> Does your Mustang manage frame drops on home screen on dual core 1GHz SoC? >100ms audio lag? >100ms input lag? Android does.
Only if you run with stock CPU governor.
For a quick & dirty experiment to show what your phone is genuinely capable of if you were to root it and reflash with a kernel that supports a governor based on the 'interactive' strategy, try this: root your phone, then install SetCPU. Choose 'performance', which forces the phone to run at max speed. Keep in mind that with most stock kernels, you're still running single-core at this point. Nevertheless, compare your experience to what you normally see. Beware: it's very, VERY hard to letting any CPU governor run once you've gotten spoiled by 'performance' mode. But you'll have to, because performance mode will nuke most batteries in an hour or two.
I have three Android phones: a HTC Hero (CDMA), a Samsung Epic4G (Galaxy S), and a Motorola Photon. The Hero, overclocked and locked to ~711MHz, feels smoother and faster than my allegedly 1-GHz dual-core Photon. Graffiti input is flawless and error-free. On my Photon, I can hardly ever enter 10+ consecutive characters without recognition errors, partly because Motorola's stock kernel (which the bastards won't allow me to modify) thrashes the CPU speed all over the place (but mostly down) and confuses the hell out of Graffiti. It's really sad that a 16Mhz Palm Pilot circa 1997 could achieve nearly 100% accuracy, but a phone that's nominally 62.5 times faster can barely tell the difference between 'C' and "O". Locking the Photon's CPU to 100% (the one thing Moto's kernel grudgingly permits) fixes the problem totally. At least, until the battery dies an hour or two later.
I wish Google would find a way to force anybody who wants to sell a phone with Android Market (oops, I mean "Google Play". God, I hate that name...) to include a CPU governor based on 'interactive'. What differentiates 'interactive' from 'ondemand' (the one, and often the only, governor in most stock kernels)? Interactive kicks the CPU up to 100% speed whenever you do something that indicates that you're interacting with the phone... when you've hit the power button to turn on the screen and wake up the phone, when you're touching the screen or pressing a button, etc. It also increases CPU speed rapidly in response to load, and backs off on speed VERY slowly. The goal of 'interactive' scheduling is that you should never be forced to wait for anything to happen just because the CPU is running at less than full speed. It allows it to slow down when the phone is genuinely inactive, but prioritizes snappiness and lag-free use over a few minutes of battery life.
In contrast, the stock 'ondemand' governor hesitates to increase speed, and slows down the phone at the first hint of an excuse. Ondemand governors are the reason why we have phones with laggy lockscreens and lurching homescreen animations. The truth is, 99% of the lag complaints people have with Android are the fault of the 'ondemand' governor and its derivatives. Governors based on 'interactive' provide a MUCH nicer user experience, with minimal impact on battery life (the phone can still slow down to 100-200MHz when you turn off the display, or when it's just sitting on your homescreen and you haven't touched it in 5 minutes). 'Ondemand' == "evil".
Maybe Best Buy going 'bye-bye' will be the impetus that *finally* gets Fry's to open a few stores in Florida. Preferably, one of which will be near Fort Lauderdale on the Seminole Indian reservation (3 freeway exits southwest of downtown, with lots of prime Big Box-ready urban land next door to the Hard Rock hotel & casinino. No Florida sales tax, just whatever % the Seminoles negotiate with Fry's:-)
Ok, put an ad on Craigslist & become the local "cable whore" who does outcalls. If women & men can get away with running online escort services, you should be able to pimp Hdmi cables. Maybe even offer a free "happy ending" and connect it for them, too.;-)
> That, alone, will desktop kill Linux for non-techies.
OK, reality check... how many real, honest-to-god NON-techies do you know of who've EVER installed Linux on their own PC on their own initiative and without having a hardcore techie standing over them while they did it? Yeah, there are a few... but if you eliminate the non-techie users who didn't get introduced to Linux via bootable CD/DVD/flashdrive, those who had it installed for them, and those who otherwise ended up with it on their computer without actually doing the installation themselves, you're left with so few users that adding a requirement to do something like pull/place a mobo jumper would have statistically zero effect, anyway. And before anyone criticizes me for saying Linux is too hard for non-techie users to install, how many of those same non-techie users could install WINDOWS on a computer with a bare hard drive and nothing more than a retail Windows CD? Maybe 3 or 4 out of a hundred?
The truth is, we're lucky to have gotten what we did. Microsoft COULD have left the door open for mischief by allowing motherboard or PC makers to sell unlock codes as a "value-added" feature, or only available with special "developer" models (like Motorola's "Developer Edition" Razr, which has earned them plenty of justified hate since you can only buy it for full inflated price, can't buy it in America, and can't have an existing Razr converted over even though there's no technical reason why they couldn't do it.)
In the real world, you have to pick your battles and save your strength for the ones that matter. In a real sense, we've basically won this round -- PC/Mobo makers MUST give us a copy of the metaphorical key, period. And they have to do it for free. That's an epic win. Instead of grousing about Intel architecture keys, we SHOULD be fighting to get the same deal with ARM hardware.
A major problem is that many capabilities of modern graphic cards are as much about software as they are about hardware. Think back 15 years, to a period when dialup modems still mattered. Remember the hell and grief Linux users went through over "Winmodems"? Here's the punchline -- the hardware itself actually WAS abundantly well-documented. For the most part, a HSP winmodem is nothing more than a cheap soundcard with an RJ-11 jack and some parts to match the signal level between TTL logic and a live phone line. Or, if it was a higher-end Lucent card, it had a fairly generic DSP whose own datasheets were easily available. The problem is, knowing how that soundcard (or DSP) works is 1% of the job in writing "Linmodem" drivers, because everything past that point is software. That's part of the reason why the Asterisk project had a relatively easy time repurposing Winmodems into "phone interface cards" for interactive voice response systems -- they didn't TRY to be modems, and literally used them only as simple soundcards.
If nVidia came out with a new, totally alien GPU architecture, then personally handed Linux Torvalds a 3,000-page datasheet with register reference and a brief "theory of operation" section -- but no working open-source reference driver, and no working sample code, it would be about as useful for the development of a modern open-source 3D driver as the latest New York telephone book. Even if they ended up with working drivers, they wouldn't hold a candle to nVidia's own binaries, because the people writing the drivers would have only the most minimal idea of how to actually USE the raw bare-metal hardware sitting under them to achieve the desired 3D outcomes.
Microsoft has a knack for foisting inappropriate user interfaces on victims. Witness Windows Mobile, which was actually a very good mobile OS underneath its hideously dysfunctional skin, but was completely unusable "out of the box" in any efficient or pleasant manner. Nontechnical users bought WinMo phones, fought with them for a few hours, then angrily took them back to the store. More motivated users spent a month tweaking them, and eventually ended up with a phone that was quite nice & a definite step up from both the early iPhones and the last PalmOS phones. Web developers had nightmares about the possibility that Microsoft could have purchased Macromedia before Adobe did, because they knew Microsoft would have completely destroyed Dreamweaver (the one truly heavy-duty web editor out there, designed for the needs of users who go to work and spend their day doing high-end web design and NOT designed to accommodate middle-management users who want to "edit web pages" and have it work like Word) if they ever got their hands on it. And now, the way Microsoft is trying to foist Metro off on PC users.
If Windows 8 had a nice way to run Metro apps (maybe adding a narrow vertical taskbar-like area to one side of the screen) in a way that didn't interfere with core Windows use or try to make it the official Microsoft-sanctioned way of writing Windows apps, nobody would mind Metro. The problem is, they're trying to cripple DESKTOP Windows down to the level of a tablet.
IMHO, it's a suicidal strategy for Windows whose only "achievement" will be the rapid deprecation of the one realm where Microsoft enjoys unquestioned and nearly-total market domination. If Microsoft focused on keeping PCs as PCs, they'd eventually lose market share to tablets (possibly running IOS, Android, or Chrome) for users who aren't content creators, but they'd still totally own desktop PCs and could use Metro as a way of grabbing a share of the tablet market as well. As it stands, some year around 2018 very well might end up being the "Year of Desktop Linux", because they're going to end up driving content-creators and developers (at least, developers who aren't developing for Windows itself) away from Windows.
In the long run, I believe this will prove to be the great strength of open source software and operating systems like Linux. Ubuntu's corporate overlords might be equally insane, but there's nothing to stop Ubuntu users from kicking Unity to the curb and replacing it with KDE, Gnome, LXDE, or any other desktop environment of choice. If the divergence between the officially-sanctioned user environment and what actual users want becomes too great, users can give Ubuntu the finger and go their own way. That's not the case with Windows. If Microsoft decides that Windows 10 will eliminate the desktop PC metaphor altogether... it's gone, and the only alternative will be to stick with an older version of Windows until Microsoft takes the activation servers offline and quits reactivating old copies. Speaking of which... check the EULAs for Windows 8, 9, and beyond carefully.
I'm expecting Microsoft to change their licensing model at any time to one where you're merely licensing "Windows" for a specific PC, and they have the right to refuse to activate old copies of Windows at all after some date in return for free upgrades to newer versions of Windows. Regulators will love it, because it would technically be a great deal for average consumers. Microsoft would love it, because it would mean they could write off and officially quit supporting old versions of Windows a year or two after they come out with a new version... and prevent a future repeat of their "XP Problem" from happening, by ensuring that XP installations on computers that die "stay dead" instead of getting resurrected on a new Mobo, or surviving a major videocard & hard drive upgrade.
> Just don't buy lamps with any of those problems. Buy the ones without.
Easier said than done. It's not like low-quality CFLs have an "ingredient" list that includes entries like, "electrolytic capacitors made with low-quality electrolyte that are 58% more likely to burst within 18 months than the leading brand!", and plenty of expensive ones use the same crap capacitors and die just as quickly. Reviews don't help much, because a lot of these products are made by the same 2 or 3 factories in China and rebranded by everyone, so not even the company selling them whose name is on the package really, truly knows for sure about the quality of the parts used to make them. Chinese companies in particular are NOTORIOUS for "quality fade", where the first few batches are pristine and high-quality, then they start cutting corners and substituting components until you notice and scream... then they behave for a while, and go right back at it again.
