for address space anyway. The big benefit is that for the first few years/decades/centuries of use, nobody would ever actually HAVE to deal with a full-blown 128-bit address... IPV6 would have a de-facto 64-bit address space, ready to become 127 bits when the need finally arises. IP addresses would double in length, but 8 bytes are still within most people's capabilities. For the foreseeable future, nobody outside of Cisco or academia would actually have to bother with 32-byte IP addresses.
Short of that, the only way ANYONE will be able to deal with 128-bit addresses is if someone comes up with a scheme for constructing plausible nonsense words that unambiguously map to real 128-bit addresses, using letter combos to encode 2 to 4 bits per letter. Say, compressing big repeating chunks of zeroes and ones with escape patterns down to 40 or 50 bits, then using those bits to specify grapheme (morpheme? "nonsense word"?) consonant-vowel combos vaguely resembling romanicized Japanese, like "fa muka gu zade no kwatule yo". It would look weird, but by re-establishing some kind of contextual link to plausible language, people would at least have SOME chance of (mostly) getting it right from memory.
The whole idea of using normal power lines for data transmission is just plain silly, at least in urban areas. When you factor in the costs of all the extra hardware needed to inject and extract the data while blocking the electricity itself, then add the problems inherent in trying to send data down what's literally an unshielded bare wire, then compare the cost of doing that to, say, hanging a fibre bundle on the pole, the whole rationale for using the power lines goes down the toilet.
The ONLY reason American power companies care about broadband-over-power lines is because they want to be the third data carrier in their market, standing alongside the phone and cable companies. The catch is, due to government regulation, getting permission to hang fibre bundles next to power lines (with zero mutual interference) is a major problem, vulnerable to challenges by the incumbent phone and cable companies. But if they can somehow squeeze data down the power lines, it's easier to slide it past the regulators.
I guarantee -- the moment the FCC officially codifies the right of power companies to offer broadband and makes it an unalienable right not subject to refusal or regulation by states, every last power company in America will instantly abandon broadband over power lines and run fibre instead unless they're explicitly prohibited from doing so.
Yeah, but having your airbags triggered by a thief is several orders of magnitude worse than having a window broken. By law, the car commits electronic suicide the moment the airbags deploy and can't be driven, period, until a few thousand dollars worth of repair work has been done. Not just the ECU and airbags... the wires and sensors need to be replaced too. And then there's the matter of legal liability... any repair shop has to buy an insurance policy against someone who tries to sue in the future claiming the recertified airbag system deployed unnecessarily or failed to deploy when necessary due to some fault of theirs.
For all intents and purposes (in America, at least), a car whose airbags have deployed is effectively "totaled" because, from the insurance company's perspective, it's cheaper to pay the claim as a total loss and sell the car to a broker for export to some third-world country where the car can legally be repaired without the airbags and recertification than it is to pay to have it repaired, recertified, and liability-insured for use in the US.
EXACTLY what happened in Miami after Andrew. People who otherwise wouldn't have been particularly affected by Andrew turned on CNN, saw reports that people were paying $10 for a bag of ice in Miami, and promptly ran out, rented a U-Haul truck, filled it with bagged ice, and headed down to Miami.
When they arrived, they quickly discovered that they weren't alone. In fact, *SO MANY* people had the exact same idea that the price of ice fell to $5/bag within a day, then $3, then $2, then $1. People can bitch about how terrible it is to exploit disaster victims, but the fact is, due to greed, a HELL of a lot of bagged ice became available in Miami... ice that wouldn't otherwise have been available. While government agencies and relief organizations were holding meetings to plan their strategic response, millions of bags of ice were already on their way.
The truth is, guys were selling bagged ice for $2-3 out of trucks RIGHT ACROSS THE STREET from government aid caravans giving it away for free. The difference was, getting free ice meant waiting in line for an hour or two. Buying it from the guy with a truck might have cost ten bucks, but it meant you could be heading home in 30 seconds. Later, as competition increased, the ice guys started driving up and down streets like ice cream trucks and offered to-the-door delivery.
Personally, I wish stores WOULD jack up the cost of things like D cells and plywood before the storm... it would stop a$$#0735 from trying to buy a six-week supply at the last minute, wiping out the stores and ensuring that most people can't buy them at ANY price. If people KNEW that D cells would cost four times as much the day before a storm hits, and twice as much for a day or two afterward, they'd be more likely to stock up on them ahead of time... or at least, buy just enough to get through the next few days.
They were/are called "Field Emission Displays" (FED), if I recall correctly. The idea was to take a flat plate of glass (or equivalent) with normal CRT phosphors, but behind each phosphor put multiple tiny solid-state electron emitters. In other words, instead of taking a single electron beam and sweeping it repeatedly across the screen, put multiple weak electron beams behind each subpixel. The rationale for using multiple beams was avoidance of dead pixels. If there were, say, six emitters behind each subpixel, one or two of them could fail outright or progressively, and the viewer would likely never notice the difference... and the few that WERE bad enough to be noticeable could be programmed around by simply increasing the intensity of that specific subpixel.
I first found out about FED displays (yeah, I know "FED Display" is redundant) when I went to CES in Orlando in 1996(?). I thought for sure FED would rule the day (not for laptops, of course, because they used too much power... but for home TVs). I'm not really sure whether the concept proved to be unfeasible, or whether FED actually mutated something else. My guess is that cheap, hi-res DLP, far bigger and cheaper TFT panels than anyone dared to predict a decade ago, and cheaper & less-fragile plasma all chipped away at the manufacturers' motives for funding it, and they all went after one or more of the other technologies instead.
IMHO, it's a pity. DLP has good intensity, is indestructable, and high resolution, but suffers from either rainbows or high cost & convergence problems. TFT has high resolution, but is limited in brightness when used with projectors, and both size & pixel-perfection in direct displays. Brand new plasma sets look impressive, but they're the lowest-rez sets of all, and I'd personally rather watch a 34" CRT than endure ugly stretched images necessitated by plasma's vulnerability to burn-in due to pillarboxing instead. CRT... well, CRT has probably the best overall color saturation, but is size-limited by the tube itself (unless, of course, your living room is big enough to drive a forklift in).
Sigh. I really, REALLY like the concept of FED displays... they have/had the potential to give us the best features from all the other alternatives (though possibly at a staggering cost).
Fine... earn more revenue by putting power outlets *everywhere* with coin slots that give 10 minutes of power for a quarter instead of hiding and disabling every visible outlet in the terminal...
The perverse irony of power outlets in terminals is that the newest terminals have the fewest and most carefully-hidden outlets of all. Old terminals from the 1960s have outlets everywhere, because they wanted to make life easy for the janitorial staff.
The Copyright and Patent offices.
I hope to god PalmSource ends up acquiring Tapwave's IP, makes the DRM to use its features optional (enabling free open-source software to take advantage of it), and formally adopts it as the reference hardware platform for the next generation of PalmOS devices (particularly phones). With BOTH free AND commercial apps available, and Palm's official blessing as an optional standard (but for all intents and purposes mandatory for new PalmOS hardware going forward), it would be hard for them to fail. Everyone knows PocketPC hardware thrashes most Palm hardware... but people still buy Palm devices because it's a friendlier, more hackable platform that occasionally enables someone to do something cool and creative that Palm themselves never envisioned (can anyone even FATHOM Microsoft officially blessing a PocketPC equivalent of Hackmaster, let alone mentioning its existence and documenting its API?)