The truth is, it's damn near impossible to buy consistently high-quality products anymore, because it's impossible for even an astute & motivated consumer to find any meaningful information about the quality of what he's buying. And no guarantee that the exact same brand & model won't be totally different the next time he goes shopping for a replacement.
Here's an easy challenge, just to show how hard it is: go buy a thirdparty micro-USB charger for a cell phone that implements the official charging standard AND can deliver the full 1.5 (let alone 2.1) amperes such a charger is theoretically supposed to be able to sustain. I guarantee that nearly every one you buy will fry itself if you try to actually draw that much power continuously for more than a few seconds or minutes. Most can sustain 700mA. A few can sustain 1000mA. But nowhere on ANY package will you see something like, "power delivery, 750mA continuous, 1000mA for up to 30 minutes when ambient temperature is below 40C". Going by what's printed on the package, they're all identical. They output 5v with some real (but unknown) upper limit to what they can sustain, and you'll never know what it is until you buy it and destroy it by drawing too much power for too long.
Touch lamps also rarely get along well with compact fluorescent bulbs, because most of them are built with SCRs and depend upon having a tiny bit of current flowing through a resistive load at all times to keep them powered up. That makes CFLs flicker visibly in a dark room when turned "off", and often causes lamps with them to spontaneously turn on.
X-10 resistive-load dimming controllers have the exact same problem. Their APPLIANCE controllers have a similar problem unless you open it up and cut the jumper that enables the "turn it off and on twice to turn it on from the local switch" feature. The phenomenon that causes the "off" CFL to flicker makes the circuit think you turned it off and on twice. Sometimes, you can fool it by plugging the lamp into an extension cord that ALSO has a nightlight with 2-watt incandescent bulb plugged in and turned on, but that's an ugly kludge that doesn't always work.
Let's complain about something more fundamental... CD jewelboxes. Worst. Design. Ever. At least 1/3 of my old CDs from high school and college have broken tabs on their jewelboxes. And the amount of space they take up is unholy. I actually CELEBRATED when some labels ditched jewelboxes in favor of album-like cardboard envelopes.
Then, let's complain about the construction of the discs themselves. I have discs from 25 years ago that barely have a visible scratch. I have a disc I bought last summer that almost became unplayable after I dropped it onto a tile floor in the living room. They've definitely lowered their durability and manufacturing standards over the past few years. It seems like new CDs are as fragile now as LPs used to be, if not more.
The power LED on my Mitsubishi DLP TV is the WORST. If you unplug the TV for more than "a few minutes", it has to cold boot when you plug it in and turn it on for the first time. For about two MINUTES, there's literally no sign of life -- no power LED, no sound, light engine off. Then, the power LED starts blinking. For about 5 minutes. FINALLY, right around the point when you're convinced the bulb burned out, the color wheel is broken, or a capacitor in the power supply has burst... it visibly powers on.
My Toshiba HD-DVD player was equally bad. You'd turn it on, and get NO SIGN OF LIFE for about 20 seconds. Apparently, whomever wrote the firmware forgot to toggle the bit that lights up the power LED. Jesus. And even then, it wasn't in a state where you could open and close the drawer for at least a minute or two. So, you'd press power, angrily tap your foot while counting to 30, press 'open' after seeing the light come on, then wait another 30-60 seconds for the drawer to actually slide open so you can insert the disc... which begins playing about a minute or two later.
Then, there's my parents' old DVD recorder... which didn't even HAVE a LED anywhere on it. Literally. Not even a red power LED. If you got it into a state where it was upconverting to a video standard your TV couldn't handle, you were basically screwed. Besides hearing the disc spin up, there was no way to know it was even alive by looking at it. And to think I bitched 20 years ago when the first VCRs came out that eliminated LED channel numbers and mode indicators, and went entirely to OSD. At least back then, you knew that if all else failed, you had to connect it to "composite", or turn the TV to channel 3 or 4. Now, if you fsck the menu settings, you could get it into a mode where you'll literally never see any sign of life unless you systematically try HDMI, component video, s-video, and composite. Oh, did I mention that the piece of junk can't even simultaneously output to composite and s-video? For a while, I *dreaded* going to see my parents, because EVERY SINGLE TIME I went to see them, their DVD recorder was in a dysfunctional state & they needed me to burn my afternoon trying to make it work again.
Well, I'm a bad person to debate the merits of wireless connectivity with. ~12 years ago, I had a Handspring Visor tethered to my bland Samsung cell phone with a $90 30" semi-custom cable that probably 3 other people in America owned. Every minute of 'online' time counted as a minute of airtime (well, not counting ~11pm-6am and weekends), it was a brisk 9600kbps (circuit-switched CDMA), and the browser was a cruel joke on a 160x160 1-bit monochrome display, but I still carried both the Visor and the serial cable with me everywhere I went, and ran my battery into the ground at least once or twice daily on weekends when I wasn't being metered.
On the other hand, if DirecTV took away my service tomorrow, I probably wouldn't notice for a few weeks. And if AT&T shut off my landline, I wouldn't notice until we either had a hurricane that blew down Sprint's tower (or their backup battery ran out, or their generator ran out of gas) or someone came to visit and the guardhouse didn't bother to call the second number in their "list of numbers to call when someone comes to visit" (my cell phone). But when Comcast has a cable internet outage while I'm at home, I usually know within a matter of seconds.;-)
> You're right, this boneheaded move by Microsoft is the best help they could possibly give for Linux on the desktop. > Of course, that just not let Microsoft off the hook for antitrust violations, specifically abusing its market power. > I can smell a new EU action on the the way, at the very least.
Unless I'm misunderstanding UEFI, that's not quite right. Contrary to the headline-hype, I believe Microsoft's OTHER explicit requirement for certification is that end users must be furnished with a way to disable it that's impossible to do by mistake, but entirely possible to do voluntarily. For example, flip a DIP switch, place or pull a jumper, enter a 32-character encryption code printed on a tiny sticker permanently affixed to the motherboard, etc.
Put another way, the UEFI rules won't stop a single Slashdot user from using Linux. Redhat is paying Microsoft for explicit approval so it can sell Redhat Linux to the OTHER potential Linux users who don't WANT to go through that much trouble to unlock their PC.
I'm sure Microsoft's motives with UEFI aren't entirely pure & MUST be scrutinized constantly, but so far, they've played everything by the book. They've guaranteed that we'll get a copy of the keys to our own systems, even if we'll have to get our hands slightly dirty to actually USE them.
Truth be told, I fear Microsoft less than the possibility of TiVO-ized Linux. God forbid, if someone decided to start giving away free laptops that are bootloader-locked to an Ubuntu variant and have advertising & "analytics" baked into the kernel & network stack, and eventually induce others to do the same thing, we're screwed. By 2020, we'll be in a position where a "free" PC hardwired to ad-supported Linux is "free", but a "non-free" "unlocked" PC costs $2,000... and can't play rented movies, run half the commercial applications out there, or access some paranoid bank web sites because it it's "untrusted". *THAT* is the scenario we have to fight like crazy and ensure never happens.
For the most part, Microsoft DOES behave itself in public. It might be grudgingly-good behavior, and it probably has plenty of impure thoughts, but as long as the EU and US are keeping an eye on it, it's unlikely to try anything blatant that would give it a permanent "hard" monopoly over x86 computing architecture.
As long as anybody can download Ubuntu and install it over a "free" copy of Windows, Microsoft is legally off the hook (in the US, at least), regardless of how few people actually *do* it. Microsoft would have to be completely *insane* to give up that magic "See, we aren't a real monopoly after all because end users can theoretically install Linux!" get-out-of-jail-free card. Linux is USEFUL to them. In the phone arena, Linux is practically a cash cow for Microsoft... they make more in royalties from the sale of an Android phone than they do in licensing fees when a phone running Windows gets sold.
> The goal though is the clients are fully generic.
Yeah, they're fully generic, as long as they're running Windows XP, a specific version of IE6, and have precisely the correct DLLs installed to handle the stupid "thin client browser-based application" the company had written during its thin client fetish circa 2001.
Or the slightly more modern one that would only work as long as your computer had a specific old version of Java installed, and would intentionally commit suicide if it detected that yours was any newer -- partly Sun's fault(*), mostly the development team's fault... they stupidly used the CLSID that means, "Ignore the JPI control panel settings and blindly use the latest one available on the computer", then followed it with a dozen if/then Javascript statements that probed the fsckin' applet for the version of the JPI that was running it and KILLED IT if ($version > 20). Jesus. God. Naked. On. A. Motorcycle. The breathtaking incompetence was unbelievable. Of course, this made things even MORE fun when YOU had a development role (unrelated applications), needed a newer JDK, and had to run the application under VMware so it could have its old JRE without screwing up Eclipse. Or when the old versions of Java became known to be vulnerable to a major exploit, and the entire company was prevented from upgrading for weeks because it would have broken that one application.
(*) I partly blame Sun, because their official documentation has ALWAYS done a totally crap job of documenting how JPI CLSID values work, and as a result the majority of JPI-dependent applets end up getting launched with CLSID values that are sub-optimal at best, and completely dysfunctional at worst. Hell, I've been using JPI for years, and pre-StackOverflow, *I* barely had a coherent understanding of what different values did and the contexts under which it was appropriate to use them.
You're assuming the data plan exists only to service Google Maps, as opposed to something you'd have anyway so you can be online 24/7 everywhere you go, with Google Maps as a freebie afterthought.
Personally, I get more net benefit from being able to go online and grab near-realtime TDWR weather radar maps. The weather maps put out by the media are almost ALWAYS 20-30, sometimes 60+, minutes behind. They show what the weather was like a half hour ago. TDWR is the weather radar used to determine whether it's safe to land a 747 on runway 17 *RIGHT NOW*. Between composite reflectivity and tilt-1 velocity, In can figure out in ~30 seconds whether or not it's worth waiting for the rain to let up before running across the parking lot.
Seriously. Last Sunday, I parked at Home Depot in the middle of a *torrential* downpour that had been going on for at least 10 minutes before I got there and parked. In the past, I would have run inside and gotten drenched. Instead, I pulled out PYKL3 (for NEXRAD3... not hi-res, but a more well-rounded app), then pulled out Tempest (free, and does TDWR). After zooming in on the TDWR display, I could see that I was literally on the trailing edge of it, and that it should be stopping any second. 2 minutes later, it stopped raining. Awesome.
Then, last night, after I got into my car at the office, I pulled it out to check the weather... and noticed a rapidly-approaching blob that went straight to red. I looked up, and less than a half mile away, I could see the waterfall-like deluge approaching (my office is next to an airport, so my normal line of sight would be about a mile anyway). I ended up sitting there for 10 minutes, getting electric thrills from the little white blobs (white == hail) nearby and a formation about 20 miles away (about 15 miles west of Boca Raton, still over the Everglades) that I thought for *sure* was a hook echo (hook== funnel cloud), but none of the spotters on the network said anything, so I guess it was a false alarm. (giddy laugh). Thanks to Android and realtime weather radar, torrential downpours and massive thunderstorms are now *fun* and exciting.:-)
Let's be honest, though... how many sales does TomTom lose to people running open map software, vs people with Android phones who use Google Maps because it's free? What compelling advantage to TomTom devices have over Android Phones with Google Maps? The irony of the article is that TomTom criticizes open maps for focusing on things like hiking trails, but that's actually one of the only areas where TomTom has any (temporary, lingering) potential advantage (because those same areas tend to have piss poor 3G coverage, so a device optimized for offline use might be a tiny bit more useful. Though Google Maps' new ability to cache maps for offline use kind of neutralizes that one, too.)
They also don't use NTFS or EXT2/3/4 as their live filesystems, and they rarely get their power cut without warning. They use filesystems like YAFFS, manage their flash directly, and have battery power that cuts the display and sends a "shut down NOW" signal long (in CPU terms) before the last drop of power literally goes away.
Contrast that with a desktop PC. There's no tradition of making the UPS part of the power supply, and quite a few run without them. So they really CAN lose power in an instant, with zero warning. Lots of SSDs tried to shave a buck off the manufacturing cost by eliminating the supercapacitor the engineers who did the reference designs inevitably made sure was there... and the reference firmware usually assumed there WAS a supercapacitor present, and made no attempt to deal gracefully with complete power loss because they made sure that all write operations could be completed in the time that supercapacitor was guaranteed to keep the drive alive if power were cut.
Now add the fact that they're running over SATA, which is really PATA with transparent seralization and deserialization taking place in the background and some extensions that get ignored when you're running in a backwards-compatible legacy mode. And the fact that Windows does everything it can to avoid letting you know when a drive is actually running in that legacy compatibility mode. Finally, add NTFS, FAT32(X), and EXT2/3/4 -- all of which were designed for a universe where hard drive failure tends to be progressive, rather than something that happens in an instant.
In short, half the problem with SSDs is due to the fact that we insist on treating them like something they aren't, run them on computers without backup power, and use filesystems that were optimized to avoid problems that are rare on SSDs, while totally ignoring real-world problems that are suicidally common with SSDs. And the other half is due to either bad design decisions (mandatory encryption, even if you value recoverability more than absolute security) and cost-cutting (elimination of supercapacitors the original engineers viewed as mandatory & non-negotiable).
The power issue is part of the reason why desktop PCs have more SSD-related problems than laptops. In most cases, unless the battery is defective/counterfeit, Windows will realize the battery is almost dead and forcibly shut down the computer long before the battery truly runs out of power. Once again, contrast that with a desktop PC, that can lose power without warning at any time.
The "legacy controller mode" problem is why more and more high-end SSDs for desktop PCs have abandoned SATA, and gotten their own embedded controllers on their own PCIexpress cards. It allows the maker to eliminate another major cause of problems by removing a few dozen abstraction layers and allows the controller to manage its flash directly, just like on a phone.
As a practical matter, if you want to use a SSD with a desktop PC, there are basically two sane ways to do it today: any SSD used purely as a write-through cache for a spinning disk, or MAYBE a recent GenuineIntel drive on a PCIexpress card running GenuineIntel firmware. Before you blindly run for Intel, though, be warned... they benchmark badly, precisely because they don't take reckless risks that multiply benchmark scores at the cost of having no "Plan B" if something goes wrong (*cough* Sandforce *cough*), and the real performance you'll see from an expensive Intel drive isn't likely to be much better than you'd get by pairing any cheap SSD with a spinning disk and using it as a write-through cache. Also, be aware that an Intel chipset is NOT good enough on its own -- thirdparty licensees will cut corners and modify the firmware in ways that eliminate most of Intel's added margin of safety. If you can't download your drive's firmware directly from intel.com, it's not really an "Intel" drive.
> I wonder if this problem could be worked around in Linux using RAID1. But instead of RAIDing it with another SSD, RAID it with a regular hard drive.
Um, yeah. It's called "using the SSD as write-through cache", and is pretty much what anybody who cares about not having his hard drives get corrupted every 3-6 months, but still wants SSD-ish performance, does now. Writes are still slow (because their speed is determined by the hard drive), but reads come straight off the SSD & the hard drive itself gets totally ignored. This is kind of what Seagate Momentus XT drives do, but on a whole-drive scale rather than just a few thousand of its most frequently-accessed sectors. There's at least one company that makes a controller for this specific purpose (you connect both the SSD and the traditional drive (which can be larger than the SSD; any additional space can be used for more partitions), and I believe OCZ makes an entire line of SSDs now whose only job in life is to be the write-through cache for a regular hard drive.
The obvious advantage over all-SSD use is that when the SSD takes a crap, the controller can just ignore the SSD, do its reads from the real hard drive, and notify the driver that the SSD has failed (yet again).
> The main downside, of course, is that at the moment, SSD has a significantly higher cost-per-byte than rotational storage.
That, and the fact that the embedded controllers in many of them spontaneously commit data-suicide every few weeks or months in a way that leaves your data practically impossible to recover (Sandforce controllers, in particular, encrypt everything, and automatically blow away the encryption key if you reflash the firmware, so if buggy firmware borks your data, you can't recover it with newer firmware, and you can't even recover it by having a data-recovery firm unsolder the chips and rip the bits from them, because everything is encrypted).
Of course, these aren't officially classified as "failures" for warranty purposes, because "securely erasing" and reformatting the drive makes it as "good" as new. The hardware itself is fine -- it's just the design that's fundamentally flawed in a way that leaves them vulnerable to instant, unrecoverable data loss (especially if Windows decides to put the computer into a sleep state at *exactly* the wrong moment). After you've gotten burned by it 3 times in a f**king year, it's enough to make you swear off of SSDs for a LONG time as anything more than a write-through cache.
Go ahead... find any site dedicated to high-performance hardware, find a reviewer who's been reviewing hardware for at least 2 years, and find ANYTHING he's written within the past 6 months that has anything good to say about SSDs. You'll find plenty of eager newbies raving about speed, but the people who've been getting burned by them for the past 2-4 years have just had it with them, and refuse to even take them seriously as real storage devices anymore. One reviewer compared SSDs to heroin, saying they're painful and hurt, but trying to do without their speed hurts even more, so you keep going back for more and more even though you know it's killing your data bit by bit.
Most cruelly of all, RAID won't save you from SSD failure, because their failure is due to events that trigger software bugs, and more often than not, if you have 3 identical SSDs doing RAID5 and something triggers that fatal condition, it's likely to trigger it in ALL of the drives simultaneously. Or at least two of them, which still leaves you SOL.
I'll buy my next SSD when they make encryption optional, and provide a recovery mode that lets me rip the raw bits off and reconstruct them offline with appropriate free recovery software. There's no technical reason why they can't do it, and abundant real-world data loss reasons why they MUST if they're going to remain viable as primary storage media. SSDs are too unreliable in their current state to trust with anything that has to persist longer than 5 minutes.
> Windows is the land of the "installer". Nothing is stopping you from bundling a JRE in your installer and installing it alongside your software if needed.
Um, actually, Sun DID restrict installers from downloading and installing Java, as well as restrict the bundling of a JRE installer with your app. It was their primary way of making money on Java. Java was "free as in beer" to anybody who personally went to java.sun.com, downloaded it, and installed it himself on his own computer, but if you wanted to automate the download & installation of a jdk in even the most trivial way, you had to pay MAJOR cash to Sun. Same as MySQL. If the end user downloaded and installed it, it was "free" (as in both liberty AND beer). But if you wanted to enable your software (which depended upon it) to install it automatically, you had two choices: release your software under the GPL, or pay MySQL a shit-ton of money for the right to automate its download & installation as part of your own software's installation. The main difference was that you COULD automate MySQL's installation by releasing your software as GPL, whereas (AFAIK), there was NOTHING that would allow you to automate Java's installation without paying cold, hard cash to Sun.
It was actually a brilliant strategy for Sun. Sneaky... but brilliant, because it meant that 99% of the programmers who developed applications in Java had NO IDEA that the restrictions even existed, because they were written in a way that didn't affect them at all. It wasn't until they formed a company and went to sell their program to naive end users that they ran into Sun's bank vault wall and found out that it was going to cost MAJOR amounts of money to automate the installation process for their users.
Oracle merely maintained the status quo that existed long before the purchase.
> If MS sticks to its guns and tries to force Metro on everyone, I think it really will be the end of them, at least in their current monolithic form. > I don't think they can afford another Vista or another poor assault on the mobile space, and Windows 8 has the potential to be both at the same time.
If Ubuntu weren't equally determined to cut its wrists and ram Unity down people's throats, I might be inclined to agree with you. The problem we face now is that all three mainstream desktop platforms (Windows, OS/X, and Ubuntu Linux) are going down the shithole together. It's already shredded Ubuntu's viability. Stop and reflect about this for a moment -- in less than a year, "Ubuntu" has managed to go from being practically a synonym for "desktop Linux", to a second-place contender to a distro 99% its users had never even HEARD of a year ago ("Mint"). If Microsoft had any sanity, that kind of nearly-overnight loss of market share would scare the bejesus out of them. Not because Microsoft has to worry about people switching, but rather because by now it should be quite painfully aware of how Windows users express their contempt and displeasure: by rejecting new versions of Windows and installing older ones, instead.
XP is a perfect example of how it can become a meme that explodes wildly out of control and takes on a life of its own. Truth be told, 94% of the people who express a preference for XP over Vista can't actually give any coherent reason for their decision, or at least not any reason that later releases didn't resolve. It doesn't matter. Once the meme that "Vista == bad" took root, it became impossible for even Microsoft to overcome. There are people who wouldn't have upgraded from XP to Vista, even if Microsoft gave away the discs for free just so they could wash their hands of XP forever. As it stands, Microsoft is setting itself up for a repeat of its Vista debacle with Windows 8.
As further proof that Microsoft has gone insane, I offer another story from Slashdot a few days ago, announcing Microsoft's decision to "de-prettify" Windows 8's desktop. Or, put another way, Microsoft is determined to take another jab at anyone who hates Metro and punish them by making its desktop environment a step backwards from Windows 7. Jesus Christ naked on a unicycle. Microsoft really, truly DOES appear determined to ensure that nobody who's running Windows 7 has any sane reason to ever get Windows 8. And if Microsoft quits selling Windows 7 & basically forces people to pirate it to use it instead of Windows 8, they'll probably have the gall to chalk Windows 8's failure to "inadequate DRM" instead of "it really fsck'ing SUCKED compared to what it replaced".
Windows developers have almost the exact same problem as Android developers. There's exactly one officially-blessed IDE, and just about every book, blog, howto, and forum post assumes you're using it.
NBandroid is a noble project that many people work hard maintaining. Unfortunately, it gets zero love from Google, and as a result, support for the latest and greatest Android SDK tends to arrive about a month or two after one of its developers gets a new phone that uses it. Like Eclipse, it has some bugs. Truthfully, most of them are minor... IF you've completely mastered Netbeans, Eclipse, Android development, and the use of build scripts. Otherwise, it'll probably stop you dead in your tracks, with little hope of moving forward any time soon unless you throw in the towel, move everything over to Eclipse, and hope that the situation improves for your next Android project.
The story with Windows is more or less the same. If you have a problem building a C# program under VS10, you can find four hundred resources online to help troubleshoot it in 18 seconds with Google. Have a similar problem with something like SharpDevelop, Eclipse, Netbeans, or another non-VS IDE, and you'll probably be looking for the answer for quite a while.
It's even worse if your native language isn't English. Visual Studio is so pervasive worldwide, even people who speak regional languages can find abundant help in their own language. There might even be one or more entire BOOKS about Visual Studio in it. Deviate from Microsoft's chosen path, and you'd better be fluent in English. OK, I'm exaggerating a little... lots of the independent IDEs are written by authors in non-English-speaking countries, and provide support in their own language as well.
At one time, I would have been optimistic and said that Microsoft's future lack of free support might encourage more progress with free alternatives. Three years of Android development have disillusioned me. NBandroid has come a long way and made enormous amounts of progress, but thanks to Google's total contempt for Netbeans, it still ends up holding *me* back whenever I try using it, and there's no way in hell I could recommend it to somebody who's learning Android programming for the first time. And we're talking about a Java development ecosystem that has historically had only TWO viable free IDEs, both of which were widely viewed as the two best IDEs available, period. Compare that to Windows, where NONE of the alternatives has market share that would count as "sloppy seconds" compared to the overwhelming dominance of Visual Studio, and all of which have real drawbacks and disadvantages compared to Visual Studio.
At the end of the day, Visual Studio is kind of like a 97 year old benevolent dictator of a prosperous country who's been ruling since he was a teenager -- people might have complaints, especially if he starts getting senile in his old age... but he's been the only government anybody in that country has ever known, and not even his fiercest opponents can really see themselves taking his place, because over the past 80-odd years, he's basically become synonymous with the country, its government, and the cultural identity of the people who live there. With the possible exception of Commodore 64 BASIC, it's hard to think of any development environment that's been more dominant and pervasive within its platform than Visual Studio within its platform and era.
Actually, there's an even better reason to stick with cheap Android tablets from China -- the market for LCDs in particular is *extremely* volatile. It's almost suicidal to design a product around a LCD screen you haven't already bought sufficient quantities of to build every unit you plan to build, because any given screen could become unavailable at any time without warning.
At least with Android tablets, you have a relatively stable programming environment. And if you really want to get down & dirty with the bare metal, any company in Shenzhen will sell you Android tablets that are pre-rooted, have totally open bootloaders, and possibly come ready to triple-boot pre-installed copies of Android, Ubuntu, and Windows Mobile (yeah, I said "Windows Mobile"). Sticking with Android tablets means not having to worry much about your supply chain. Low-end Chinese Android devices are about as close to commodities as you can get -- most use the same Rockchip innards, and run minimally-tweaked copies of Rockchip's reference distro of Android. If one supplier goes away, you can almost certainly find 20 more selling compatible tablets for $10 less.
If the tablet has a micro-SD card and you need it to use as an embedded controller for something, you can even repurpose the microSD card slot as an accessible SPI port with a few extra GPIOs. Remove the tablet from its case, put it in a new case of your own, connect it to your circuit with the nasty hacked microSD-to-SPI patch cord you had someone solder together, and you're good to go. It wouldn't even surprise me if commodity Chinese Android tablets HAVE additional headers on their circuit boards to expose things like SPI, I2C, GPIO, and UARTs to other products that might use them as a component.
If I simply said that in my post-IPV6 network my 192.168.100.1 PC was 2001:55c:37:e94::100, somebody would have felt compelled to point out that ipv6 addresses are hexadecimal and missed the point that there's no requirement that you actually USE digits A-F.
Likewise, if I didn't include the background about zeroconf and the obligatory note about a router not being strictly needed in a purely-ipv6 network, there would have been a half-dozen replies pointing out that you don't actually *have* to do anything NAT_like with IPv6.
It's unfortunately, but you aren't allowed to make IPv6 sound human-manageable in polite company. Its supporters will shout you down and make it sound like the theory behind fractal image comppression. And if that fails, they'll beat you up and tell you to use DNS instead of raw IP addresses, totally missing the point (raw IP addresses always work, even when DNS is totally borked).
Culturally, IPv6 is very much the realm of academics who refuse to tolerate the equivalent of Cliff's notes. If you don't exhaustively document your arguments and prove that you did your homework, they'll attack and criticize you. It's a shame. Especially now that Comcast managed to rip away the shroud of academic abstraction that formerly clouded the whole topic, and has come up with a viable way to let their customers (for the next few years, at least) have their cake & eat it too... a personal/64 ipV6 network prefix to call your own, plus a real IPv4 address to fall back on until the day it becomes as moot as a fax modem (when you realize you set up the computer weeks/months ago, and never bothered to install the drivers for it).;-)
> Lets see YOU sir figure up an IP V6 address map for...lets say a 40 person small business, in your head.
Actually, it doesn't *have* to be as bad as you think, regardless of what some of the more ardent IPv6 advocates might say (IMHO, they're their own worst enemies who manage to forge arguments & disagreement that's completely unnecessary for the sake of pedantic academic abstraction).
Let's look at how things work now... you basically have two chunks of data to remember:
1. Your public IP address: a.b.c.d
2. Your private IP addresses, all of which almost certainly fall within a block of 253 between 1 and 254 of some C-like/24 network. In most cases, every computer you deal with has the same first three values, and differs only with respect to the last. Going a step further, you probably don't randomly use the whole range... your router is probably x.y.z.1, your DHCP is probably x.y.z.200-x.y.z.250, you probably have a bunch of virtual port-forwarded servers clumped between x.y.z.10 and x.y.z.20, and the PC you personally use is probably x.y.z.100 (just to make your life easy).
Now... let's suppose you're a Comcast customer. In all likelihood, the first two chunks of your IPv6 address will be 2001:055c (or, if you prefer, 2001:55c).
As a practical matter, most IP addresses you're likely to see for the next 10 years will probably begin with 2001, or one of a half-dozen or so values common to nearly everyone in America. It doesn't have to be that way, and won't be that way forever, but for now, think of the old pre-cellphone-number-portability days when 99% of the people you called on a daily basis were in one, maybe... MAYBE two, area codes. And it never even OCCURRED to you that "1" was America's country code... it was just the digit you dialed first when making a long-distance call (and later, the digit that affirmed your understanding that the call might not be "free").
Anyway, that brings us to 055c -- Comcast's first netblock. Every IP address that begins with 2001:055c belongs to Comcast. In all likelihood, AT&T will have a similar block for DSL & U-Verse, and Verizon will have one for DSL & FIOS. In theory, they could have one for DSL and a different one for FIOS, but keep in mind the next two chunks, which will specifically define the second half of the 64-bit prefix assigned to you. Those two blocks -- 32 bits -- include more valid prefixes that can be assigned to customers than there are usable addresses in the entire IPv4 address space. So, there's plenty to go around.
Let's suppose for a moment that you get lucky, and as an early-adopter, you end up with a prefix like 2001:055c:0037:0e94. If you prefer, it can be written as 2001:55c:37:e94. Now, compare that to memorizing 64.192.87.239. The "2001" fades into the background noise (for now), because just about every address you'll see begins with it anyway. That leaves 55c:37:394. Is that really any harder to memorize than 64.192.87.239?
"But wait," you protest... what about the last 64 bits? No human can ever remember some god-awful address like 2001:55c:37:e94:8d1f:ca90:ff03:2109!"
And you're absolutely right. Which is why your desktop PC, whose internal IP address is currently 192.168.100.100, has as its fixed IP address 2001:55c:37:e94::100
"But... But... I read somewhere that the last 64 bits would be detmined by the MAC address!". And you're right... if you leave Windows up to its defaults. That's how zeroconf works. And the nice thing is, as a computer-savvy invididual, you don't have to do it that way. There's absolutely NOTHING to stop you from going into Windows' network control panel and explicitly setting your ethernet adapter's IPv6 address to 2001:055c:0037:0e94:0000:0000:0000:0100 (which can be equally represented as "2001:55c:37:e94::100").
Much nicer, no?
Of course, the trailing "100" isn't "really" a 100... it's actually 256... but neither your computer nor Comcast has to know that. You're perfectly free to limit 16-bit chun
Slashdot Mirror
> Does your Mustang manage frame drops on home screen on dual core 1GHz SoC? >100ms audio lag? >100ms input lag? Android does.
Only if you run with stock CPU governor.
For a quick & dirty experiment to show what your phone is genuinely capable of if you were to root it and reflash with a kernel that supports a governor based on the 'interactive' strategy, try this: root your phone, then install SetCPU. Choose 'performance', which forces the phone to run at max speed. Keep in mind that with most stock kernels, you're still running single-core at this point. Nevertheless, compare your experience to what you normally see. Beware: it's very, VERY hard to letting any CPU governor run once you've gotten spoiled by 'performance' mode. But you'll have to, because performance mode will nuke most batteries in an hour or two.
I have three Android phones: a HTC Hero (CDMA), a Samsung Epic4G (Galaxy S), and a Motorola Photon. The Hero, overclocked and locked to ~711MHz, feels smoother and faster than my allegedly 1-GHz dual-core Photon. Graffiti input is flawless and error-free. On my Photon, I can hardly ever enter 10+ consecutive characters without recognition errors, partly because Motorola's stock kernel (which the bastards won't allow me to modify) thrashes the CPU speed all over the place (but mostly down) and confuses the hell out of Graffiti. It's really sad that a 16Mhz Palm Pilot circa 1997 could achieve nearly 100% accuracy, but a phone that's nominally 62.5 times faster can barely tell the difference between 'C' and "O". Locking the Photon's CPU to 100% (the one thing Moto's kernel grudgingly permits) fixes the problem totally. At least, until the battery dies an hour or two later.
I wish Google would find a way to force anybody who wants to sell a phone with Android Market (oops, I mean "Google Play". God, I hate that name...) to include a CPU governor based on 'interactive'. What differentiates 'interactive' from 'ondemand' (the one, and often the only, governor in most stock kernels)? Interactive kicks the CPU up to 100% speed whenever you do something that indicates that you're interacting with the phone... when you've hit the power button to turn on the screen and wake up the phone, when you're touching the screen or pressing a button, etc. It also increases CPU speed rapidly in response to load, and backs off on speed VERY slowly. The goal of 'interactive' scheduling is that you should never be forced to wait for anything to happen just because the CPU is running at less than full speed. It allows it to slow down when the phone is genuinely inactive, but prioritizes snappiness and lag-free use over a few minutes of battery life.
In contrast, the stock 'ondemand' governor hesitates to increase speed, and slows down the phone at the first hint of an excuse. Ondemand governors are the reason why we have phones with laggy lockscreens and lurching homescreen animations. The truth is, 99% of the lag complaints people have with Android are the fault of the 'ondemand' governor and its derivatives. Governors based on 'interactive' provide a MUCH nicer user experience, with minimal impact on battery life (the phone can still slow down to 100-200MHz when you turn off the display, or when it's just sitting on your homescreen and you haven't touched it in 5 minutes). 'Ondemand' == "evil".
Maybe Best Buy going 'bye-bye' will be the impetus that *finally* gets Fry's to open a few stores in Florida. Preferably, one of which will be near Fort Lauderdale on the Seminole Indian reservation (3 freeway exits southwest of downtown, with lots of prime Big Box-ready urban land next door to the Hard Rock hotel & casinino. No Florida sales tax, just whatever % the Seminoles negotiate with Fry's :-)
Ok, put an ad on Craigslist & become the local "cable whore" who does outcalls. If women & men can get away with running online escort services, you should be able to pimp Hdmi cables. Maybe even offer a free "happy ending" and connect it for them, too. ;-)
> That, alone, will desktop kill Linux for non-techies.
OK, reality check... how many real, honest-to-god NON-techies do you know of who've EVER installed Linux on their own PC on their own initiative and without having a hardcore techie standing over them while they did it? Yeah, there are a few... but if you eliminate the non-techie users who didn't get introduced to Linux via bootable CD/DVD/flashdrive, those who had it installed for them, and those who otherwise ended up with it on their computer without actually doing the installation themselves, you're left with so few users that adding a requirement to do something like pull/place a mobo jumper would have statistically zero effect, anyway. And before anyone criticizes me for saying Linux is too hard for non-techie users to install, how many of those same non-techie users could install WINDOWS on a computer with a bare hard drive and nothing more than a retail Windows CD? Maybe 3 or 4 out of a hundred?
The truth is, we're lucky to have gotten what we did. Microsoft COULD have left the door open for mischief by allowing motherboard or PC makers to sell unlock codes as a "value-added" feature, or only available with special "developer" models (like Motorola's "Developer Edition" Razr, which has earned them plenty of justified hate since you can only buy it for full inflated price, can't buy it in America, and can't have an existing Razr converted over even though there's no technical reason why they couldn't do it.)
In the real world, you have to pick your battles and save your strength for the ones that matter. In a real sense, we've basically won this round -- PC/Mobo makers MUST give us a copy of the metaphorical key, period. And they have to do it for free. That's an epic win. Instead of grousing about Intel architecture keys, we SHOULD be fighting to get the same deal with ARM hardware.
A major problem is that many capabilities of modern graphic cards are as much about software as they are about hardware. Think back 15 years, to a period when dialup modems still mattered. Remember the hell and grief Linux users went through over "Winmodems"? Here's the punchline -- the hardware itself actually WAS abundantly well-documented. For the most part, a HSP winmodem is nothing more than a cheap soundcard with an RJ-11 jack and some parts to match the signal level between TTL logic and a live phone line. Or, if it was a higher-end Lucent card, it had a fairly generic DSP whose own datasheets were easily available. The problem is, knowing how that soundcard (or DSP) works is 1% of the job in writing "Linmodem" drivers, because everything past that point is software. That's part of the reason why the Asterisk project had a relatively easy time repurposing Winmodems into "phone interface cards" for interactive voice response systems -- they didn't TRY to be modems, and literally used them only as simple soundcards.
If nVidia came out with a new, totally alien GPU architecture, then personally handed Linux Torvalds a 3,000-page datasheet with register reference and a brief "theory of operation" section -- but no working open-source reference driver, and no working sample code, it would be about as useful for the development of a modern open-source 3D driver as the latest New York telephone book. Even if they ended up with working drivers, they wouldn't hold a candle to nVidia's own binaries, because the people writing the drivers would have only the most minimal idea of how to actually USE the raw bare-metal hardware sitting under them to achieve the desired 3D outcomes.
Microsoft has a knack for foisting inappropriate user interfaces on victims. Witness Windows Mobile, which was actually a very good mobile OS underneath its hideously dysfunctional skin, but was completely unusable "out of the box" in any efficient or pleasant manner. Nontechnical users bought WinMo phones, fought with them for a few hours, then angrily took them back to the store. More motivated users spent a month tweaking them, and eventually ended up with a phone that was quite nice & a definite step up from both the early iPhones and the last PalmOS phones. Web developers had nightmares about the possibility that Microsoft could have purchased Macromedia before Adobe did, because they knew Microsoft would have completely destroyed Dreamweaver (the one truly heavy-duty web editor out there, designed for the needs of users who go to work and spend their day doing high-end web design and NOT designed to accommodate middle-management users who want to "edit web pages" and have it work like Word) if they ever got their hands on it. And now, the way Microsoft is trying to foist Metro off on PC users.
If Windows 8 had a nice way to run Metro apps (maybe adding a narrow vertical taskbar-like area to one side of the screen) in a way that didn't interfere with core Windows use or try to make it the official Microsoft-sanctioned way of writing Windows apps, nobody would mind Metro. The problem is, they're trying to cripple DESKTOP Windows down to the level of a tablet.
IMHO, it's a suicidal strategy for Windows whose only "achievement" will be the rapid deprecation of the one realm where Microsoft enjoys unquestioned and nearly-total market domination. If Microsoft focused on keeping PCs as PCs, they'd eventually lose market share to tablets (possibly running IOS, Android, or Chrome) for users who aren't content creators, but they'd still totally own desktop PCs and could use Metro as a way of grabbing a share of the tablet market as well. As it stands, some year around 2018 very well might end up being the "Year of Desktop Linux", because they're going to end up driving content-creators and developers (at least, developers who aren't developing for Windows itself) away from Windows.
In the long run, I believe this will prove to be the great strength of open source software and operating systems like Linux. Ubuntu's corporate overlords might be equally insane, but there's nothing to stop Ubuntu users from kicking Unity to the curb and replacing it with KDE, Gnome, LXDE, or any other desktop environment of choice. If the divergence between the officially-sanctioned user environment and what actual users want becomes too great, users can give Ubuntu the finger and go their own way. That's not the case with Windows. If Microsoft decides that Windows 10 will eliminate the desktop PC metaphor altogether... it's gone, and the only alternative will be to stick with an older version of Windows until Microsoft takes the activation servers offline and quits reactivating old copies. Speaking of which... check the EULAs for Windows 8, 9, and beyond carefully.
I'm expecting Microsoft to change their licensing model at any time to one where you're merely licensing "Windows" for a specific PC, and they have the right to refuse to activate old copies of Windows at all after some date in return for free upgrades to newer versions of Windows. Regulators will love it, because it would technically be a great deal for average consumers. Microsoft would love it, because it would mean they could write off and officially quit supporting old versions of Windows a year or two after they come out with a new version... and prevent a future repeat of their "XP Problem" from happening, by ensuring that XP installations on computers that die "stay dead" instead of getting resurrected on a new Mobo, or surviving a major videocard & hard drive upgrade.
> Just don't buy lamps with any of those problems. Buy the ones without.
Easier said than done. It's not like low-quality CFLs have an "ingredient" list that includes entries like, "electrolytic capacitors made with low-quality electrolyte that are 58% more likely to burst within 18 months than the leading brand!", and plenty of expensive ones use the same crap capacitors and die just as quickly. Reviews don't help much, because a lot of these products are made by the same 2 or 3 factories in China and rebranded by everyone, so not even the company selling them whose name is on the package really, truly knows for sure about the quality of the parts used to make them. Chinese companies in particular are NOTORIOUS for "quality fade", where the first few batches are pristine and high-quality, then they start cutting corners and substituting components until you notice and scream... then they behave for a while, and go right back at it again.
The truth is, it's damn near impossible to buy consistently high-quality products anymore, because it's impossible for even an astute & motivated consumer to find any meaningful information about the quality of what he's buying. And no guarantee that the exact same brand & model won't be totally different the next time he goes shopping for a replacement.
Here's an easy challenge, just to show how hard it is: go buy a thirdparty micro-USB charger for a cell phone that implements the official charging standard AND can deliver the full 1.5 (let alone 2.1) amperes such a charger is theoretically supposed to be able to sustain. I guarantee that nearly every one you buy will fry itself if you try to actually draw that much power continuously for more than a few seconds or minutes. Most can sustain 700mA. A few can sustain 1000mA. But nowhere on ANY package will you see something like, "power delivery, 750mA continuous, 1000mA for up to 30 minutes when ambient temperature is below 40C". Going by what's printed on the package, they're all identical. They output 5v with some real (but unknown) upper limit to what they can sustain, and you'll never know what it is until you buy it and destroy it by drawing too much power for too long.
Touch lamps also rarely get along well with compact fluorescent bulbs, because most of them are built with SCRs and depend upon having a tiny bit of current flowing through a resistive load at all times to keep them powered up. That makes CFLs flicker visibly in a dark room when turned "off", and often causes lamps with them to spontaneously turn on.
X-10 resistive-load dimming controllers have the exact same problem. Their APPLIANCE controllers have a similar problem unless you open it up and cut the jumper that enables the "turn it off and on twice to turn it on from the local switch" feature. The phenomenon that causes the "off" CFL to flicker makes the circuit think you turned it off and on twice. Sometimes, you can fool it by plugging the lamp into an extension cord that ALSO has a nightlight with 2-watt incandescent bulb plugged in and turned on, but that's an ugly kludge that doesn't always work.
Let's complain about something more fundamental... CD jewelboxes. Worst. Design. Ever. At least 1/3 of my old CDs from high school and college have broken tabs on their jewelboxes. And the amount of space they take up is unholy. I actually CELEBRATED when some labels ditched jewelboxes in favor of album-like cardboard envelopes.
Then, let's complain about the construction of the discs themselves. I have discs from 25 years ago that barely have a visible scratch. I have a disc I bought last summer that almost became unplayable after I dropped it onto a tile floor in the living room. They've definitely lowered their durability and manufacturing standards over the past few years. It seems like new CDs are as fragile now as LPs used to be, if not more.
The power LED on my Mitsubishi DLP TV is the WORST. If you unplug the TV for more than "a few minutes", it has to cold boot when you plug it in and turn it on for the first time. For about two MINUTES, there's literally no sign of life -- no power LED, no sound, light engine off. Then, the power LED starts blinking. For about 5 minutes. FINALLY, right around the point when you're convinced the bulb burned out, the color wheel is broken, or a capacitor in the power supply has burst... it visibly powers on.
My Toshiba HD-DVD player was equally bad. You'd turn it on, and get NO SIGN OF LIFE for about 20 seconds. Apparently, whomever wrote the firmware forgot to toggle the bit that lights up the power LED. Jesus. And even then, it wasn't in a state where you could open and close the drawer for at least a minute or two. So, you'd press power, angrily tap your foot while counting to 30, press 'open' after seeing the light come on, then wait another 30-60 seconds for the drawer to actually slide open so you can insert the disc... which begins playing about a minute or two later.
Then, there's my parents' old DVD recorder... which didn't even HAVE a LED anywhere on it. Literally. Not even a red power LED. If you got it into a state where it was upconverting to a video standard your TV couldn't handle, you were basically screwed. Besides hearing the disc spin up, there was no way to know it was even alive by looking at it. And to think I bitched 20 years ago when the first VCRs came out that eliminated LED channel numbers and mode indicators, and went entirely to OSD. At least back then, you knew that if all else failed, you had to connect it to "composite", or turn the TV to channel 3 or 4. Now, if you fsck the menu settings, you could get it into a mode where you'll literally never see any sign of life unless you systematically try HDMI, component video, s-video, and composite. Oh, did I mention that the piece of junk can't even simultaneously output to composite and s-video? For a while, I *dreaded* going to see my parents, because EVERY SINGLE TIME I went to see them, their DVD recorder was in a dysfunctional state & they needed me to burn my afternoon trying to make it work again.
> Open means you cooked up something with a few preselected other vendors, in secrecy, and then released it
You're right. Thank ${deity} Google is so open with Android, and enthusiastically embraces community development and contributions.
Oh, wait a minute.
Sigh.
Well, I'm a bad person to debate the merits of wireless connectivity with. ~12 years ago, I had a Handspring Visor tethered to my bland Samsung cell phone with a $90 30" semi-custom cable that probably 3 other people in America owned. Every minute of 'online' time counted as a minute of airtime (well, not counting ~11pm-6am and weekends), it was a brisk 9600kbps (circuit-switched CDMA), and the browser was a cruel joke on a 160x160 1-bit monochrome display, but I still carried both the Visor and the serial cable with me everywhere I went, and ran my battery into the ground at least once or twice daily on weekends when I wasn't being metered.
On the other hand, if DirecTV took away my service tomorrow, I probably wouldn't notice for a few weeks. And if AT&T shut off my landline, I wouldn't notice until we either had a hurricane that blew down Sprint's tower (or their backup battery ran out, or their generator ran out of gas) or someone came to visit and the guardhouse didn't bother to call the second number in their "list of numbers to call when someone comes to visit" (my cell phone). But when Comcast has a cable internet outage while I'm at home, I usually know within a matter of seconds. ;-)
> You're right, this boneheaded move by Microsoft is the best help they could possibly give for Linux on the desktop.
> Of course, that just not let Microsoft off the hook for antitrust violations, specifically abusing its market power.
> I can smell a new EU action on the the way, at the very least.
Unless I'm misunderstanding UEFI, that's not quite right. Contrary to the headline-hype, I believe Microsoft's OTHER explicit requirement for certification is that end users must be furnished with a way to disable it that's impossible to do by mistake, but entirely possible to do voluntarily. For example, flip a DIP switch, place or pull a jumper, enter a 32-character encryption code printed on a tiny sticker permanently affixed to the motherboard, etc.
Put another way, the UEFI rules won't stop a single Slashdot user from using Linux. Redhat is paying Microsoft for explicit approval so it can sell Redhat Linux to the OTHER potential Linux users who don't WANT to go through that much trouble to unlock their PC.
I'm sure Microsoft's motives with UEFI aren't entirely pure & MUST be scrutinized constantly, but so far, they've played everything by the book. They've guaranteed that we'll get a copy of the keys to our own systems, even if we'll have to get our hands slightly dirty to actually USE them.
Truth be told, I fear Microsoft less than the possibility of TiVO-ized Linux. God forbid, if someone decided to start giving away free laptops that are bootloader-locked to an Ubuntu variant and have advertising & "analytics" baked into the kernel & network stack, and eventually induce others to do the same thing, we're screwed. By 2020, we'll be in a position where a "free" PC hardwired to ad-supported Linux is "free", but a "non-free" "unlocked" PC costs $2,000... and can't play rented movies, run half the commercial applications out there, or access some paranoid bank web sites because it it's "untrusted". *THAT* is the scenario we have to fight like crazy and ensure never happens.
For the most part, Microsoft DOES behave itself in public. It might be grudgingly-good behavior, and it probably has plenty of impure thoughts, but as long as the EU and US are keeping an eye on it, it's unlikely to try anything blatant that would give it a permanent "hard" monopoly over x86 computing architecture.
As long as anybody can download Ubuntu and install it over a "free" copy of Windows, Microsoft is legally off the hook (in the US, at least), regardless of how few people actually *do* it. Microsoft would have to be completely *insane* to give up that magic "See, we aren't a real monopoly after all because end users can theoretically install Linux!" get-out-of-jail-free card. Linux is USEFUL to them. In the phone arena, Linux is practically a cash cow for Microsoft... they make more in royalties from the sale of an Android phone than they do in licensing fees when a phone running Windows gets sold.
> The goal though is the clients are fully generic.
Yeah, they're fully generic, as long as they're running Windows XP, a specific version of IE6, and have precisely the correct DLLs installed to handle the stupid "thin client browser-based application" the company had written during its thin client fetish circa 2001.
Or the slightly more modern one that would only work as long as your computer had a specific old version of Java installed, and would intentionally commit suicide if it detected that yours was any newer -- partly Sun's fault(*), mostly the development team's fault... they stupidly used the CLSID that means, "Ignore the JPI control panel settings and blindly use the latest one available on the computer", then followed it with a dozen if/then Javascript statements that probed the fsckin' applet for the version of the JPI that was running it and KILLED IT if ($version > 20). Jesus. God. Naked. On. A. Motorcycle. The breathtaking incompetence was unbelievable. Of course, this made things even MORE fun when YOU had a development role (unrelated applications), needed a newer JDK, and had to run the application under VMware so it could have its old JRE without screwing up Eclipse. Or when the old versions of Java became known to be vulnerable to a major exploit, and the entire company was prevented from upgrading for weeks because it would have broken that one application.
(*) I partly blame Sun, because their official documentation has ALWAYS done a totally crap job of documenting how JPI CLSID values work, and as a result the majority of JPI-dependent applets end up getting launched with CLSID values that are sub-optimal at best, and completely dysfunctional at worst. Hell, I've been using JPI for years, and pre-StackOverflow, *I* barely had a coherent understanding of what different values did and the contexts under which it was appropriate to use them.
You're assuming the data plan exists only to service Google Maps, as opposed to something you'd have anyway so you can be online 24/7 everywhere you go, with Google Maps as a freebie afterthought.
Personally, I get more net benefit from being able to go online and grab near-realtime TDWR weather radar maps. The weather maps put out by the media are almost ALWAYS 20-30, sometimes 60+, minutes behind. They show what the weather was like a half hour ago. TDWR is the weather radar used to determine whether it's safe to land a 747 on runway 17 *RIGHT NOW*. Between composite reflectivity and tilt-1 velocity, In can figure out in ~30 seconds whether or not it's worth waiting for the rain to let up before running across the parking lot.
Seriously. Last Sunday, I parked at Home Depot in the middle of a *torrential* downpour that had been going on for at least 10 minutes before I got there and parked. In the past, I would have run inside and gotten drenched. Instead, I pulled out PYKL3 (for NEXRAD3... not hi-res, but a more well-rounded app), then pulled out Tempest (free, and does TDWR). After zooming in on the TDWR display, I could see that I was literally on the trailing edge of it, and that it should be stopping any second. 2 minutes later, it stopped raining. Awesome.
Then, last night, after I got into my car at the office, I pulled it out to check the weather... and noticed a rapidly-approaching blob that went straight to red. I looked up, and less than a half mile away, I could see the waterfall-like deluge approaching (my office is next to an airport, so my normal line of sight would be about a mile anyway). I ended up sitting there for 10 minutes, getting electric thrills from the little white blobs (white == hail) nearby and a formation about 20 miles away (about 15 miles west of Boca Raton, still over the Everglades) that I thought for *sure* was a hook echo (hook== funnel cloud), but none of the spotters on the network said anything, so I guess it was a false alarm. (giddy laugh). Thanks to Android and realtime weather radar, torrential downpours and massive thunderstorms are now *fun* and exciting. :-)
Let's be honest, though... how many sales does TomTom lose to people running open map software, vs people with Android phones who use Google Maps because it's free? What compelling advantage to TomTom devices have over Android Phones with Google Maps? The irony of the article is that TomTom criticizes open maps for focusing on things like hiking trails, but that's actually one of the only areas where TomTom has any (temporary, lingering) potential advantage (because those same areas tend to have piss poor 3G coverage, so a device optimized for offline use might be a tiny bit more useful. Though Google Maps' new ability to cache maps for offline use kind of neutralizes that one, too.)
They also don't use NTFS or EXT2/3/4 as their live filesystems, and they rarely get their power cut without warning. They use filesystems like YAFFS, manage their flash directly, and have battery power that cuts the display and sends a "shut down NOW" signal long (in CPU terms) before the last drop of power literally goes away.
Contrast that with a desktop PC. There's no tradition of making the UPS part of the power supply, and quite a few run without them. So they really CAN lose power in an instant, with zero warning. Lots of SSDs tried to shave a buck off the manufacturing cost by eliminating the supercapacitor the engineers who did the reference designs inevitably made sure was there... and the reference firmware usually assumed there WAS a supercapacitor present, and made no attempt to deal gracefully with complete power loss because they made sure that all write operations could be completed in the time that supercapacitor was guaranteed to keep the drive alive if power were cut.
Now add the fact that they're running over SATA, which is really PATA with transparent seralization and deserialization taking place in the background and some extensions that get ignored when you're running in a backwards-compatible legacy mode. And the fact that Windows does everything it can to avoid letting you know when a drive is actually running in that legacy compatibility mode. Finally, add NTFS, FAT32(X), and EXT2/3/4 -- all of which were designed for a universe where hard drive failure tends to be progressive, rather than something that happens in an instant.
In short, half the problem with SSDs is due to the fact that we insist on treating them like something they aren't, run them on computers without backup power, and use filesystems that were optimized to avoid problems that are rare on SSDs, while totally ignoring real-world problems that are suicidally common with SSDs. And the other half is due to either bad design decisions (mandatory encryption, even if you value recoverability more than absolute security) and cost-cutting (elimination of supercapacitors the original engineers viewed as mandatory & non-negotiable).
The power issue is part of the reason why desktop PCs have more SSD-related problems than laptops. In most cases, unless the battery is defective/counterfeit, Windows will realize the battery is almost dead and forcibly shut down the computer long before the battery truly runs out of power. Once again, contrast that with a desktop PC, that can lose power without warning at any time.
The "legacy controller mode" problem is why more and more high-end SSDs for desktop PCs have abandoned SATA, and gotten their own embedded controllers on their own PCIexpress cards. It allows the maker to eliminate another major cause of problems by removing a few dozen abstraction layers and allows the controller to manage its flash directly, just like on a phone.
As a practical matter, if you want to use a SSD with a desktop PC, there are basically two sane ways to do it today: any SSD used purely as a write-through cache for a spinning disk, or MAYBE a recent GenuineIntel drive on a PCIexpress card running GenuineIntel firmware. Before you blindly run for Intel, though, be warned... they benchmark badly, precisely because they don't take reckless risks that multiply benchmark scores at the cost of having no "Plan B" if something goes wrong (*cough* Sandforce *cough*), and the real performance you'll see from an expensive Intel drive isn't likely to be much better than you'd get by pairing any cheap SSD with a spinning disk and using it as a write-through cache. Also, be aware that an Intel chipset is NOT good enough on its own -- thirdparty licensees will cut corners and modify the firmware in ways that eliminate most of Intel's added margin of safety. If you can't download your drive's firmware directly from intel.com, it's not really an "Intel" drive.
> I wonder if this problem could be worked around in Linux using RAID1. But instead of RAIDing it with another SSD, RAID it with a regular hard drive.
Um, yeah. It's called "using the SSD as write-through cache", and is pretty much what anybody who cares about not having his hard drives get corrupted every 3-6 months, but still wants SSD-ish performance, does now. Writes are still slow (because their speed is determined by the hard drive), but reads come straight off the SSD & the hard drive itself gets totally ignored. This is kind of what Seagate Momentus XT drives do, but on a whole-drive scale rather than just a few thousand of its most frequently-accessed sectors. There's at least one company that makes a controller for this specific purpose (you connect both the SSD and the traditional drive (which can be larger than the SSD; any additional space can be used for more partitions), and I believe OCZ makes an entire line of SSDs now whose only job in life is to be the write-through cache for a regular hard drive.
The obvious advantage over all-SSD use is that when the SSD takes a crap, the controller can just ignore the SSD, do its reads from the real hard drive, and notify the driver that the SSD has failed (yet again).
> The main downside, of course, is that at the moment, SSD has a significantly higher cost-per-byte than rotational storage.
That, and the fact that the embedded controllers in many of them spontaneously commit data-suicide every few weeks or months in a way that leaves your data practically impossible to recover (Sandforce controllers, in particular, encrypt everything, and automatically blow away the encryption key if you reflash the firmware, so if buggy firmware borks your data, you can't recover it with newer firmware, and you can't even recover it by having a data-recovery firm unsolder the chips and rip the bits from them, because everything is encrypted).
Of course, these aren't officially classified as "failures" for warranty purposes, because "securely erasing" and reformatting the drive makes it as "good" as new. The hardware itself is fine -- it's just the design that's fundamentally flawed in a way that leaves them vulnerable to instant, unrecoverable data loss (especially if Windows decides to put the computer into a sleep state at *exactly* the wrong moment). After you've gotten burned by it 3 times in a f**king year, it's enough to make you swear off of SSDs for a LONG time as anything more than a write-through cache.
Go ahead... find any site dedicated to high-performance hardware, find a reviewer who's been reviewing hardware for at least 2 years, and find ANYTHING he's written within the past 6 months that has anything good to say about SSDs. You'll find plenty of eager newbies raving about speed, but the people who've been getting burned by them for the past 2-4 years have just had it with them, and refuse to even take them seriously as real storage devices anymore. One reviewer compared SSDs to heroin, saying they're painful and hurt, but trying to do without their speed hurts even more, so you keep going back for more and more even though you know it's killing your data bit by bit.
Most cruelly of all, RAID won't save you from SSD failure, because their failure is due to events that trigger software bugs, and more often than not, if you have 3 identical SSDs doing RAID5 and something triggers that fatal condition, it's likely to trigger it in ALL of the drives simultaneously. Or at least two of them, which still leaves you SOL.
I'll buy my next SSD when they make encryption optional, and provide a recovery mode that lets me rip the raw bits off and reconstruct them offline with appropriate free recovery software. There's no technical reason why they can't do it, and abundant real-world data loss reasons why they MUST if they're going to remain viable as primary storage media. SSDs are too unreliable in their current state to trust with anything that has to persist longer than 5 minutes.
> Windows is the land of the "installer". Nothing is stopping you from bundling a JRE in your installer and installing it alongside your software if needed.
Um, actually, Sun DID restrict installers from downloading and installing Java, as well as restrict the bundling of a JRE installer with your app. It was their primary way of making money on Java. Java was "free as in beer" to anybody who personally went to java.sun.com, downloaded it, and installed it himself on his own computer, but if you wanted to automate the download & installation of a jdk in even the most trivial way, you had to pay MAJOR cash to Sun. Same as MySQL. If the end user downloaded and installed it, it was "free" (as in both liberty AND beer). But if you wanted to enable your software (which depended upon it) to install it automatically, you had two choices: release your software under the GPL, or pay MySQL a shit-ton of money for the right to automate its download & installation as part of your own software's installation. The main difference was that you COULD automate MySQL's installation by releasing your software as GPL, whereas (AFAIK), there was NOTHING that would allow you to automate Java's installation without paying cold, hard cash to Sun.
It was actually a brilliant strategy for Sun. Sneaky... but brilliant, because it meant that 99% of the programmers who developed applications in Java had NO IDEA that the restrictions even existed, because they were written in a way that didn't affect them at all. It wasn't until they formed a company and went to sell their program to naive end users that they ran into Sun's bank vault wall and found out that it was going to cost MAJOR amounts of money to automate the installation process for their users.
Oracle merely maintained the status quo that existed long before the purchase.
> If MS sticks to its guns and tries to force Metro on everyone, I think it really will be the end of them, at least in their current monolithic form.
> I don't think they can afford another Vista or another poor assault on the mobile space, and Windows 8 has the potential to be both at the same time.
If Ubuntu weren't equally determined to cut its wrists and ram Unity down people's throats, I might be inclined to agree with you. The problem we face now is that all three mainstream desktop platforms (Windows, OS/X, and Ubuntu Linux) are going down the shithole together. It's already shredded Ubuntu's viability. Stop and reflect about this for a moment -- in less than a year, "Ubuntu" has managed to go from being practically a synonym for "desktop Linux", to a second-place contender to a distro 99% its users had never even HEARD of a year ago ("Mint"). If Microsoft had any sanity, that kind of nearly-overnight loss of market share would scare the bejesus out of them. Not because Microsoft has to worry about people switching, but rather because by now it should be quite painfully aware of how Windows users express their contempt and displeasure: by rejecting new versions of Windows and installing older ones, instead.
XP is a perfect example of how it can become a meme that explodes wildly out of control and takes on a life of its own. Truth be told, 94% of the people who express a preference for XP over Vista can't actually give any coherent reason for their decision, or at least not any reason that later releases didn't resolve. It doesn't matter. Once the meme that "Vista == bad" took root, it became impossible for even Microsoft to overcome. There are people who wouldn't have upgraded from XP to Vista, even if Microsoft gave away the discs for free just so they could wash their hands of XP forever. As it stands, Microsoft is setting itself up for a repeat of its Vista debacle with Windows 8.
As further proof that Microsoft has gone insane, I offer another story from Slashdot a few days ago, announcing Microsoft's decision to "de-prettify" Windows 8's desktop. Or, put another way, Microsoft is determined to take another jab at anyone who hates Metro and punish them by making its desktop environment a step backwards from Windows 7. Jesus Christ naked on a unicycle. Microsoft really, truly DOES appear determined to ensure that nobody who's running Windows 7 has any sane reason to ever get Windows 8. And if Microsoft quits selling Windows 7 & basically forces people to pirate it to use it instead of Windows 8, they'll probably have the gall to chalk Windows 8's failure to "inadequate DRM" instead of "it really fsck'ing SUCKED compared to what it replaced".
Windows developers have almost the exact same problem as Android developers. There's exactly one officially-blessed IDE, and just about every book, blog, howto, and forum post assumes you're using it.
NBandroid is a noble project that many people work hard maintaining. Unfortunately, it gets zero love from Google, and as a result, support for the latest and greatest Android SDK tends to arrive about a month or two after one of its developers gets a new phone that uses it. Like Eclipse, it has some bugs. Truthfully, most of them are minor... IF you've completely mastered Netbeans, Eclipse, Android development, and the use of build scripts. Otherwise, it'll probably stop you dead in your tracks, with little hope of moving forward any time soon unless you throw in the towel, move everything over to Eclipse, and hope that the situation improves for your next Android project.
The story with Windows is more or less the same. If you have a problem building a C# program under VS10, you can find four hundred resources online to help troubleshoot it in 18 seconds with Google. Have a similar problem with something like SharpDevelop, Eclipse, Netbeans, or another non-VS IDE, and you'll probably be looking for the answer for quite a while.
It's even worse if your native language isn't English. Visual Studio is so pervasive worldwide, even people who speak regional languages can find abundant help in their own language. There might even be one or more entire BOOKS about Visual Studio in it. Deviate from Microsoft's chosen path, and you'd better be fluent in English. OK, I'm exaggerating a little... lots of the independent IDEs are written by authors in non-English-speaking countries, and provide support in their own language as well.
At one time, I would have been optimistic and said that Microsoft's future lack of free support might encourage more progress with free alternatives. Three years of Android development have disillusioned me. NBandroid has come a long way and made enormous amounts of progress, but thanks to Google's total contempt for Netbeans, it still ends up holding *me* back whenever I try using it, and there's no way in hell I could recommend it to somebody who's learning Android programming for the first time. And we're talking about a Java development ecosystem that has historically had only TWO viable free IDEs, both of which were widely viewed as the two best IDEs available, period. Compare that to Windows, where NONE of the alternatives has market share that would count as "sloppy seconds" compared to the overwhelming dominance of Visual Studio, and all of which have real drawbacks and disadvantages compared to Visual Studio.
At the end of the day, Visual Studio is kind of like a 97 year old benevolent dictator of a prosperous country who's been ruling since he was a teenager -- people might have complaints, especially if he starts getting senile in his old age... but he's been the only government anybody in that country has ever known, and not even his fiercest opponents can really see themselves taking his place, because over the past 80-odd years, he's basically become synonymous with the country, its government, and the cultural identity of the people who live there. With the possible exception of Commodore 64 BASIC, it's hard to think of any development environment that's been more dominant and pervasive within its platform than Visual Studio within its platform and era.
Actually, there's an even better reason to stick with cheap Android tablets from China -- the market for LCDs in particular is *extremely* volatile. It's almost suicidal to design a product around a LCD screen you haven't already bought sufficient quantities of to build every unit you plan to build, because any given screen could become unavailable at any time without warning.
At least with Android tablets, you have a relatively stable programming environment. And if you really want to get down & dirty with the bare metal, any company in Shenzhen will sell you Android tablets that are pre-rooted, have totally open bootloaders, and possibly come ready to triple-boot pre-installed copies of Android, Ubuntu, and Windows Mobile (yeah, I said "Windows Mobile"). Sticking with Android tablets means not having to worry much about your supply chain. Low-end Chinese Android devices are about as close to commodities as you can get -- most use the same Rockchip innards, and run minimally-tweaked copies of Rockchip's reference distro of Android. If one supplier goes away, you can almost certainly find 20 more selling compatible tablets for $10 less.
If the tablet has a micro-SD card and you need it to use as an embedded controller for something, you can even repurpose the microSD card slot as an accessible SPI port with a few extra GPIOs. Remove the tablet from its case, put it in a new case of your own, connect it to your circuit with the nasty hacked microSD-to-SPI patch cord you had someone solder together, and you're good to go. It wouldn't even surprise me if commodity Chinese Android tablets HAVE additional headers on their circuit boards to expose things like SPI, I2C, GPIO, and UARTs to other products that might use them as a component.
2/3 the length was self-defense.
If I simply said that in my post-IPV6 network my 192.168.100.1 PC was 2001:55c:37:e94::100, somebody would have felt compelled to point out that ipv6 addresses are hexadecimal and missed the point that there's no requirement that you actually USE digits A-F.
Likewise, if I didn't include the background about zeroconf and the obligatory note about a router not being strictly needed in a purely-ipv6 network, there would have been a half-dozen replies pointing out that you don't actually *have* to do anything NAT_like with IPv6.
It's unfortunately, but you aren't allowed to make IPv6 sound human-manageable in polite company. Its supporters will shout you down and make it sound like the theory behind fractal image comppression. And if that fails, they'll beat you up and tell you to use DNS instead of raw IP addresses, totally missing the point (raw IP addresses always work, even when DNS is totally borked).
Culturally, IPv6 is very much the realm of academics who refuse to tolerate the equivalent of Cliff's notes. If you don't exhaustively document your arguments and prove that you did your homework, they'll attack and criticize you. It's a shame. Especially now that Comcast managed to rip away the shroud of academic abstraction that formerly clouded the whole topic, and has come up with a viable way to let their customers (for the next few years, at least) have their cake & eat it too... a personal /64 ipV6 network prefix to call your own, plus a real IPv4 address to fall back on until the day it becomes as moot as a fax modem (when you realize you set up the computer weeks/months ago, and never bothered to install the drivers for it). ;-)
> Lets see YOU sir figure up an IP V6 address map for...lets say a 40 person small business, in your head.
Actually, it doesn't *have* to be as bad as you think, regardless of what some of the more ardent IPv6 advocates might say (IMHO, they're their own worst enemies who manage to forge arguments & disagreement that's completely unnecessary for the sake of pedantic academic abstraction).
Let's look at how things work now... you basically have two chunks of data to remember:
1. Your public IP address: a.b.c.d
2. Your private IP addresses, all of which almost certainly fall within a block of 253 between 1 and 254 of some C-like /24 network. In most cases, every computer you deal with has the same first three values, and differs only with respect to the last. Going a step further, you probably don't randomly use the whole range... your router is probably x.y.z.1, your DHCP is probably x.y.z.200-x.y.z.250, you probably have a bunch of virtual port-forwarded servers clumped between x.y.z.10 and x.y.z.20, and the PC you personally use is probably x.y.z.100 (just to make your life easy).
Now... let's suppose you're a Comcast customer. In all likelihood, the first two chunks of your IPv6 address will be 2001:055c (or, if you prefer, 2001:55c).
As a practical matter, most IP addresses you're likely to see for the next 10 years will probably begin with 2001, or one of a half-dozen or so values common to nearly everyone in America. It doesn't have to be that way, and won't be that way forever, but for now, think of the old pre-cellphone-number-portability days when 99% of the people you called on a daily basis were in one, maybe... MAYBE two, area codes. And it never even OCCURRED to you that "1" was America's country code... it was just the digit you dialed first when making a long-distance call (and later, the digit that affirmed your understanding that the call might not be "free").
Anyway, that brings us to 055c -- Comcast's first netblock. Every IP address that begins with 2001:055c belongs to Comcast. In all likelihood, AT&T will have a similar block for DSL & U-Verse, and Verizon will have one for DSL & FIOS. In theory, they could have one for DSL and a different one for FIOS, but keep in mind the next two chunks, which will specifically define the second half of the 64-bit prefix assigned to you. Those two blocks -- 32 bits -- include more valid prefixes that can be assigned to customers than there are usable addresses in the entire IPv4 address space. So, there's plenty to go around.
Let's suppose for a moment that you get lucky, and as an early-adopter, you end up with a prefix like 2001:055c:0037:0e94. If you prefer, it can be written as 2001:55c:37:e94. Now, compare that to memorizing 64.192.87.239. The "2001" fades into the background noise (for now), because just about every address you'll see begins with it anyway. That leaves 55c:37:394. Is that really any harder to memorize than 64.192.87.239?
"But wait," you protest... what about the last 64 bits? No human can ever remember some god-awful address like 2001:55c:37:e94:8d1f:ca90:ff03:2109!"
And you're absolutely right. Which is why your desktop PC, whose internal IP address is currently 192.168.100.100, has as its fixed IP address 2001:55c:37:e94::100
"But... But... I read somewhere that the last 64 bits would be detmined by the MAC address!". And you're right... if you leave Windows up to its defaults. That's how zeroconf works. And the nice thing is, as a computer-savvy invididual, you don't have to do it that way. There's absolutely NOTHING to stop you from going into Windows' network control panel and explicitly setting your ethernet adapter's IPv6 address to 2001:055c:0037:0e94:0000:0000:0000:0100 (which can be equally represented as "2001:55c:37:e94::100").
Much nicer, no?
Of course, the trailing "100" isn't "really" a 100... it's actually 256... but neither your computer nor Comcast has to know that. You're perfectly free to limit 16-bit chun